Content of education

There were no trials located in newly diagnosed people with Type 1 diabetes specifically, or concerned with the initial content of education. The American Diabetes Association guidelines suggest that as part of initial visit people should be referred to a diabetes educator if education is not provided by the physician or practice staff, but content of this education is not defined (IV).³²

Educational setting

One small randomised controlled trial comparing the efficacy of classroom teaching of diabetes skills, compared to individualised learning, found that classroom teaching led to a greater level of awareness about diabetes self-care.⁷⁷ However, there was no significant difference in terms of the level of use of self-care practices. Furthermore, the two education techniques provided no different outcome of levels of technical skill in self-care. However this study made no analysis of comparability of study groups at baseline and was not blinded (Ib).

Technology interventions

One randomised controlled study compared two interactive computer schemes to reinforce an educational video. The first gave additional feedback and information on the correct answers, the second only the correct answers.²⁵⁷ People with diabetes in the interactive group scored significantly better in a follow-up test of diabetes knowledge than those following the standard scheme. There were no significant differences in user ratings for the two software packages, but the people in the additional feedback group had a better diabetes knowledge at baseline, so the results may be biased by this confounding factor (Ib).

Guidelines for self-management education

An update of the US standards for diabetes self-care management based on a literature review covered the organisation of diabetes self-management education, its content and provision.³³¹ A multiprofessional task force encompassing all the major interested stakeholders agreed the following standards (IV).

• Education and information-giving will involve the interaction of the individual with diabetes with a multifaceted education instructional team, which may include a behaviourist, exercise physiologist, ophthalmologist, optometrist, pharmacist, physician, podiatrist, registered dietitian, registered nurse, other healthcare professionals, and paraprofessionals.
• Instructors will obtain regular continuing education in the areas of diabetes management, behavioural interventions, teaching and learning skills, and counselling skills.
• Assessed needs of the individual will determine which content areas listed below are delivered:

  –

  describing the diabetes disease process and treatment options

  –

  incorporating appropriate nutritional management

  –

  incorporating physical activity into lifestyle

  –

  utilising medications (if applicable) for therapeutic effectiveness

  –

  monitoring blood glucose, urine ketones (where appropriate) and using results to improve control

  –

  preventing, detecting, and treating acute complications

  –

  preventing (through risk reduction behaviour), detecting, and treating chronic complications

  –

  goal-setting to promote health, and problem-solving for daily living

  –

  integrating psychosocial adjustment to daily life

  –

  promoting preconception care, management during pregnancy, and gestational diabetes management (if applicable).

• An individualised assessment, development of an education plan and periodic reassessment between participant and instructor will direct the selection of appropriate educational materials and interventions.
• The assessment includes relevant medical history, cultural influences, health beliefs and attitudes, diabetes knowledge, self-management skills and behaviours, readiness to learn, cognitive ability, physical limitations, family support, and financial status.
• There shall be documentation of the individual's assessment, education plan, intervention, evaluation, and follow-up in the permanent confidential education record.

General education programmes

Within an overall review of patient education models for diabetes (not type-specific) one health technology appraisal reviewing four controlled trials of a range of education programmes including items of self-management, self-monitoring, diet, the effects of insulin and exercise, taught by a variety of staff or self-taught, and as an initial intense course or as ongoing programmes reported a variety of positive outcomes compared to normal care.⁴¹⁹ This review found that one study had demonstrated improvements over 10 years in diabetic control, in terms of reduced HbA_1c levels. In another study an intensive five-day training course was found to be effective in reducing HbA_1c levels. In one study there was no difference in blood glucose control with education compared to usual care, while there were no between-group comparisons made in another other study. Education was also shown to improve blood pressure. There is limited evidence to suggest a reduced rate of ketoacidosis and reduced hospitalisation. However, there was no evidence to indicate that education can reduce body mass index. There is some data to suggest increased incidence of hypoglycaemic episodes. Long-term outcomes of retinopathy, or neuropathy were not found to be significantly affected by education, but there is some limited evidence to suggest nephropathy incidence is improved, although rates were low. Unsurprisingly, diabetes knowledge was significantly improved with education, although this was not true of quality of life. Overall the included trials were of moderate methodological rigour. Three of the trials included were investigating education in the context of intensification of treatment compared to normal care, and it is difficult to be sure that the benefits reported are directly attributable to the education aspect of the intervention (NICE).

Metabolic control and quality of life were not found to be significantly affected by a structured outpatient programme of education led by a nurse, dietitian, and other people with diabetes over 4 weeks in a large randomised trial as compared to conventional care (Ib).¹²²

A medium sized randomised controlled trial of a monthly education programme at which different aspects of diabetes treatment and technical skills were considered found that after one year of education HbA_1c levels were reduced compared with normal clinical care in people with Type 1 diabetes.²⁹³ However, age differences between the control and intervention groups at baseline mean that this study is possibly methodologically limited.

Another moderate sized systematic review of eight trials encompassing over 3,000 patients with either Type 1 or Type 2 diabetes, found that intensive versus brief education on foot care provided a significant decrease in incidence of foot ulcers, and in one trial amputations, but no difference in the same outcomes over seven years in another study.⁵¹⁰ This is despite three trials reporting successful uptake of messages regarding foot care behaviour. Another trial reported in this review found that an intensive educational intervention including both people with diabetes and doctors improved the prevalence of serious foot lesions compared to usual care, although the composite outcome of all foot lesions and amputations was not significantly improved. Authors of the review noted methodological limitations of the included studies, and outcome reporting times varied between individual trials (Ia).

Diabetes self-management education

Evaluation, in a large systematic review, of a range of diabetes self-management education programmes (DSME) compared to normal routine levels in populations of people with diabetes found that interventions based in community gathering places were able to reduce blood glycated haemoglobin (GHb) and fasting blood glucose levels.³⁷⁷ There is some evidence that they can also improve diabetes knowledge and improve physical activity (minutes of walking). Other trials reviewed that were based in the home setting - half of which included children or adolescents -showed a significant decrease in GHb after DSME, and a borderline beneficial effect on weight for people undergoing DSME as compared to conventional care. Specific analysis in patients with Type 1 diabetes found no significant change in diabetes knowledge with such programmes (Ia).

Other educational interventions

A small randomised controlled trial in people with Type 1 diabetes found that an intervention whereby patients received regular telephone contact with a diabetes nurse to alter insulin regimen decreased HbA_1c over six months compared to usual care.⁴⁹⁷ This difference was not found to be affected by age, sex, or type of diabetes (Ib).

