NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Women's and Children's Health (UK). Diabetes in Pregnancy: Management of Diabetes and Its Complications from Preconception to the Postnatal Period. London: National Institute for Health and Care Excellence (UK); 2015 Feb. (NICE Guideline, No. 3.)

  • August 2018: Footnotes were added by NICE to recommendation 1.1.10 to clarify BMI in different ethnic groups, and to recommendation 1.2.11 to provide a link to the DVLA. The footnotes on glibenclamide and on thresholds for defining risk of developing type 2 diabetes were updated.

August 2018: Footnotes were added by NICE to recommendation 1.1.10 to clarify BMI in different ethnic groups, and to recommendation 1.2.11 to provide a link to the DVLA. The footnotes on glibenclamide and on thresholds for defining risk of developing type 2 diabetes were updated.

Cover of Diabetes in Pregnancy

Diabetes in Pregnancy: Management of Diabetes and Its Complications from Preconception to the Postnatal Period.

Show details

5Antenatal care

5.1. Monitoring blood glucose and ketones during pregnancy

5.1.1. Blood glucose monitoring

5.1.1.1. Review question

What is the effectiveness of blood glucose monitoring in predicting adverse outcomes in women with type 1, type 2 or gestational diabetes during pregnancy?

5.1.1.2. Introduction

The aim of this review was to evaluate the effectiveness of monitoring blood glucose in pregnant women with type 1, type 2 or gestational diabetes.

In the previous guideline on diabetes in pregnancy, 2 recommendations were made to inform how self monitoring of intermittent capillary blood glucose should be performed. During pregnancy, women with diabetes were to be advised to test fasting blood glucose levels and blood glucose levels 1 hour after every meal and women with insulin-treated diabetes were to be advised to additionally test blood glucose levels before going to bed at night.

The review question in this update does not examine the evidence available for the performance of self monitoring, but specifically focuses on the frequency of monitoring blood glucose and timing relative to meals.

5.1.1.3. Description of included studies

Nine studies were included in the current review (Bancroft et al., 2000; Espersen et al., 1985; Goldberg et al., 1986; Hawkins et al., 2009; Langer et al., 1994; Manderson et al., 2003; de Veciana et al., 1995; Weisz et al., 2005) that examined 5 comparisons of self monitoring strategies.

Four studies were randomised controlled trials (RCTs) (Bancroft et al., 2000; Manderson et al., 2003; de Veciana et al., 1995; Varner et al., 1983), 3 were prospective cohort studies (Espersen et al., 1985 [had historical controls]; Weisz et al., 2005; Langer et al., 1994), 1 study was a retrospective cohort (Hawkins et al., 2009) and 1 was a retrospective case-control study (Goldberg et al., 1986).

The studies were conducted in the UK (Bancroft et al., 2000; Manderson et al., 2003), Denmark (Espersen et al.), the USA (Goldberg et al., 1986; Hawkins et al., 2009; de Veciana et al., 1995; Langer et al., 1994; Varner et al., 1983) and Israel (Weisz et al., 2005). The number of women in the studies ranged from 28 (Varner et al., 1983) to 2,461 (Langer et al., 1994).

Six studies reported on women with gestational diabetes (Goldberg et al., 1986; Bancroft et al., 2000; Hawkins et al., 2009; Langer et al., 1994; de Veciana et al., 1995; Weisz et al., 2005), 2 studies reported on women with type 1 diabetes (Manderson et al., 2003; Varner et al., 1983) and 1 study reported on women with insulin dependent diabetes mellitus (Espersen et al., 1985).

Two studies compared monitoring with no monitoring (Bancroft et al., 2000; Espersen et al., 1985), 3 studies compared daily self monitoring with weekly monitoring in clinic (Goldberg et al., 1986; Hawkins et al., 2009; Varner et al., 1983), 2 studies compared preprandial with postprandial monitoring (Manderson et al., 2003; de Veciana et al., 1995), 1 prospective cohort study compared 1 hour postprandial to 2 hours postprandial capillary blood glucose monitoring (Weisz et al., 2005) and the final study compared monitoring 4 times a day (conventional strategy involving fasting and 2 hour post prandial sampling after each meal) to monitoring 7 times a day (intensified strategy involving a fasting sample and sampling before breakfast, preprandial, 2 hour post prandial and at bedtime) (Langer et al., 1994).

Table 45. Description of the methods used for monitoring of blood glucose in each study and other interventions as detailed.

Table 45

Description of the methods used for monitoring of blood glucose in each study and other interventions as detailed.

5.1.1.4. Evidence profile

The GRADE profiles for this review question are presented in Tables 46 to 50.

Table 46. GRADE profile for monitoring of blood glucose versus no monitoring of blood glucose.

Table 46

GRADE profile for monitoring of blood glucose versus no monitoring of blood glucose.

Table 47. GRADE profile for daily monitoring versus weekly testing of blood glucose.

Table 47

GRADE profile for daily monitoring versus weekly testing of blood glucose.

Table 48. GRADE profile for preprandial monitoring versus postprandial monitoring of blood glucose.

Table 48

GRADE profile for preprandial monitoring versus postprandial monitoring of blood glucose.

Table 49. GRADE profile for 1 hour postprandial monitoring versus 2 hour postprandial monitoring of blood glucose.

Table 49

GRADE profile for 1 hour postprandial monitoring versus 2 hour postprandial monitoring of blood glucose.

Table 50. GRADE profile for 4 daily measurements versus 7 daily measurements of blood glucose.

Table 50

GRADE profile for 4 daily measurements versus 7 daily measurements of blood glucose.

5.1.1.5. Evidence statements

5.1.1.5.1. Monitoring versus no monitoring

One study (n=68) found no difference in the number of vaginal births (relative risk [RR] 1.0, 95% confidence interval [CI] 0.7 to 1.4) or caesarean sections (RR 1.0, 95% CI 0.5 to 2.1) when comparing women with type 1, type 2 or gestational diabetes who had their blood glucose monitored with those who did not. There was also no difference between groups whose HbA1c values were measured at 28 weeks (MD −0.6, 95% CI −1.5 to 0.3), 32 weeks (MD 0.2, 95% CI 0.5 to 0.9), 36 weeks (MD −0.3, 95% CI −0.8 to 0.2)), 38 weeks (MD −0.2 95% CI −0.7 to 0.3) or at term (MD −0.4, 95% CI −1.2 to 0.4). The evidence for this outcome was of moderate quality.

One study (n=68) found no difference in the number of neonates born large for gestational age (RR 1.3, 95% CI 0.5 to 3.2), the incidence of shoulder dystocia (RR 0.4, 95% CI 0.0 to 8.9) or the incidence of neonatal hypoglycaemia (RR 0.4, 95% CI 0.1 to 1.7) when comparing women who had their blood glucose monitored with those who did not. The quality of this evidence was low.

One study (n=123) found no difference in the risk of large for gestational age neonates born to women with type 1, type 2 or gestational diabetes between groups associated with monitoring compared with no monitoring of blood glucose. The quality of this evidence was very low.

5.1.1.5.2. Monitoring strategies
Daily monitoring versus weekly monitoring

Two studies (n=990; n=116) found no difference between groups in the number of vaginal births (including births with forceps) associated with women with type 1, type 2 or gestational diabetes who received blood glucose monitoring compared to those who did not (RR 0.94, 95% CI 0.85 to 1.04; RR 1.40, 95% CI 0.56 to 3.50, respectively). The quality of the evidence was very low.

Two studies (n=116; n=990) found no difference between groups in the number of vaginal births with forceps associated with women with type 1, type 2 or gestational diabetes who received blood glucose monitoring compared to those who did not (odds ratio [OR] 2.77, 95% CI 0.9 to 8.4; RR 0.6, 95% CI 0.3 to 1.4, respectively). The quality of the evidence was very low.

One study (n=116) found no difference between groups for vaginal births without forceps (OR 0.49, 95% CI 0.24 to 1.04) when comparing women with type 1, type 2 or gestational diabetes who monitored their blood glucose daily with women who had weekly monitoring of their blood glucose. The quality of the evidence was very low.

Three studies (n=116; n=990; n=28) found no difference between groups for the risk of caesarean sections when comparing women with type 1, type 2 or gestational diabetes (respectively OR 1.41, 95% CI 0.6 to 3.2; RR 1.12 95% CI 0.9 to 1.3; RR 0.78, 95% CI 0.39 to 1.54). The quality of the evidence was very low.

One study (n=990) found a lower risk of neonates large for gestational age (on the 90th centile or above) being born to women with type 1, type 2 or gestational diabetes who monitored their blood glucose daily compared with women who performed weekly monitoring of their blood glucose (RR 0.7, 95% CI 0.5 to 0.9). The quality of the evidence was very low.

One study (n=116) found a reduced risk of neonatal hypoglycaemia (OR 0.19, 95% CI 0.08 to 0.5), while 2 studies (n=990; n=28) found no difference in the risk of neonatal hypoglycaemia (RR 1.6, 95% CI 1.0 to 2.8; RR 0.57, 95% CI 0.20 to 1.59 respectively) when comparing women with type 1, type 2 or gestational diabetes. The quality of the evidence for this outcomes was very low.

One study (n=990) found no difference in the risk of shoulder dystocia between groups (RR 0.8, 0.3 to 2.3) when comparing women with type 1, type 2 or gestational diabetes. The quality of the evidence for this outcome was very low.

Preprandial versus postprandial monitoring

Two studies (n=61; n=66) found no difference between groups in the risk of caesarean section associated with women with type 1, type 2 or gestational diabetes who received preprandial monitoring compared with those who did not (RR 1.45, 95% CI 0.9 to 2.3; RR 1.63, 95% CI 0.8 to 3.4). The evidence for this outcome was low quality.

One study (n=61) found no difference in the final HbA1c value between groups (MD 0.3, 95% CI −0.1 to 0.7) or in the change in HbA1c value from booking (MD 0.1, 95% CI −0.5 to 0.7) between women with type 1, type 2 or gestational diabetes who received postprandial monitoring compared with preprandial measurements to monitor their blood glucose. The evidence for this outcome was of moderate quality.

One study (n=66) found an increased risk of neonates large for gestational age (greater than 90th centile) associated with the group of women who underwent preprandial compared with postprandial measurements to monitor their blood glucose (RR 1.2, 95% CI 0.7 to 1.9). The quality of evidence for this outcome was low.

One study (n=66) found an increased risk of large for gestational age (greater than 90th percentile) neonates associated with the group of women who underwent preprandial compared with postprandial measurements to monitor their blood glucose (RR 3.5, 95% CI 1.3 to 9.5). The quality of evidence for this outcome was moderate.

One study (n=66) found no increased risk of shoulder dystocia associated with the group of women who underwent preprandial compared with postprandial measurements to monitor their blood glucose (RR 6.0, 95% CI 0.8 to 47.1). The quality of evidence for this outcome was moderate.

Two studies (n=61; n=66) found no difference between groups in the risk of neonatal hypoglycaemia in women with type 1, type 2 or gestational diabetes who underwent preprandial compared with postprandial measurements to monitor their blood glucose (RR 1.1, 95% CI 0.5 to 2.5; RR 7.0, 95% CI 0.9 to 53.8, respectively). The quality of evidence for this outcome was low.

Two studies (n=62; n=66) found no difference between groups in the risk of neonatal hypoglycaemia in women with type 1, type 2 or gestational diabetes who underwent preprandial compared with postprandial measurements to monitor their blood glucose (RR 2.8, 95% CI 0.1 to 66.6; RR 3, 95% CI 0.1 to 71.1, respectively). The quality of evidence for this outcome was low.

One hour postprandial versus 2 hours postprandial monitoring

One study (n=112) found no difference between groups in the risk of caesarean section in women with type 1, type 2 or gestational diabetes who received 1 hour postprandial monitoring compared with 2 hour postprandial measurements to monitor their blood glucose (RR 0.8, 95% CI 0.4 to 1.4). The quality of evidence for this outcome was very low.

One study (n=112) found no difference between groups in the risk of large for gestational age neonates in women with type 1, type 2 or gestational diabetes who received 1 hour postprandial monitoring compared with 2 hour postprandial measurements to monitor their blood glucose (RR 0.5, 95% CI 0.2 to 1.5). The quality of evidence for this outcome was very low.

Four daily (fasting and 3 post prandial measurements) versus 7 daily measurements

One study (n=2461) found an increased risk of caesarean section in women with type 1, type 2 or gestational diabetes associated with 4 daily measurements compared with 7 daily measurement of blood glucose (RR 1.4, 95% CI 1.2 to 1.7). The quality of this evidence was very low.

One study (n=2461) found an increased length of neonatal intensive care unit (NICU) length of stay associated with women with type 1, type 2 or gestational diabetes who had 4 daily measurements compared with 7 daily measurements of blood glucose (MD 1.7, 95% CI 1.5 to 1.9). The quality of this evidence was very low.

One study (n=2461) found an increased risk of large for gestational age (greater than 90th centile) neonates in women with type 1, type 2 or gestational diabetes associated with 4 daily measurements compared with 7 daily measurements of blood glucose (RR 1.5, 95% CI 1.3 to 1.9). The quality of this evidence was very low.

One study (n=2461) found an increased risk of shoulder dystocia in the neonates of women with type 1, type 2 or gestational diabetes who had 4 daily measurements compared with 7 daily measurement of blood glucose (RR 3.1, 95% CI 1.2 to 8.4). The quality of this evidence was very low.

One study (n=2461) found an increased risk of neonatal hypoglycaemia associated with women with type 1, type 2 or gestational diabetes who had 4 daily measurements compared with 7 daily measurements of blood glucose (RR 5.2, 95% CI 3.8 to 7.1). The quality of this evidence was very low.

One study (n=2000) found no difference between groups of women with type 1, type 2 or gestational diabetes who had 4 daily measurements compared with 7 daily measurements of blood glucose for the outcomes of stillbirth (per 1000) (RR 4, 95% CI 0.5 to 35.7) or neonatal death (per 1000) (RR 0.7, 95% CI 0.1 to 4.0). The quality of this evidence was very low.

5.1.1.6. Health economics profile

No health economic evidence was identified that considered blood glucose monitoring in predicting adverse outcomes in women with type 1, type 2 or gestational diabetes during pregnancy.

De novo analysis was not undertaken for this question as it was not considered as high priority as other issues within the guideline.

5.1.1.7. Evidence to recommendations

5.1.1.7.1. Relative value placed on the outcomes considered

The guideline development group prioritised the following maternal outcomes for this review:

  • mode of birth (spontaneous vaginal, operative vaginal, caesarean section [elective or emergency])
  • HbA1c % (as a measure of glycaemic control during pregnancy)
  • hypoglycaemic episodes during pregnancy (another measure of glycaemic control during pregnancy).

The group also prioritised the following neonatal outcomes:

  • large for gestational age (however defined in the study, for example using a customised measure based on gestational age and population norms; dichotomous data preferred)
  • neonatal intensive care unit length of stay greater than 24 hours
  • shoulder dystocia (as a specific example of birth trauma)
  • neonatal hypoglycaemia (however defined)
  • mortality.

The guideline development group prioritised mode of birth as an outcome for this review and noted that babies of women with diabetes whose blood glucose levels were tested weekly or less frequently had a higher incidence of instrumental birth.

The gorup argued that maternal hypoglycaemia was an important outcome that was distinct from estimates of glycaemic control measured by HbA1c percentage, but this outcome was not reported in any of the studies.

The group prioritised large for gestational age and noted that babies of women with diabetes who are tested frequently tend to be smaller than babies of women who are tested less frequently.

The group also considered that NICU stay of more than 24 hours was a surrogate measure of significant neonatal problems, such as prolonged neonatal hypoglycaemia or respiratory distress. They noted that this stay was significantly shorter in babies of women who had monitoring 7 times daily compared with those who had less frequent monitoring.

5.1.1.7.2. Consideration of clinical benefits and harms

The guideline development group noted that, overall, the evidence supported the view that more frequent testing of blood glucose (and subsequent adjustment of treatment) led to better outcomes.

Intermittent glucose monitoring requires pricking a finger for a capillary blood sample several times during the day and using a meter to measure blood glucose. This is disruptive and requires a significant commitment from the woman. The group believed that self monitoring of blood glucose would be especially helpful to women who are more likely to experience hypoglycaemic episodes (such as those who experience wide variability in their glucose regulation, or who are on insulin or glibenclamide, or who may have hypoglycaemia unawareness). The 2008 guideline on diabetes in pregnancy recommended this as the standard method of glucose monitoring for all women with type 1 and those with type 2 diabetes on insulin therapy.

The guideline development group recognised that frequent self monitoring can also provoke anxiety in that some women who may feel pressure to manipulate their treatment regimen or achieve overly tight regulation.

Thus, the group believed that women's individual perceptions of the greater likelihood of good outcomes would be an important motivation in their commitment to and satisfaction with blood glucose self monitoring. However, the group also noted that if an adverse outcome occurs despite a woman's adherence to the monitoring strategy and subsequent modification of her glycaemic control, this might have a negative impact on her experience of pregnancy and adversely affect her engagement in any future pregnancy.

Finally, the guideline development group was of the view that the frequency of monitoring should reflect the severity of the disease and its treatment and to improve compliance in women with less severe disease. Hence they felt that it would be reasonable to write recommendations for 3 different categories of women with diabetes in pregnancy in decreasing order of severity: women with type 1 diabetes; women with type 2 or gestational diabetes on a multiple insulin dose regimen; and women with type 2 or gestational diabetes who were on diet and exercise therapy only, or taking oral therapy or a single daily dose of intermediate-acting or long-acting insulin.

5.1.1.7.3. Consideration of health benefits and resource uses

The guideline development group noted that self monitoring of blood glucose is part of standard NHS treatment for people with diabetes. Any increase in frequency of testing during pregnancy will incur an additional cost. However, because tight blood glucose control is particularly important for improving pregnancy outcomes, the benefits of additional testing are likely to outweigh testing costs.

5.1.1.7.4. Quality of evidence

The quality of the evidence ranged from moderate to very low.

There is no new evidence regarding women with type 2 diabetes. The majority of the reported evidence regards women with gestational diabetes, with only 1 study reporting on women with type 1 diabetes.

Monitoring versus no-monitoring

Because monitoring and adjusting treatment to target values is central to clinical management in diabetic pregnancies, the guideline development group recognised that the comparison of ‘monitoring versus no monitoring’ was, arguably, inappropriate in this review. There were no significant differences in any outcomes when self monitoring was compared to no self monitoring and the evidence was moderate to very low in quality. Specifically, the Bancroft (2000) study included women with impaired glucose tolerance with a wide range of gestation up to 38 weeks who may have been studied too late to be able to demonstrate any effect on fetal growth and maternal delivery. Given the small size of the study, the group felt it was underpowered to detect significant differences in outcomes. In addition, the Espersen (1985) study was considered to use outdated self monitoring methods and a schedule of monitoring that was insufficiently intensive to be adequately reflective of current practice.

Daily monitoring versus weekly monitoring

The evidence for the comparison of daily monitoring versus weekly testing was of low and very low quality. The guideline development group considered that the 2 studies examining this comparison did not contain useful data because weekly testing was performed in the clinic setting by a healthcare professional and this was not self monitoring. As such, it is not a practical or cost-effective option for pregnant women with diabetes.

Preprandial versus postprandial testing

There were 2 trials that compared pre- and postprandial testing in women with type1 diabetes and in women with insulin-requiring gestational diabetes respectively, which were moderate to low in quality. The guideline development group noted that in these studies the postprandial groups received more insulin and had smaller babies than the preprandial groups.

The 2008 guideline recommended that fasting and postprandial testing be performed by women with diabetes and those requiring insulin should also test at bedtime. The guideline development group recognised that postprandial testing is important because it correlates with fetal growth in the third trimester. However, they considered that women on basal bolus dosage regimes could not adjust their insulin dosage without knowing their preprandial blood glucose values as well.

The guideline development group was aware that women with type 1 diabetes who are not pregnant are required to test preprandially and at bedtime as part of intensification of glucose control which also involves carbohydrate counting and adjustment of insulin dosage as a result. Further, the group recognised that the carbohydrate/insulin ratio differs for different meals through the day and also changes throughout pregnancy, and that both preprandial and postprandial testing would therefore be needed.

One hour postprandial versus 2 hour postprandial comparison

One study compared 1 hour with 2 hour postprandial testing and provided very low quality evidence of no significant difference in the 2 outcomes examined (caesarean section and large for gestational age). The guideline development group commented on the lack of detail reported in this study regarding the exact timing of testing; notably was it from the beginning or from the end of each meal? The group noted from the review of continuous glucose monitoring that the postprandial peak in glucose is likely to be 60 to 90 minutes after meals, though again it was not clear whether this was from the start or the end of the meal. The group also considered that it may be more convenient and women may be more likely to remember to perform testing if it was performed sooner (at 1 hour) rather than later (2 hours) after the meal finished.

Four measurements per day versus 7 measurements per day

One prospective study compared 4 daily measurements with 7 daily measurements, providing evidence of very low quality. In terms of the outcomes considered in this review, 7 measurements per day conferred more benefit to women and their babies than 4, although it was not entirely clear that the benefit came solely from monitoring alone, given the different management protocols used in the study (different clinics and teams, different interventions in clinics and different monitoring methods). The guideline development group further noted that the paper was not likely to have been peer reviewed, the 2 groups were not truly randomised, as randomisation depended on availability of the memory meters, and the intensive therapy group (who had memory meters) only performed 5 plus or minus 2 home glucose tests each day rather than 7 tests.

The guideline development group had concerns that advising women to test 7 times a day would represent a change in practice for women with gestational diabetes and might prove unpopular.

5.1.1.7.5. Other considerations

The guideline development group had some concern that the Latina population in the last study might limit the applicability of the findings to a UK population.

The group also expressed uncertainty regarding whether overall glucose control or peak glucose values influence fetal growth. Although there have been suggestions that postprandial peaks are more important, the HAPO 2008 study suggests that both fasting and postprandial glucose concentrations influence fetal growth.

5.1.1.8. Key conclusions

The guideline development group concluded that for all women with diabetes both pre- and postprandial testing was important during pregnancy and that it should be performed 7 times a day for women with type 1 or insulin-requiring type 2 or gestational diabetes.

Women who achieved glucose regulation using diet or oral therapy or single dose intermediate or long-lasting insulin did not need to test preprandially and testing could be limited to a fasting sample and samples at 1 hour after meals every day.

5.1.1.9. Recommendations

60.

Advise pregnant women with type 1 diabetes to test their fasting, pre-meal, 1-hour post-meal and bedtime blood glucose levels daily during pregnancy. [new 2015]

61.

Advise pregnant women with type 2 diabetes or gestational diabetes who are on a multiple daily insulin injection regimen to test their fasting, pre-meal, 1-hour post-meal and bedtime blood glucose levels daily during pregnancy. [new 2015]

62.

Advise pregnant women with type 2 diabetes or gestational diabetes to test their fasting and 1-hour post-meal blood glucose levels daily during pregnancy if they are:

  • on diet and exercise therapy or
  • taking oral therapy (with or without diet and exercise therapy) or single-dose intermediate-acting or long-acting insulin. [new 2015]

5.1.1.10. Research recommendations

19. Post-meal blood glucose testing in women with diabetes in pregnancy: is the 1 hour test more acceptable than the 2 hour test?
Why this is important

Self-monitoring of blood glucose is an important tool in the management of diabetes in pregnancy. Many studies have shown that post prandial hyperglycaemia is a predictor for fetal macrosomia and may contribute to neonatal hypoglycaemia. Current recommendations state that tests should be performed at either one or two hours post meals. Studies have demonstrated however, that the 1hour post prandial test is more likely to detect abnormal values which may require treatment and helps the person understand the relationship between food and blood glucose levels. Identifying acceptability of blood monitoring regimes using qualitative studies may improve both compliance and accuracy of testing and optimise pregnancy outcomes.

20. What is the optimum frequency of blood glucose testing in pregnancy in women with pre-existing diabetes who are not taking insulin?

The optimum frequency of blood glucose testing in pregnancy in women with pre-exsiting diabetes who are not taking insulin is unknown. While daily fasting blood glucose values in women on insulin are required to optimize the basal insulin dose and avoid nocturnal hypoglycaemia for women not taking insulin a daily fasting glucose is less informative as there is little day-to-day variability. Unlike women taking insulin there is no need to perform a pre-bedtime glucose value to lessen the risk of nocturnal hypoglycaemia. The frequency of blood glucose tests for other times in the day are currently recommended to be the same as for women on insulin. Randomised control trials are required to inform on the optimum frequency of blood glucose testing in pregnancy in women who are not taking insulin

5.1.2. Ketone monitoring

5.1.2.1. Review question

What is the effectiveness of blood ketone monitoring compared with urine ketone monitoring for women with type 1, type 2 or gestational diabetes during pregnancy?

5.1.2.2. Introduction

Ketones are derived from the breakdown of fat and can be used as a source of energy, for example during starvation. Ketones are usually present in very low concentrations in urine and blood in non-diabetic populations. The concentration of ketones in blood is normally less than 0.3 mmol/litre and they are usually undetectable by routine urine tests. Various factors can contribute to a raised concentration of ketones in the blood or urine, including metabolic disorders (such as uncontrolled diabetes or weight loss), dehydration, low carbohydrate intake and individual variations in the threshold for ketonuria.

In someone with diabetes, increased ketone concentrations might indicate impending or established ketoacidosis (DKA). This serious condition can occur at relatively low blood glucose concentrations in pregnant women and requires urgent medical attention because of an increased risk of harm to the fetus. Although ketoacidosis is more common in women with type 1 diabetes, it has also been described in women with type 2 and gestational diabetes. Because DKA can profoundly compromise the wellbeing of both the woman and her baby (including maternal and fetal death), the 2008 guideline recommended that women with type 1 diabetes who are planning to become pregnant should be offered ketone testing strips and advised to test for ketonuria or ketonaemia if they become hyperglycaemic or unwell. In the absence of any evidence, this recommendation was based on a consensus of the guideline development group's knowledge and the best clinical practice at the time.

The 2008 guideline did not include research recommendations related to ketone monitoring in the preconception period.

5.1.2.3. Description of included studies

No studies were identified that assessed how blood ketones should be monitored during pregnancy.

5.1.2.4. Health economics profile

No health economic evidence was identified that considered blood ketone monitoring compared with urine ketone monitoring for women with type 1, type 2 or gestational diabetes during pregnancy.

This question was not prioritised for health economic evaluation as the guideline development group thought there were more important priorities.