Behavioural and education interventions

There are no systematic reviews and few prospective randomised studies that report on methods to improve concordance in self-management in people with Type 1 diabetes. One small unblinded study, which was methodologically limited owing to high drop-out rates and inequalities in patient characteristics at baseline, found that an intervention of a self-taught study programme to improve self-control behaviour was able to demonstrate improved adherence to goals of self-monitoring of blood glucose level over 12 weeks.²⁴⁵ The intervention included a wide range of educational and behavioural choice items, and the relative effectiveness of any of these is hard to define. The methodological limitations of the study would not form a rigorous basis for recommending such an approach (Ib).

A similar intervention among adolescents (mean age 18 years) in India enrolled in a prospective randomised trial with an intervention of 15 hours of individualised learning over three months comprising both behavioural and cognitive strategies based on an operant learning model, found improved adherence on a composite three-item scale, compared to usual care.³¹² This improved adherence was mirrored in significantly improved blood glucose level compared to people in the control group. However this study had a small sample size and was unblinded, and it was not possible to determine whether benefits persist after the cessation of the intervention (Ib).

Education interventions

One small-to medium-sized randomised trial of a specialist education programme delivered to people with Type 1 diabetes by a team of physicians, dietitians, and specialist nurses found there to be no statistically significant differences in diabetes knowledge or adherence to dietary advice compared to a control group who received conventional diabetes education. Both groups improved in both measures immediately after the completion of the education intervention but then knowledge and adherence fell away with time. This trial was sited in Finland and there may be differences in content of conventional diabetes education compared to that of the UK care setting (Ib).²⁷¹

Monitoring devices

There were no significant differences in adherence to glucose self-monitoring or in blood glucose levels reported at six months between two interventions with novel glucose monitoring devices and control with a standard device from a medium-sized multicentre randomised trial.¹⁹⁵ The trial included a population of people with Type 1 diabetes who had had the condition for an average of 14 years. The study was blinded between the two novel monitoring machines, but the people in the control group would have been aware that they were not receiving the intervention as they continued to use their usual machine. To evaluate adherence all patients were asked to keep diaries of self-monitoring behaviour and this may have stimulated greater adherence even in the control group than under normal everyday self-monitoring conditions (Ib).

Reliability and validity

Papers contained within a systematic review suggest that the evidence on issues of observer training, interdevice variability, the effects of long-term use and patient acceptability have not been adequately addressed (IV).⁹⁸

One study within the systematic review comparing self-reported readings against a memory meter showed that inaccuracies in readings were common.⁹⁸ This was due to rounding of values, omission of outlying values and reporting of results when no test had been performed. These findings were confirmed in another reviewed study of 14 people who recorded lower blood glucose values in logbook records than meter memories (Ia).

Reported within the systematic review, one trial suggested that patients needed to be informed of the memory capacity of their meters to improve accuracy.⁹⁸ A further study reported in the review argued that the true diurnal variability in glycaemia in people with Type 1 diabetes is too great to be measured, even when self-monitoring of blood glucose (SMBG) is repeated seven times daily (Ia).

Patient factors (as described below) were shown to have an impact (both positive and negative) on the reliability of monitoring in five studies.⁹⁸

Reliability can be improved through proper training of patients, and was shown in sub-group analysis to be equally as good in older people, and people with visual impairments (on condition that extensive instruction has been provided).

One study concluded that as impairment of colour vision affects the ability to interpret self-monitoring with visually read strips, suggesting that all patients should be screened for colour vision before self-monitoring begins (Ia).

Clinical effectiveness of blood glucose monitoring

Four trials contained within a systematic review failed to show with sufficient power a demonstrated effect of SMBG on blood glucose control (Ia).⁹⁸

Two trials comparing urine and blood testing showed no clinical difference in the two tests (Ia).⁹⁸

A systematic review reported on patient preferences for different monitoring techniques.⁹⁸ One trial reported patients preferring blood testing, or a combination of blood and urine testing, compared to urine testing alone. No preference was stated for visual strips or strips with meters (Ia).

One methodologically limited comparative study comparing blood glucose meters with visual test strips showed patients found the two techniques equally convenient to use, although overall more patients preferred the blood glucose meter.¹⁶⁴

Preferences were based on: accuracy, confidence in test result, no judgement by patient, inability to cheat with result and use of the built in timer (Ib).

One methodologically limited comparative study showed that fructosamine self-test results correlated well with laboratory test with very low bias.¹³⁷ Imprecision of the self-test was higher than the laboratory test, but could still identify patients with good versus poor glycaemic control (DS).

A further methodologically limited diagnostic study in people with Type 2 diabetes showed self-testing of fructosamine to be comparable in accuracy to laboratory fructosamine and GHb values (DS).⁷⁸

One trial with 25 patients showed no significant difference in glucose control or patient practice based on frequency of testing.¹⁸⁰ The authors stated that they are unable to identify any optimal frequency for blood glucose self-monitoring in typical diabetic population. There is little or no relationship between the frequency of blood glucose monitoring, the frequency of insulin dose adjustments and the level of metabolic control (Ib).

A study of the preferences of 18 patients within a systematic review reported a preference for testing four times daily, twice weekly or four times daily once a week, compared to twice daily every day of the week (Ib).⁹⁸

One study from a systematic review reported fasting plasma glucose to be less useful as an accurate mode of monitoring in insulin treated people with diabetes than in other people (IIa).⁹⁸

Changes to diet

Four small randomised controlled trials were identified examining different diet regimens in people with type 1 diabetes.^{82,167,198,323} One randomised controlled study found that a high fibre diet (50g/day) for 24 weeks compared to a low fibre diet (15 g/day) improved blood glucose profile, and number of hypoglycaemic events, although HbA1c, cholesterol, body weight, and insulin dose were not affected (Ib).¹⁶⁷

A high carbohydrate, high fibre and low fat diet, compared to conventional low carbohydrate diet, taught by a dietitian in an unblinded randomised controlled trial was seen at 12 months to improve HbA1c (Ib).³²³

The addition of vitamin E to the normal diet has been shown to provide no benefit in terms of cholesterol level, HbA1c, body mass index (BMI), insulin dose, or blood pressure over a 12 month period (Ib).⁸²

There were significant improvements in glomerular filtration rate, and a decline in albuminuria after 4 weeks of a low protein diet compared to a normal protein diet in a randomised prospective trial in people with overt diabetic nephropathy.¹⁹⁸ Outcomes of urinary sodium excretion, blood pressure, BMI, and HbA1c were not significantly different between the diets (Ib).

Therapy adjustment for normal eating

Canadian clinical practice guidelines recommend that all people with diabetes on fixed-dose insulin regimen should have an individualised meal and activity plan developed.⁵³⁵ Two studies showed that patients should be taught how to adjust insulin dosage, diet and physical activity in response to blood glucose levels, to reduce incidence of hypoglycaemia (Ia).

A medium-sized randomised controlled trial of a five-day outpatient programme to enable patients to replace insulin by matching it to desired carbohydrate intake amongst adults with type 1 diabetes found that the intervention improved HbA1c compared to a control of normal care to six months.³³ Positive effects were also seen in weighted quality of life and total well-being. There was no effect on incidence of severe hypoglycaemia, weight or total cholesterol. This trial enrolled people with poorly controlled diabetes (Ib).

A similar small trial in which intensified insulin plus simplified diet was compared to conventional therapy and diet found HbA1c to be significantly reduced, although there was no difference between the study groups for outcomes of body weight, BMI, cholesterol, or triglycerides (Ib).⁸²

Undefined diet

A large cohort study comparing degree of liberalisation of diet away from a specific controlled diet after a treatment and teaching programme with estimation of carbohydrate intake and subsequent insulin self-adjustment found that there was no significant relationship between BMI and degree of diet liberalisation.³⁵⁰ In addition there was no relationship with HbA1c level or severe hypoglycaemia. However there was a relationship between liberalised diet and higher cholesterol levels, and an inverse relationship with tendency to monitor blood glucose more than three times a day (IIa).

Other evidence

The recent evidence-based guidelines for nutrition principles developed by the ADA, provide a broad overview of research in the area of improved diabetes care for people with Type 1 diabetes through beneficial nutritional therapies.³² There are recommendations based on well-performed RCTs showing significant effectiveness of interventions for areas such as carbohydrates, dietary fat, energy balance and obesity, nutritional therapy for the treatment or prevention of acute complications, and hypertension. Recommendations in other key areas are based on cohort or uncontrolled studies (Ia).

Glycated haemoglobin testing

A Diabetes UK consensus statement recommended that only HbA1c should be used in the monitoring of blood glucose control. Other studies reported within a systematic review have shown discrepancies in the classification of patients between HbA1c and HbA1 assays (IV).⁹⁸

Two studies in a systematic review showed high intra-individual variability for GHb assays in non-diabetic and in diabetic subjects with stable or variable control.⁹⁸ One of these studies suggested an association between clinical control and sampling interval (IIa).

The same systematic review reported on randomised controlled trial evidence supporting the use of GHb measurements, in particular results cited from the DCCT demonstrated the usefulness of these assays in contributing to improved long-term blood glucose control and a reduction in morbidity (Ia).⁹⁸

A Danish systematic review reported that HbA1c values allowed clinicians to identify patients with poor glycaemic control, concluding that GHb is the most clinically appropriate test of long-term glycaemia and should be used in routine management of Type 1 diabetes (Ia).²⁸²

Frequency of monitoring

The optimal frequency of testing has not been established.

One study within a systematic review recommended that no more than six GHb assays were necessary in a given year (IV).⁹⁸

ADA recommendations advise GHb measurements are performed in accordance with clinical judgements.³² ADA consensus recommends GHb testing at least twice a year in patients with stable glycaemic control who are meeting treatment goals. Testing should be more frequent (quarterly) in patients whose therapy has changed or who are not meeting glycaemic control targets (IV).

Fructosamine testing

There are discrepancies in the evidence surrounding the use of fructosamine testing.

One study within a systematic review reported fructosamine testing as able to detect shorter or more recent fluctuations in blood glucose compared to GHb.⁹⁸ Fructosamine testing does not have the problems of standardisation associated with GHb, thus results are comparable between laboratories (IIa).

Two studies within a systematic review described a high correlation between fructosamine and HbA1c, however, later studies debated this claim.⁹⁸ One study suggested that although fructosamine correlates with HbA1c, the value of HbA1c in an individual has been shown to not routinely be inferred with reliability from the level of fructosamine (IIa).

Two studies contained within a systematic review, in patients with renal failure and elderly Type 2 diabetes patients with liver cirrhosis and nephrotic syndrome, suggest the influence of chronic conditions rather than metabolic control on fructosamine levels is the source of unreliability in test result.⁹⁸ The systematic review concludes that more evidence is needed to resolve these issues (IV).

One correlation study within a review showed no significant correlation between HbA1c and fructosamine results over a six month follow up (III).⁹⁸

Frequency of monitoring

ADA recommendations state that assays of glycated serum protein would have to be performed on a monthly basis to gather the same information as measured in GHb three to four times per year (IV).³²

A systematic review urges caution in using fructosamine testing, in light of the fact that fructosamine values can be improved by increased concordance a week or two before testing (IV).⁹⁸

Another study found that fasting blood glucose levels (FBG) and serum fructosamine are not as useful as HbA1c for monitoring diabetic control, but are additional extras for assessing control over short and long periods (IIa).⁹⁸

Continuous glucose monitoring systems (CGMS)

Three observational studies compared continuous glucose monitoring systems (CGMS) with SMBG.^186,187,308 Studies demonstrated good correlation of CGMS with plasma and capillary measures of blood glucose over a range of blood glucose values. Error grid analysis showed the majority of readings fell within a clinically acceptable margin of error across all studies (III).

One study reported acceptable level of comfort with CGMS. However, none of these studies address viable outcomes of glycaemic control or long-term use.³⁰⁸ Study methodology is not clearly reported.

Near patient testing

In this guideline, ‘near patient testing’ is defined as a biochemical or other test at or near (in time and place) the clinical consultation, such that the result is available at the consultation.

One controlled trial within a systematic review demonstrated that near patient testing led to an increase in management changes for patients with poor glucose control.¹⁸⁴ Near patient testing for HbA1c improved the process of care of patients (IIa).

In the same review, questionnaires recording patient satisfaction of near patient testing concluded that the introduction of near patient testing for HbA1c improves the likelihood of monitoring and discussion of glycaemic control at patient visits.¹⁸⁴ Patients reported that this was important to them and resulted in greater satisfaction with the test information provided (III).