5.1.2.5. Evidence to recommendations

5.1.2.5.1. Relative value placed on the outcomes considered

The guideline development group prioritised the following maternal outcomes for this review:

  • preterm birth (birth before 37+0 weeks' gestation; dichotomous or continuous data)
  • non-routine hospital contact or assessment for ketosis (ketonaemia or ketonuria, however defined), including phone contact
  • hospital admission for diabetic ketoacidosis
  • maternal satisfaction.

The group also prioritised these fetal and neonatal outcomes:

  • mortality – perinatal and neonatal death
  • neonatal intensive care unit (NICU) length of stay greater than 24 hours.

The outcomes chosen were considered to be clinically meaningful for the woman and baby, could be reliably assessed in clinical research studies and were expected to be commonly reported in the evidence available for inclusion.

The guideline development group decided to prioritise non-routine hospital contact or assessment for ketosis as an outcome because pregnant women with diabetes will be tested routinely for ketones.

The group recognised that maternal mortality in association with diabetic ketoacidosis is a possibility, but agreed that it is a rare event in UK clinical settings and so it was not prioritised for this review.

5.1.2.5.2. Consideration of clinical benefits and harms

The guideline development group recognised the potential health risk to babies from high concentrations of ketones in blood or urine of pregnant women with diabetes. Although high maternal ketone levels during pregnancy have not been proven conclusively to be dangerous to the fetus, neither have they have been proven to be harmless. The 2008 guideline assumed that there was a benefit in measuring ketones in women who are pregnant if they become hyperglycaemic or unwell. This was consistent with the NICE clinical guideline on type 1 diabetes which recommends monitoring blood or urine for the presence of ketones. The guideline development group for this guideline agreed with the previous guideline that there are potential benefits in the ability to measure blood ketones in women who are pregnant if the woman is unwell or has very high blood glucose values. The group believed the benefits in terms of prompt recognition of DKA and its treatment were greater than the harms of unnecessary testing.

The group recognised that one of the main advantages of using a blood ketone test is that it can provide an accurate, convenient and timely assessment of ketosis. In contrast, urine testing only provides a qualitative measure of any ketosis over the preceding period since the woman last passed urine. (This is discussed at greater length in the section on biochemical issues below). Furthermore, it does not precisely correlate with blood ketone concentration which is more likely to reflect the severity of DKA. Finally, blood ketone levels increase before urine ketone levels, allowing an earlier identification of any metabolic deterioration.

In summary, the guideline development group felt that blood ketone tests give a specific value that more accurately reflects the level of ketosis and its severity, thus leading to more timely recognition of DKA and earlier treatment.

5.1.2.5.3. Consideration of health benefits and resource uses

Although the guideline development group noted that blood testing strips are more expensive than those used for urine testing, and notwithstanding that a full health economic analysis was not undertaken, they were of the view that the convenience of undertaking blood testing for ketones (which can be performed at the same time as testing blood glucose levels and with the same device) would result in greater patient compliance than using urine testing strips. In turn this would lead to a more prompt response and treatment following an abnormal result, which should result in a lower overall cost as suggested by a study in young people with type 1 diabetes (Laffel et al., 2006). In this study it was shown that the higher costs of blood ketone testing were offset by reduced treatment costs for DKA as a result of lower rates of hospitalisation.

The guideline development group noted that although blood ketone testing meters are not universally available for all diabetic patients, in general many patients with an increased risk of DKA (such as women with type 1 diabetes and unstable glucose control, and those on insulin pumps) would already have been issued with one.

The group was of the view, because of the serious risk to the fetus of DKA, that all pregnant women with type 1 diabetes should be issued with blood ketone monitoring equipment and receive appropriate training on its use and advice on when to test. Most meters are able to measure both blood glucose and blood ketone concentrations.

However, the guideline development group felt that, although DKA has been reported in pregnant women with type 2 diabetes and gestational diabetes, this was less common than in women with type 1 diabetes. They did not, therefore, feel that it was justified to issue ketone monitoring equipment routinely to women with type 2 diabetes or gestational diabetes, although they recommended that women should be advised to seek prompt medical attention if they became hyperglycaemic or unwell in order to exclude DKA.

The group noted that no new evidence was found to establish the most effective method for monitoring ketones in women with type 1, type 2 and gestational diabetes during pregnancy. In the absence of pregnancy-specific evidence, the group relied in part on consensus opinion and current best clinical practice.

5.1.2.5.4. Other considerations
Practical issues

The guideline development group noted that for many patients with diabetes the convenience of testing blood for both glucose and ketones at the same time from a single capillary sample represented a distinct advantage of blood testing for ketones. Also, many people with diabetes (particularly the young) find urine testing inconvenient and unpleasant, and would rather avoid doing it.

Biochemical factors

Urine ketone testing strips are based on the nitroprusside reaction, which primarily detects aceto-acetate and acetone. The strips are read visually, comparing the colour obtained with a colour-coded chart, so do not require instrumentation for automatic reading. With this method the presence and quantity of ketones is reported subjectively either as ‘negative’, ‘small’, ‘moderate’ or ‘large’.

Blood ketone testing strips measure beta-hydroxybutyric acid, which is the predominant ketone body in ketoacidosis. The strip is read by a meter using a chemical process that does not require a colour chart and gives an accurate blood concentration. Moreover, blood beta-hydroxybutyric acid measurements are used to assess the severity of DKA and inform insulin and fluid replacement, and help monitor response to treatment.

The guideline development group also noted that urine strips degrade over time and their accuracy is reduced after 6 months. In addition, urine strips can give a false-negative reading, either because:

  • they have been exposed to the air for long periods
  • the urine specimen is highly acidic
  • the women is using certain prescription medicines (such L-DOPA metabolites)
  • there are high levels of phenylketones.
Other guidelines in development

During the development of this guideline, NICE established liaison between the guideline development groups that were concurrently updating several diabetic guidelines, with the aim of aligning recommendations (the other guidelines were on type 1 diabetes in adults, type 2 diabetes in adults, and diabetes in children and young people). As a result, the guideline development group for this guideline was aware that the group working on the guideline for diabetes in children and young people was recommending blood ketone testing rather than urine testing and that the group working on the guideline for type 1 diabetes in adults was recommending it for in-patients.

5.1.2.6. Key conclusions

Due to the lack of new evidence, the guideline development group used their knowledge and understanding of best clinical practice at the time to review the 2008 recommendation. The discussions and conclusions of the guideline development groups working on the guidelines for type 1 diabetes in adults and diabetes in children and young people were also noted and discussed. The group working on this guideline agreed that because blood ketone testing can provide more accurate information about the severity of ketosis, as well as helping monitor the response to therapy, it should be recommended in pregnant women with type 1 diabetes if they become hyperglycaemic or unwell in preference to urine testing. The group also argued that blood testing ketones could be logistically easier than urine in that it could be carried out at the same time as a capillary test was performed for glucose testing. The group recognised that it is important to inform pregnant women with type 2 diabetes and gestational diabetes to seek urgent medical advice if they become hyperglycaemic or unwell, in order to detect possible DKA and avoid further metabolic deterioration that could lead to significant maternal and fetal morbidity and even mortality. For the same reason, the group believed that it is essential for healthcare professionals to test routinely for ketonaemia in all pregnant women with diabetes who present with hyperglycaemia or who are otherwise unwell.

5.1.2.7. Recommendations

63.

Offer pregnant women with type 1 diabetes blood ketone testing strips and a meter, and advise them to test for ketonaemia and to seek urgent medical advice if they become hyperglycaemic or unwell. [new 2015]

64.

Advise pregnant women with type 2 diabetes or gestational diabetes to seek urgent medical advice if they become hyperglycaemic or unwell. [new 2015]

65.

Test urgently for ketonaemia if a pregnant woman with any form of diabetes presents with hyperglycaemia or is unwell, to exclude diabetic ketoacidosis. [new 2015]

66.

During pregnancy, admit immediately women who are suspected of having diabetic ketoacidosis for level 2 critical caremm, where they can receive both medical and obstetric care. [2008]

5.1.2.8. Research recommendations

21. What is the value of ketone testing in pregnancy in women with type 2 diabetes or GDM?
Why this is important

Ketoacidosis develops more rapidly with hyperglycaemia in pregnancy and also with lower levels of hyperglycaemia. If it occurs it requires high dependency care and is associated with a significant risk of fetal death. Women with type I diabetes are advised to check for ketones in their blood if they have sustained hyperglycaemia or feel unwell to check whether ketoacidosis is developing. However this advice is traditionally not usually given to women with type 2 diabetes, but reports do show that these women and even occasionally women with gestational diabetes may develop ketoacidosis.

Accordingly there is a need to perform studies on the prevalence of significant ketonaemia and ketonuria in women with type 2 diabetes and gestational diabetes and to randomise women who meet the pre-specified diagnostic criteria to either an intervention or routine management to determine whether this will reduce the risk of diabetic ketoacidosis.

5.2. Target blood glucose values for women with type 1, type 2 or gestational diabetes during pregnancy

5.2.1. Review question

What are the target ranges for blood glucose in women with type 1, type 2 or gestational diabetes during pregnancy?

5.2.2. Introduction

The purpose of this review was to determine the optimal target values for blood glucose in women with type 1, type 2 or gestational diabetes during pregnancy. The search for this question included randomised controlled trials (RCTs), systematic reviews and comparative observational studies. Non-comparative observational studies were to be included only if no comparative studies were identified. The same search was used to identify studies for this review and the reviews of target values for blood glucose pre-conception, target values for HbA1c pre-conception and during pregnancy, and for blood glucose and HbA1c monitoring during pregnancy.

The guideline development group defined 8 maternal and neonatal priority outcomes.

Maternal outcomes were:

  • mode of birth (spontaneous vaginal, operative vaginal or elective or emergency caesarean section)
  • pre-eclampsia
  • HbA1c levels at any time during pregnancy
  • hypoglycaemic episodes at any time during pregnancy.

Neonatal outcomes were:

  • large for gestational age
  • neonatal intensive care unit (NICU) length of stay greater than 24 hours
  • shoulder dystocia
  • mortality, whether perinatal (stillbirth and death up to 7 days after birth) or neonatal (death up to 28 days after birth).

The original review question in the 2008 guideline was “What are the target ranges for blood glucose during pregnancy?” Studies that examined glycaemic control using blood glucose or HbA1c measurements were included as evidence in the chapter.

A more specific approach has been taken for this update. Four separate review questions have been stipulated to examine blood glucose or HbA1c measurements prior to conception and during pregnancy.

Sixteen studies were included in the chapter in the original guideline on target values during pregnancy. The majority of these studies examined HbA1c and were considered as part of the current review protocol. Six of the 16 studies were assessed for their relevance to this review. One was included (Landon et al., 1987) and five were excluded (Evers et al., 2002, Langer et al., 1994, Jovanovic et al., 1981, Jovonovic Peterson et al., 1991, Karlsson & Kjellmer 1972).

5.2.3. Description of included studies

Overall, 6 studies were identified for inclusion in this review. Three studies were RCTs (DeMarini et al., 1994; Farrag, 1987; Sacks et al., 2006), 1 a secondary analysis of RCT data (Rowan et al., 2010) and 2 were retrospective cohort studies (Combs et al., 1992; Landon et al., 1987). Locations included the USA (Combs et al., 1992; DeMarini et al., 1994; Landon et al., 1987; Sacks et al., 2006), Australia and New Zealand (Rowan et al., 2010) and Saudia Arabia (Farrag, 1987). The number of participants ranged from 22 to 724. Participants had White class diabetes B to D (Landon et al., 1986), White class diabetes B or C (Farrag, 1987), gestational diabetes (Rowan et al., 2010), White class diabetes B to RF (Combs et al., 1992), White class B to RT (DeMarini et al., 1994) and type 1 diabetes mellitus (Sacks et al., 2006). The ethnicity of participants in the 2 studies where it was reported was primarily white (Rowan et al., 2010; Sacks et al., 2006).

Optimal blood glucose control ranged from 5.3 mmol/litre to 9.7 mmol/litre, depending upon the type of blood glucose measurement taken. Three studies used postprandial blood glucose (Combs et al., 1992; DeMarini et al., 1994; Rowan et al., 2010), 2 studies used mean blood glucose, which included fasting plasma glucose measurements (Landon et al., 1987; Sacks et al., 2006) and 1 study reported fasting blood glucose (Rowan et al., 2010).

One study did not specify the blood glucose measurements to which targets were related (Farrag, 1987) but it was assumed that targets related to fasting blood glucose due to the low values assigned. Two studies did not implement specific blood glucose target values for participants to reach (Landon et al., 1987; Rowan et al., 2010), instead using thresholds applied post hoc for optimal glucose control. When included as an outcome, 2 studies measured HbA1 rather than HbA1c (DeMarini et al., 1994; Landon et al., 1987). Mean HbA1 values were converted to HbA1c using the Michigan formula.

Of the guideline development group's priority outcomes, evidence was available for mode of birth (Landon et al., 1987; Sacks et al., 2006), pre-eclampsia (Rowan et al., 2010), HbA1c during pregnancy (DeMarini et al., 1994; Landon et al., 1987; Rowan et al., 2010; Sacks et al., 2006), maternal hypoglycaemic episodes (Farrag, 1987), perinatal mortality (Farrag, 1987) and large for gestational age (Combs et al., 1992; Landon et al., 1987; Rowan et al., 2010). One study (DeMarini et al., 1994) used the term ‘glycosylated haemoglobin’ rather than HbA1c and 1 study specified that HbA1 was measured (Landon et al., 1987).

5.2.4. Evidence profile

GRADE profiles are presented according to glucose thresholds. Reasons for the use of each threshold are given in Table 51.

Table 51. Blood glucose thresholds for optimal control used in GRADE profiles with associated reasons.

Table 51

Blood glucose thresholds for optimal control used in GRADE profiles with associated reasons.

The GRADE profiles for this question are presented in Tables 52 to 58.

Table 52. GRADE profile for comparison of fasting blood glucose less than 5.3 mmol/litre versus 5.3 mmol/litre or more in women with gestational diabetes.

Table 52

GRADE profile for comparison of fasting blood glucose less than 5.3 mmol/litre versus 5.3 mmol/litre or more in women with gestational diabetes.

Table 53. GRADE profile for comparison of fasting blood glucose less than 5.6 mmol/litre versus 5.6 mmol/litre or more in women with White class diabetes B and C (type 1 diabetes).

Table 53

GRADE profile for comparison of fasting blood glucose less than 5.6 mmol/litre versus 5.6 mmol/litre or more in women with White class diabetes B and C (type 1 diabetes).

Table 54. GRADE profile for comparison of mean capillary blood glucose less than 6.1 mmol/litre in women with White class diabetes B to D.

Table 54

GRADE profile for comparison of mean capillary blood glucose less than 6.1 mmol/litre in women with White class diabetes B to D.

Table 55. GRADE profile for comparison of 2 hour postprandial blood glucose less than 6.4 mmol/litre versus 6.4 mmol/litre or more in women with gestational diabetes.

Table 55

GRADE profile for comparison of 2 hour postprandial blood glucose less than 6.4 mmol/litre versus 6.4 mmol/litre or more in women with gestational diabetes.

Table 56. GRADE profile for comparison of strict control of 1.5 hour postprandial blood glucose (less than 6.7mmol/litre) versus customary control (less than 7.8 mmol/litre) in women with type 1 diabetes (White class diabetes B to RT).

Table 56

GRADE profile for comparison of strict control of 1.5 hour postprandial blood glucose (less than 6.7mmol/litre) versus customary control (less than 7.8 mmol/litre) in women with type 1 diabetes (White class diabetes B to RT).

Table 57. GRADE profile for comparison of 1 to 2 hour postprandial blood glucose of 7.8 mmol/litre or less versus more than 7.8 mmol/litre in women with White class diabetes B to RF.

Table 57

GRADE profile for comparison of 1 to 2 hour postprandial blood glucose of 7.8 mmol/litre or less versus more than 7.8 mmol/litre in women with White class diabetes B to RF.

Table 58. GRADE profile for comparison of mean blood glucosea of 7.8 mmol/litre or less versus 9.7 mmol/litre or less in women with type 1 diabetes mellitus.

Table 58

GRADE profile for comparison of mean blood glucosea of 7.8 mmol/litre or less versus 9.7 mmol/litre or less in women with type 1 diabetes mellitus.

5.2.5. Evidence statements

5.2.5.1. Fasting blood glucose levels less than 5.3 mmol/litre versus 5.3 mmol/litre or more

One secondary analysis of trial data (n=724) found a reduction in risk of pre-eclampsia (RR 0.47, 95% CI 0.27 to 0.83) and of large for gestational age in babies (RR 0.48, 95% CI 0.35 to 0.67) of women with fasting blood glucose less than 5.3 mmol/litre compared with women with fasting blood glucose of 5.3 mmol/litre or more. The quality of evidence was very low.

5.2.5.2. Fasting blood glucose levels less than 5.6 mmol/litre versus 5.6 mmol/litre or more

One randomised controlled trial (n=60) found an increase in maternal hypoglycaemic episodes (RR 39.71, 95% CI 2.26 to 697.01) in women with fasting blood glucose less than 5.6 mmol/litre compared with women with fasting blood glucose of 5.6 mmol/litre or more. The same study found no evidence of a reduction in risk for pre-eclampsia (RR 0.92, 95% CI 0.10 to 8.59), caesarean delivery (RR 0.62, 95% CI 0.15 to 2.64), large for gestational age (RR 0.10, 95% CI 0.006 to 1.68) or perinatal mortality (RR 0.53, 95% CI 0.03 to 11.14) in women or the babies of women with fasting blood glucose less than 5.6 mmol/litre. The quality of the evidence for these outcomes was very low.

5.2.5.3. Mean capillary blood glucose levels less than 6.1 mmol/litre versus 6.1 mmol/litre or more

One retrospective cohort study (n=75) found a reduction in mean HbA1c levels in the third trimester (MD −1.6, 95% CI −2.1 to −1.1) as well as a reduced risk of large for gestational age (RR 0.27, 95% CI 0.09 to 0.77) in babies of women with mean capillary blood glucose less than 6.1 mmol/litre compared with women with mean capillary blood glucose greater than 6.1 mmol/litre. The same study found no evidence for an effect of mean capillary blood glucose less than 6.1 mmol/litre on risk of caearean section (RR 0.93, 95% CI 0.58 to 1.49). Mean capillary blood glucose was calculated from a minimum of 16 weeks of values from daily fasting and 3 preprandial sampling (11am, before dinner and at bedtime) throughout the second and third trimesters. This amounted to more than 450 samples per patient. The quality of evidence was very low.

5.2.5.4. Postprandial blood glucose levels less than 6.4 mmol/litre versus 6.4 mmol/litre or more

One secondary analysis of trial data (n=724) found a reduction in risk of pre-eclampsia (RR 0.36, 95% CI 0.30 to 0.43) and of large for gestational age (RR 0.46, 95% CI 0.33 to 0.64) in babies of women with postprandial blood glucose less than 6.4 mmol/litre compared with women with postprandial blood glucose of 6.4 mmol/litre or above. The quality of evidence was very low.

5.2.5.5. Postprandial blood glucose levels less than 6.7 mmol/litre versus more than 7.8 mmol/litre

One randomised controlled trial (n=137) found no evidence of a reduction in mean HbA1c levels in the first trimester (MD 0.0, 95% CI -0.6 to 0.6), second trimester (MD 0.1, 95% CI −0.4 to 0.6) or third trimester (MD −0.1, 95% CI −0.5 to 0.3) in women with postprandial blood glucose less than 6.7 mmol/litre compared with women with postprandial blood glucose more than 7.8 mmol/litre. The quality of evidence was very low.

5.2.5.6. Postprandial blood glucose levels 7.8 mmol/litre or less versus more than 7.8 mmol/litre

One retrospective cohort study (n=111) found a reduced risk of large for gestational age (RR 0.53 95% CI 0.29 to 0.95) in babies of women with postprandial blood glucose less than 7.8mmol/litre compared with women with postprandial blood glucose greater than 7.8mmol/litre. The quality of the evidence was very low.

5.2.5.7. Mean blood glucose levels of 7.8 mmol/litre or less versus 9.7 mmol/litre or less

One randomised controlled trial (n=22) found reduced mean HbA1c values during the first trimester (MD −1.2, 95% CI −2.32 to −0.08) in women with mean blood glucose values less than 7.8 mmol/litre compared with women with a mean blood glucose values less than 9.7 mmol/litre. The same study found no evidence of an effect of mean blood glucose on mean HbA1c values during the second (MD −0.5, 95% CI −1.12 to 0.12) or third (MD −0.3 95%, CI −0.95 to 0.35) trimesters or on caesarean section rates (RR 0.92, 95% CI 0.49 to 1.73). Mean blood glucose values were derived from results of capillary plasma glucose self-monitoring performed 7 times a day, before and 1 hour after the first bite of each meal and at bedtime using memor- based portable glucose meters. The quality of the evidence was very low.

No evidence was reported for either shoulder dystocia or NICU length of stay in any of the included studies.

5.2.6. Health economics profile

No health economic advice was identified that compared target ranges for blood glucose in women with type 1, type 2 or gestational diabetes during pregnancy.

This was not prioritised for health economic analysis. This was because although a target may affect the interventions and management used to assist the patient in achieving that target, the target does not incur an opportunity cost.

5.2.7. Evidence to recommendations

5.2.7.1. Relative value placed on the outcomes considered

The guideline development group prioritised pre-eclampsia and maternal hypoglycaemic episodes for maternal outcomes and large for gestational age and shoulder dystocia for neonatal outcomes.

The group noted that data from the HAPO study (HAPO Study Cooperative Research Group et al., 2008) demonstrated a linear relationship between maternal blood glucose and the risk of complications such as macrosomia. Thus, in theory, blood glucose values in women with any form of diabetes should be kept as close to the non-diabetic range as possible. However, the group acknowledged the difficulties of safely achieving this in practice because of the risk of hypoglycaemia.

Therefore, in making recommendations about target values for women with diabetes in pregnancy, the group inclined to use those values for which the evidence showed some benefit. Accordingly, from the evidence they suggested that the following would be reasonable targets:

  • fasting level – less than 5.3 or 5.6 mmol/litre (Rowan et al. [2010] reported a lower incidence of pre-eclampsia and large for gestational age with a target threshold of 5.3 mmol/litre, but Farrag [1987] reported a higher incidence of maternal hypoglycaemic episodes with a target threshold of 5.6 mmol/litre.)
  • 1 hour value – less than 7.8 mmol/litre (In a study of women who largely measured the 1 hour values, Combs et al. [1992] reported a lower incidence of large for gestational age with a target threshold of 7.8 mmol/litre.)
  • 2 hour value – less than 6.4 mmol/litre (Rowan et al. [2010] reported a lower incidence of pre-eclampsia and large for gestational age with a target threshold of 6.4 mmol/litre.)

5.2.7.2. Consideration of clinical benefits and harms

Ideally, women should strive for blood glucose levels as near to normal as is safely achievable. For women taking insulin and glibenclamide there is a risk of hypoglycaemia and the guideline development group felt that it would be sensible to provide a limit for the lower level of blood glucose for women on these treatments. However, there was no evidence identified in the review that could inform the group of this limit. The group therefore chose 4.0 mmol/litre because this was the ‘safe’ lower target value recommended by Diabetes UK. For women on diet and exercise or metformin the risk of hypoglycaemia was very low and the group did not feel that it was necessary to set a lower limit for women on these treatments.

The group commented that as the normal blood glucose values in non-diabetic women change in pregnancy, perhaps the targets ranges for diabetic women should be different at different stages of pregnancy. However, they acknowledged that there were no data to inform such guidance and that the target ranges recommended would inevitably have to apply to the whole of pregnancy. They noted that some of the studies used very short gestational intervals. Targets may have to be adjusted for individual women depending upon their personal circumstances and treatment. Women with gestational diabetes receiving diet and exercise therapy and/or metformin should be able to achieve near normal blood glucose values. That is less likely to be possible in women with type 1 or 2 diabetes or gestational diabetes who are receiving insulin or glibenclamide.

The other concern of setting low or near normal targets for women with diabetes in pregnancy which are difficult to achieve is that it could be setting women up to fail and result in them losing confidence in their ability to self manage their diabetes. This could potentially result in less engagement and worse control, perhaps resulting in worse outcomes.

If a woman presents with gestational diabetes at 30 weeks and is set targets, it may be too late to prevent some poor outcomes and she may still have a large for gestational age baby or develop pre-eclampsia. If a woman has a high blood glucose at 30 weeks it is likely that it was also high at 20 weeks. This is an issue which is especially relevant for women with gestational diabetes. Existing guidance means that diagnosis of gestational diabetes is often not made until the third trimester. By this stage the argument that it is ‘too late to affect adverse outcomes’ may apply. This concern was specifically expressed in relation to the study by Rowan et al. (2010).

5.2.7.3. Quality of evidence

The outcomes from each study were generally of very low quality. In the study of Combs et al. (1992) the postprandial value of less than or equal to 7.8 mmol/litre was measured at 1 hour in most patients but some women had samples taken between 1 and 2 hours. The authors of the study by Farrag (1987) did not specify the timing of the blood glucose measurements to which targets were related. It was assumed that targets related to fasting blood glucose, given the low values, and in accordance with a Cochrane review (Middleton et al., 2012) which included this study.

5.2.7.4. Other considerations

The inclusion of studies that reported laboratory measured blood glucose was considered by the guideline development group to be valid, even though women would be monitoring capillary samples. Clinical target ranges would have to be based upon capillary blood glucose concentrations.

5.2.8. Recommendations

67.

Agree individualised targets for self-monitoring of blood glucose with women with diabetes in pregnancy, taking into account the risk of hypoglycaemia. [2008]

68.

Advise pregnant women with any form of diabetes to maintain their capillary plasma glucose below the following target levels, if these are achievable without causing problematic hypoglycaemia:

  • fasting: 5.3 mmol/litre
    and
  • 1 hour after meals: 7.8 mmol/litre or
  • 2 hours after meals: 6.4 mmol/litre. [new 2015]
69.