Within the health technology assessment a retrospective cohort study showed that, after allowing for confounding factors, mean HbA1c level was lower following near patient testing and the immediate availability of results.¹⁸⁴ In order to precisely quantify the effect of the testing system on HbA1c level, further, prospective studies are required (IIa).

A systematic review reported four studies on the effectiveness of benchtop analysers compared with traditional laboratory methods.⁹⁸ Two studies showed comparable results between the two techniques when operated by non-medical personnel. One study found that the benchtop analyser, although reliable, tended to slightly underestimate HbA1c, compared with high performance liquid chromatography (HPLC) (IIa).

6.1.4.1. Recommendations

R38: Clinical monitoring of blood glucose levels by high precision DCCT-aligned methods of haemoglobin A1c (HbA1c) should be performed every two to six months depending on:

• achieved level of blood glucose control
• stability of blood glucose control
• change in insulin dose or regimen (D).

R39: Site-of-care measurement, or before clinical consultation measurement, should be provided (D).

R40: HbA1c results should be communicated to the person with Type 1 diabetes after each measurement. The term “A1c” can be used for simplicity (D).

R42: Fructosamine should not be used as a routine substitute for HbA1c estimation (B).

R43: Continuous glucose monitoring systems have a role in the assessment of glucose profiles in adults with consistent glucose control problems on insulin therapy, notably:

• repeated hyper- or hypo-glycaemia at the same time of day
• hypoglycaemia unawareness, unresponsive to conventional insulin dose adjustment (B).

Evidence: narrative

The UKPDS showed that the reduction over a median of 10 years in HbA1c from 7.9 to 7.0 % using sulphonylureas or insulin provided much of the benefit that could be expected from that degree of improved glycaemic control. However it also illustrated the difficulties in being able to reach this level (7.0 %) in a substantial proportion of people. Thus providing only one target is likely to encounter a significant number of people who ‘fail’ to meet that target. Similarly for some individuals an even lower target is desirable as they may have additional risk factors that necessitates even tighter blood glucose control. The UKPDS also suggested that there were no thresholds for cessation of benefit and that the lower the level of mean HbA1c the better.

Working group commentary

The Working group tried to reflect these issues when deciding upon a target HbA1c. They concluded that a range was the best option, recognising the difficulty in achieving a low target whilst recognising the importance of trying to achieve as near normal an HbA1c level as possible, and in particular recognising that additional risk factors made the lower limit even more important for many individuals. While no study suggests clear thresholds, the group noted on the basis of the epidemiological evidence in the DCCT (Type 1 diabetes) and UKPDS that microvascular risk was low once average HbA1c was around 7.0-8.0 % while arterial risk continued to fall down to 6.0 to 7.0 % (DCCT standardised).

Thus the target for each individual should be set which fully takes into account: their assessed risk factors, including: age, BMI, blood pressure and lipid status; side effects of therapy, other individual factors, patient choice (NICE).³²⁶

The NICE Type 2 diabetes guidelines therefore recommended 6.5 to 7.5 % as ideal targets, individualized by balance of macrovascular (tend to 6.5 %) and microvascular (7.5 %) risk (NICE).

The ADA has republished its recommendations yearly.³² These choose a ‘glycaemic goal’ of HbA1c <7.0 % for adults (type of diabetes not specified), equating this to pre-prandial <7.2 mmol/l and peak post-prandial <10.0 mmol/l. However, a table in the same paper suggests that an HbA1c of 7.0 % equates to mean self-monitored plasma glucose of 9.5 mmol/l116 (IV).

However in the same issue (January 2003) the ADA notes in a chapter on ‘Implications of the DCCT’ that the level of glucose control to be sought under ideal circumstances is an HbA1c of around 7.2% (average glucose 8.6 mmol/l).³² This argument is, however, based on that achieved in the DCCT, and is thus not theoretically justified (IV).

The microvascular risk threshold is what determines the diagnostic threshold for diabetes. In theory, oral glucose tolerance test (OGTT) findings should give some guidance as to this threshold. Unfortunately these are mainly based on non-physiological glucose load findings, and set a top limit of risk for 2h post-prandial levels. Fasting levels have been set as a microvascular threshold of 7.0 mmol/l (based on epidemiological equivalence with 2h OGTT levels), which would map to a DCCT-harmonized HbA1c of about 7.7 % (IV).

Glucose equivalents

Two non-randomised controlled studies report the relationship between HbA1c and self-monitored pre- and post-prandial glucose levels.^355,438 The reports are consistent and can be related to DCCT-harmonised assays. It must be noted that these studies used pre-determined self-monitored profiles taken from memory meters, and cannot easily be translated into patient-selected estimations, or only pre-prandial monitoring. They also omit the effects of night-time glucose profiles between bedtime and pre-breakfast readings. These data give the most robust evidence of the relationship between HbA1c and the toxic glucose concentrations which actually cause the microvascular damage (IIa).

6.2.3.1. Consideration

There must be a threshold for glucose control and the development of microvascular complications, or non-diabetic people would get complications. Indeed this threshold must be well above the normal range as people with impaired glucose tolerance (IGT) do not (by definition) get microvascular complications. As people with IGT have HbA1c levels of up to 7.0 %, this by itself sets a lower limit of microvascular risk.

The microvascular thresholds of HbA1c 7.5 % set around 10 years ago have stood the test of all data published since. If anything the DCCT, Krolewski and McCance data suggest a figure closer to 8.0 %.

Recommendations from the ADA (7.0 %) and American College of Endocrinologists (6.5 %) are not type of diabetes specific; data does suggest macrovascular protection is gained by lowering blood glucose levels into the normal range, and the NICE (inherited) guidelines for Type 2 diabetes go for HbA1c 6.5 % in these higher arterial risk individuals.

Some people with Type 1 diabetes are at higher arterial risk, notably those with developing nephropathy. This can be identified by increased albumin excretion rate. The presence of features of the metabolic syndrome will also predict higher arterial risk. It may be appropriate to consider tighter targets for glucose control (if feasible) in people in these categories.

However these levels are better considered as assessment levels, to be used in setting realistic targets for the individual. Major diabetes services in Europe currently only get about 20 % of people with Type 1 diabetes into the sub-7.5 % bracket. UK composite data (UKDIABS) shows some services doing better, but this may only represent non-standardised GHb estimation.