Advise pregnant women with diabetes who are on insulin or glibenclamide to maintain their capillary plasma glucose level above 4 mmol/litre. [new 2015]

5.2.9. Research recommendations

22. What is the role of CGM in helping women achieve blood glucose targets in pregnancy?

Why this is important

Continuous glucose monitoring (CGM) is a technology for measuring subcutaneous interstitial fluid glucose concentrations every few minutes. It is often used in conjunction with continuous subcutaneous insulin infusion (CSII) using an insulin pump. In combination, these technologies have been shown to improve glycaemic control by reducing glucose variability and the number and severity of hypoglycaemic episodes in non-pregnant adults. However there have been few systematic studies using the latest real-time devices in pregnant women. One of the main complications of diabetic pregnancy is fetal macrosomia and its associated problems of difficult delivery and neonatal hypoglycaemia. These problems are closely related to mean glucose levels in the late second and third trimesters of pregnancy. However, it remains uncertain whether the main driver for fetal growth is the fasting or post prandial blood glucose or the magnitude of the glycaemic variation. If CGM (± CSII) can be shown to improve glucose control in later pregnancy there is a real prospect of reducing a common and serious complication of diabetic pregnancy. An RCT of CGM versus conventional intermittent capillary blood glucose monitoring in pregnant women is urgently required. A pilot study looking at intermediate outcomes such as the incidence of fasting and postprandial hyperglycaemia and glycaemic variability could be followed with a much larger study exploring macrosomia rates if CGM was shown to be effective in terms of improved blood glucose control.

23. What is the role of telemedicine in helping women achieve blood glucose targets in pregnancy?

Why this is important

These is extensive well documented research to show that good glycaemic control in pregnancy reduces adverse pregnancy outcome such as macrosomia, operative delivery, instrumental delivery, shoulder dystocia, neonatal hypoglycaemia and admission to the neonatal intensive care unit. Research to investigate the role of telemedicine is assisting women in achieving target blood glucose levels is required and needs to explore women's access to healthcare through telemedicine and the acceptability of this method of communication to both the woman and healthcare professionals. An important goal of telemedicine is to improve both the women with pre-existing diabetes and healthcare professional's satisfaction with care. Aspects of satisfaction include acceptance, of the equipment, and the woman healthcare professionals' interaction. Studies are required to examine the clinical outcomes of women using telemedicine compared to women who did not

Randomised controlled trials of support using telemedicine versus conventional support for diabetic women during pregnancy would be the best way of exploring the value of these new technologies.

24. What sequence and/or combinations of therapies best enable women to achieve blood glucose targets?

Why this is important

Tight glycaemic control in pregnancy is necessary to reduce fetal malformation rates and macrosomia. For women with type 1 diabetes this is usually achieved with multiple insulin injections at meal times together with intermediate or long-acting insulin at night (so called basal-bolus or MDI). For women with type 2 diabetes on oral therapy, or gestational diabetes the best sequence and/or combination of treatments is less clear. There is now widespread agreement that metformin is first line treatment but it is unclear whether the addition of glibenclamide or basal or meal-time insulin should be the next therapeutic intervention. An RCT of glibenclamide versus basal versus meal-time insulin would answer this important question. Initially, glycaemic targets would be the intermediate outcome but this could be expanded to look at fetal macrosomia rates and neonatal complications.

25. What are the barriers that women experience to achieving blood glucose targets?

Why this is important

Achieving good blood glucose control both before and during pregnancy in women with pre-existing diabetes is vital for normal fetal development in the first trimester. Good control also helps to prevent macrosomia and other complications in the third trimester in women with pre-existing or gestational diabetes. Whereas many women manage to achieve these targets, a proportion of women continue to find it difficult to do so. A number of factors could be involved, such as health beliefs, a poor understanding of the importance of good blood glucose control, an inability to be able to comply with a demanding regimen of up to 7-times daily blood glucose testing, and the need to adjust insulin dosage. A better understanding of the barriers in this cohort of women is needed so that healthcare professionals can work to overcome them. Robust qualitative studies are needed to explore these barriers, with the aim of improving blood glucose control and fetal outcomes in pregnancy for women with pre-existing diabetes and women with gestational diabetes.

5.3. HbA1c values for women with type 1, type 2 or gestational diabetes during pregnancy – monitoring and target values

5.3.1. Monitoring

5.3.1.1. Review question

What is the effectiveness of HbA1c monitoring in predicting adverse outcomes in women with type 1, type 2 or gestational diabetes during pregnancy?

5.3.1.2. Introduction

In the 2008 guideline a recommendation was made regarding HbA1c monitoring in pregnancy which stated that HbA1c should not be used routinely for assessing glycaemic control in the second and third trimesters of pregnancy. This does not rule out monitoring HbA1c if clinically indicated and is not explicit about whether or not to monitor HbA1c in the first trimester, although this is generally considered to be useful in advising the woman of the risk of adverse pregnancy outcome.

The review question in this update evaluates the effectiveness of monitoring HbA1c in pregnant women with type 1, type 2 or gestational diabetes, specifically in the context of whether the 2008 guideline recommendation not to monitor HbA1c routinely in the second and third trimesters of pregnancy should be changed.

The review also considers the frequency of monitoring HbA1c, whether monitoring HbA1c is more effective than monitoring blood glucose alone and whether different monitoring strategies are appropriate in women with type 1, type 2 and gestational diabetes.

5.3.1.3. Description of included studies

No relevant studies were identified for inclusion in the current review.

5.3.1.4. Evidence profile

There is no GRADE profile as no relevant studies were identified.

5.3.1.5. Evidence statements

No relevant studies were identified for inclusion in this review.

5.3.1.6. Health economics profile

No health economic evidence was identified that addressed the effectiveness of HbA1c monitoring in predicting adverse outcomes in women with type 1, type 2 or gestational diabetes during pregnancy.

This was not prioritised for health economic analysis as the guideline development group considered that any recommendation which departed from current practice would only have a relatively small cost impact and that there would be very limited clinical data to inform such an analysis.

5.3.1.7. Evidence to recommendations

5.3.1.7.1. Relative value placed on the outcomes considered

The guideline development group prioritised the following maternal outcomes for this review:

  • mode of birth (spontaneous vaginal, operative vaginal, caesarean section [elective or emergency])
  • pre-eclampsia (HbA1c may predict this)

The group also prioritised the following neonatal outcomes:

  • large for gestational age (however defined in the study, for example using a customised measure based on gestational age and population norms; dichotomous data were preferred)
  • neonatal intensive care unit (NICU) length of stay greater than 24 hours
  • shoulder dystocia (as a specific example of birth trauma)
  • neonatal hypoglycaemia (however defined)
  • any congenital abnormality, regardless of gestational age
  • mortality

The guideline development group prioritised any congenital abnormalities over preterm birth as an outcome because such abnormalities arise very early in pregnancy.

The group prioritised large for gestational age because babies of women with diabetes who are tested frequently are more likely to be smaller than babies of women who are tested less frequently.

The group recognised that there would be some overlap between NICU length of stay greater than 24 hours and presence of neonatal hypoglycaemia. The group prioritised neonatal hypoglycaemia because they considered that it was a more clinically useful outcome than the presence of neonatal hyperinsulinaemia or hyper C-peptide-aemia. However, they acknowledged that hyper C-peptide-aemia may be important in defining future research priorities.

5.3.1.7.2. Consideration of clinical benefits and harms

The guideline development group considered that a benefit of HbA1c monitoring is that it represents a retrospective average measure of glycaemic control. They wanted to establish whether HbA1c, particularly first-trimester HbA1c, could be useful (for example for counselling, fetal monitoring during pregnancy and evaluating the likelihood of needing neonatal intensive care).

The group noted several situations where HBA1c testing could be of particular value. For example, results of blood glucose monitoring may not be available in the clinic because women do not always bring their meters or diaries with them. Where this information is not available, knowledge of HbA1c results provides reassurance to clinicians and women in validating women's assertions that their glucose control is stable or improving. Also, where excessive fetal growth is of concern, HbA1c monitoring might highlight a need for more intensive glucose monitoring to enable more informed treatment adjustments or can confirm that there is no need to adjust insulin treatment to control growth if results correspond well to those for blood glucose. Another clinical scenario that the group identified related to late pregnancy. Where a women presents with a normal oral glucose tolerance test (OGTT) but glycosuria or a fetus that is assessed as being large for gestational age for late second trimester, an HbA1c result can be useful to determine whether there has been any glucose intolerance over the past 10 to 12 weeks, notwithstanding the OGTT result.

The group recognised that HbA1c has limitations in pregnancy because of changes in red cell turnover and the frequent occurrence of iron deficiency. Specifically, HbA1c tends to decrease in pregnancy. Also, in iron deficiency anaemia it can be less reliable. However, it can have value in being used as an audit tool of the process of care by checking once in each trimester. It has also been linked in population-based studies to pregnancy outcomes for mother and baby and forms part of the ongoing National Pregnancy in Diabetes audit (Murphy et al., 2013), and can be correlated to other outcomes.

The group was aware that there are alternatives to HbA1c monitoring, such as glycated albumin, fructosamine or 1,5-Anhydroglucitol monitoring. However, these assays are not widely available and there is limited robust data showing any advantage over HbA1c.

5.3.1.7.3. Consideration of health benefits and resource uses

The guideline development group considered whether HbA1c monitoring results would be useful as a biochemical test to determine which women should have a diagnostic OGTT. They concluded that HbA1c could be useful to identify those women with undiagnosed type 2 diabetes and provide an estimate for how long glucose dysregulation may have been present. This might inform discussions around increased surveillance for congenital anomaly if the hyperglycaemic period included embryogenesis. However, they acknowledged that there would be cost implications for more widespread HbA1c monitoring.

5.3.1.7.4. Other considerations

Variation in practice and the guidance from the 2008 guideline is not being followed for a variety of reasons.

The 2008 guideline noted that HbA1c had not been validated as a marker of average glycaemia in the second and third trimesters of pregnancy and because of physiological changes that occur in all pregnant women it could not be recommended for routine assessment of blood glucose control. There is a risk that an observed reduction in HbA1c in women with diabetes during the second and third trimesters of pregnancy might reflect changes in red cell production or anaemia and would not necessarily indicate improved glycaemic control, and thus could be falsely reassuring. These concerns were the basis for the previous recommendation.

The group acknowledged that is difficult to conduct the study that would suggest target values for HbA1c. However, they were aware of several observational studies in large cohorts of women with pre-existing diabetes where there were associations between increasing levels of HbA1c and worsening outcomes for women and their babies, including stillbirth (Tennant et al., 2014; Glinianaia et al., 2012; Murphy et al., 2011; Murphy et al., 2013). In other words, the group was of the view that although there was no evidence that routine HbA1c testing in pregnancy would be useful in assessing blood glucose control, it was, nonetheless, a marker of risk of adverse outcome and could be of value in practice for that purpose. It is difficult to establish the normal reference ranges for pregnancy because of the impact of anaemia and increased red cell turnover, but the data from the aforementioned observational studies indicates that an HbA1c value in pregnancy above 48 mmol/mol (6.5%) is associated with an increasing risk of adverse outcome.

In the light of these considerations the group decided to recommend that HbA1c should not be used in a diabetic pregnancy to assess glucose control: however, it could be used in specific circumstances to assess the risk in those pregnancies with 48 mmol/mol (6.5%) as a threshold.

Furthermore, the group felt that a high HbA1c at the time of diagnosis in a woman with apparent gestational diabetes in early pregnancy would identify women at increased risk of type 2 diabetes. This was important as the management might be different from that offered to women with gestational diabetes. This might include undertaking retinal and renal screening, greater attention to the results of blood glucose monitoring, with an increased vigilance for the need for pharmacological treatment, and a different monitoring and surveillance strategy after delivery.

5.3.1.8. Key conclusions

The guideline development group noted that HbA1c monitoring in pregnancy was not recommended in the previous guideline. The current guideline development group agreed that there is no evidence to recommend its routine use in pregnancy as a measure of glucose control. It is not currently possible to advocate an alternative measure of average glycaemic control.

The group believed that HbA1c was an indicator of risk of adverse outcome in a diabetic pregnancy, with that risk increasing progressively above 48 mmol/mol (6.5%).

The group also believed that an increased HbA1c at the time of diagnosis of gestational diabetes in early pregnancy raised the possibilty that the woman had previously unrecognised type 2 diabetes.

5.3.1.9. Recommendations

70.

Measure HbA1c levels in all pregnant women with pre-existing diabetes at the booking appointment to determine the level of risk for the pregnancy. [new 2015]

71.

Consider measuring HbA1c levels in the second and third trimesters of pregnancy for women with pre-existing diabetes to assess the level of risk for the pregnancy. [new 2015]

72.

Be aware that level of risk for the pregnancy for women with pre-existing diabetes increases with an HbA1c level above 48 mmol/mol (6.5 %). [new 2015]

73.

Measure HbA1c levels in all women with gestational diabetes at the time of diagnosis to identify those who may have pre-existing type 2 diabetes. [new 2015]

74.

Do not use HbA1c levels routinely to assess a woman's blood glucose control in the second and third trimesters of pregnancy. [2008]

5.3.2. Target values

The purpose of this review is to identify target values for HbA1c in women with type 1, type 2 or gestational diabetes during pregnancy. The search for this study included RCTs, systematic reviews and comparative observational studies. Non-comparative observational studies were to be included only if no comparative studies were identified. The same search was used to identify studies for this review and the reviews of target values for blood glucose pre-pregnancy and during pregnancy, target values for HbA1c pre-pregnancy and for blood glucose and HbA1c monitoring during pregnancy.

The guideline development group defined 8 priority outcomes for this review. Maternal outcomes were:

  • mode of birth (spontaneous vaginal, operative vaginal or elective or emergency caesarean section)
  • pre-eclampsia
  • hypoglycaemic episodes at any time during pregnancy.

Neonatal outcomes were:

  • large for gestational age
  • neonatal hypoglycaemia
  • neonatal intensive care unit (NICU) length of stay greater than 24 hours
  • shoulder dystocia
  • mortality (defined as perinatal [stillbirth and death up to 7 days after birth] or neonatal [death up to 28 days after birth]).

The original review question in the 2008 guideline was “What are the target ranges for blood glucose during pregnancy?” Studies that examined glycaemic control using blood glucose or HbA1c measurements were included as evidence in the chapter. A more specific approach has been taken in this update. Four separate review questions have been stipulated to examine blood glucose or HbA1c measurements prior to conception and during pregnancy.

Sixteen studies were included in the previous guideline on target values during pregnancy. The majority of these studies examined HbA1c and were considered for inclusion in this review.

5.3.2.1. Review question

What is the target value for HbA1c in women with type 1, type 2 or gestational diabetes during pregnancy?

5.3.2.2. Description of included studies

Four studies met the inclusion criteria for this review. One was a prospective cohort study (Ekbom et al., 2008) and 3 were retrospective cohort studies (Barnes et al., 2013; Mikkelsen et al., 2011; Vaarasmaki et al., 2000). Studies were carried out in Denmark (Ekbom et al., 2008; Mikkelsen et al., 2011), Australia (Barnes et al., 2013) and Finland (Vaarasmaki et al., 2000).

Numbers of participants ranged from 148 to 1695. Women had gestational diabetes (Barnes et al., 2013; Mikkelsen et al., 2011), type 1 diabetes (Ekbom et al., 2008) and type 1 and 2 diabetes (White class B to R) (Vaarasmaki et al., 2000). HbA1c thresholds used as cut-offs for optimal control were 37 mmol/mol (5.5%) (Barnes et al., 2013), 38 mmol/mol (5.6%) (Mikkelsen et al., 2011), 48 mmol/mol (6.5%) (Ekbom et al., 2008) and at different thresholds of either between 20 and 42 mmol/mol (4.0% and 6.0%) or less than 64 mmol/mol (8.0%), depending upon the time period when women were treated (Vaarasmaki et al., 2000). The first 3 studies used HbA1c, but the latter study used the term ‘glycosylated haemoglobin’ throughout (Vaarasmaki et al., 2000). The higher value of 64 mmol/mol (8.0%) used in this study was considered to be based on HbA1 due to the time period that data were collected. This value was therefore converted to HbA1c using the Michigan formula. Only 1 of the 4 studies set specific target values for women to achieve and reported the numbers who achieved the targets (Mikkelsen et al., 2011). One of the studies referred to the use of DCCT-aligned (Diabetes Control and Complications Trial) values for HbA1c in 10% of the women included in the study (Ekbom et al., 2008). The remaining 2 studies did not refer to the use of DCCT-alignment.

Of the guideline development group priority outcomes, 2 studies reported evidence for large for gestational age (Barnes et al., 2013; Mikkelsen et al., 2011), 1 for pre-eclampsia, shoulder dystocia and mode of delivery (Mikkelsen et al., 2011), 1 for neonatal hypoglycaemia (Mikkelsen et al., 2011), 1 for maternal hypoglycaemia (Ekbom et al., 2008) and 1 for NICU stay (Vaarasmaki et al., 2000).

5.3.2.3. Evidence profile

GRADE profiles are presented according to HbA1c thresholds. Reasons for the use of each threshold are given in Table 59.

Table 59. HbA1c thresholds for optimal control used in GRADE profiles with reasons for their use.

Table 59

HbA1c thresholds for optimal control used in GRADE profiles with reasons for their use.

The evidence for this profile is presented in Tables 60 to 63.

Table 60. GRADE profile for comparison of HbA1c of 37 mmol/mol (5.5%) or less with HbA1c greater than 37 mmol/mol (5.5%) during pregnancy in women with gestational diabetes mellitus.

Table 60

GRADE profile for comparison of HbA1c of 37 mmol/mol (5.5%) or less with HbA1c greater than 37 mmol/mol (5.5%) during pregnancy in women with gestational diabetes mellitus.

Table 61. GRADE profile for comparison of HbA1c of 38 mmol/mol (5.6%) or less with HbA1c greater than 38 mmol/mol (5.6%) during pregnancy in women with gestational diabetes mellitus.

Table 61

GRADE profile for comparison of HbA1c of 38 mmol/mol (5.6%) or less with HbA1c greater than 38 mmol/mol (5.6%) during pregnancy in women with gestational diabetes mellitus.

Table 62. GRADE profile for comparison of HbA1c of 48 mmol/mol (6.5%) or less with HbA1c greater than 48 mmol/mol (6.5%) during pregnancy in women with type 1 diabetes mellitus.

Table 62

GRADE profile for comparison of HbA1c of 48 mmol/mol (6.5%) or less with HbA1c greater than 48 mmol/mol (6.5%) during pregnancy in women with type 1 diabetes mellitus.

Table 63. GRADE profile for comparison of HbA1c between 20 and 42 mmol/mol (4.0% and 6.0%) or less than 56 mmol/mol (7.3%) with HbA1c greater than 42 mmol/mol (6.0%) or 56 mmol/mol (7.3%) or more during pregnancy in women with White class diabetes B to R.

Table 63

GRADE profile for comparison of HbA1c between 20 and 42 mmol/mol (4.0% and 6.0%) or less than 56 mmol/mol (7.3%) with HbA1c greater than 42 mmol/mol (6.0%) or 56 mmol/mol (7.3%) or more during pregnancy in women with White class diabetes B to R.

5.3.2.4. Evidence statements

5.3.2.4.1. HbA1c level 37 mmol/mol (5.5%) versus more than 37 mmol/mol (5.5%)

One study (number not reported) found an increase in the risk of large for gestational age neonates in women with an HbA1c greater than 37 mmol/mol (5.5%) compared with women with an HbA1c less than or equal to 37 mmol/mol (5.5%) (OR 1.38, 95% CI 1.01 to 1.90). The quality of the evidence for this outcome was very low.

5.3.2.4.2. HbA1c level 38 mmol/mol (5.6%) versus more than 38 mmol/mol (5.6%)

One study (n=148) found a reduction in risk of being large for gestational age (RR 0.47, 95% 0.27 to 0.81) and of hypoglycaemia (RR 0.30, 95% 0.15 to 0.60) in the neonates of women with gestational diabetes mellitus who obtained a treatment goal of HbA1c of 38 mmol/mol (5.6%) or less compared with those who did not achieve the treatment goal. The quality of evidence for these outcomes was very low and moderate, respectively. The same study found no difference between groups for the outcomes of pre-eclampsia (RR 1.23, 95% CI 0.33 to 4.56), shoulder dystocia (RR 2.65, 95% CI 0.13 to 54.18) or mode of delivery (RR 1.05, 95% CI 0.47 to 1.72). The quality of evidence for these outcomes was very low.

5.3.2.4.3. HbA1c level 48 mmol/mol (6.5%) versus more than 48 mmol/mol (6.5%)

One study (n=213) found no difference between groups in the risk of maternal hypoglycaemia in women with type 1 diabetes mellitus whose HbA1c levels were less than or equal to 48 mmol/mol (6.5%) compared with those whose HbA1c levels were greater than 48 mmol/mol (6.5%) RR 1.25 95% CI 0.65 to 2.44). The quality of evidence for this outcome was very low.

5.3.2.4.4. HbA1c levels 20–42 mmol/mol (4.0% to 6.0%) or less than 56 mmol/mol (7.3%) versus more than 42 mmol/mol (6.0%) or 56 mmol/mol (7.3%)

One study (n=84) found a reduced risk of neonatal unit stay of longer than 10 days associated with women with White class diabetes B to R who have HbA1c levels of 20–42 mmol/mol (4.0% to 6.0%) or less than 56 mmol/mol (7.3%) (converted from a reported HbA1 value of 8.0%) compared with those who have HbA1c levels greater than 42 mmol/mol (6.0%) or 56 mmol/mol (7.3%) or more (RR 0.14, 95% CI 0.03 to 0.59). The quality of evidence for this outcome was very low.

No evidence was identified for neonatal mortality in any of the studies included in this review.

5.3.2.5. Health economics profile

No health economic evidence was identified that addressed the target value for HbA1c in women with type 1, type 2 or gestational diabetes during pregnancy.

This was not prioritised for health economic analysis because although a target may affect the interventions and management used to assist the patient in achieving that target, a target of itself does not incur an opportunity cost.

5.3.2.6. Evidence to recommendations

5.3.2.6.1. Relative value placed on the outcomes considered

The guideline development group prioritised mode of delivery and pre-eclampsia for the maternal outcomes and large for gestational age and shoulder dystocia for the neonatal outcomes.

5.3.2.6.2. Consideration of clinical benefits and harms

The clinical benefits and harms of setting HbA1c targets in pregnancy were not fully clarified by this review, with the 4 studies all having significant limitations and variable findings.

Both the Barnes et al. (2013) study and the Mikkelson et al. (2011) study suggested that a threshold of 37–38 mmol/mol (5.5–5.6%) did reduce the incidence of large for gestational age babies, but in the Mikkelson et al. (2011) study there was no reduction in the incidence of shoulder dystocia. Another apparent fetal/neonatal benefit of this target threshold was a significant reduction of neonatal hypoglycaemia. The guideline development group commented, however, that the population in this study may not have been reflective of all women with gestational diabetes, with the profile being skewed towards the more severe end of the spectrum, and this might explain some of the benefits to the infant. Nevertheless, there were benefits. There was no benefit to the woman in either the incidence of pre-eclampsia or caesarean section rates. But the group commented that this might be influenced by the fact that most of the women were recruited relatively late in pregnancy and the potential for improved glycaemic control improving outcome would be limited, and acknowledged that the incidence of large for gestational age babies was reduced..

The Ekbom et al. (2008) study used a higher target threshold (48 mmol/mol (6.5%) or less) and although no differences in outcomes were found, it was reassuring that the incidence of maternal hypoglycaemia was not increased in the group achieving ‘tighter’ control. However, it should be acknowledged that 45 mmol/mol (6.3%) is a relatively ‘lenient’ threshold, being above the upper limit of ‘normal’ for a person without diabetes. It is possible that if a lower threshold had been used then maternal hypoglycaemia might have occurred.

Finally, the Varaasmaki et al. (2000) study showed a lower incidence of stay greater than 10 days in the NICU. The group questioned the reasons the authors chose this outcome when such a length of stay is an uncommon outcome for infants of diabetic women. The more common outcome is a stay of less than 48 hours for neonatal hypoglycaemia. The paper did not provide the reasons for choosing this outcome nor the indications for NICU admission.

5.3.2.6.3. Consideration of health benefits and resource uses

Obtaining an HbA1c level incurs an opportunity cost, both in terms of laboratory analysis and staff time. There is uncertainty about what would be a normal range of HbA1c in pregnancy and how it may vary across different trimesters. There is a lack of good quality data on the use of HbA1c in pregnancy and therefore its routine use to assess glycaemic control does not currently justify the opportunity cost.

5.3.2.6.4. Quality of evidence

Overall the studies were of very low quality. The reasons for this have been discussed above.

The Mikkelsen et al. (2011) study recruited women late in pregnancy, which limits achievable improvement in outcome. Also, this study seems to include women from the more severe end of the gestational diabetes spectrum, and it cannot be assumed that data would apply to all women with gestational diabetes.

In the Ekbom et al. (2008) study maternal hypoglycaemia did not increase despite tighter control. However, hypoglycaemia was not defined and the chosen HbA1c threshold was not particularly low.

Although the guideline development group decided to retain the Vaarasmaki et al. (2000) study, it was very indirect, both for outcome of interest (the protocol agreed outcome for NICU stay was over 24 hours not 10 days) and differing HbA1c thresholds which were merged in the analysis. Finally, glycaemic control was defined according to values that were not within the reference range based on the time period when data were collected (1986–92). In the light of these factors, the group decided to not to place too much weight on this study.

5.3.2.6.5. Other considerations

The guideline development group was concerned that normal ranges of HbA1c in pregnancy have not been established and different thresholds probably should be set for different trimesters. But there were no data to inform such a recommendation. Also, HbA1c values are lower if the woman has iron deficiency anaemia.

The review did not look at the use of HbA1c in women with gestational diabetes diagnosed in the first trimester to identify those who might actually have type 2 diabetes. The group was aware that the IADPSG diagnostic criteria suggest that a value of more than 6.5% at the first prenatal visit would identify women with ‘overt diabetes’. As a consequence, the group felt that this might provide a specific indication for targeted HbA1c testing.

Given the paucity of good quality data, the guideline development group concurred with the 2008 guideline that HbA1c should not be used routinely to assess glycaemic control in diabetic pregnancies. However, the guideline development group for this guideline agreed, on the basis of their clinical experience and the reasoning above, that selective monitoring of HbA1c can be useful during pregnancy, although healthcare professionals need to be aware of its potential drawbacks. It is not currently possible to advocate an alternative measure of average glycaemic control.

Examples of when the test could be used as an adjunct to assessment of glycaemic control in selected cases included when women do not bring in their glucose monitor, in the absence of extensive home blood glucose monitoring and to motivate or reassure women.

The guideline development group agreed that the HbA1c value at booking was important to inform about the risk of fetal and neonatal outcome. Even if we do not have evidence of the usefulness of HbA1c after the first trimester, measuring HbA1c could be useful to inform the management of diabetes in pregnancy to avoid negative maternal or fetal outcomes.

In view of the lack of evidence about what normal values for HbA1c should be during pregnancy, the guideline development group felt unable to make any recommendation regarding target values for HbA1c.

5.3.2.6.6. Recommendations

The guideline development group did not make any recommendation regarding target values for HbA1c.

5.3.2.7. Research recommendations

26. What are the normal ranges for HbA1c in non-diabetic pregnancy?
Why this is important

HbA1c is an important widely used indicator of glycaemic control, indicating glycaemic control over the preceding 10-12 weeks, and is used outside pregnancy to indicate the quality of glycaemic control in the medium term. HbA1c is lower in pregnancy for a variety of reasons inluding increased red cell turnover and relative iron deficiency. Prospective longitudinal observational studies in women, confirmed by glucose tolerance testing to not have diabetes, documenting the normal ranges of HbA1c in pregnancy in the non-diabetic population are needed. Once these normal ranges for HbA1c during pregnancy are established, that will allow them to be used in assessing the quality of control in women with diabetes.

27. Which is the optimum timing of the post-prandial blood glucose test in pregnancy – 1, 1.5 or 2 hours?
Why this is important

The optimum timing of the post-prandial blood glucose test is the time of the peak blood glucose values following a meal. This result helps to inform insulin does adjustment for the prandial insulin as well as the assessment of overall glycaemic control. The timing of the peak of the blood glucose rise after meals is dependent on a number of factors including gastric emptying times and the glycaemic index of the diet and its fat content. Gastric emptying tends to be slower in pregnancy and in women with diabetic gastroparesis. It is uncertain to what extent changes in gastric emptying effect the timing of the glucose peak after meals throughout pregnancy. Studies need to be performed to determine the optimum time for the post-prandial blood glucose test in pregnancy as to ascertain whether this should be the same for each trimester.

28. What are the barriers to testing blood glucose frequently in pregnancy?
Why this is important

Achieving optimal glycaemic control during pregnancy in women with diabetes lessens the likelihood of adverse outcomes especially with respect to the fetus/newborn. Whilst many women manage to achieve these targets, there is a proportion of women who find it difficult to do so. A number of contributing factors could be involved, such as disbelief in the health benefits, a poor understanding of the importance of good blood glucose control, an inability to be able to comply with the a demanding regime of up to 7-times daily blood glucose testing. A better understanding of the barriers to testing blood glucose in this cohort of women is needed in order that health care professionals to can work with women to overcome them. Robust qualitative studies are needed to explore these barriers, with the aim of improving glycaemic control and fetal outcomes in pregnancy for women with pre-existing diabetes and gestational diabetes.

29. Are other glycosylated molecules better than HbA1c at summarising blood glucose control in pregnancy?
Why this is important

Increasing glycosylated haemoglobin in early pregnancy is associated with an increasing risk of congenital malformations and miscarriage and in late pregnancy with an increased risk of excessive fetal growth and neonatal morbidity. However, it assesses diabetes control over the previous 10-12 weeks and hence anomalous results are sometimes found which do not seem to concur with glycaemic control as observed on a day to day basis. Other molecules are also glycosylated but reflect glucose control over a shorter time period. It may be that these are of more value in advising clinicians about glucose control and hence the risks to the fetus. Prospective observational studies are required which compare glycosylated haemoglobin with other glycosylated molecules, with glucose control being assessed by continuous glucose monitoring and standard maternal and neonatal outcomes being studied.

5.4. Management of diabetes during pregnancy

5.4.1. Description of the evidence

5.4.1.1. Hypoglycaemia

Hypoglycaemia significantly affects maternal quality of life and increases the risk of physical injury. During pregnancy the frequency of hypoglycaemia may increase due to intensification of treatment, an impairment of counter-regulatory hormonal responses and an increased risk of hypoglycaemia unawareness.214,215 Pregnancy nausea and vomiting can also contribute to hypoglycaemia due to fluctuations in carbohydrate ingestion. [EL = 2+]

A cohort study of 84 pregnant women with type 1 diabetes undergoing intensified treatment68 found hypoglycaemia requiring third-party assistance occurred in 71% of women, with a peak incidence between 10–15 weeks of gestation. Consequences of maternal hypoglycaemia included several grand mal seizures, five episodes of cerebral oedema, and two road traffic accidents and comminuted fracture of the tibia and fibula. [EL = 2++]

5.4.1.2. Hyperemesis gravidarum

Severe nausea and vomiting in pregnant women with diabetes can lead to ketoacidosis, and DKA during pregnancy carries a risk of fetal death.

Two case studies were identified that reported healthy live births to women with diabetes and hyperemesis gravidarum following treatment with parenteral nutrition.216,217 One case study reported a fetal death in a woman with hyperemesis gravidarum and DKA following a delay in treatment.218 [EL = 3]

5.4.1.3. Diabetic ketoacidosis

A case series of 37 women admitted with DKA219 concluded that vomiting and the use of betamimetic medicines were the primary cause in 57% of cases. Non-adherence to treatment and physician management errors were the primary cause in 24% of cases and contributory in 16%. Common physician management errors included the use of urine instead of blood to monitor maternal glucose control, failure to adhere to pregnancy standards of glucose control and failure to employ home blood glucose monitoring. [EL = 3]

A cohort study of 257 people220 with DKA admitted to a large urban teaching hospital compared outcomes in people treated by a general physician (n = 224) with those in people treated by a physician with subspecialty training in diabetes (n = 33). People treated by a diabetes specialist had shorter length of stay (3.3 versus 4.9 days, P < 0.0043) and incurred lower hospital charges ($5,463 versus $10,109, P < 0.0001). Plasma glucose in generalist-treated people took longer to fall to less than 11.1 mmol/litre and they had a higher rate of readmission for DKA than the specialist-treated people (6% versus 2%, P = 0.03). [EL = 2+]Rapid-acting insulin analogues

Rapid-acting insulin analogues (aspart and lispro) confer the following benefits compared with regular insulin outside pregnancy:221

  • fewer episodes of hypoglycaemia
  • a reduction in postprandial glucose excursions
  • an improvement in overall glycaemic control
  • an improvement in patient satisfaction.

These benefits have also been demonstrated in the pregnant population (see Section 3.8).

5.4.1.4. Long-acting insulin analogues

The NICE guideline for the management of type 1 diabetes recommends the long-acting insulin analogue glargine for use outside of pregnancy.221 However, no clinical trials have as yet been published for their use in pregnancy (see Section 3.9).

5.4.1.5. Four-times-daily versus twice-daily insulin regimens

An open label RCT compared glycaemic control and perinatal outcomes in pregnant women with diabetes using two different insulin regimens.222 One hundred and thirty-eight women with gestational diabetes and 58 with pre-existing diabetes received insulin four times daily, and 136 women with gestational diabetes and 60 with pre-existing diabetes received insulin twice daily. Glycaemic control was better with the four-times-daily regimen than with the twice-daily regimen. In women with gestational diabetes the four-times-daily regimen resulted in a lower rate of overall neonatal morbidity than the twice-daily regimen. Four-times-daily rather than twice- daily insulin improved glycaemic control and perinatal outcomes without increasing the risks of maternal hypoglycaemia and caesarean section. [EL = 1++]

5.4.1.6. Continuous subcutaneous insulin infusion (insulin pump therapy)

The most widespread method of administering insulin is via subcutaneous insulin injections using a basal/bolus regimen consisting of a basal dose of long-acting insulin, usually administered with a pen before bed, and bolus of rapid-acting insulin given before meals. This is often referred to as a multiple daily injection (MDI) regimen. Insulin can also be administered using CSII (also known as insulin pump therapy). Both regular insulin and rapid-acting insulin can be administered by pump. The potential benefits of CSII are reduced risk of hypoglycaemia, decreased risk of fasting hyperglycaemia and improved adherence as the woman does not have to constantly inject insulin.223

NICE guidance for the non-pregnant population concluded that, compared with MDI regimens, CSII results in a modest but worthwhile improvement in blood glucose control and quality of life.14

A systematic review224 investigated the effectiveness of insulin delivery via CSII as compared with MDI regimens for the treatment of diabetes during pregnancy in women with pre-existing diabetes or gestational diabetes. Only two studies were included in the review, neither included women with gestational diabetes. There was a significant increase in mean birthweight associated with CSII as opposed to MDI (two trials, 61 participants, weighted mean difference (WMD) 220.56, 95% CI −2.09 to 443.20). However, taking into consideration the lack of significant difference in rate of macrosomia (birthweight greater than 4000 g; RR 3.20, 95% CI 0.14 to 72.62), this finding was not viewed by the authors as being clinically significant. There were no significant differences in perinatal outcomes between CSII and MDI (perinatal mortality, including stillbirths from 24 weeks of gestation and neonatal deaths up to 7 days of life, RR 2.00, 95% CI 0.20 to 19.91; fetal anomaly, RR 1.07, 95% CI 0.07 to 15.54; gestational age at birth, WMD 0.63, 95% CI −4.87 to 6.13; neonatal hypoglycaemia, RR 1.00, 95% CI 0.07 to 14.64; and SGA, RR 1.55, 95% CI 0.27 to 9.00). Neither were there any significant differences in maternal outcomes between CSII and MDI (caesarean section rate, RR 1.03, 95% CI 0.57 to 1.84; mean maternal HbA1c; 24 hour mean blood glucose level in each trimester; hypoglycaemia; or hyperglycaemia). [EL1+]

Three further RCTs of CSII in women with type 1 diabetes during pregnancy were identified.225227 The studies included a total of 200 women. There were no significant differences between groups in glycaemic control or in obstetric or neonatal outcomes. There were four cases of ketoacidosis in women using pumps. This was attributed to catheter occlusion (one case), catheter leakage (one case) and pump failure (one case). Another case was reported but without attribution. [EL = 1++]

One RCT was designed to assess the effect of CSII on retinopathy.226 This study of 40 women with type 1 diabetes reported progression to proliferative retinopathy in two women using CSII. This was attributed to rapid and significant improvement in glycaemic control. [EL = 1+]

Three cohort studies compared outcomes in women using CSII with those in women using MDI regimens. [EL = 2+]228 [EL = 2+]229 [EL = 2++]230 In each study women were offered pump therapy due to difficulties achieving glycaemic control. All studies reported good glycaemic control and obstetric and neonatal outcomes.

5.4.1.7. Current practice

The CEMACH enquiry (comparison of women with type 1 and type 2 diabetes) reported that women with type 1 diabetes were more likely to experience recurrent episodes of hypoglycaemia than women with type 2 diabetes (P < 0.001), with 61% (105/171) of women with type 1 diabetes and 21% (25/121) of the women with type 2 diabetes having recurrent episodes of hypoglycaemia.33 One or more episodes of hypoglycaemia required help in 25% (33/133) of the women with type 1 diabetes and 4% (4/102) of the women with type 2 diabetes. [EL = 3−4]

5.4.2. Existing guidance

The NICE technology appraisal relating to insulin pump therapy (CSII) for people with type 1 diabetes states that insulin pumps can be used in pregnancy even if there is good glycaemic control on MDI regimens.14

The Driver and Vehicle Licensing Agency (DVLA) medical rules for drivers do not include any special considerations for diabetes in pregnancy. Fitness to drive is assessed on the basis of the risk of hypoglycaemia, regardless of whether or not the driver is pregnant (see www.dvla.gov.uk/medical.aspx and www.direct.gov.uk/en/Motoring/DriverLicensing/MedicalRulesForDrivers/index.htm) and https://www.gov.uk/diabetes-driving.

5.4.3. Evidence statement

During pregnancy women with diabetes treated using insulin are at an increased risk of hypoglycaemia and hypoglycaemia unawareness.

Rapid-acting insulin analogues (aspart and lispro) are associated with fewer episodes of hypoglycaemia compared with regular human insulin. When compared with regular human insulin the use of rapid-acting insulin analogues during pregnancy has also been associated with a reduction in postprandial glucose excursions, an improvement in overall glycaemic control and an improvement in patient satisfaction.

Ketoacidosis is a complication that can result in fetal death. Outcomes may be improved with prompt assessment and treatment by a health professional with specialist diabetes training.

RCTs have shown similar outcomes in women using CSII and MDI regimens. Ketoacidosis may result from pump failure. Cohort studies have reported good outcomes in women offered pump therapy because of difficulty achieving glycaemic control using MDI regimens.

5.4.4. From evidence to recommendations

Women and their partners should be informed of the increased risk of hypoglycaemia and hypoglycaemia unawareness during pregnancy, and information about prevention, recognition and treatment (including the provision of a concentrated glucose solution and, if they have type 1 diabetes, glucagon, and education in their use) should be reinforced in women with insulin- treated diabetes who are pregnant. Women with insulin-treated diabetes should also be advised of the consequences of hypoglycaemia and the dangers associated with driving during periods of hypoglycaemia unawareness. They should be encouraged to carry something that identifies them as having diabetes so they can be treated promptly if disabling hypoglycaemia occurs.

The evidence supports the use of the rapid-acting insulin analogues aspart and lispro in women with diabetes in pregnancy, and also insulin pump therapy (CSII) in women who have difficulty achieving glycaemic control without disabling hypoglycaemia.

Since DKA can be accelerated in pregnancy and is associated with serious maternal and fetal adverse outcomes (including fetal death), the 2008 GDG's consensus view was that current best practice should be followed namely that DKA should be excluded in women with type 1 diabetes who become unwell in pregnancy and pregnant women with DKA should be admitted immediately for level 2nn critical care where they can receive medical and obstetric care. However, in addition, the 2014 GDG also noted that, though DKA was primarily a complication of women with type 1 diabetes, there were a number of case reports that indicated that DKA could also occur in women with either type 2 diabetes or gestational diabetes and they felt that the recommendation should not be confined to women with Type 1 diabetes.

5.4.5. Recommendations

Insulin treatment and risks of hypoglycaemia

75.

Be aware that the rapid-acting insulin analogues (aspart and lispro) have advantages over soluble human insulin during pregnancy and consider their use. [2008]

76.

Advise women with insulin-treated diabetes of the risks of hypoglycaemia and impaired awareness of hypoglycaemia in pregnancy, particularly in the first trimester. [2008]

77.

Advise pregnant women with insulin-treated diabetes to always have available a fast-acting form of glucose (for example, dextrose tablets or glucose-containing drinks). [2008, amended 2015]

78.

Provide glucagon to pregnant women with type 1 diabetes for use if needed. Instruct the woman and her partner or other family members in its use. [2008, amended 2015]

79.

Offer women with insulin-treated diabetes continuous subcutaneous insulin infusion (CSII; also known as insulin pump therapy) during pregnancy if adequate blood glucose control is not obtained by multiple daily injections of insulin without significant disabling hypoglycaemiaoo. [2008]

5.4.6. Research recommendations

30. Do new-generation CSII pumps offer an advantage over traditional intermittent insulin injections in terms of pregnancy outcomes in women with type 1 diabetes? [2008]

Why this is important

Randomised controlled trials have shown no advantage or disadvantage of using continuous subcutaneous insulin infusion (CSII) pumps over multiple intermittent insulin injections in pregnancy in terms of glycaemic control or maternal and fetal outcomes. The new generation of CSII pumps (± continuous glucose monitoring – CGM) with more sophisticated technologies enabling more precise insulin dosing offer technological advantages that would make a randomised controlled trial appropriate.

5.5. Continuous glucose monitoring

5.5.1. Introduction

Continuous glucose monitoring (CGM) has been available for over 12 years. Initially, the technology was retrospective and patients were unaware of their glucose values. More recently ‘real time’ glucose monitoring has been used, often in combination with continuous subcutaneous insulin infusion (CSII, or insulin pump) therapy. Although there are differences between the available systems the basic principles are the same. A glucose sensor is implanted subcutaneously and a small transmitter is clipped onto it on the skin surface. The sensor comprises a thin needle that is covered in glucose oxidase enzyme; this reacts with glucose in the subcutaneous tissue, freeing electrons which produce a small charge that is picked up by the transmitter. Interstitial fluid (not blood) glucose levels are measured every few minutes and the results sent via the transmitter, either to an insulin pump or a separate monitor. Results can be blinded or available real time. Sensors last for 5–7 days and then have to be replaced. The transmitters need replacing every year. Some sensors need calibrating with a capillary blood glucose sample once or twice a day.

There is a variable (but on average good) correlation between interstitial fluid and blood glucose, but the precision is not as good in the hypoglycaemic range. There is also a lag time for equilibration between blood and interstitial glucose concentrations which varies between individuals and according to the rate of change of blood glucose. This lag can be up to 45 minutes but is less with newer sensors.

Clinical studies of CGM have shown a modest improvement in HbA1c of approximately 5 mmol/mol (0.43%) but no consistent benefit in terms of rates of severe hypoglycaemia, largely because patients at higher risk have often been excluded from the trials.

5.5.2. Review question

What is the effectiveness of continuous glucose monitoring in pregnant women with diabetes compared with intermittent capillary blood glucose monitoring?

5.5.3. Objective

The objective of this review question is to assess whether continuous glucose monitoring during pregnancy is more effective than intermittent capillary blood glucose monitoring for improving both glycaemic control and other maternal and fetal/neonatal outcomes.

5.5.4. Description of included studies

Five studies were included in the current review (Kestila et al., 2007; Kerssen et al., 2006; Murphy et al., 2008; Secher et al., 2013; Yogev et al., 2003). Three studies were randomised controlled trials (Kestila et al., 2007; Murphy et al., 2008; Secher et al., 2013) and 2 were within-participant comparisons (Kerssen et al., 2006; Yogev et al., 2003).

The studies were conducted in the UK (Murphy et al., 2008), the Netherlands (Kerssen et al., 2006), Finland (Kestila et al., 2007), Denmark (Secher et al., 2013) and Israel (Yogev et al., 2003). The number of women in the studies ranged from 34 (Yogev et al., 2003) to 154 (Secher et al., 2013). Two studies reported on women with type 1 diabetes (Yogev et al., 2003; Kerssen et al., 2006), 2 studies reported on women with type 1 and women with type 2 diabetes (Murphy et al., 2008; Secher et al., 2013) and 1 study reported on women with gestational diabetes (Kestila et al., 2007).

Table 64. Description of the methods used for continuous and intermittent measurements of blood glucose in each study.

Table 64

Description of the methods used for continuous and intermittent measurements of blood glucose in each study.

5.5.5. Evidence profile

The GRADE profile for this review question is presented in Table 65 below.

Table 65. GRADE profile for effectiveness of continuous glucose monitoring in pregnancy women with diabetes compared with intermittent capillary blood glucose monitoring.

Table 65

GRADE profile for effectiveness of continuous glucose monitoring in pregnancy women with diabetes compared with intermittent capillary blood glucose monitoring.

5.5.6. Evidence statements

5.5.6.1. Maternal outcomes

When women who used intermittent glucose monitoring were compared with those who used continuous glucose monitoring, there were no differences in rates of unassisted vaginal birth (RR 0.9, 95% CI 0.7 to 1.2, 2 RCTs, n=144), assisted vaginal birth (RR 1.0, 95% CI 0.2 to 4.8, 1 RCT, n=73) or caesarean section (RR 1.0, 95% CI 0.8 to 1.3, 3 RCTs, n=298). There were also no differences in the risk of pre-term birth (RR 1.1, 95% CI 0.7 to 1.9, 3 RCTs, n=298) or in the mean gestational age at birth (MD −0.4 weeks, 95% CI −1.0 to 0.2, 1 RCT n=73). The evidence for these findings was of very low to low quality.

There were no differences in mean HbA1c values in women who used continuous monitoring compared with women who used intermittent monitoring at 8 weeks' gestation (MD −0.2, 95% CI not calculable [NC], p=0.72, 1 RCT n=149), 28 to 32 weeks' gestation (MD −0.3, 95% CI −0.6 to 0.03, 1 RCT, n=71), 33 weeks' gestation (MD 0.0, 95% CI NC, p=0.39, 1 RCT, n=149) and 36 weeks' gestation (MD −0.1, 95% CI NC, p=0.63, 1 RCT, n=149),. However, 1 RCT found that at 32 to 36 weeks' gestation the mean HbA1c value was reduced in women who used continuous monitoring compared with intermittent monitoring (MD −0.6, 95% CI −0.9 to −0.3, n=71). In contrast, results from a within-participants study reported higher mean glucose levels detected with continuous monitoring compared with intermittent monitoring (MD 1.1, 95% CI 0.8. to 1.5, n=34). A second within-participants study reported no differences in mean glucose levels detected in women using continuous glucose monitoring compared with those using intermittent monitoring at measurement frequencies of 4 to 5 readings a day (MD 0.1, 95% CI NC, p=not specified [NS], n=86), 6 to 9 readings a day (MD −0.2, 95% CI NC, p=NS, n=86) or 10 or more readings a day (MD −0.2, 95% CI NC, p=NS, n=86). The evidence for these findings was of very low to moderate quality.

One RCT reported no difference in the risk of women having at least 1 severe hypoglycaemic episode in the continuous monitoring group compared with the intermittent monitoring group (RR 1.0, 95% CI 0.5 to 2.1, 1 RCT, n=154). The evidence for this finding was of moderate quality. A within-participants study reported no differences in the mean hypoglycaemic episodes detected in women using continuous glucose monitoring compared with those using intermittent monitoring at measurement frequencies of 4 to 5 readings a day (MD 1.7, 95% CI NC, p=NS, n=86), 6 to 9 readings a day (MD −1.3, 95% CI NC, p=NS, n=86) or 10 or more readings a day (MD 1.0, 95% CI NC, p=NS, n=86). The evidence for this finding was of very low quality. None of the studies reported maternal satisfaction or maternal mortality.

5.5.6.2. Neonatal outcomes

When women who used intermittent glucose monitoring were compared with those who used continuous glucose monitoring, there were no differences in the risks of miscarriage (RR 1.4, 95% CI 0.2 to 8.3, 1 RCT, n=154), early neonatal death (RR 0.9, 95% CI 0.1 to 13.0, 1 RCT, n=72), large for gestational age babies (90th centile or above) (RR 1.02, 95% CI 0.74 to 1.40, 3 RCTs, n=289) or extremely large for gestational age babies (97.7th centile or above) (RR 0.5, 95% CI 0.2 to 1.3, 1 RCT, n=72). The evidence for these findings was of very low to low quality.

Two RCTs reported no differences in the risk of NICU transferral when comparing women using continuous or intermittent monitoring (combined results: RR 0.9, 95% CI 0.5 to 1.6, n=145). However, 1 RCT in women with gestational diabetes reported that of the babies that were transferred, the babies of women using continuous monitoring spent a shorter time in the NICU than the babies of women using intermittent monitoring (RR 0.8, 95% CI −1.6 to −0.1, 2 RCTs, n=73). The evidence for these findings was of very low to low quality.

5.5.7. Health economics profile

A review of the literature did not find any health economic evidence on continuous glucose monitoring in pregnant women with diabetes compared with intermittent capillary blood glucose monitoring.

This was originally prioritised as a question for health economic analysis. However, the clinical review did not find sufficient effectiveness evidence that would support the routine use of continuous glucose monitoring in diabetic pregnancy. Therefore, in the absence of clinical evidence of effectiveness, a formal economic analysis was not needed to demonstrate that continuous glucose monitoring in diabetic pregnancy could not currently be justified on cost effectiveness grounds.

5.5.8. Evidence to recommendations

5.5.8.1. Relative value placed on the outcomes considered

The guideline development group considered that the primary purpose of monitoring was to prevent morbidity or mortality associated with hypo or hyper glycaemia. Hence they prioritised maternal glycaemic control as the principal outcome for the comparison of these different monitoring techniques, but noted that incidence of hypoglycaemic episodes was an important aspect of this. Maternal satisfaction was also an essential consideration because the group considered that uptake of continuous glucose monitoring (CGM) might be determined by the individual woman's perception of the risk benefit of an invasive intervention.

The group felt that the likelihood of a large for gestational age baby in women with diabetes might be determined early in the second trimester (especially in women with gestational diabetes) and that the technique used for monitoring may not have an effect on this outcome because it was introduced too late. NICU length of stay was chosen because it is a useful outcome to integrate into health economics, but the group acknowledged that maternal glucose control during labour and not just the control during the rest of pregnancy can also affect the length of stay in NICU.

5.5.8.2. Consideration of clinical benefits and harms

Continuous glucose monitoring requires the insertion of a sensor under the skin that records interstitial fluid glucose levels at frequent times (for example every 5 minutes) throughout the day. An alarm sounds to alert the user when blood glucose levels fall below or exceed agreed thresholds. The guideline development group believed that although this technology may be helpful to women who are more likely to experience hypoglycaemic episodes (such as those who experience wide variability in their glucose regulation or who may have hypoglycaemia unawareness), it could also provoke anxiety for some women who may feel pressure to manipulate their treatment regimen too frequently in order to achieve overly tight regulation.

The continuous glucose monitoring device requires an initial calibration based on a steady state of glucose control recorded over a 20 minute period. This can be difficult for some women (for example with type 1 diabetes) to achieve. Thereafter measurements are made at regular intervals throughout the day. These readings are downloaded and the trace discussed with the woman at clinic. This provides the opportunity to reflect on dietary behaviour or glycaemic management and to modify if necessary.

Intermittent glucose monitoring requires a capillary blood sample several times during the day and use of a meter to measure blood glucose. This is a disruptive undertaking, requiring commitment to purpose, and that can also provoke anxiety. It is, however, the standard method of glucose monitoring for all with type 1 diabetes and those with type 2 diabetes on insulin therapy outside of pregnancy. The recommended frequency of testing is, however, greater during pregnancy.

The guideline development group therefore considered both monitoring techniques to be invasive and disruptive and that women's individual perceptions of the probability of negative outcomes would be important in their commitment to and satisfaction with the technique used. However, the group also noted that if an adverse outcome occurs despite a woman's adherence to the monitoring strategy and subsequent modification of her glycaemic control, this might have a negative impact on her experience of pregnancy and adversely affect her engagement in any future pregnancy.

5.5.8.3. Consideration of health benefits and resource uses

The guideline development group noted that continuous glucose monitoring is not universally available and that expertise in and funding of its usage is currently limited to specialist centres.

5.5.8.4. Quality of evidence

There was no strong evidence of benefit or harm of continuous monitoring over intermittent monitoring. But the quality of the evidence was not good.

The evidence ranged from very low to moderate in quality. None of the studies included more than 154 women. Two of the studies were within-participant comparisons. The guideline development group remarked that the included studies followed different research and intervention protocols, that there was variable adherence to these and that no studies examined the use of continuous glucose monitoring throughout pregnancy. Only 1 study reflected current clinical management by analysing continuous glucose monitoring data in real time. The remaining 4 studies analysed the data retrospectively and were downgraded because of this limitation. The group noted further limitations in the data, including the quality of glucose control in the intermittent glucose monitoring groups and a paucity of information regarding other aspects of pregnancy care, for example in labour.