That current technologies of diabetes care markedly limit the proportion of people on insulin who were able to manage themselves to ideal levels was not seen as a bar to setting such assessment levels. It was noted that arterial risk would be likely to have a different relationship in this regard from microvascular risk, and that for the former there was little direct information available for people with Type 1 diabetes, but that the understandings gained in Type 2 diabetes and people without diabetes gave strong guidance in this respect. It was felt that as the assessment of the evidence available pointed to target definition in the same range as other published guidelines, and in particular the NICE inherited guidelines for Type 2 diabetes, there was practical utility for practice of care in having matching recommendations. Lastly the problem of hypoglycaemia in limiting was what achievable in any individual should be addressed within any recommendations, to assuage inappropriate attempts to achieve tight control and counter impressions of failure if targets are not attained.

Insulin and insulin analogues

Insulin with the molecular structure of human and animal insulins is currently available. Evidence from the majority of studies reports no significant differences in hypoglycaemic episodes and glycaemic control between the insulin of human and animal chemical structures (Ia).^163,249,434

Conventional two-dose insulin regimens may result in a high frequency of nocturnal hypoglycaemia. Intensified three-dose insulin regimens improves glycaemic control, but often do not improve morning blood glucose (Ia).¹³⁹

Continuous subcutaneous insulin infusion (CSII) improves nocturnal and morning glycaemic control compared with multiple daily injection (MDI) regimens. With multiple injection regimens the morning injection must not be delayed. Total and bolus insulin doses required are lower with CSII compared with MDI (Ib).¹⁹³

Mortality from acute metabolic causes (ketoacidosis) was reported as significantly increased with intensified treatment; odds ratio 7.20 (pumps) 1.13 (multiple daily injection).¹³⁹ The pump data is however based on early pump technologies (Ia).

Similar glycaemic control results from either lente or isophane (NPH) insulin when used as basal insulin for multiple injection regimens together with a short-acting insulin preparation before meals (Ib).⁵⁰⁰

On the balance of effectiveness and cost-effectiveness evidence, insulin glargine, which has a peakless action profile, is also recommended as a long-acting preparation for people with Type 1 diabetes³⁶³; some studies in this review show significantly lower fasting blood glucose with insulin glargine than isophane (NPH) insulin and others suggest that people on insulin glargine may experience fewer hypoglycaemic events than people receiving once-daily isophane (NPH) insulin (NICE).³⁶³

Evidence from a large multicentered study suggests that people commonly inject insulin closer to meal-time than the recommended 30 minutes. Due to slow absorption and delayed action, the use of unmodified (‘soluble’) human insulin as pre-meal dose results in high and variable post-breakfast blood glucose concentrations, which together with the incidence of later hypoglycaemia suggests that this regimen does not give satisfactory post-prandial blood glucose control in many patients (Ib).⁵¹³

Rapid acting insulin analogues allow injection closer to meal times due to their pharmacokinetic profile (Ib).^123,299,375

A meta-analysis¹¹⁵ and several open-label trials^{39,136,160,279,405,431,436,437,513} show that Insulin lispro is more effective than unmodified (‘soluble’) human insulin in improving post-prandial glucose control, without an increase in the rate of hypoglycaemic episodes(Ia).

Two studies^25,222 show reduced frequency of nocturnal hypoglycaemia⁷⁰ with insulin lispro compared to unmodified (‘soluble’) human insulin (Ib).

Two studies show reduced frequency of severe hypoglycaemia with insulin lispro compared to unmodified (‘soluble’) human insulin (Ia).^70,463

Patients perceive an improvement in their well-being and quality of life with rapid-acting insulin analogues due to flexibility of injection times and less frequent hypoglycaemic reactions (Ib).^222,249,431

The effects of insulin lispro on HbA1c levels (overall glycaemic control) have not been firmly established.^{291, 1064 275 115,463,513} The long-term safety profile is as yet unknown (Ia).

Two multicentre randomised studies^225,463 and one RCT²⁹⁹ showed insulin aspart to improve post-prandial glucose control more effectively than unmodified (‘soluble’) human insulin, without an increase in the rate of hypoglycaemic episodes. Fewer major hypoglycaemic episodes were observed (Ia).

A before-and-after study has shown that a lower dose of meal time insulin can be taken along with an increase in basal dose, with no increase in hypoglycaemic episodes when insulin lispro is used as a replacement for human insulin as meal-time injection therapy (IIb).¹³⁶

Two randomised trials have shown that it is possible to replace mealtime unmodified (‘soluble’) human insulin with insulin lispro or insulin aspart without detriment to glycaemic control if care is taken to replace basal insulin delivery more physiologically (Ib).^212,543

A multi-arm randomised trial found that adding a few units of isophane (NPH) insulin to insulin lispro at each meal, in combination with bed-time isophane (NPH) insulin improves blood glucose concentrations compared to an unmodified (“soluble”) human insulin regimen in the a multidose regimen (Ib).¹²³

Splitting the evening administration of insulin to short-acting insulin at dinner and isophane (NPH) insulin at bedtime has a number of advantages over mixed administration of short-acting insulin and isophane (NPH) at dinner. Compared with the mixed mealtime regimen, the evening split regimen reduced by more than 60% the risk of nocturnal hypoglycaemia; improved long-term control of blood glucose levels; decreased variability of blood glucose levels in fasting state and led to improvement in preserved hormonal, symptom and cognitive function responses to hypoglycaemia (Ib).^146,476

When basal insulin replacement is by either continuous subcutaneous insulin infusion (CSII) or multiple daily administrations of isophane (NPH) insulin, the long term administration of lispro at mealtime reduces HbA1c; however, compared with multiple daily injections, patients using continuous subcutaneous administration of insulin (mainly those using older systems) have been at a significantly higher risk of ketoacidosis (Ib).¹⁹³

Frequency of hypoglycaemic reactions were found to be similar on patient-mixed and premixed insulins.^135,436 One randomised controlled trial showed premixed preparations of insulin analogues to be well suited for those who wish to limit the number of daily injections; 83% of people expressed a preference for premixed insulins throughout the trial (Ib).¹³⁵

Few studies have addressed the needs of people with diabetes with suboptimal glucose control, and none of suitable design from the evidence hierarchy were found for review.

In a group of people with Type 1 diabetes with poor glucose control, the introduction of more intensive insulin regimens may lead to high loss to follow-up.¹²⁵

Poor outcome appears to be due to the people refusing the constraints of multiple daily injections, effective blood glucose self-monitoring, and regular clinic visits at short time intervals. It was suggested that people should be given clear and concise information on treatment goals and the ways in which these goals are to be attained as well as an explanation of the advantages and disadvantages (IV).