There was only 1 study that included women with gestational diabetes and in this study, plus 1 further study of women with type 1 diabetes, monitoring was primarily used as a decision tool for treatment initiation or adjustment.

The group considered that glucose variability, not surprisingly, was less marked in gestational diabetes, compared to type 1 or 2 diabetes. Hence the outcomes were considered individually by study population and also in meta analyses across mixed populations where available.

Two studies of women with type 1 and type 2 diabetes reported results for glycaemic control using HbA1c, although there was only 1 isolated statistically significant finding demonstrating that HbA1c was lower when continuous glucose monitoring was used compared with intermittent glucose monitoring, and then only late in the third trimester.

The guideline development group considered maternal satisfaction outcomes which were not presented quantitatively but for which some information was available in all 5 studies. There were no adverse events associated with use of the CGM systems (such infection at sensor insertion site) and 2 studies reported that it was generally well tolerated. However, the group noted that in 1 study only two-thirds of participants in the continuous glucose monitoring group complied with protocols (due to persistent alarms) but agreed that this might be less of a problem with newer technology. There was no corresponding maternal satisfaction information regarding intermittent capillary glucose monitoring.

There were no data pertaining to admission to NICU and only very limited data on NICU length of stay from 1 study of gestational diabetes, with few details provided on intrapartum care, criteria for admission to NICU or definition of level of care.

There were no significant differences in mode of birth or preterm birth rates in trials including women with gestational diabetes or type 1 and 2 diabetes, or for the neonatal outcomes of mortality or large for gestational age.

5.5.8.5. Other considerations

The guideline development group considered that factors such as ethnicity and social status might affect the uptake and acceptability of continuous glucose monitoring. However, the reported data did not include this information.

The group noted that continuous glucose monitoring was an evolving technology and that 4 of the studies used technology that would be considered outdated compared with current practice.

Finally, it was stressed that although intermittent monitoring could be undertaken in a wide variety of health settings, continuous glucose monitoring was only available in relatively few centres. Because of this, the group felt that women using CGM should have access to help and advice from a member of the diabetes care team who had the appropriate experience and expertise in its use.

5.5.9. Key conclusions

The guideline development group concluded that given the available evidence it is not possible to recommend CGM routinely in diabetic pregnancy.

CGM might be useful in specific circumstances, as demonstrated in studies in non-pregnant populations (Pickup et al., 2011, Choudhary et al., 2013). These would include women with hypoglycemia unawareness or unstable glycaemia, or to gain a better understanding of an individual patient's glycaemic control.

5.5.10. Recommendations

80.

Do not offer continuous glucose monitoring routinely to pregnant women with diabetes. [new 2015]

81.

Consider continuous glucose monitoring for pregnant women on insulin therapy:

  • who have problematic severe hypoglycaemia (with or without impaired awareness of hypoglycaemia) or
  • who have unstable blood glucose levels (to minimise variability) or
  • to gain information about variability in blood glucose levels. [new 2015]
82.

Ensure that support is available for pregnant women who are using continuous glucose monitoring from a member of the joint diabetes and antenatal care team with expertise in its use. [new 2015]

5.5.11. Research recommendations

31. What is the role of continuous glucose monitoring in women with type 1 and 2 diabetes in preparation for pregnancy?

Why this is important

Continuous glucose monitoring (CGM) is a technology for measuring subcutaneous interstitial fluid glucose concentrations every few minutes. It is often used in conjunction with continuous subcutaneous insulin infusion (CSII) using an insulin pump. In combination, these technologies have been shown to improve glycaemic control by reducing glucose variability and the number and severity of hypoglycaemic episodes. However there have been few systematic studies using the latest real-time devices in women who are planning a pregnancy or who are already pregnant. There is a nearly four-fold increase in major congenital malformations in babies born to mothers who have diabetes and this risk is strongly correlated to glycaemic control before and at the time of conception. If CGM (± CSII) can be shown to improve blood glucose control at this crucial time there is a real prospect of reducing one of the most feared and serious complications of diabetic pregnancy. An RCT of CGM versus conventional intermittent capillary blood glucose monitoring in women planning a pregnancy is urgently required. A pilot study looking at intermediate outcomes such as glycaemia could be followed with a much larger study exploring malformation rates if CGM was shown to be effective in improving blood glucose control.

32. How should continuous glucose monitoring be used in women during pregnancy with type 1 and 2 diabetes who have recurrent severe hypoglycaemia or hypoglycaemia unawareness?

Why this important

Continuous glucose monitoring (CGM) is a technology for measuring subcutaneous interstitial fluid glucose concentrations every few minutes. It is often used in conjunction with continuous subcutaneous insulin infusion (CSII) using an insulin pump. In combination, these technologies have been shown to improve glycaemic control by reducing glucose variability and the number and severity of hypoglycaemic episodes. However there have been few systematic studies using the latest real-time devices in women who are planning a pregnancy or who are already pregnant. There is a nearly four-fold increase in major congenital malformations in babies born to mothers who have diabetes and this risk is strongly correlated to glycaemic control before and at the time of conception. Because of this, strict targets for glycaemic control have been recommended before and during pregnancy with a concomitant increased risk of severe hypoglycaemia. This is a particular problem for women with long standing type 1 diabetes who often have blunted or absent warning signs of low blood glucose. These individuals are at increased risk of sudden death due to unrecognised hypoglycaemia. If CGM (± CSII) can be shown to improve glycaemia without increasing the rate of hypoglycaemia at this crucial time there is a real prospect of enabling women to achieve their glycaemic target without increasing serious hypoglycaemia rates. An RCT of CGM versus conventional intermittent capillary blood glucose monitoring in women with hypoglycaemia unawareness and who are planning a pregnancy is urgently required.

33. Is continuous glucose monitoring acceptable to women to manage diabetes in pregnancy compared to conventional care?

Why this is important

For the main non-pregnant patient population, the question of the effectiveness of CGM has been addressed in several studies. However, this question remains unanswered in sub-groups such as those receiving pre-pregnancy care and women at high risk of hypoglycaemia. Further studies have demonstrated that CGM is associated with improved maternal glycaemic control and reduced risk of macrosomia. In addition these favourable effects on pregnancy complications potentially offer longer term health benefits for the infants. Though continuous glucose monitoring appeared to be tolerated well in general compliance is critical. Qualitative studies addressing what contributes to improved compliance and acceptance of CGM in pregnancy may provide a platform for improved care and outcomes of pregnancy.

5.6. Retinal assessment during pregnancy

There are two widely used classifications for grading the levels of diabetic retinopathy. The English classification for sight-threatening diabetic retinopathy is used for two-field diabetic retinopathy screening (see the National Screening Committee's diabetic retinopathy screening programme for England and Wales, available at www.retinalscreening.nhs.uk/). The English classification progresses from no diabetic retinopathy, to background diabetic retinopathy, to pre-proliferative diabetic retinopathy, to proliferative diabetic retinopathy (PDR). The Early Treatment Diabetic Retinopathy Study (ETDRS) classification has provided an evidence base for progression of diabetic retinopathy based on grading of seven-field stereo-photographs. The ETDRS classification progresses from no diabetic retinopathy, to mild non-proliferative diabetic retinopathy (NPDR), to moderate NPDR, to moderately severe NPDR, to severe NPDR, to PDR. The lesions found within the two classifications follow a common language: background diabetic retinopathy is equivalent to mild NPDR and is characterised by increased vascular permeability; pre-proliferative diabetic retinopathy is equivalent to moderate NPDR, moderately severe NPDR and severe NPDR; in PDR the development of new blood vessels can significantly reduce vision. The progression from no diabetic retinopathy to PDR normally occurs over a period of years, but sudden worsening may occur in pregnancy.

Duration of diabetes is known to be an important factor in the progression of diabetic retinopathy and in the development of PDR in people with diabetes. Data from an epidemiological study showed that PDR varied from 1.2% to 67% in people who had had diabetes for less than 10 years and more than 35 years, respectively.231

5.6.1. Description of the evidence

A cohort study232 determined the prevalence of retinopathy characteristically seen in people with diabetes and IGT, and in people with new-onset diabetes of known duration in the Diabetes Prevention Program (DPP) cohort. The DPP recruited and followed people with elevated fasting glucose (5.3–6.9 mmol/litre) and IGT with no history of diabetes other than gestational diabetes that did not persist after pregnancy. A random sample of 302 participants who developed diabetes and those who remained free from diabetes after 3 years follow-up was used for the retinopathy study. Retinopathy consistent with diabetic retinopathy was detected in 12.6% of people with diabetes and 7.9% of people without diabetes (P = 0.03). The study suggests that retinopathy is present in people with elevated fasting glucose and IGT with no known history of diabetes. [EL = 2+]

5.6.1.1. Progression of diabetic retinopathy

Three cohort studies were identified that found pregnancy to be independently associated with progression of diabetic retinopathy.

One study compared 180 women who became pregnant during an RCT of intensive versus conventional treatment of type 1 diabetes with women who did not become pregnant.233 In the intensive treatment group 693/2950 (23%) non-pregnant women had progression of retinopathy compared with 39/124 (31%) pregnant women (OR 1.62, 95% CI 1.01 to 2.59, P < 0.05). In the conventional treatment group 1742/5605 (31%) non-pregnant women had progression of retinopathy compared with 37/73 (50.3%) pregnant women (OR 2.54, 95% CI 1.59 to 4.03, P < 0.0001). [EL = 2++]

A cohort study compared 60 pregnant women with type 1 diabetes to 80 non-pregnant women with type 1 diabetes.234 Progression of retinopathy occurred in 10/35 women with pre-existing retinopathy in the pregnant group. There was no progression of retinopathy in the non-pregnant controls (24 had pre-existing retinopathy). [EL = 2+]

A cohort study compared 171 pregnant women with type 1 diabetes to 298 non-pregnant women with type 1 diabetes.235 A multivariate analysis (pregnancy, HbA1c, blood pressure, number of previous pregnancies and duration of diabetes) found pregnancy to be independently associated with progression of retinopathy (OR 1.8, 95% CI 1.1 to 2.8, P < 0.02). [EL = 2++]

5.6.1.2. Severity of retinopathy at conception

Four studies found progression of retinopathy during pregnancy to be associated with severity of retinopathy at conception.

A cohort study of 155 pregnant women with type 1 diabetes found women with more severe retinopathy at conception were more likely to show progression during pregnancy (χ2 for trend, P < 0.001).236 The study found progression of two steps or more in 4/39 (10.3%) women with no retinopathy, 8/38 (21.1%) women with microaneurysms only, 5/32 (18.8%) women with mild NPDR and 17/31 (54.8%) women with moderate NPDR. Women with no retinopathy or only microaneurysms at conception did not develop PDR. PDR developed in 2/32 (6%) women with mild NPDR and 9/31 (29%) women with moderate NPDR at conception. [EL = 2++]

A cohort study of 35 women with type 1 diabetes found progression during pregnancy in 3/10 women with no retinopathy at baseline and in 3/20 with background retinopathy (2/20 developed proliferative retinopathy).237 Diabetic retinopathy deteriorated during pregnancy in all five women with proliferative retinopathy at baseline. [EL = 2+]

A cohort study evaluated 65 women with type 1 diabetes before pregnancy, during each trimester and 12 months postpartum.238 The study found 38 women had no retinopathy at conception, 28 (74%) showed no progression and ten (26%) progressed to mild NPDR. Twenty-two women had NPDR at conception, 5 (22.5%) showed no progression, 12 (55%) had NPDR progression and 5 (22.5%) progressed to PDR necessitating photocoagulation. The difference in progression of retinopathy between these two groups was statistically significant (P = 0.0001). [EL = 2+]

A cohort study followed 154 women with type 1 diabetes.239 Twenty-three percent (18/78) of women with no retinopathy in the first trimester progressed; 28/68 (41%) women with NPDR in the first trimester progressed; and 5/8 (63%) women with PDR in the first trimester progressed (P = 0.01). [EL = 2++]

5.6.1.3. Duration of diabetes

Six cohort studies of women with type 1 diabetes found progression of retinopathy during pregnancy was associated with duration of diabetes.234,236,238241

The effect of duration of diabetes on progression of retinopathy during pregnancy is difficult to separate from the effect of the severity of retinopathy at conception as the two are correlated. A study of 155 pregnant women with type 1 diabetes236 found that in women with moderate or more severe retinopathy at baseline, retinopathy progressed by two steps or more in 55% of women with 15 years or less of duration of diabetes and 50% of women with more than 15 years of duration of diabetes. However, PDR developed in only 18% of women with 15 years or less of duration of diabetes compared with 39% of women with more than 15 years of duration of diabetes. This suggests that severity of retinopathy at conception is more important than duration of diabetes for the progression of diabetes during pregnancy, but that duration of diabetes may be an important factor in the development of PDR. [EL = 2++]

A recent cohort study of 179 pregnancies in 139 women with type 1 diabetes241 found progression of retinopathy was significantly increased in women with duration of diabetes of 10– 19 years compared with duration less than 10 years (8/80 versus 0/71, P = 0.007) and in women with moderate to severe NPDR at booking (6/163 versus 3/10, P = 0.01). The study included 20 pregnancies in women with duration of diabetes more than 20 years who had no or mild retinopathy at booking and of these only one progressed. This suggests that severity of retinopathy at conception may be more important than duration of diabetes in the progression of retinopathy during pregnancy. [EL = 2++]

5.6.1.4. Glycaemic control

Seven studies considered the effect of glycaemic control on the progression of diabetic retinopathy during pregnancy.233,235239,242 All studies found poor glycaemic control to be associated with progression of retinopathy during pregnancy.

5.6.1.5. Magnitude of improvement in glycaemic control

Three studies233,237,239 found that a large improvement in glycaemic control in the first trimester was associated with progression of retinopathy. The effect of large improvement of glycaemic control is difficult to separate from poor glycaemic control as women with the largest improvement were those who had poor initial control.236 [EL = 2++]

Progression of retinopathy following commencement of intensive treatment has also been observed in non-pregnant adults with diabetes.243246 In the DCCT study70,243 involving 1441 people with type 1 diabetes (726 with no retinopathy at baseline) progression of retinopathy was observed at the 6 and/or 12 month visit in 13.1% of people in the intensive group compared with 7.6% in the conventional group (P < 0.001). Among people who had experienced early worsening of retinopathy, 69% in the intensive group and 57% in the conventional group had shown complete recovery by the 18 month visit. Overall people with early worsening of retinopathy in the intensive group had a 74% reduction in the risk of subsequent progression as compared with people with early worsening who received conventional treatment (P < 0.001). [EL = 1++]

Logistic regression incorporating both initial HbA1c and the change between initial HbA1c and 4 month HbA1c found the latter to be the dominant factor for early worsening of retinopathy. There was no evidence that people with more rapid reduction of HbA1c had a greater risk of early worsening of retinopathy than people with more gradual reduction when the reductions were of similar magnitude.243 [EL = 1++]

5.6.1.6. Hypertension

A cohort study of 154 women with type 1 diabetes examined the effect of hypertension on the progression of retinopathy during pregnancy.239 Multiple regression found pregnancy-induced hypertension (P = 0.01) and chronic hypertension (P = 0.02) to be associated with progression of retinopathy. [EL = 2++]

A cohort study of 65 pregnant women with type 1 diabetes238 found systolic blood pressure to be higher in women who showed progression of retinopathy during pregnancy than in those who did not (P < 0.005). [EL = 2++]

5.6.1.7. Postpartum regression

A cohort study followed 154 women with type 1 diabetes through pregnancy to 12 weeks postpartum.239 Fifty-one women had progression of retinopathy during pregnancy of which seven developed PDR. Thirteen women experienced postpartum regression. None of the women who developed PDR during pregnancy experienced postpartum regression. [EL = 2++]

A cohort study of 65 women with type 1 diabetes were followed until 12 months postpartum.238 Thirty-eight women had no retinopathy at conception. Of these 28 showed no change during pregnancy. Ten showed mild progression during pregnancy, of which five showed complete postpartum regression. There was no development of PDR in the group with no retinopathy at conception. Twenty-two women had NPDR at conception: five of these women experienced no change during pregnancy; twelve progressed from mild to severe NPDR, of which two showed regression postpartum; five progressed to PDR. [EL = 2+]

5.6.1.8. Laser treatment for diabetic macular oedema

A large multicentre RCT247 in which people with macular oedema and mild or moderate diabetic retinopathy in one or both eyes were randomly assigned to focal argon laser photocoagulation (754 eyes) or deferred photocoagulation (1490 eyes) showed that focal photocoagulation substantially reduced the risk of visual loss (12% versus 24% at 3 year follow-up). However this RCT did not mention whether pregnant women were included. [EL = 1+] A further report248 from the RCT described treatment techniques in detail. It defined the concepts of ‘clinically significant macular oedema’ and ‘treatable lesions’. [EL = 3–4]

A subsequent RCT249 compared eyes selected for early photocoagulation in the first RCT247 by treating with one of four combinations of scatter (panretinal) and focal treatment. It was found that for eyes with macular oedema, focal photocoagulation was effective in reducing the risk of moderate visual loss but that scatter photocoagulation was not. Focal treatment also increased the chance of visual improvement, decreased the frequency of persistent macular oedema and caused only minor visual field losses. [EL = 1+]

An audit by the UK National Diabetic Retinopathy Laser Treatment group250,251 was conducted in 546 people undergoing their first photocoagulation treatment for maculopathy. At 9 month follow-up, the results showed that 9.2% had a deterioration in visual acuity equivalent to a doubling of the visual angle and 3.3% of eyes had a visual acuity less than 6/60. Improvement in the macular oedema occurred in 64.6% and exudates in 77.3%. [EL = 3–4]

5.6.1.9. Laser treatment for proliferatvie diabetic retinopathy

A study of 55 pregnant women with type 1 diabetes found that progression of retinal disease was arrested with photocoagulation during pregnancy in four women with proliferative retinopathy.240 [EL = 3]

The Diabetic Retinopathy Study group252 recommend treatment for control eyes with ‘high risk characteristics’. They reported four retinopathy factors that increase the 2 year risk of developing severe visual loss: presence of vitreous or preretinal haemorrhage; presence of new vessels; location of new vessels on or near the optic disc and severity of new vessels. [EL = 3–4]

An RCT253 that compared photocoagulation with no treatment found that photocoagulation reduced the risk of severe visual loss by 50% or more. The 2 year risk of severe visual loss without treatment outweighed the risk of harmful treatment effects for eyes with new vessels and preretinal or vitreous haemorrhage and for eyes with new vessels on or within one disc diameter of the optic disc (NVD) equalling or exceeding one-quarter to one-third of the disc area, even in the absence of preretinal or vitreous haemorrhage. [EL = 1+]

The UK National Diabetic Retinopathy Laser Treatment audit251 was conducted on 546 people undergoing their first photocoagulation treatment for maculopathy. At 9 month follow-up neovascularisation had regressed fully in 50.8% of cases with proliferative retinopathy, and there was no change or deterioration in 10.3%. This audit showed that regression of neovascularisation was associated with greater areas of retinal ablation at the initial treatment session. [EL = 3–4]

5.6.1.10. Effect of blood pressure on macular oedema and diabetic retinopathy

A study254 investigated the relationship between blood pressure and diabetic retinopathy in 249 young people with type 1 diabetes. Retinopathy was present in 63% of young people and hypertension in 2%. The presence of high-normal blood pressure (> 90th percentile but less than 141/90 mm Hg) resulted in a prospectively higher occurrence of retinopathy and of progression of pre-existing retinopathy. [EL 3–4]

A cross-sectional study255 in Norway of 600 people with a mean age of 19.8 years evaluated the association of various risk factors with retinopathy. In a multiple logistic regression model, age (P = 0.0001), higher mean HbA1c (P = 0.009 ), duration of diabetes (P = 0.0001) and mean arterial blood pressure (P = 0.0001) were significantly associated with retinopathy. [EL 2++]

A cross-sectional study256 in the USA of 634 people with type 1 diabetes diagnosed before age 30 years evaluated retinopathy after 14 years. Progression was more likely with higher HbA1c or diastolic blood pressure at baseline, an increase in the HbA1c level and an increase in diastolic blood pressure level from the baseline to the 4 year follow-up. The increased risk of proliferative retinopathy was associated with the presence of hypertension at baseline, whereas the increased risk of a person developing macular oedema was associated with the presence of gross proteinuria at baseline. [EL 2+]

An RCT257 investigated the effect of tight blood pressure control and risk of microvascular complications in people with type 2 diabetes. After 9 years follow-up the group assigned to tight blood pressure control had a 34% reduction in risk in the proportion of participants with deterioration of retinopathy by two steps (99% CI 11% to 50%, P = 0.0004) and a 47% reduced risk (99% CI 7% to 70%, P = 0.004) of deterioration in visual acuity by three lines of the ETDRS chart. [EL 1++]

A cross-sectional study258 investigated risk factors related to the incidence and progression of diabetic retinopathy from diagnosis over 6 years in 1919 people with type 2 diabetes. Development of retinopathy (incidence) was strongly associated with baseline glycaemia, glycaemic exposure over 6 years, higher blood pressure and with not smoking. In those who already had retinopathy, progression was associated with older age, male sex, hyperglycaemia (higher HbA1c) and with not smoking. [EL 2++]

A prospective RCT259 compared the effects of intensive and moderate blood pressure control on the incidence and progression of type 2 diabetic complications in 470 people. At 5.3 years follow-up no difference was found in the incidence between the intensive and moderate groups with regard to the progression of diabetic retinopathy. [EL 1+]

An RCT260 compared tight blood pressure control (blood pressure less than 150/85) with less tight blood pressure control (blood pressure less than 180/105) and its relationship with diabetic retinopathy in 1148 people with diabetes. At 4.5 years follow-up people allocated to tight blood pressure control were less likely to undergo photocoagulation (RR 0.65, P = 0.03). This difference was driven by a difference in photocoagulation due to maculopathy (RR 0.58, P = 0.02). [EL = 1+]

5.6.2. Current practice

The CEMACH enquiry reported that a detailed retinal assessment was recorded in the woman's notes at least once during pregnancy in 79.9% of women with pre-existing diabetes.2 The CEMACH case– control study reported that women with poor pregnancy outcome were as likely not to have a retinal assessment during the first trimester or at booking if later (36% [70/194]) than women who had a good pregnancy outcome (27% [49/183], OR 1.4, 95% CI 0.9 to 2.2, adjusted for maternal age and deprivation).33 Only 55% of the 258 assessments were recorded to have been done through dilated pupils and for 40% of women details about the retinal assessment procedure were not documented. The most common concern noted by the CEMACH enquiry panels over sub-optimal diabetes care in pregnancy was sub-optimal retinal function monitoring and management. [EL = 3–4]

The CEMACH enquiry (comparison of women with type 1 and type 2 diabetes) reported that women with type 1 diabetes were more likely to have retinopathy than women with type 2 diabetes (P < 0.001), with 36% (50/138) of women with type 1 diabetes and 9% (9/96) of the women with type 2 diabetes having retinopathy in pregnancy.33 This was a new finding in 26% (13/50) of the women with type 1 diabetes and 56% (5/9) of the women with type 2 diabetes. Of the women with pre-existing retinopathy there was evidence of deterioration in 18% of the women with type 1 diabetes and 11% of the women with type 2 diabetes. Women with type 1 diabetes were more likely to have a retinal assessment compared to women with type 2 diabetes (78% versus 64%, P = 0.02). Where retinopathy was found both groups of women were as likely to be referred to an ophthalmologist (35% versus 44%, P = 0.62). [EL = 3–4]

5.6.3. Existing guidance

The NSF for diabetes recommends full retinal assessment in all women with pre-existing diabetes during the first trimester (or at booking if this is later).20

5.6.4. Evidence statement

In some women pregnancy may accelerate progression of diabetic retinopathy. This is more likely in women with more severe diabetic retinopathy, poor glycaemic control and hypertension. Some diabetic retinopathy may regress spontaneously after the woman has given birth.

It is difficult to separate the influence of different factors which have been found to be associated with progression of retinopathy during pregnancy. The magnitude of improvement in glycaemic control is associated with glycaemic control prior to conception (which in turn is associated with duration of diabetes and severity of diabetes at conception). The DCCT found the magnitude, but not the rapidity, of the reduction in HbA1c during the first 6 months of intensive treatment to be an important risk factor for early worsening of diabetic retinopathy. Whether or not the risk of retinopathy progression can be reduced by more gradual reduction in glycaemic control can be resolved only by an RCT.

Evidence supports the use of laser treatment in diabetic macular oedema. Further evidence shows that control of blood pressure has a positive effect on macular oedema and progression of diabetic retinopathy.

5.6.5. From evidence to recommendations

Given the evidence of a rapid change in diabetic retinopathy during pregnancy (because of persistent hyperglycaemia) the GDG's view is that healthcare professionals should err on the side of caution by offering increased frequency of surveillance in the preconception period and throughout pregnancy to women with long-standing poor glycaemic control and pre-proliferative diabetic retinopathy or PDR, and by treating pre-existing diabetic retinopathy before conception.

There is evidence that rapid optimisation of glycaemic control can worsen diabetic retinopathy. However, it is the GDG's view that the benefits to the fetus of good glycaemic control outweigh the risks to the woman (early worsening of diabetic retinopathy). Healthcare professionals should, therefore, encourage improvement in glycaemic control in pregnancy and address ophthalmological complications of diabetes during pregnancy if they occur. In most women, laser treatment can be performed during pregnancy and will reduce the risks of sight loss as a result of progression of diabetic retinopathy. Careful control of blood pressure will also have a positive effect on sight-threatening diabetic retinopathy. Only in very rare circumstances might early birth be considered to reduce the risks of vision loss in pregnancy.

The GDG's view is that retinal assessment during pregnancy for women with diabetes should be conducted in accordance with the recommendations of the National Screening Committee's diabetic retinopathy screening programme (that is, retinal assessment should be performed using digital imaging with mydriasis (dilation of the pupils) using tropicamide). In making its recommendations, the GDG has also noted the report of the CEMACH diabetes in pregnancy audit, which highlighted that only 55% of women with pre-existing diabetes were documented to have received retinal assessment through dilated pupils.