Few studies addressed the needs of people newly diagnosed with diabetes and none of suitable design from the evidence hierarchy were found for review.

Acarbose and insulin combination therapy

Four randomised controlled trials, two large parallel group,^221,433 and two small crossover designs^310,514 were identified that examined the use of acarbose in conjunction with insulin therapy compared to insulin and placebo in each case, in people with Type 1 diabetes. A multicentred study with variable doses titrated up to 300 mg three times a day for 24 weeks, found a significant reduction in HbA1c levels with acarbose compared to placebo, and decreases in fasting and post-prandial glucose levels to two hours.²²¹ There were no differences between groups for daily insulin dose or hypoglycaemic events, although adverse events of abdominal pain, diarrhoea, and flatulence were more common with acarbose. This led to more frequent treatment discontinuation in the acarbose group than the placebo group. A similar Italian trial with up to 100 mg acarbose three times daily for 24 weeks found no difference in HbA1c levels, daily insulin dose, fasting glycaemia, and total cholesterol.⁴³³ However, a significant decrease was found in two-hour post-prandial plasma glucose level, and HDL cholesterol levels were lower in people on acarbose than placebo. Again minor adverse events were more common in the acarbose group, but hypoglycaemic episodes were similar in both groups. Although care was taken not to alter baseline insulin doses, this could be adjusted if glucose levels exceeded 11.1 mmol/l or reduced with hypoglycaemic episodes (Ib).

The two cross-over trials with 100 mg acarbose three times a day over relatively short time period did not assess requirement for wash out periods (although analysis in one found no effect of treatment order) and did not account for study withdrawals. One study found a benefit in terms of HbA1c with acarbose,³¹⁰ while the other found no significant differences between groups.⁵¹⁴ Potential methodological limitations of these trials would not permit them to be used as an evidence base to inform recommendations in this area (Ib).

Sulfonylurea and insulin combination therapy

Two small randomised controlled trials investigated the use of glibenclamide (called ‘glyburide’ as the trials were conducted in the USA) in the therapy for Type 1 diabetics. A study using 5 mg glyburide (orally) for 12 weeks compared to placebo after a 12-week open label insulin stabilisation run-in period found fasting blood glucose declined significantly at 12 weeks from baseline, although no comparison was made between groups.¹⁹¹ No differences were found in daily insulin dose or glycated haemoglobin levels at any stage of the study. A randomised study without comparison between groups at baseline with 5 mg glyburide daily for 24 weeks compared to placebo found no differences in plasma C-peptide levels between groups, nor difference in plasma glucose concentrations at any time point.¹⁷⁴ Although HbA1c levels were reported to have changed more from baseline in the glyburide treated group at six weeks, potential methodological limitations of these trials would not permit them to be used as an evidence base to inform recommendations in this area (Ib).

Comparison of 15 mg of glibenclamide daily with placebo in addition to insulin therapy in a small sample of people with Type 1 diabetes in a randomised double-blind cross-over study found mean blood glucose level, HbA1c, and blood glucose variability to be significantly lower with the intervention among people who retained endogenous insulin production.⁷¹ No such differences were found in a subgroup who were C-peptide negative. Although the study had a medium term intervention period of three months, it did not provide analysis of the cohort as a whole for glibenclamide vs placebo and thus cannot be used for recommendations given the small sample sizes of the subgroups, and the inherent difficulties of extrapolating such findings to a wider population (Ib).

A reduced insulin requirement at 18 months was found in patients given 80 mg gliclazide twice a day compared to placebo in a small sample in a long-term study.¹⁴⁵ Although glycated haemoglobin both fasting and one hour post-breakfast were found to be very similar in both groups, the gliclazide group had C-peptide levels significantly higher than people on placebo for the same test times of the day, at six-monthly assessment points to 18 months. This study only applies to people with retained endogenous insulin secretion, and thus not the overwhelming majority of people with Type 1 diabetes (Ib).

Metformin

A medium-sized randomised controlled study found that the addition of metformin to an insulin regimen provided by CSII was able to reduce the total IR required by the person with Type 1 diabetes (including reduced basal therapy) as compared to placebo over a period of six months. This was achieved without significant change to HBA1c or increased incidence of hypoglycaemia (Ib).

Monitoring

The major risk factors for foot complications have been identified in several systematic reviews as history of ulceration and lack of sensation.^361,454

The NICE Clinical guidelines for Type 2 diabetes reported inconsistent evidence of markers associated with foot complications from nine studies using a range of methods and patient data.³⁶¹ These included: old age, duration of diabetes, neuropathy, peripheral vascular disease, renal disease, foot deformities, plantar callus, previous ulceration or amputation, poor vision, poor footwear, cigarette smoking, social deprivation and social isolation (NICE).

The NICE guideline also reported five surveys investigating additional risk factors for the elderly, concluded that suboptimal supervision of elderly patients in hospital, residential care and general practice increases their risk of ulceration and amputation (NICE).³⁶¹

Organisation of screening programmes

The SIGN guidelines note that absence of reliable symptoms and the high prevalence of asymptomatic disease make foot screening essential (IV).⁴⁵⁴

One large comparative trial in a systematic review of a combined screening and foot protection programme reported a statistically significant reduction in major amputations over a two-year period compared to normal organisation of care (Ia).³⁸¹

The NICE clinical guidelines report a Cochrane review comparing trials of general practice vs hospital care for recall and review of foot problems, and conclude that despite the methodological flaws in these trials a system of shared care – joint participation between hospitals and general practices – provides levels of surveillance as good as hospital diabetic clinic attendance alone (NICE).³⁶¹

Information exchange between specialists is advocated in one review in the NICE Type 2 diabetes guidelines.³⁶¹ However, no evidence exists to specify the components of these procedures (NICE).

The guidelines foot care working party also endorsed the findings of Diabetes UK that a multidisciplinary team of professionals should be available to promptly provide the full range of appropriate foot care services to patients (NICE).³⁶¹

Detection of loss of foot sensation

SIGN guidelines concluded from three studies that neuropathy screening performed by using clinical neuropathy disability scores, 10 g monofilaments or vibration perception thresholds, alone or in combination, have benefits in selecting patients at increased risk of foot ulceration (DS).⁴⁵⁴

Additional techniques available for assessing neuropathic deficit that are considered in SIGN guidelines include tactile circumferential discriminator, the graduated tuning fork, thermal discrimination devices and others.⁴⁵⁴ These techniques have not been prospectively evaluated but generally compare with other techniques for detection of ulcers (IV).