If diabetic retinopathy is found to be present in early pregnancy, referral should usually be guided by the standard referral criteria (see the National Screening Committee's website), and women should be seen by an ophthalmologist within 4 weeks, except for PDR when urgent referral is required. As with preconception care, if there are concerns in relation to the possible worsening of diabetic retinopathy with imminent improvement of very poor blood glucose control then referral with lesser degrees of retinopathy may be considered.

The recommendations in relation to retinal assessment in the preconception period are presented in Section 3.13.

5.6.6. Recommendations

83.

Offer pregnant women with pre-existing diabetes retinal assessment by digital imaging with mydriasis using tropicamide following their first antenatal clinic appointment (unless they have had a retinal assessment in the last 3 months), and again at 28 weeks. If any diabetic retinopathy is present at booking, perform an additional retinal assessment at 16-20 weeks. [2008, amended 2015]

84.

Diabetic retinopathy should not be considered a contraindication to rapid optimisation of blood glucose control in women who present with a high HbA1c in early pregnancy. [2008]

85.

Ensure that women who have preproliferative diabetic retinopathy or any form of referable retinopathy diagnosed during pregnancy have ophthalmological follow-up for at least 6 months after the birth of the baby. [2008, amended 2015]

86.

Diabetic retinopathy should not be considered a contraindication to vaginal birth. [2008]

5.6.7. Research recommendations

34. Should retinal assessment during pregnancy be offered to women diagnosed with gestational diabetes who are suspected of having pre-existing diabetes?

Why this is important

Women with gestational diabetes may have previously unrecognised type 2 diabetes with retinopathy. At present this is not screened for because of the difficulty in identifying these women amongst the larger group who have reversible and self-limiting gestational diabetes. The benefit of recognising such women is that treatment for diabetic retinopathy is available and can prevent short and long-term deterioration of visual acuity. An observational study of retinal photography assessment in women newly diagnosed with gestational diabetes would determine whether the prevalence is high enough to justify routine screening.

5.7. Renal assessment during pregnancy

Diabetic nephropathy is a progressive disease that can be divided into the following stages:261 [EL = 4]

  • microalbuminuria (incipient nephropathy) – small amounts of albumin are excreted in the urine
  • macroalbuminuria or proteinuria (overt nephropathy) – widespread glomerular sclerosis resulting in progressively larger amounts of protein excreted in the urine
  • end-stage renal disease – decreasing creatinine clearance, increasing serum creatinine and uraemia.

5.7.1. Description of the evidence

5.7.1.1. Effect of pregnancy on progression of nephropathy

A systematic review262 considered the effects of pregnancy on diabetic nephropathy. The review included 11 longitudinal studies involving a total of 201 people. Only one study had a non- pregnant control group. The other studies compared the average rate of decline in renal function with the expected rate of decline in the general non-pregnant population of people with diabetic nephropathy. The review found that most studies suggest that pregnancy is not associated with development of nephropathy or with accelerated progression of pre-existing nephropathy, with the exception of women with moderate to advanced disease where pregnancy may accelerate progression to end-stage renal disease. [EL = 2++]

5.7.1.2. Effect of nephropathy on pregnancy outcome

A systematic review262 which included 11 studies and 681 people found that women with diabetic nephropathy were at increased risk of adverse pregnancy outcomes, in particular fetal growth restriction , chronic hypertension, pre-eclampsia and preterm birth (see Table 66). Pre-eclampsia and preterm birth were associated with incipient nephropathy (microalbuminuria) as well as overt nephropathy. [EL = 1++]

Table 66. Outcome of pregnancy in women with diabetic nephropathy.

Table 66

Outcome of pregnancy in women with diabetic nephropathy.

A cohort study was identified that had been published since the systematic review. The cohort study considered pregnancy outcome in women with type 1 diabetes and microalbuminuria.263 Of 240 consecutive pregnancies, 203 women (85%) had normal urinary albumin excretion, 26 (11%) had microalbuminuria and 11 (5%) had diabetic nephropathy. In this study normal urinary albumin excretion was defined as less than 30 mg/24 hours, microalbuminuria was defined as urinary albumin excretion 30–300 mg/24 hours and diabetic nephropathy was defined as urinary albumin excretion more than 300 mg/24 hours. The incidence of pre-eclampsia was 6% in women with normal urinary albumin excretion, 42% in women with microalbuminuria and 64% in women with nephropathy (P < 0.001). The incidence of preterm birth (before 34 weeks) was 6% in women with normal urinary albumin excretion, 23% in women with microalbuminuria and 45% in women with diabetic nephropathy (P < 0.001). The incidence of SGA babies was 2% in women with normal urinary albumin excretion, 4% in women with microalbuminuria and 45% in women with diabetic nephropathy (P < 0.001). [EL = 2++]

5.7.1.3. Antihypertensive treatment for microalbuminuria

A cohort study involving 46 women evaluated the impact of antihypertensive treatment with methyldopa in normotensive pregnant women with type 1 diabetes and microalbuminuria.264 The women were similar in terms of age, diabetes duration, pre-pregnancy BMI, HbA1c and blood pressure, and all were referred before 17 weeks of gestation. The prevalence of preterm birth before 34 weeks of gestation was reduced from 23% to 0% (P = 0.02); the prevalence of preterm birth before 37 weeks of gestation was reduced from 62% to 40% (P = 0.15); and the prevalence of pre-eclampsia was reduced from 42% to 20% (P = 0.11). Perinatal mortality occurred in 4% versus 0%. [EL = 2++]

5.7.2. Current practice

The CEMACH enquiry reported that women who had a poor pregnancy outcome were more likely not to have monitoring for nephropathy (22% [46/209]) than women who had a good pregnancy outcome (13% [26/206], OR 1.9, 95% CI 1.1 to 3.3, adjusted for maternal age and deprivation). In an additional case–control analysis lack of monitoring for nephropathy was associated only with fetal congenital anomaly and not with fetal or neonatal death after 20 weeks of gestation; it is therefore unlikely to have been causative for poor pregnancy outcome. Nephropathy itself was not associated with poor pregnancy outcome. One of the most common concerns noted by the CEMACH enquiry panels over sub-optimal diabetes care in pregnancy was sub-optimal renal function monitoring and management.33 [EL = 3–4]

The CEMACH enquiry (comparison of women with type 1 and type 2 diabetes) reported that there was no significant difference in the rate on nephropathy in pregnancy in women with type 1 or type 2 diabetes, with 8% (12/148) of women with type 1 diabetes and 5% (6/119) of the women with type 2 diabetes having nephropathy during their pregnancy.33 Women with type 1 diabetes were as likely to have monitoring for nephropathy as women with type 2 diabetes (86% versus 79%, P = 0.60). Where nephropathy was found both groups of women were as likely to have a test of renal function (75% versus 50%, P = 0.29). [EL = 3–4]

The CEMACH enquiry did not state an explicit standard of monitoring for nephropathy, but it recommended that appropriate monitoring included testing for microalbuminuria (incipient nephropathy) via protein dipstick testing of urine or serum creatinine.33 [EL = 3–4]

5.7.3. Existing guidance

The NICE guideline for type 2 diabetes defines microalbuminuria as albumin : creatinine ratio 3.5 mg/mmol or more (for women) or albumin concentration 20 mg/litre or more. Macroalbuminuria is defined as albumin : creatinine ratio 30 mg/mmol or more or albumin concentration 200 mg/litre or more.8

The NICE guidelines for type 1 and type 2 diabetes in adults recommend annual testing for nephropathy using urine albumin : creatinine ratio and serum creatinine. It is recommended that people with nephropathy have measurements of urine albumin and serum creatinine levels at each visit.7,8

The NICE guidance on “Hypertension in pregnancy” (CG 107) recommends the use of low dose aspirin in women with diabetes to reduce the risk of hypertensive disorders in pregnancy:

Advise women at high risk of pre-eclampsia to take 75 mg of aspirin* daily from 12 weeks until the birth of the baby. Women at high risk are those with any of the following:

  • hypertensive disease during a previous pregnancy
  • chronic kidney disease
  • autoimmune disease such as systemic lupus erythematosis or antiphospholipid syndrome
  • type 1 or type 2 diabetes
  • chronic hypertension, [1.1.2.1].

5.7.4. Evidence statement

In the majority of studies pregnancy has not been associated with the development of nephropathy or with accelerated progression of pre-existing nephropathy. Data from three studies suggest that in women with moderate to advanced disease pregnancy may accelerate progression to end- stage renal disease.

All stages of nephropathy, including microalbuminuria, are associated with adverse pregnancy outcomes, especially fetal growth restriction , pre-eclampsia and preterm birth.

A small cohort study suggested that antihypertensive treatment with methyldopa in women with type 1 diabetes and microalbuminuria reduced the risk of preterm birth (before 34 weeks of gestation).

No evidence was identified in relation to thromboprophylaxis in the presence of macro-albuminuria.

5.7.5. From evidence to recommendations

NICE recommends that renal assessment outside pregnancy should use urine albumin : creatinine ratio and serum creatinine. Estimated glomerular filtration rate (eGFR) should not be used during pregnancy as it underestimates the glomerular filtration rate.439 There is no evidence on the optimal assessment schedule during pregnancy. As both microalbuminuria and macroalbuminuria are associated with adverse outcomes the GDG recommends assessment in the preconception period or at the first presentation after conception. All pregnant women should have their urine tested for proteinuria as part of routine antenatal care (see the NICE antenatal care guideline).9 If serum creatinine is abnormal (120 micromol/litre or more) or if total protein excretion exceeds 2 g/day, referral to a nephrologist should be considered.

No evidence was identified in relation to thromboprophylaxis in the presence of macroalbuminuria. The GDG's consensus view is that healthcare professionals should follow best current practice in terms of thromboprophylaxis for women with diabetes and macroalbuminuria (antenatal administration of aspirin for proteinuria less than 5 mg/day and heparin for proteinuria more than 5 mg/day; planned early birth may need to be considered because of the risk of developing pre-eclampsia).

The recommendations in relation to renal assessment in the preconception period are presented in Section 3.14.

Note added for 2015 guideline

The guideline development group felt that the diagnostic criteria for severe pre-existing renal disease used in the recommendations should conform to those recommended in the NICE clinical guideline on chronic renal disease. A cohort study was identified that had been published since the systematic review. The cohort study considered pregnancy outcomes in women with type 1 diabetes and microalbuminuria.263 Of 240 consecutive pregnancies, 203 women (85%) had normal urinary albumin excretion, 26 (11%) had microalbuminuria and 11 (5%) had diabetic nephropathy. In this study normal urinary albumin excretion was defined as less than 30 mg per 24 hours, microalbuminuria was defined as urinary albumin excretion 30–300 mg per 24 hours and diabetic nephropathy was defined as urinary albumin excretion more than 300 mg per 24 hours. The incidence of pre-eclampsia was 6% in women with normal urinary albumin excretion, 42% in women with microalbuminuria and 64% in women with nephropathy (p< p0.001). The incidence of preterm birth (before 34 weeks) was 6% in women with normal urinary albumin excretion, 23% in women with microalbuminuria and 45% in women with diabetic nephropathy (p<0.001). The incidence of SGA babies was 2% in women with normal urinary albumin excretion, 4% in women with microalbuminuria and 45% in women with diabetic nephropathy (p<0.001). [EL = 2++]

5.7.5.1. Antihypertensive treatment for microalbuminuria

A cohort study involving 46 women evaluated the impact of antihypertensive treatment with methyldopa in normotensive pregnant women with type 1 diabetes and microalbuminuria.264 The women were similar in terms of age, diabetes duration, pre-pregnancy BMI, HbA1c and blood pressure, and all were referred before 17 weeks of gestation. The prevalence of preterm birth before 34 weeks of gestation was reduced from 23% to 0% (P = 0.02); the prevalence of preterm birth before 37 weeks of gestation was reduced from 62% to 40% (P = 0.15); and the prevalence of pre-eclampsia was reduced from 42% to 20% (P = 0.11). Perinatal mortality occurred in 4% versus 0%. [EL = 2++]

5.7.6. Current practice

The CEMACH enquiry reported that women who had a poor pregnancy outcome were more likely not to have monitoring for nephropathy (22% [46/209]) than women who had a good pregnancy outcome (13% [26/206], OR 1.9, 95% CI 1.1 to 3.3, adjusted for maternal age and deprivation). In an additional case–control analysis lack of monitoring for nephropathy was associated only with fetal congenital anomaly and not with fetal or neonatal death after 20 weeks of gestation; it is therefore unlikely to have been causative for poor pregnancy outcome. Nephropathy itself was not associated with poor pregnancy outcome. One of the most common concerns noted by the CEMACH enquiry panels over sub-optimal diabetes care in pregnancy was sub-optimal renal function monitoring and management.33 [EL = 3–4]

The CEMACH enquiry (comparison of women with type 1 and type 2 diabetes) reported that there was no significant difference in the rate on nephropathy in pregnancy in women with type 1 or type 2 diabetes, with 8% (12/148) of women with type 1 diabetes and 5% (6/119) of the women with type 2 diabetes having nephropathy during their pregnancy.33 Women with type 1 diabetes were as likely to have monitoring for nephropathy as women with type 2 diabetes (86% versus 79%, P = 0.60). Where nephropathy was found both groups of women were as likely to have a test of renal function (75% versus 50%, P = 0.29). [EL = 3–4]

The CEMACH enquiry did not state an explicit standard of monitoring for nephropathy, but it recommended that appropriate monitoring included testing for microalbuminuria (incipient nephropathy) via protein dipstick testing of urine or serum creatinine.33 [EL = 3–4]

5.7.7. Existing guidance

The NICE guideline for type 2 diabetes defines microalbuminuria as albumin : creatinine ratio 3.5 mg/mmol or more (for women) or albumin concentration 20 mg/litre or more. Macroalbuminuria is defined as albumin : creatinine ratio 30 mg/mmol or more or albumin concentration 200 mg/litre or more.8

The NICE guidelines for type 1 and type 2 diabetes in adults recommend annual testing for nephropathy using urine albumin : creatinine ratio and serum creatinine. It is recommended that people with nephropathy have measurements of urine albumin and serum creatinine levels at each visit.7,8

5.7.8. Evidence statement

In the majority of studies pregnancy has not been associated with the development of nephropathy or with accelerated progression of pre-existing nephropathy. Data from three studies suggest that in women with moderate to advanced disease pregnancy may accelerate progression to end- stage renal disease.

All stages of nephropathy, including microalbuminuria, are associated with adverse pregnancy outcomes, especially fetal growth restriction , pre-eclampsia and preterm birth.

A small cohort study suggested that antihypertensive treatment with methyldopa in women with type 1 diabetes and microalbuminuria reduced the risk of preterm birth (before 34 weeks of gestation).

No evidence was identified in relation to thromboprophylaxis in the presence of macro-albuminuria.

5.7.9. From evidence to recommendations

NICE recommends that renal assessment outside pregnancy should use urine albumin: creatinine ratio and serum creatinine. Estimated glomerular filtration rate (eGFR) should not be used duringpregnancy as it underestimates the glomerular filtration rate.439 There is no evidence on the optimal assessment schedule during pregnancy. As both microalbuminuria and macroalbuminuria are associated with adverse outcomes the GDG recommends assessment in the preconception period or at the first presentation after conception. All pregnant women should have their urine tested for proteinuria as part of routine antenatal care (see the NICE antenatal care guideline).9 If serum creatinine is abnormal (120 micromol/litre or more) or if total protein excretion exceeds 2 g/day, referral to a nephrologist should be considered.

No evidence was identified in relation to thromboprophylaxis in the presence of macroalbuminuria. The GDG's consensus view is that healthcare professionals should follow best current practice in terms of thromboprophylaxis for women with diabetes and macroalbuminuria (antenatal administration of aspirin for proteinuria less than 5 mg/day and heparin for proteinuria more than 5 mg/day; planned early birth may need to be considered because of the risk of developing pre-eclampsia).

The recommendations in relation to renal assessment in the preconception period are presented in Section 3.14.

NOTE ADDED FOR 2015 GUIDELINE: The GDG felt that the diagnostic criteria for severe pre-existing renal disease used in the recommendations should conform to those recommended in Chronic Renal Disease (NICE Clinical Guideline 182)

5.7.10. Recommendations

87.

If renal assessment has not been undertaken in the preceding 3 months in women with pre-existing diabetes, arrange it at the first contact in pregnancy. If the serum creatinine is abnormal (120 micromol/litre or more), the urinary albumin:creatinine ratio is greater than 30 mg/mmol or total protein excretion exceeds 2 g/day, referral to a nephrologist should be considered (eGFR should not be used during pregnancy). Thromboprophylaxis should be considered for women with proteinuria above 5 g/day (macroalbuminuria). [2008, amdended 2015]

88.

For guidance on using antiplatelet agents to reduce the risk of pre-eclampsia in pregnant women with diabetes, see recommendation 1.1.2.1 in the NICE guideline on Hypertension in pregnancy. [new 2015]

5.7.11. Research recommendations

35. Does identification of microalbuminuria during pregnancy offer the opportunity for appropriate pharmacological treatment to prevent progression to pre-eclampsia in women with pre-existing diabetes?

Why this is important

Microalbuminuria testing is available, but it is not performed routinely in antenatal clinics for women with pre-existing diabetes because a place for prophylactic treatment of pre-eclampsia in microalbuminuria-positive women has not been investigated. The benefit of clinically and cost-effective prophylactic treatment would be to significantly improve pregnancy outcomes in this group of women.

5.8. Screening for congenital malformations

5.8.1. Description of the evidence

Women with diabetes have an increased risk of having a baby with a congenital malformation. Major congenital malformations affecting babies of women with diabetes include cardiac, neural tube and genitourinary anomalies. Table 67 lists anomalies associated with diabetes as well as the estimated prevalence and RR compared to women without diabetes, as reported in published studies.265

Table 67. Detectable major congenital malformations in babies of women with pre-existing diabetes.

Table 67

Detectable major congenital malformations in babies of women with pre-existing diabetes.

More recent data from the CEMACH enquiry found the prevalence of confirmed major anomalies to be 41.8 per 1000 total births (live and stillborn).2 Separate rates for babies of women with type 1 diabetes (n = 1707) and type 2 diabetes (n = 652) born between 1 March 2002 and 28 February 2003 are summarised in Table 68.266 Women with type 2 diabetes were more likely to come from a Black, Asian or Other Minority Ethnic group (type 1 diabetes 9.1%, type 2 diabetes 48.8%). Perinatal mortality in babies of women with diabetes was 31.8 per 1000 births, nearly four times higher than the general maternity population. One hundred and ninety-seven major congenital anomalies were confirmed in 148 babies. The prevalence of major congenital anomaly was 46 per 1000 births in women with diabetes (48 per 1000 births for type 1 diabetes, 43 per 1000 births for type 2 diabetes), more than twice the expected rate. The increase was mainly due to an increase in neural tube defects (4.2-fold) and congenital heart disease (3.4-fold). Anomalies in 65% (71/109) of babies were diagnosed antenatally. Congenital heart disease was diagnosed antenatally in 54.8% (23/42) of babies. Anomalies other than congenital heart disease were diagnosed antenatally in 71.6% (48/67) of babies. [EL = 3–4]

Table 68. Observed and expected prevalence of congenital malformations in babies of women with type 1 and type 2 diabetes (from CEMACH).

Table 68

Observed and expected prevalence of congenital malformations in babies of women with type 1 and type 2 diabetes (from CEMACH).

The benefits of screening for congenital malformations include the opportunity for counselling, enabling families time to prepare, allowing antenatal treatment, and ensuring appropriate obstetric management.

According to the NICE antenatal care guideline,9 all pregnant women should be offered screening for congenital malformations at 18–20 weeks of gestation as part of routine antenatal care. This section considers what additional screening should be offered to women with diabetes.

5.8.2. First-trimester screening for chromosomal anomalies

The NICE antenatal care guideline9 recommends that all pregnant women should be offered screening for Down's syndrome. Women should understand that it is their choice to embark on screening for Down's syndrome. Screening should be performed by the end of the first trimester (14 weeks of gestation), but provision should be made to allow later screening (up to 20 weeks of gestation) for women booking later in pregnancy. The screening test offered should be the ‘combined test’ (nuchal translucency (NT), beta human chorionic gonadotropin [β-hCG] and pregnancy-associated plasma protein-A (PAPP-A)) at 11–14 weeks of gestation. At 15–20 weeks of gestation the most clinically effective and cost-effective serum screening test should be offered, namely the ‘triple test’ or ‘quadruple test’ (hCG, alpha fetoprotein (AFP), unconjugated estriol (uE3) and inhibin A). The integrated test should not be routinely used as a screening test for Down's syndrome. Information about screening options for Down's syndrome that can be understood by all women, including those whose first language is not English, should be given to women as early as possible and ideally before the booking visit, allowing the opportunity for further discussion before embarking on screening. If a woman receives a screen-positive result, she should have rapid access to appropriate counselling by trained staff. The second-trimester ultrasound scan (at 18–20 weeks of) should not be routinely used for Down's syndrome screening using soft markers. The presence of an isolated soft marker with an exception of increased nuchal fold noted on the routine anomaly scan (at 18–20 weeks of gestation) should not be used to adjust the a priori risk for Down's syndrome. The presence of an increased nuchal fold or two or more soft markers should prompt the offer of fetal medicine referral.

Women with diabetes do not have an increased risk of chromosomal anomalies, however published studies have shown that some biochemical markers tend to be lower in women with type 1 diabetes than in women without diabetes. Therefore, clinical practice has been to make adjustments when calculating the risk of anomalies for women with type 1 diabetes to take account of these differences.

A meta-analysis268 included published studies of differences in AFP (14 studies, 253 women with type 1 diabetes), uE3 (six studies, 687 women with type 1 diabetes), total hCG (nine studies, 1350 women with type 1 diabetes), free β-hCG (one study, 251 women) and inhibin (three studies, 445 women). The weight-corrected median multiple-of-median (MoM) was 0.92 for AFP, 0.94 for uE3, 0.96 for total hCG, 0.96 for free β-hCG and 1.03 for inhibin (MoM values are ratios of median MoM in women with type 1 diabetes to median MoM in women without diabetes). No CIs or tests of statistical significance were presented in the meta-analysis. [EL = 2+]

Since publication of the meta-analysis, one study has been published on free β-hCG.269 The study compared 79 women with type 1 diabetes to 16366 women without diabetes. There were no significant differences in weight-corrected free β-hCG (type 1 diabetes MoM 0.87, 95% CI 0.75 to 1.16, women without diabetes MoM 1.00, P = 0.52). [EL = 2+]

Two studies were identified that compared levels of PAPP-A in women with and without diabetes during pregnancy. One study compared 79 women with type 1 diabetes to 93 pregnant women without diabetes.270 Levels of PAPP-A were significantly lower in women with type 1 diabetes (P = 0.024). [EL = 2+]

The second study compared PAPP-A levels in 79 women with type 1 diabetes to those in 16 366 women without diabetes.269 There was no significant difference in PAPP-A levels (type 1 diabetes MoM 1.02, 95% CI 0.83 to 1.05, women without diabetes MoM 1.01, P = 0.36). [EL = 2+]

One study was identified that compared NT results in 195 women with type 1 diabetes to those in 33 301 women without diabetes.269 There was no difference in mean NT between the two groups (0.0358 mm versus 0.0002 mm, P = 0.418). [EL = 2++]

5.8.3. Second-trimester ultrasound screening for structural anomalies

The NICE antenatal care guideline9 recommends that ‘ultrasound screening for fetal anomalies should be routinely offered, normally between 18 weeks 0 days and 20 weeks 6 days.’

A cohort study compared 130 women with diabetes (85 type 1 diabetes, 45 type 2 diabetes) with 12 169 low-risk pregnant women for the same period.271 All women had routine ultrasound at 16–24 weeks of gestation. A total of ten major anomalies (7.7%) and three minor anomalies (2.3%) were present in the fetuses of women with diabetes. The incidence of major congenital malformations was greater in the women with diabetes than in the low-risk control group (8% versus 1.4%, P < 0.001). The detection rate was significantly lower in the women with diabetes (30% versus 73%, P < 0.01) and the mean BMI was significantly higher (29 kg/m2 versus 23 kg/ m2). Thirty-seven percent (48/130) of scans undertaken in women with diabetes were judged to be unsatisfactory, mainly because of maternal obesity (45/48). The majority (86% [19/22]) of repeat scans were also unsatisfactory. Of the 82 women with diabetes who had satisfactory images, two had congenital malformations. Both were detected antenatally (detection rate 100%). Of the 48 whose image quality was judged to be unsatisfactory there were eight major congenital anomalies. Only one was detected antenatally (detection rate 12.5%). [EL = 2++]

A cohort study considered 432 women with type 1 diabetes who underwent ultrasound screening between 12 and 23 weeks of gestation.272 The ultrasound included four chambers of the heart and the great vessels. At birth 32 babies had 38 major congenital malformations, 52% (18/32) of which were detected antenatally. There were eight heart anomalies of which five were detected antenatally. All six CNS abnormalities were detected antenatally. The lesions most commonly missed by sonography were ventricular septal defect, abnormal hand or foot, unilateral renal abnormality, and cleft palate without cleft lip. The test performance was: sensitivity 56%, specificity 99.5%, PPV 90%, NPV 97%. [EL = 2++]

In a study of 289 women with diabetes273 comprehensive ultrasound including a four chamber view undertaken at 18 weeks of gestation by a perinatologist had a test performance for detection of non-cardiac anomalies as follows: sensitivity 59%, specificity 100%, PPV 100%, NPV 98%. The test performance of the standard four chamber view was: sensitivity 33%, specificity 100%, PPV 100%, NPV 97%. In comparison the test performance for echocardiogram was: sensitivity 92%, specificity 99%, PPV 92%, NPV 99%. [EL = 2++]

A cohort study reported on 250 women274 with pre-existing diabetes who underwent fetal echocardiogram at 20–22 weeks of gestation. Views included the four chamber view, the left ventricular long-axis view with visualisation of the aortic outflow tract, the short-axis view with visualisation of the pulmonary outflow tract and ductus arteriosus, and longitudinal view of the aortic arch. All examinations were undertaken by three experienced ultrasonographers. There were eight cardiac anomalies (3.2%), six of which were detected antenatally by echocardiogram. There was one false-negative result and one false-positive result. One fetus had an apparently normal heart at 21 weeks of gestation but was found to have a small atrial-septal defect at birth. The false positive was a case of perimembranous ventricular-septal defect. The test sensitivity was 85.7% and specificity was 99.5%. [EL = 2++]

A study of 223 women with insulin-requiring diabetes (128 type 1 diabetes, 47 type 2 diabetes, 48 gestational diabetes) considered the utility of different echocardiogram views.275 There were 11 heart defects, nine of which were detected antenatally. The two missed cases were in women who were obese. Seven defects occurred in women with type 1 diabetes, three in women with type 2 diabetes and one in a woman with insulin-requiring gestational diabetes. The sensitivity of the four chamber view was 73% (8/11) and specificity was 100%. The sensitivity of the four chamber view and aortic outflow tract was 82% (9/11) and the specificity was 100%. Other views did not contribute to detection of a defect. The two missed cases (pulmonary atresia with a ventricular septal defect and an isolated ventricular septal defect) could theoretically have been detected on the four chamber view. [EL = 2++]

One study276 examined 725 women (with or without diabetes) who had been referred for echocardiogram following a comprehensive anatomy ultrasound that included a four chamber/ left ventricular outflow tract view. The indications for referral included pre-existing diabetes (without additional indication, n = 226), fetal anomaly seen on anatomy ultrasound (n = 130) and family history of congenital heart disease (n = 133). Twenty-nine echocardiograms were reported as abnormal (4%). The indications for referral in these cases were an abnormal four chamber/left ventricular outflow tract view at ultrasound (66%), aneuploidy (14%) other fetal anomaly (17%) and fetal arrhythmia (3%). No abnormal fetal echocardiograms were reported in women with isolated pre-existing diabetes (i.e. with a normal four chamber/left ventricular outflow tract view at ultrasound). [EL = 2++]

5.8.4. Evidence statement

A number of studies have found no significant differences between women with type 1 diabetes and women without diabetes in terms of NT and weight-corrected total hCG, β-hCG and inhibin. On this basis it can be advised that no adjustment is required in these biochemical markers when calculating risks for congenital abnormalities in the fetuses of women with diabetes. A meta-analysis found weight-corrected AFP to be approximately 8% lower in women with type 1 diabetes and weight-corrected uE3 to be 6% lower in women with type 1 diabetes and therefore adjustments should be applied accordingly. Two studies have found conflicting results with regard to levels of PAPP-A. Therefore, until further evidence is available, adjustments should continue to be applied.

A number of well-designed observational studies have found more congenital anomalies are detected antenatally in women with diabetes when antenatal examination includes views of the four chambers of the fetal heart and outflow tracts. No more anomalies are detected with additional views. One study found detection rates were significantly worse in women with diabetes compared with low-risk women. This was largely attributed to obesity in women with diabetes resulting in unsatisfactory images.

5.8.5. Cost-effectiveness

The effectiveness of methods of screening for congenital cardiac malformations in women with diabetes was identified by the GDG as a priority for health economic analysis. The methods and results from the health economic modelling are summarised here; further details are provided in Appendix N.

Women with diabetes are at increased risk of having a baby with a cardiac malformation (the risk being approximately five times that of the general maternity population). Therefore, the GDG considered that this was an area where a different screening programme from that used in routine antenatal care might be justified on health economic grounds. An economic model was used to compare the cost-effectiveness of screening for congenital cardiac malformations using the four chamber plus outflow tracts view versus the four chamber view alone, which represents current practice. The baseline model suggested that the four chamber plus outflow tracts view was highly cost-effective in pregnant women with diabetes with a cost per QALY of approximately £4,000. One-way sensitivity analysis showed that four chamber plus outflow tracts view continued to be cost-effective when parameter values were varied within plausible ranges.

5.8.6. Existing guidance

As noted above, the NICE antenatal care guideline9 recommends that all pregnant women should be offered ultrasound screening for congenital malformations (ideally at 18–20 weeks of gestation) using the four chamber plus outflow tracts view as part of routine antenatal care. Women should be given information regarding the purpose and implications of the anomaly scan in order to enable them make an informed choice as to whether or not to have the scan. The guideline recommends that all pregnant women should be offered screening for Down's syndrome and that women should understand that it is their choice to embark on screening for Down's syndrome. If a woman receives a screen positive result, she should have rapid access to appropriate counselling by trained staff.

5.8.7. From evidence to recommendations

A health economic model demonstrated the cost-effectiveness of screening for congenital cardiac malformations based on the four chamber view of the fetal heart and outflow tracts relative to current practice of screening using the four chamber view alone. Data from European Surveillance of Congenital Anomalies (EUROCAT) and published literature suggest that an antenatal diagnosis of TGA may reduce neonatal mortality. These data, together with the higher prevalence of cardiac malformations in pregnant women with diabetes compared to pregnant women without diabetes underpin this result. For this reason the GDG identified screening for congenital cardiac malformations using the four chamber plus outflow tracts view as a key priority for implementation for women with diabetes (this form of screening is recommended as part of routine antenatal care but it does not form a key priority for implementation in the NICE antenatal care guideline9). There may be additional benefits of screening not taken into account in the model, the existence of which would tend to further improve the relative cost-effectiveness of screening based on the four chamber plus outflow tracts view.

The GDG's view is that a specialist cardiac scan should be offered at 22 weeks of gestation only if the results of the four chamber plus outflow tracts view are abnormal or if there is a relevant history of cardiac malformations. This is likely to bring a cost saving to the NHS because there is currently a tendency to offer a specialist cardiac scan to many women with diabetes.

2015 Update: The recommendations relating to screening for congenital malformations were edited by the GDG for the 2015 update to provide greater clarity (see section 5.11.7.5).

5.8.8. Recommendations

89.

Offer women with diabetes an ultrasound scan for detecting fetal structural abnormalities, including examination of the fetal heart (4 chambers, outflow tracts and 3 vessels), at 20 weeks. [2008, amended 2015]

5.8.9. Research recommendations

36. How reliable is first-trimester screening for Down's syndrome incorporating levels of pregnancy- associated plasma protein (PAPP-A) in women with pre-existing diabetes?

Why this is important

Several screening tests for Down's syndrome incorporate measurements of PAPP-A. However, two clinical studies have reported conflicting results in terms of whether levels of PAPP-A in women with type 1 diabetes are lower than those in other women. Current practice is to adjust PAPP-A measurements in women with diabetes on the assumption that their PAPP-A levels are indeed lower than those of other women. Further research is, therefore, needed to evaluate the diagnostic accuracy and effect on pregnancy outcomes of screening tests for Down's syndrome incorporating measurements of PAPP-A in women with pre-existing diabetes.

37. How effective is transvaginal ultrasound for the detection of congenital malformations in women with diabetes and coexisting obesity?

Why this is important

Women with diabetes are at increased risk of having a baby with congenital malformations and current recommendations advise detailed ultrasound surveillance of the fetus at 20 weeks. Obstetric ultrasound signals are attenuated by the woman's abdominal wall fat. The incidence of obesity in pregnancy is increasing, and many women with diabetes (particularly women with type 2 diabetes) are obese, and this may limit the sensitivity of abdominal ultrasound screening for congenital malformations. Vaginal ultrasound, in theory, is not affected in this way. However, there is currently no evidence that fetal anatomical surveillance undertaken at about 13 weeks is more effective than abdominal ultrasound at 20 weeks. Comparative studies are, therefore, needed to evaluate the relative diagnostic accuracy of vaginal ultrasound at 13 weeks and abdominal ultrasound 20 weeks in the same group of women with diabetes in pregnancy and coexisting obesity.

5.9. Monitoring fetal growth and wellbeing

5.9.1. Description of the evidence

5.9.1.1. Fetal growth

Women with gestational diabetes and pre-existing diabetes are at increased risk of having a baby with macrosomia (see Sections 4.5 and 5.2). Macrosomia is defined in terms of absolute birthweight (usually more than 4000 g) or birthweight percentile for gestational age (usually ≥ 90th percentile), also referred to as LGA. Macrosomia is a risk factor for shoulder dystocia, brachial plexus injury, asphyxia or prolonged labour, operative delivery and postpartum haemorrhage (see Section 6.3 and Chapter 7).

Women with diabetes are also at risk of having a baby that is SGA. The risks associated with a baby that is SGA are not as well documented as for macrosomia, but at least one study was identified that suggested that babies who were SGA (< 10th percentile for gestational age) have an increased risk of perinatal morbidity and mortality.277

There is no clear consensus for monitoring fetal size in pregnant women with diabetes.278 Clinical assessment of fetal size is by measurement of the symphysis–fundal height. Fetal size can also be measured by sonography. The two main ultrasonic methods for predicting birthweight are estimated fetal weight (EFW) and abdominal circumference of the fetus. EFW uses a combination of parameters, for example, the Hadlock formula279 uses femur length, biparietal diameter, head circumference and abdominal circumference. Mean errors in estimating fetal weight are between 8% and 15% of actual birthweight.279 EFW increases the rate of caesarean section in false positives (AGA babies incorrectly diagnosed as LGA).280,281 Accuracy of estimated fetal weight is worse in women with diabetes282 and for macrosomic babies.283

Compared with babies of women with diabetes who are AGA, LGA babies have accelerated growth of insulin-sensitive tissue such as abdominal wall fat.284 Abdominal circumference is therefore considered to be a more relevant measure of diabetes-related macrosomia and the risk of shoulder dystocia. Abdominal circumference also has the advantage of being a single measure that is accessible even when the head is engaged in the pelvis.

A systematic review of 63 studies (51 evaluating the accuracy of estimated fetal weight and 12 the accuracy of fetal abdominal circumference) involving 19 117 women pooled data to produce summary receiver operating characteristic (sROC) curves for studies with various test thresholds.278 Summary likelihood ratios (LRs) for positive and negative test results were generated for an estimated fetal weight of 4000 g and an abdominal circumference of 36 cm for predicting birthweight over 4000 g. The sROC curve area for estimated fetal weight was not different from the area for fetal abdominal circumference (0.87 versus 0.85, P = 0.91). For predicting a birthweight of over 4000 g the summary LR was 5.7 (95% CI 4.3 to 7.6) for a positive test and 0.48 (95% CI 0.38 to 0.60) for a negative test. For ultrasound fetal abdominal circumference of 36 cm the LR for a positive test for predicting birthweight over 4000 g was 6.9 (95% CI 5.2 to 9.0) and the LR for a negative test was 0.37 (95% CI 0.30 to 0.45). There was no difference in accuracy between estimated fetal weight and abdominal circumference in the prediction of macrosomia at birth. The LRs suggest that both tests are only moderately useful at best. A positive test result is more accurate for ruling in macrosomia than is a negative test for ruling it out. [EL = 1++]

A diagnostic accuracy study compared 31 published formulas for estimated fetal weight in predicting macrosomia (birthweight 4000 g or more) in babies of women with diabetes.285 One hundred and sixty-five women with pre-existing diabetes or gestational diabetes who had sonograms to estimate fetal weight after 36 weeks of gestation and within 2 weeks of birth were included in the study. Three measures of accuracy were compared: area under the ROC curve relating estimated fetal weight to macrosomia; systematic error; and absolute error. All 31 formulas for estimating fetal weight had similarly poor accuracy for prediction of macrosomia. [EL = 2+]

A cohort study evaluated the reliability of ultrasound estimation of fetal weight in 1117 women (48 with gestational diabetes) with a singleton pregnancy who had undergone ultrasound estimation of fetal weight less than 7 days before a term birth (at or later than 37 weeks of gestation).286 Both large and normal weight babies of women with diabetes tended to have their weight underestimated. Given that reliability of ultrasound estimation of fetal weight to detect larger babies was poor, the study suggests that ultrasound use in the management of suspected macrosomia should be discouraged. [EL = 2+]

A retrospective cohort study investigated the association between ultrasound fetal biometry and amniotic fluid insulin levels at birth in 93 pregnant women with pre-existing diabetes or IGT.287 Babies of women with pre-existing diabetes had significantly greater mean growth velocity (1.39, 95% CI 0.43 to 2.23 versus 0.39, 95% CI −01.7 to 0.95, P = 0.04), significantly greater mean estimated fetal weight and greater mean birthweight centile than those with gestational diabetes or IGT. Amniotic fluid insulin levels demonstrated a similar significant difference between women with pre-existing diabetes and those with gestational diabetes or IGT. The study demonstrated that ultrasound measures of fetal size and growth are not sufficiently accurate to predict those babies likely to be at risk from the effects of fetal hyperinsulinaemia. [EL = 2+]

A retrospective cohort study involving 242 pregnant women with IGT evaluated the performance of estimated fetal weight and fetal growth velocity in the prediction of birthweight.288 The study showed that estimated fetal weight and fetal growth velocity have limited utility in predicting LGA babies. Estimated fetal weight and fetal growth velocity did not predict neonatal hypoglycaemia. [EL = 2+]

A prospective study of 181 women with diabetes (133 pre-existing type 1 diabetes, 48 gestational diabetes) compared the prediction power, at different gestational ages, of clinical and ultrasound measurements for fetal size.289 Clinical and ultrasound estimates were made at 28, 34 and 38 weeks of gestation or before birth. The study found all measurements were poor predictors of eventual standardised birthweight. Prediction improved with closeness to birth. Adding ultrasound to clinical information improved prediction, but only to a small extent. There was no difference in the prediction power for macrosomia between clinical and ultrasound measurements. [EL = 2++]

5.9.1.2. Fetal wellbeing

Three main tests are used by obstetricians to monitor fetal wellbeing. These are umbilical artery Doppler ultrasound velocimetry, fetal cardiotocography (non-stress test) and the biophysical profile. Monitoring for fetal wellbeing assumes that fetal compromise can be identified and that appropriately timed intervention (induction of labour or caesarean section) may reduce the risk of perinatal morbidity, admission to neonatal intensive care, asphyxia and fetal death.290

Doppler ultrasound

Doppler ultrasound uses sound waves to detect the movement of blood in the umbilical artery. It is used during pregnancy to assess fetus–placenta and/or uterus–placenta circulation.

A systematic review considered the effectiveness of Doppler ultrasound in high-risk pregnancies.291 The review included 11 studies involving 7000 women. Compared with no Doppler ultrasound, Doppler ultrasound in high-risk pregnancies (especially those complicated by hypertension or presumed impaired fetal growth) was associated with fewer perinatal deaths, fewer inductions of labour (OR 0.83, 95% CI 0.74 to 0.93) and fewer admissions to hospital (OR 0.56, 95% CI 0.43 to 0.72) without adverse effects. [EL = 1++] However, Doppler ultrasound offers no benefits to the low-risk population.292 [EL = 1++]

Abnormal umbilical Doppler ultrasound results are associated with chronic placental insufficiency, as occurs in pregnancies complicated by pre-eclampsia and fetal growth restriction . Although women with diabetes are at increased risk for these conditions, the majority of adverse outcomes in pregnancies complicated by diabetes are not associated with placental insufficiency.293 Small studies of women with gestational diabetes or pre-existing diabetes with good glycaemic control have considered the performance of Doppler ultrasound in predicting any adverse pregnancy outcome and have reported low sensitivities294297 [EL = 2++ to 2+]

Nonetheless a study has reported that Doppler ultrasound is better than fetal cardiotocography or biophysical profile in pregnant women with diabetes.298 In this study involving 207 women with diabetes, all three tests were performed concurrently within 1 week of birth. An adverse pregnancy outcome was defined as a pregnancy in which the baby was born before 37 weeks of gestation or had at least one of the following: growth restriction, hypocalcaemia, hypoglycaemia, hyperbilirubinaemia, respiratory distress syndrome or fetal risk requiring caesarean section. There were no perinatal deaths in this series. The performance of the three tests is summarised in Table 69. [EL = 2++]

Table 69. Performance of umbilical artery Doppler ultrasound, fetal cardiotocography and biophysical profile in predicting overall adverse pregnancy outcome; data from Bracero et al (1996).

Table 69

Performance of umbilical artery Doppler ultrasound, fetal cardiotocography and biophysical profile in predicting overall adverse pregnancy outcome; data from Bracero et al (1996).

A prospective double-blind randomised study was performed between 28 and 40 weeks of gestation in 92 pregnant women with diabetes to evaluate a random single Doppler ultrasound measurement of the systolic : diastolic ratio of the umbilical artery as a predictor of perinatal outcome in pregnancies complicated by diabetes.299 The performance of the Doppler ultrasound measurement as a predictor of poor perinatal outcome was: sensitivity 39%, specificity 92%, PPV 54%, NPV 86%. The data suggest that the systolic : diastolic ratio of the umbilical artery offers no advantage over other well-established tests in the management of pregnancy in women with diabetes. [EL = 1+]

Sixty-five pregnant women with diabetes were examined in a cohort study to evaluate the clinical usefulness of Doppler ultrasound flow velocity waveform analysis in such pregnancies.300 Umbilical and uterine artery flow velocity waveforms were obtained during the third trimester with a continuous wave Doppler ultrasound device. There was no difference in various clinical and Doppler ultrasound parameters between women with good glycaemic control and those with poor control. In contrast, the clinical and Doppler parameters were significantly different in women with pre-eclampsia than in those without pre-eclampsia, regardless of glycaemic control. There was a weak positive linear correlation (r = 0.30, P < 0.02) between maternal HbA1c and umbilical artery flow velocity waveforms (systolic : diastolic ratio). Proteinuria correlated better with umbilical artery systolic : diastolic ratio (r = 0.49, P < 0.001). The study suggests that Doppler ultrasound flow velocity waveform analysis may be clinically useful only in pregnancies complicated by diabetes and coexisting pre-eclampsia. [EL = 2+]

A prospective cohort study investigated Doppler ultrasound measurement of the fetal umbilical artery velocimetry in 56 women with diabetes, of whom 14 had varying degrees of vascular complications.301 The mean Doppler ultrasound values were higher in women with diabetes and vasculopathy than in women without diabetes and women with diabetes but no vasculopathy. The third-trimester systolic : diastolic ratio was greater than 3 in almost 50% of women with vasculopathy. A tendency towards adverse outcomes was observed at systolic : diastolic ratios approaching 4. Statistically significant correlations were found between elevated Doppler indices and maternal vasculopathy associated with hypertension and worsening renal insufficiency. Fetal growth restriction and neonatal metabolic complications were also significantly correlated with elevated Doppler indices. The data indicate an increased resistance circuit among women with diabetes and vasculopathy, which may reflect a relative reduction in basal uteroplacental blood flow and the need for cautious interpretation of Doppler indices in these women. [EL = 2+]

Another prospective cohort study was conducted to determine whether fetal aortic velocity waveforms were correlated with fetal outcome in pregnancies complicated by type 1 diabetes.302 Fetal aortic blood flow was assessed in 30 pregnant women with type 1 diabetes. The babies demonstrated no evidence of fetal distress at birth and there was no relationship between the mean third-trimester fetal aortic systolic : diastolic ratios and perinatal death, preterm deliveries, birthweight, Apgar scores at 1 minute and 5 minutes, or neonatal metabolic abnormalities. The data demonstrate a poor correlation between fetal aortic Doppler waveform analysis and fetal outcome. [EL = 2+]

5.9.1.3. Current practice

The CEMACH enquiry found that 21% of singleton births with a known birthweight had a birthweight of 4000 g or more in women with poor pregnancy outcomes.33 This was higher than the national average of 11%. A total of 5.7% births were severely macrosomic singleton births (a birthweight of 4500 g or over). The CEMACH enquiry case–control study reported that antenatal evidence of fetal growth restriction was associated with poor pregnancy outcome (OR 2.9, 95% CI 1.4 to 6.3, adjusted for maternal age and deprivation), but antenatal evidence of fetal macrosomia was not (OR 0.8, 95% CI 0.5 to 1.3, adjusted for maternal age and deprivation).33 Fetal surveillance was sub-optimal for 20% of 37 babies with antenatal evidence of fetal growth restriction and for 45% of 129 babies with antenatal evidence of macrosomia. For babies with antenatal evidence of macrosomia, sub-optimal fetal surveillance was associated with poor pregnancy outcome (OR 5.3, 95% CI 2.4 to 12.0, adjusted for maternal age and deprivation). Additional case–control analysis showed an association with fetal and neonatal death after 20 weeks of gestation, but not with fetal anomaly. [EL = 3–4]

The CEMACH enquiry found no difference in the proportion of women with type 1 or type 2 diabetes with antenatal evidence of macrosomia (P = 0.99) or fetal growth restriction (P = 0.31).33 [EL = 3–4]

The CEMACH enquiry found that shoulder dystocia was documented in 7.9% of vaginal births. The rate of shoulder dystocia was related to birthweight with 0.9% of babies weighing less than 2500 g, 4.7% of babies 2500–3999 g, 22.0% of babies 4000–4249 g, 25% of babies 4250–4499 g and 42.9% of babies 4500 g or more. The CEMACH enquiry found that Erb palsy occurred in 4.5 per 100 births; this is greater than the incidence of 0.42 per 1000 live births reported in the general population.2 [EL = 3]

The CEMACH enquiry found 0.9% of singleton babies born to women with type 1 diabetes and 1.3% singleton babies born to women with type 2 diabetes were less than 1000 g; this is higher than the national average for England and Wales (0.5%).2 [EL = 3]

The main concerns of the enquiry panels regarding surveillance of macrosomic and growth-restricted babies was lack of timely follow-up (affecting approximately 80% of babies). For macrosomic babies, there were also concerns about poor interpretation of ultrasound scans and about actions taken in response to tests. [EL = 3–4]

5.9.2. Evidence statement

The main ultrasonic methods for predicting birthweight (EFW and abdominal circumference) perform similarly in terms of diagnostic accuracy in women with diabetes. However, no clinical studies were identified that compared clinical outcomes using the two methods.

Umbilical artery Doppler ultrasound has better diagnostic accuracy as a test of fetal wellbeing in pregnant women with diabetes than has fetal cardiotocography or biophysical profile. Doppler ultrasound is also a better predictor of adverse maternal and neonatal outcomes, but its effectiveness is limited to high-risk pregnancies defined in terms of fetal growth restriction and/or pre-eclampsia, rather than diabetes per se.

5.9.3. Cost-effectiveness

The effectiveness of methods for monitoring fetal growth and wellbeing in women with diabetes was identified by the GDG as a priority for health economic analysis.

The lack of comparative data in relation to clinical outcomes resulting from ultrasonic methods for assessing fetal growth (for example, fetal abdominal circumference alone versus abdominal circumference plus fetal head circumference) precluded formal cost-effectiveness analysis. The GDG's discussions included consideration of the frequency of ultrasound assessment of fetal growth and the implications for cost-effectiveness. The GDG's view was that three scans should be offered (rather than two): this would allow healthcare professionals to advise women with diabetes on the direction of pregnancy, rather than providing estimates of fetal growth that might be masked by measurement error; it would also allow assessment of the need for, and response to, insulin therapy. Nevertheless, three scans at 4-weekly intervals starting at 28 weeks of gestation was thought to represent a reduction in the frequency of growth scans compared with current clinical practice that would, therefore, bring a cost saving to the NHS.

The clinical evidence in relation to monitoring fetal wellbeing showed that umbilical artery Doppler ultrasound is more effective in predicting adverse outcomes in women with diabetes than fetal cardiotocography or biophysical profile. Given that the clinical effectiveness of Doppler ultrasound is limited to women with other risk factors (notably fetal growth restriction and/or pre-eclampsia), and that current practice involves routine use of Doppler ultrasound to monitor fetal wellbeing in women with diabetes, a recommendation not to monitor fetal wellbeing routinely before 38 weeks of gestation was considered likely to be cost-effective.

5.9.4. Existing guidance

The NICE antenatal care guideline9 recommends that symphysis–fundal height should be measured and recorded for pregnant women at each antenatal appointment from 24 weeks of gestation. A fetal growth scan to detect SGA unborn babies should be offered to women if the symphysis–fundal height measurement is at least 3 cm less than the gestational age in weeks. Ultrasound estimation of fetal size for suspected LGA unborn babies should not be undertaken in a low-risk population. Doppler ultrasound should not be used to monitor fetal growth during pregnancy. Customised fetal growth charts should not be used for screening for SGA babies.

5.9.5. From evidence to recommendations

In the absence of comparative data on the effectiveness of different methods of ultrasound monitoring of fetal growth, the GDG recommended that fetal growth and amniotic fluid volume (to detect polyhydramnios) should be monitored by ultrasound every 4 weeks from 28 weeks of gestation to 36 weeks of gestation. The GDG's view is that this would represent a change in clinical practice which would effect a reduction in the frequency of monitoring for fetal growth and amniotic fluid volume in women with diabetes and would, therefore, bring a cost saving to the NHS. Fetal growth and amniotic fluid volume should be measured in all women with pre-existing diabetes and gestational diabetes (i.e. even in women with gestational diabetes controlled by diet alone) because of the increased risk of macrosomia.

Evidence shows that monitoring for fetal wellbeing using umbilical artery Doppler ultrasound is a better predictor of pregnancy outcome than fetal cardiotocography and biophysical profile in women with diabetes. However, routine monitoring of fetal wellbeing for women with diabetes is not recommended before 38 weeks of gestation because the effectiveness of Doppler ultrasound is limited to women at risk of fetal growth restriction and/or pre-eclampsia. In making this recommendation the GDG sought to effect a change in clinical practice that would bring a cost saving to the NHS.

5.9.6. Recommendations

90.

Offer pregnant women with diabetes ultrasound monitoring of fetal growth and amniotic fluid volume every 4 weeks from 28 to 36 weeks. [2008]

91.

Routine monitoring of fetal wellbeing (using methods such as fetal umbilical artery Doppler recording, fetal heart rate recording and biophysical profile testing) before 38 weeks is not recommended in pregnant women with diabetes, unless there is a risk of fetal growth restriction. [2008, amended 2015]

92.

Provide an individualised approach to monitoring fetal growth and wellbeing for women with diabetes and a risk of fetal growth restriction (macrovascular disease and/or nephropathy). [2008, amended 2015]

5.9.7. Research recommendations

38. How can the fetus at risk of intrauterine death be identified in women with diabetes?

Why this is important

Unheralded intrauterine death remains a significant contributor to perinatal mortality in pregnancies complicated by diabetes. Conventional tests of fetal wellbeing (umbilical artery Doppler ultrasound, cardiotocography and other biophysical tests) have been shown to have poor sensitivity for predicting such events. Alternative approaches that include measurements of liquor erythropoietin and magnetic resonance imaging spectroscopy may be effective, but there is currently insufficient clinical evidence to evaluate them. Well-designed randomised controlled trials that are sufficiently powered are needed to determine whether these approaches are clinically and cost-effective.

5.10. Timetable of antenatal appointments

5.10.1. Description of the evidence

No specific searches were undertaken for this section of the guideline. The evidence is drawn from publications identified in searches for other sections.