There is general agreement in systematic reviews and guidelines that the 5.07 mono- filament (10 g) is cheap and easy to use compared to other neuropathic tests and is the recommended screening test for neuropathy as a risk factor for diabetes foot ulcers (IV/NICE).^361,454

A systematic review of a particular monofilament and other threshold tests for preventing ulceration and amputation in people with diabetes found this design of monofilament correlated best with the presence or history of an ulcer.³¹⁶ Evidence varies as to the appropriate number of sites to use with this technique, the majority of studies testing at 1 site. The plantar surface of the forefoot provides the best discrimination between those who did and did not have ulcers (III).

Four prospective studies included in a systematic review described a strong predictive ability of the monofilament test for future foot ulceration and amputation and a high reproducibility (DS).³¹⁶

Within a systematic review two non-randomised studies reported physical symptoms of tingling, burning, hyperaesthesia and other uncomfortable sensations affecting >40% of people with diabetes after diagnosis.³¹⁶ However, two separate studies reported poor correlation of pain symptoms with foot ulceration (III).

Prospective evidence is sparse for traditional clinical assessment, using pinprick, tuning fork vibration or light touch with a cotton wisp.³¹⁶ While the reproducibility of these investigations is low, replicability is slightly better for ankle jerks; however these tests are considered poor predictors of ulceration (DS).

Two-point discrimination was shown in one study in a well-produced systematic review to be more sensitive but less specific than monofilament or vibration perception threshold (VPT) testing.³¹⁶ Temperature sensation was found in two studies to be cumbersome and irritating and correlated less well with risk of ulceration compared to monofilament or VPT (DS).

One further medium-sized diagnostic study described the comparability of a new technique combining a monofilament and pinprick test to reference standard tests.³⁹¹ The new technique was found to have good correlation with VPT and a neuropathy disability score assessment, and a specificity and sensitivity of roughly 80% and 70% respectively in detecting both neuropathy disability score and VPT results identifying moderate to severe neuropathy (DS).

Detection of peripheral vascular disease

Screening for vascular insufficiency is less well documented than ulceration in existing reviews (IV).⁴⁵⁴

Two studies in the SIGN guidelines note that absence of pedal pulses can be used in first-line screening as a guide to peripheral vascular disease.⁴⁵⁴ Evidence from one study urges caution when evaluating ankle pressure and pressure indices, which can be falsely elevated in people with diabetes (DS).

A systematic review of observational studies noted a restricted accuracy of pedal pulses in identifying severe peripheral ischaemia (DS).⁶⁴

The validity of Doppler ultrasonography to determine ankle-branchial index as an indicator of peripheral blood flow was also questioned by one study in a systematic review.⁶⁴ The study noted that calcification of the media of the distal arteries, common in diabetes, may lead to artificially high systolic pressure in the ankle (DS).

Foot complication surveillance

70.

On the basis of findings from foot care surveillance, foot ulceration risk should be categorised into:

• low current risk (normal sensation and palpable pulses)
• increased risk (impaired sensory nerve function or absent pulses, or other risk factor)
• high risk (impaired sensory nerve function and absent pulses or deformity or skin changes, or previous ulcer)
• ulcer present.

Foot care management

71.

For people found to be at increased risk or high risk of foot complications:

• arrange specific assessment of other contributory risk factors including deformity, smoking, level of blood glucose control
• arrange/reinforce specific foot care education, and review those at high risk as part of a formal foot ulcer prevention programme
• consider the provision of special footwear, including insoles and orthoses, if there is a deformity, callosities or previous ulcer.

72.

For people with an ulcerated foot:

• arrange referral to a specialist diabetes foot care team incorporating specifically- trained foot care specialists (usually state-registered podiatrists) within one to two days if there is no overt infection of the ulcer or surrounding tissues, or as an emergency if such infection is present
• use antibiotics if there is any evidence of infection of the ulcer or surrounding tissues, and continue these long-term if infection is recurrent
• use foot dressings taking account of cost according to local experience, ensuring arrangements are in place to monitor and change dressings frequently (often daily) accordingly to need
• remove dead tissue from diabetic foot ulcers
• consider the use of off-loading techniques (such as contact casting) for people with neuropathic foot ulcers
• do not use cultured human dermis (or equivalent), hyperbaric oxygen therapy, topical ketanserin or growth factors in routine foot ulcer management
• consider ensuring complete and effective foot education through the use of graphic visualisations of the consequences of ill-managed foot ulceration in people with recurrent ulceration or previous amputation
• review progress in ulcer healing frequently (daily to monthly) according to need
• if peripheral vascular disease is detected, refer for early assessment by a specialist vascular team.

Charcot osteoarthropathy

73.

Adults with suspected or diagnosed Charcot osteoarthropathy should be referred immediately to a multidisciplinary diabetes foot care team.

Significance of patient-reported sexual symptomatology in predicting actual physiological measures of sexual dysfunction

A medium-sized cross-sectional cohort study in people with diabetes mellitus evaluated the significance of patient-reported sexual symptomatology in predicting penile tip and base rigidity, tip and base duration of erectile episode.²¹ This study reports that the presence of morning erections was associated with increased Rigiscan values of tip rigidity (r=0.64), base rigidity (r=0.58), tip duration of erectile episode (r=0.65) and base duration of erectile episodes (r=0.57), all demonstrating significant relationships (IIa).

Reports of fuller erectile quality were also significantly associated with increased Rigiscan values of tip rigidity (r=0.58), base rigidity (r=0.42), tip duration of erectile episode (r=0.67) and base duration of erectile episode (r=0.71).249 Other significant associations found in this cohort study included intact ejaculatory capacity being associated with increased Rigiscan measures of tip rigidity (r=0.45). Tip duration of erectile episode (r=0.56) and base duration of erectile episode (r=0.30) were also related to Rigiscan measures in the same study.²¹

A significant inverse relationship was found between symptom frequency and the Rigiscan measure of base duration of erectile episodes, with greater symptom frequency being associated with diminished duration values of erectile episodes at the penile base (r=-0.39) (IIa).²¹

Correlations of lower limb nerve fibre abnormalities with erectile dysfunction

A medium-sized cross-sectional cohort study aimed to characterise the neuropathy in erectile dysfunction, as well as to identify nerve fibre subtypes that may be preferentially affected.⁵²⁰ Patients were evaluated with a symptom questionnaire based on the Michigan Neuropathy Screening instrument questionnaire and examined clinically. Sural and peroneal nerve-conduction studies and quantitative sensory and autonomic tests (using the staging system of Dyck) were used to detect nerve abnormalities in the lower limbs. Various methodological limitations inherent to the study limited the validity of the results derived from the trial (IIa).