5.10.1.1. Current practice

The CEMACH diabetes in pregnancy programme provides data on current practice in England, Wales and Northern Ireland in relation to antenatal care, including care plans, for women with type 1 and type 2 diabetes. The enquiry panels classified maternity care during pregnancy as sub-optimal for 58% of women who had poor pregnancy outcomes and 44% of women who had good pregnancy outcomes (OR 1.9, 95% CI 1.2 to 2.8, adjusted for maternal age and deprivation). The two most frequently cited categories for sub-optimal antenatal care were fetal surveillance (monitoring fetal growth and wellbeing; see Section 5.9) and management of maternal risks. Other categories cited included problems with the antenatal diabetes multidisciplinary team. There were no significant differences between women with type 1 and type 2 diabetes in terms of sub-optimal antenatal care.33 [EL = 3–4]

The enquiry reported that 63% of maternity units in England Wales and Northern Ireland had a full multidisciplinary team comprising an obstetrician, a diabetes physician, a diabetes specialist nurse, a diabetes specialist midwife and a dietitian. There had been an increase in provision of staff over the preceding 8 years, with the availability of a diabetes specialist midwife in the antenatal clinic increasing from 25% to 77% of units, and the availability of a dietitian increasing from 40% to 80% of units. Seventy-five percent of women were reported to have maternity and diabetes care provided in a joint clinic, although only 22% of women were reported to have the entire multidisciplinary team involved in their care.33 [EL = 3–4]

The CEMACH enquiry panels commented that infrequent clinic appointments, lack of multidisciplinary involvement and communication issues were factors in sub-optimal diabetes care in pregnancy in some of the women in the case–control study.33 [EL = 3–4]

The CEMACH enquiry recommended that an individualised care plan for pregnancy (and the postnatal period) be used and that the care plan should include, as a minimum:33

  • targets for glycaemic control
  • a schedule for retinal screening
  • a schedule for renal screening
  • a plan for fetal surveillance during birth
  • postnatal diabetes care.

It was recommended that the care plan should be implemented from the beginning of pregnancy by a multidisciplinary team present at the same time in the same clinic. [EL = 3–4]

5.10.1.2. Existing guidance

The NSF for diabetes20 recommends that antenatal care for women with diabetes should be delivered by a multidisciplinary team consisting of an obstetrician, a diabetes physician, a diabetes specialist nurse, a midwife and a dietitian.

The NICE antenatal care guideline9 contains recommendations on the schedule of appointments that should be offered as part of routine antenatal care, including recommendations about what should happen at each appointment.

5.10.2. From evidence to recommendations

The GDG's view is that women with diabetes who are pregnant should be offered immediate contact with a joint diabetes and antenatal clinic, and they should have contact with the diabetes care team for assessment of glycaemic control every 1–2 weeks throughout pregnancy.

The timing and content of antenatal care appointments for women with diabetes should follow the schedule for routine antenatal care appointments recommended in the NICE antenatal care guideline,9 except where specific additions and/or differences are indicated below to support the recommendations made elsewhere in the guideline. The main differences between routine antenatal care (as specified in the NICE antenatal care guideline)9 and antenatal care for women with diabetes are summarised in Table 70. Ongoing opportunites for accessing information, education and advice should be offered to women with diabetes throughout the antenatal period.

Table 70. Timetable of antenatal appointments for women with diabetes.

Table 70

Timetable of antenatal appointments for women with diabetes.

Evidence from the CEMACH enquiry shows that many maternity units have not yet implemented the recommendation in the NSF for diabetes to provide diabetes and maternity care in a joint diabetes/antenatal clinic delivered by a multidisciplinary team. In formulating its recommendations the GDG sought to reinforce the recommendation contained in the NSF for diabetes.

5.10.2.1. First antenatal appointment

The GDG's view is that women with diabetes should be offered confirmation of viability and gestational age at the first antenatal appointment. This is earlier than in routine antenatal care because diabetes is associated with a high rate of miscarriage (see Sections 3.1 and 5.3) and because diabetes can disrupt the menstrual cycle leading to difficulty in determining the timing of ovulation.

Women with pre-existing diabetes may already have attended for preconception care and advice. For these women, the first antenatal appointment provides an opportunity to reinforce information, education and advice in relation to achieving optimal glycaemic control (including dietary advice). Women who have not attended for preconception care and advice should be offered the corresponding information, education and advice for the first time; a clinical history should seek to establish the extent of diabetes-related complications (including neuropathy and vascular disease); medications for diabetes and its complications should also be reviewed at this time.

Women with pre-existing diabetes who have not had a retinal assessment in the previous 12 months should be offered an assessment at the first presentation in pregnancy (see Section 5.6). Women with pre-existing diabetes who have not had a renal assessment in the previous 12 months should be offered an assessment at the first presentation in pregnancy (see Section 5.7).

All women with diabetes should have contact with the diabetes care team for assessment of glycaemic control every 1–2 weeks throughout the antenatal period (this could include telephone contact) and HbA1c should be used to assess long-term glycaemic control in the first trimester of pregnancy (see Section 5.3).

The GDG's discussions included consideration of screening for Down's syndrome. Screening methods for Down's syndrome in women with diabetes are currently no different to those for women without diabetes, and so the GDG made no specific recommendations in relation to the schedule for screening for Down's syndrome.

The GDG's discussions also included consideration of surveillance for pre-eclampsia. Women with diabetes are at increased risk of pre-eclampsia (see Section 5.7), but methods for surveillance (testing for proteinuria) and management of pre-eclampsia in women with diabetes are no different to those for women without diabetes. The schedule of appointments for routine antenatal care recommended in the NICE antenatal care guideline9 includes testing urine for proteinuria at every appointment, and so the GDG made no specific recommendations in relation to surveillance for pre-eclampsia.

5.10.2.2. 16 weeks of gestation

If any retinopathy is present at booking an additional assessment should be made at 16–20 weeks of gestation for women with pre-existing diabetes (see Section 5.6).

5.10.2.3. 20 weeks of gestation

Women with diabetes should be offered an ultrasound anatomical examination of the four chamber view of the fetal heart and outflow tracts at 20 weeks of gestation because the diagnostic accuracy is better at 20 weeks of gestation than at 18–19 weeks of gestation (see Section 5.8). The GDG's view is that the routine ultrasound scan for detecting structural anomalies, which should be offered to all pregnant women, should also be performed at 20 weeks of gestation in women with diabetes because it is more convenient for the woman to have both scans at one visit.

5.10.2.4. 25 weeks of gestation

No evidence was identified to suggest that antenatal care for women with diabetes should be different to routine antenatal care at 25 weeks of gestation.

5.10.2.5. 28 weeks of gestation

Ultrasound monitoring of fetal growth (to detect LGA or SGA babies) and amniotic fluid volume (to detect polyhydramnios) should start at 28 weeks of gestation and continue at 4-weekly intervals (i.e. 32 weeks and 36 weeks; see Section 5.9). Women with pre-existing diabetes who had no diabetic retinopathy at their first antenatal clinic visit should be offered retinal assessment at 28 weeks of gestation (see Section 5.6). Women who have been diagnosed with gestational diabetes as a result of routine antenatal screening enter the care pathway at 28 weeks of gestation (see Section 4.4 and the NICE antenatal care guideline9). They should be offered information about the risks to the woman and the baby that is offered to women with pre-existing diabetes in the preconception period.

5.10.2.6. 32 weeks of gestation

Ultrasound monitoring of fetal growth and amniotic fluid volume should be offered at 32 weeks of gestation as part of 4-weekly monitoring (see Section 5.9). It is the GDG's view that, for women with diabetes, the routine investigations that would normally be offered to nulliparous pregnant women at 31 weeks of gestation should instead be offered at 32 weeks of gestation because it is more convenient for the woman to have all the investigations at one visit.

5.10.2.7. 34 weeks of gestation

No evidence was identified to suggest that antenatal care for women with diabetes should be different to routine antenatal care at 34 weeks of gestation.

5.10.2.8. 36 weeks of gestation

Ultrasound monitoring of fetal growth and amniotic fluid volume should be offered at 36 weeks of gestation as part of 4-weekly monitoring (see Section 5.9). Evidence shows that women with diabetes are likely to give birth soon after 36 weeks of gestation, either through spontaneous labour, elective induction of labour or elective caesarean section to reduce the risk of stillbirth and birth trauma associated with fetal macrosomia (see Section 6.1). Given the evidence, the GDG's view is that women with diabetes should be offered information and advice in relation to intrapartum care and postnatal care at 36 weeks of gestation. The information and advice should cover: timing, mode and management of labour and birth, including options for elective early birth (see Section 6.1); analgesia and anaesthesia (see Section 6.2); changes to hypoglycaemic therapy during and after birth (see Sections 6.3 and 8.1); management of the baby after birth, including early feeding, detection and management of neonatal hypoglycaemia and other diabetes-related complications (see Chapter 7); initiation of breastfeeding and the effect of breastfeeding on glycaemic control (see Section 8.1); and information about contraception and follow-up (see Section 8.2).

5.10.2.9. 38 weeks of gestation

Induction of labour, or caesarean section if indicated, should be offered to women with diabetes at 38 weeks of gestation (see Section 6.1). Monitoring of fetal wellbeing should be offered to women with diabetes who are awaiting spontaneous labour.

5.10.2.10. 39–41 weeks of gestation

No evidence was identified to suggest that antenatal care for women with diabetes who have not given birth by 40 weeks of gestation should be different to routine antenatal care at 40– 41 weeks of gestation. However, evidence shows that many women with diabetes give birth before 40 weeks of gestation (see Section 6.1). Monitoring of fetal wellbeing should be offered to women with diabetes who are awaiting spontaneous labour at 39–41 weeks of gestation.

The GDG's discussions also included consideration of thyroid function in women with diabetes. There was no reason to suppose that women with diabetes required testing for thyroid function.

5.10.3. Recommendations

See the end of the next section for combined recommendations

5.11. Specialist teams

5.11.1. Review question

What is the effectiveness of specialist teams for pregnant women with diabetes?

5.11.2. Introduction

The objective of this review question is to determine whether specialist care during pregnancy for women with diabetes, as recommended in the 2008 guideline, is effective. Two types of specialist care were investigated by the review: receiving care from a multidisciplinary team compared with standard antenatal care; and receiving care from a centralised hospital compared with care from a peripheral hospital.

5.11.3. Description of included studies

Five studies (2 prospective and 3 retrospective observational studies) were identified for inclusion for this review question (Dunne et al., 2009; Hadden et al., 1999; Owens et al., 2012; Traub et al., 1987; Wilson et al., 2009). One prospective (Owens et al., 2012) and 1 retrospective (Wilson et al., 2009) study compared the use of a multidisciplinary team with standard care. One prospective (Dunne et al., 2009) and 2 retrospective (Hadden et al., 1999; Traub et al., 1987) studies compared the provision of care in a centralised hospital with care given in a peripheral hospital. Of the 2 studies that reported on multidisciplinary teams, 1 included 272 pregnancies without reporting the number of women (Owens et al., 2012) and the other included 96 women who all had gestational diabetes (Wilson et al., 2009). Of the 3 studies that compared centralised and peripheral care, 1 included 104 pregnancies in 84 women (Dunne et al., 2009), another included 856 pregnancies without reporting the number of women (Hadden et al., 1999) and the third included 221 pregnancies in 187 women (Traub et al., 1987).

The guideline development group's priority outcomes that were reported in the studies were:

  • mode of birth
  • glycaemic control in pregnancy (measured using HbA1C)
  • fetal or neonatal mortality
  • the number of large for gestational age babies
  • length of stay in a neonatal intensive care unit.

The priority outcomes that were not reported in the studies were:

  • the number of preterm births
  • maternal satisfaction
  • initiation of breastfeeding.

5.11.4. Evidence profile

The GRADE profiles for this review question are presented in Tables 71 and 72.

Table 71. GRADE profile for effectiveness of multidisciplinary teams for pregnant women with diabetes.

Table 71

GRADE profile for effectiveness of multidisciplinary teams for pregnant women with diabetes.

Table 72. GRADE profile for effectiveness of centralised care for pregnant women with diabetes.

Table 72

GRADE profile for effectiveness of centralised care for pregnant women with diabetes.

5.11.5. Evidence statements

5.11.5.1. Multidisciplinary team compared to standard antenatal care

When comparing women cared for in multidisciplinary teams with those receiving standard antenatal care, there were no significant differences in the risk of women having a vaginal birth (1 study, OR 1.2, 95% CI 0.5 to 2.6, n=96), an assisted/instrumental birth (1 study, OR 0.8, 95% CI 0.2 to 3.6, n=96) or caesarean section (2 studies, OR 1.4, 95% CI 0.9 to 2.2, n=464).

There were mixed results for glycaemic control. In the first and second trimester, 1 study (n=272) reported lower mean HbA1c in women with type 1 diabetes (MD −3, 95% CI −4.5 to −0.1 and MD −1, 95% CI −1.3 to −0.7) and in women with type 2 diabetes (MD −7, 95% CI −8.4 to −5.6 and MD −5, 95% CI −5.2 to −4.8) receiving care from a multidisciplinary team. However, a second study of women with type 1 or type 2 diabetes (n=96) reported no differences in mean HbA1c in the first (MD 0, 95% CI −0.3 to 0.3) or second trimester (MD −0.2 95% CI −0.6 to 0.2) among women receiving care from a multidisciplinary team compared with women receiving standard antenatal care.

In the third trimester, 1 study (n=272) reported that mean HbA1c was lower in women with type 1 diabetes (MD −3, 95% CI −3.3 to −2.8) but higher in women with type 2 diabetes (MD 1, 95% CI 0.8 to 1.2) who received care from a multidisciplinary team compared with standard antenatal care. A second study that combined women with type 1 and type 2 diabetes into 1 group reported a significantly lower mean HbA1c in women receiving care from a multidisciplinary team (MD −0.4, 95% CI −0.7 to −0.1).

In terms of fetal and neonatal outcomes, 1 study (n=272) found no difference in the risk of perinatal death (OR 0.1, 95% CI 0.0 to 1.0) or stillbirth (OR 0.3, 95% CI 0.1 to 1.7) in the babies of women receiving multidisciplinary care compared with those receiving standard antenatal care. However, the guideline development group did think that this could have been a function of the small population size. A decreased risk of miscarriage in the group of women receiving care from a multidisciplinary team (OR 0.3, 95% CI 0.1 to 0.6) was reported.

The same study (n=272) reported no differences in the risk of babies being large for gestational age among women with type 1 (OR 0.8, 95% CI 0.5 to 1.4) or type 2 (OR 1.6, 95% CI 0.9 to 3.0) diabetes receiving standard antenatal or multidisciplinary team care, or in the risk of neonatal care unit admission (OR 0.8, 95% CI 0.5 to 1.4). A second study (n=96) did find a reduced risk of special care baby unit admission in the babies of women who received care from a multidisciplinary team compared with those receiving standard antenatal care (OR 0.3, 95% CI 0.1 to 0.7).

The quality of evidence for these outcomes was very low.

5.11.5.2. Centralised care compared to peripheral care

When comparing centralised to peripheral care, 1 study (n=160) reported a reduced risk of caesarean section in the women receiving centralised care (OR 0.5, 95% CI 0.3 to 0.9). No other maternal outcomes were reported.

In terms of neonatal and fetal outcomes, there were no differences reported in the risk of neonatal death (2 studie: n=936, OR 0.5, 95% CI 0.1 to 2.0), fetal loss (2 studies: n=936, OR 1.2, 95% CI 0.8 to 1.8), miscarriage (2 studies: n=264, OR 0.7, 95% CI 0.3 to 1.6; and 1 study: n=776, OR 1.5, 95% CI 0.9 to 2.4), stillbirth (3 studies: n=1040, OR 0.7, 95% CI 0.3 to 1.6), perinatal death (3 studies: n=1040, OR 0.6, 95% CI 0.3 to 1.3) or of babies being large for gestational age (1 study: n=104, OR 0.7, 95% CI 0.2 to 2.1). One study (n=104) found fewer admissions to the neonatal unit for babies born to mothers who received centralised care compared with those who received peripheral care (OR 0.1, 95% CI 0.0 to 0.4).

The quality of evidence for these outcomes was very low.

5.11.6. Health economics profile

A systematic review of the literature did not find any published evidence on the cost effectiveness of specialist teams for pregnant women with diabetes.

This question was initially prioritised for health economic analysis but it was a lower priority than other guideline topics and the clinical evidence was not very strong.

5.11.7. Evidence to recommendations

5.11.7.1. Relative value placed on the outcomes considered

The guideline development group placed an equal value on all of the outcomes, as they each contribute to morbidity and mortality.

5.11.7.2. Consideration of clinical benefits and harms

When comparing women receiving care from a multidisciplinary team with women receiving standard care, glycaemic control (indicated by lower HbA1c) was better in the group of women receiving care from a multidisciplinary team. For fetal outcomes, there were significantly fewer miscarriages and significantly fewer special care baby unit admissions in the group of women who received care from a multidisciplinary team.

When comparing centralised care with peripheral care, there were significantly fewer caesarean sections in the group of women receiving centralised care. There were significantly fewer admissions to the neonatal unit for babies whose mothers who received centralised care.

5.11.7.3. Consideration of health benefits and resource use

The guideline development group accepted that care from a multidisciplinary team was likely to be a more expensive way of delivering services than ‘standard care’, although the multidisciplinary approach represents current NHS practice. The group believes that a multidisciplinary approach facilitates better communication and care in a complex area of service provision. They therefore are of the view that it is likely to result in better outcomes which will justify any additional costs.

5.11.7.4. Quality of evidence

The quality of the evidence was rated as very low for all reported outcomes considered in the review.

The data could be considered to be limited as 4 out of 5 studies were undertaken in populations that were from the same geographical area (Ireland and Northern Ireland) where there may be less ethnic diversity than that encountered on the UK mainland. The fifth study was from the UK. However, the group felt that the data from all five studies were probably more relevant to the population for which the guidance was intended than studies from other parts of the world. Another limitation is that the majority of the data came from historical cohorts rather than prospective studies. Some studies were poorly reported and the analysis conducted in the papers was not always clearly reported.

5.11.7.5. Other considerations

The guideline development ggroup was aware of the need for effective communication between healthcare professionals, both in the primary and secondary sector, as recommended in the guideline on patient experience.

In the reviewing and updating the table of antenatal appointments recommendation in the original guideline, the guideline development group for this guideline was aware of some inconsistencies between the recommendations regarding the use of ultrasound to screen for structural abnormalities in the original guideline. The relevant recommendations in the original guideline were that women with diabetes should be offered antenatal examination of the 4 chamber view of the fetal heart and outflow tracts at 18–20 weeks and (in Table 5.7) to offer 4 chamber view of the fetal heart and outflow tracts plus scans that would be offered at 18–20 weeks as part of routine antenatal care.

However, Table 5.6 in the original guideline stated that the ultrasound scan for detecting structural anomalies and anatomical examination of the 4 chamber view of the fetal heart plus outflow tracts should occur at 20 weeks. This was on the basis that scanning the fetal cardiac anatomy including the 4 chamber view was better at 20 weeks than 18 weeks.

In the light of this duplication of recommendations and inconsistency in gestational age, the guideline development group felt that it would be better to bring together the separate recommendations about screening for congenital abnormalities (scanning for structural abnormalities in general, scanning the 4 chamber view of the fetal heart and performing the ultrasound scan at 20 weeks, rather than the 18 weeks in non-diabetic pregnancy) into 1 recommendation for greater clarity.The recommendation can be found in Section 5.8.8.

5.11.8. Key conclusions

The evidence was not strong enough to change existing recommendations.

5.11.9. Recommendations

93.

Offer immediate contact with a joint diabetes and antenatal clinic to women with diabetes who are pregnant. [2008]

94.

Ensure that women with diabetes have contact with the joint diabetes and antenatal clinic for assessment of blood glucose control every 1–2 weeks throughout pregnancy. [2008, amended 2015]

95.

At antenatal appointments, provide care specifically for women with diabetes, in addition to the care provided routinely for healthy pregnant women (see the NICE guideline on antenatal care). Table 73 describes how care for women with diabetes differs from routine antenatal care. At each appointment, offer the woman ongoing opportunities for information and education. [2008, amended 2015]

Table 73. Timetable of antenatal appointments.

Table 73

Timetable of antenatal appointments.

5.12. Preterm labour in women with diabetes

5.12.1. Description of the evidence

5.12.2. Incidence of preterm birth

A prospective cohort study303 examined the importance of glycaemic control and risk of preterm birth in women with type 1 diabetes who have normoalbuminuria and no pre-eclampsia during pregnancy. Seventy-one women with complete data on HbA1c, insulin dose and albumin excretion rate measured at 12 weeks of gestation and every second week thereafter were recruited and followed. The overall rate of preterm birth was 23%; women who experienced preterm birth had higher HbA1c throughout pregnancy. Regression analysis showed that HbA1c was the strongest predictor of preterm birth from 6–32 weeks of gestation and that the risk of preterm birth was more than 40% when HbA1c was above 7.7% at 8 weeks of gestation. [EL = 2+]

5.12.2.1. Antenatal steroids

Antenatal steroids are administered to women who have a spontaneous or planned preterm birth to accelerate fetal lung development and prevent respiratory distress syndrome. The use of steroids in women with diabetes is associated with a significant worsening of glycaemic control requiring an increase in insulin dose.

Two studies were identified that reported on approaches to increasing insulin in women with diabetes undergoing treatment with antenatal steroids.

The first study reported on a test of an algorithm for improved subcutaneous insulin treatment during steroid treatment (intramuscular administration of betamethasone 12 mg repeated after 24 hours).304 The algorithm was as follows:

  • on day 1 (the day on which the first betamethasone injection is given), the night insulin dose should be increased by 25%
  • on day 2, all insulin doses should be increased by 40%
  • on day 3, all insulin doses should be increased by 40%
  • on day 4, all insulin doses should be increased by 20%
  • on day 5, all insulin doses should be increased by 10–20%
  • during days 6 and 7, the insulin doses should be gradually reduced to their levels before treatment.

The study involved 16 women, eight of whom were treated before the introduction of the algorithm (cohort 1) and another eight who were treated after its introduction (cohort 2). Women in cohort 1 had insulin doses adjusted individually based on the level of blood glucose obtained. The median blood glucose over the 5 days was 6.7 mmol/litre, 14.3 mmol/litre, 12.3 mmol/litre, 7.7 mmol/litre and 7.7 mmol/litre in cohort 1 and 7.7 mmol/litre, 8.2 mmol/litre, 9.6 mmol/litre, 7.0 mmol/litre and 7.4 mmol/litre in cohort 2 (P < 0.05 for days 2 and 3). None of the women developed ketoacidosis or severe hypoglycaemia. [EL = 2+]

The second study reported on the use of a supplementary intravenous sliding scale to indicate the required dosage of supplementary insulin infusion in six women receiving antenatal steroids.305 The supplementary insulin was in addition to the woman's usual subcutaneous insulin regimen and usual dietary programme. The additional infusion was commenced immediately before the first steroid injection and continued for at least 12 hours after the second injection. If blood glucose levels were too high on the initial regimen (glucose 10.1 mmol/litre or more for 2 consecutive hours) the dosage regimen was moved up to the next level. If the blood glucose level was less than 4 mmol/litre the dosage regimen was reduced by one level. Data were collected on six women receiving dexamethasone. Significant amounts of supplementary intravenous insulin were required (median dose 74 U, range 32–88 U) in order to achieve glucose control following administration of dexamethasone. Seventy-five percent of all glucose measurements were within 4–10 mmol/litre. [EL = 3]

5.12.2.2. Tocolytic Betamimetics increase agents

Tocolytic agents are used to inhibit uterine contractions. They may help to delay birth and allow women to complete a course of antenatal steroids. Betamimetics have been widely used for tocolysis, although they are no longer recommended as the first choice for general use.306 blood glucose concentrations307309 and several cases of ketoacidosis have been reported in women with diabetes following administration of these medicines (see Section 5.1).310314

5.12.2.3. Current practice

CEMACH undertook a descriptive study of all pregnancies of women with pre-existing diabetes who gave birth or booked between 1 March 2002 and 28 February 2003.2 Of the 3474 women in this study with a continuing pregnancy at 24 weeks of 328 gave birth before 34 weeks of gestation. Thirty-five of these pregnancies resulted in a stillbirth. Of the remaining 293 women, 70.3% received a full course of antenatal steroid therapy. The most common reason given for non-administration of antenatal steroids was birth of the baby before the full course could be given. In a small group of women diabetes was considered a contraindication to antenatal steroid use. [EL = 3]

5.12.3. Evidence statement

The use of antenatal steroids for fetal lung maturation in women with diabetes is associated with a significant worsening of glycaemic control.

Two studies that reported on approaches to modifying insulin dose in women undergoing antenatal steroid treatment showed that glycaemic control could be improved by increasing the insulin dose immediately prior to and during administration of antenatal steroids. However, the two protocols evaluated were only moderately successful in keeping blood glucose levels at the desired level (less than 7 mmol/litre).

Evidence shows that administration of betamimetics to suppress labour induces hyperglycaemia and ketoacidosis.

5.12.4. From evidence to recommendations

The evidence supports the use of increased insulin dose immediately before and during antenatal administration of steroids. Since the two protocols that have been evaluated were only moderately successful in achieving glycaemic control, women receiving additional insulin during administration of antenatal steroids should be closely monitored according to an agreed protocol in case the insulin dose requires further adjustment.

When tocolysis is indicated in women with diabetes an alternative to betamimetics should be used to avoid hyperglycaemia and ketoacidosis.

5.12.5. Recommendations

96.

Diabetes should not be considered a contraindication to antenatal steroids for fetal lung maturation or to tocolysis. [2008]

97.

In women with insulin-treated diabetes who are receiving steroids for fetal lung maturation, give additional insulin according to an agreed protocol and monitor them closely. [2008, amended 2015]

98.

Do not use betamimetic medicines for tocolysis in women with diabetes. [2008]

5.12.6. Research recommendations

There were no research recommendations relating to preterm labour in women with diabetes.

Footnotes

mm

Level 2 critical care is defined as care for patients requiring detailed observation or intervention, including support for a single failing organ system or postoperative care and those ‘stepping down’ from higher levels of care.

nn

Level 2 critical care is defined as care for patients requiring detailed observation or intervention, including support for a single failing organ system or postoperative care and those ‘stepping down’ from higher levels of care.

oo

For the purpose of this guidance, ‘disabling hypoglycaemia’ means the repeated and unpredicted occurrence of hypoglycaemia requiring third-party assistance that results in continuing anxiety about recurrence and is associated with significant adverse effect on quality of life.

Copyright © 2015 National Collaborating Centre for Women's and Children's Health.
Bookshelf ID: NBK328347