Relationship of symptoms of depression, sexual dysfunction and neuropathy in women

A small cross-sectional cohort study assessed the relationship between symptoms of depression (as measured by the Beck Depression Inventory and the Hamilton Psychiatric Rating Scale), sexual dysfunction (as measured by a questionnaire which asked patients to rate their symptoms on a scale from 0 to 10), and neuropathy (as measured by the visual analogue scale).²⁸⁹ However, various methodological limitations inherent to the study limit the validity of the results derived from the trial, and should not be used as the basis for a positive recommendation (IIa).

Sildenafil

One large multicentre study of sildenafil at 100 mg/day compared to placebo in men with erectile dysfunction and Type 1 or Type 2 diabetes found significantly more men were able to achieve and to maintain erections with sildenafil than placebo at 12 weeks.⁵⁰ Another 11 outcomes from questionnaire-based evaluation of male sexual function described significant improvement with the intervention drug, however there were no differences in indices of frequency and level of sexual desire. Erectile function was improved regardless of age, duration of erectile dysfunction, duration of diabetes or type of diabetes, and the incidence of adverse arterial events was similar in both groups (Ib).

A smaller prospective study from the UK found that sildenafil at 25 mg or 50 mg, compared to placebo, significantly improved adjusted duration of penile rigidity at base and tip.⁴¹⁴ In addition, there was an improved number of erections hard enough for sexual intercourse with either dose, with no serious adverse events being related to treatment (Ib).

Anticonvulsants

One large meta-analysis found a significant benefit of at least 50% pain relief with people with anticonvulsants compared to placebo.⁹³ The relative risk estimates showed that anti- convulsants had a significantly increased incidence of adverse effects compared with placebo for minor but not major harm (Ia).

One small, randomised controlled trial of gabapentin found an improvement over placebo control on a pain questionnaire at 12 weeks but with no significant improvement on a visual analogue pain scale, or present pain intensity.¹⁸¹ No significant adverse events were reported in either study arm but minor events drowsiness, fatigue and imbalance were more common in the population on gabapentin than on placebo (Ib).

The differences in mean pain intensities between the intervention and control groups were significant after eight weeks at lamotrigine doses of 200, 300 and 400 mg in a small-scale prospective randomised trial.¹⁴⁰ This study found no significant changes in assessment of McGill Pain Questionnaire, Beck Depression Inventory and Pain Disability Index (Ib).

Antidepressants

One large meta-analysis found a significant benefit of at least 50% pain relief with people with antidepressants compared to placebo with pooled analyses of tricyclic antidepressants showing significant benefit but no benefit with selective serotonin re-uptake inhibitors.⁹³

Tricyclic antidepressants used were prescribed in doses in the low to moderate range for depression. Antidepressants had a significantly increased incidence of adverse effects compared with placebo with typical antimuscarinic effects of dry mouth, constipation and blurred vision. Also major events (leading to withdrawal from the trial) were more common with antidepressants than placebo, the number needed to harm (NNH) for a major adverse effect with antidepressants compared with placebo was 17 (Ia).

A small short-term randomised controlled trial investigating mean pain intensity diary scores in a six-week within-patient comparison, showed that desipramine was superior to placebo.³¹⁵ No significant difference between incidence of adverse events or withdrawals between desipramine and placebo groups (Ib).

Other therapies

Amantadine: A small randomised controlled trial with a one-week follow-up found amantadine infusion at 200 mg in 500 ml 0.9% saline infusion over three-hour period caused a significant clinically relevant reduction in pain score when compared with placebo, and caused a significant improvement in the neuropathy symptom score.³⁵ Following amantadine, there was a clinically significant subjective tenfold improvement in pain relief (Ib).

Capsaicin: A meta-analysis comparing a range of studies with outcomes from four to eight weeks found capsaicin cream produced significantly higher response rates than placebo cream for physician assessment of global pain in two of the trials, but not in the other two (Ia).⁵⁴²

Clonidine: No statistically significant difference between intervention and control groups in patients' pain record diary or pain intensity levels in two randomised trials of clonidine.^72,541 In the patients completing the study, dry mouth and drowsiness tended to occur more commonly with clonidine than placebo (Ib).⁵⁴¹

Gamma-linolenic acid: Compared with placebo, dietary supplementation with gamma- linolenic acid was reported as being associated with significant clinical, neuropsychological and quantitative sensory improvement in established distal diabetic polyneuropathy in the medium term.²³⁶ A significant improvement in the gamma-linolenic acid group compared with the placebo group was seen in nine variables: symptom scores, median MCV (m/s), peroneal MCV (m/s), median CMAP (mV), peroneal CMAP (mV), median SNAP (μV), sural SNAP (μV), ankle HT (°C), wrist HT (°C). This study recruited only people with Type 2 diabetes (Ib).

A second trial confirms this with gamma-linolenic acid being significantly superior in improving neuropsychological, neurological and thermal sensation parameters of diabetic neuropathy compared with placebo over a one-year period.²⁵⁰ A significant improvement in the gamma-linolenic acid group compared with the placebo group was seen in: peroneal MNCV, median MNCV, extensor digitorium brevis CMAP, thenar CMAP, sural SNAP, median SNAP, wrist heat threshold, wrist cold threshold, arm muscle strength, arm tendon reflexes, leg tendon reflexes, arm sensation, leg sensation. Subgroup analysis showed improvement of outcome parameters with the gamma-linolenic acid was greater in patients with initial HbA1c <10% than those with HbA1c >10% (Ib).

Isosorbide dinitrate (ISDN): A small crossover trial showed significant reductions in pain and burning sensation using the ISDN spray compared with placebo.⁵³⁹ During the ISDN phase of the study, two patients developed mild transient headaches, which resolved spontaneously and did not affect overall adherence with the spray (Ib).

Mexiletine: Trials of mexiletine have provided mixed results in terms of efficacy for pain reduction in people with diabetes and painful neuropathy. This difference in effect could be due to clinical differences in study populations, doses utilised or length of follow-up measured (Ib).

A significant reduction in pain during night-time (as estimated by the visual analogue scale score for pain) was observed in the mexiletine 675 mg group compared with the placebo group as was a significant reduction in sleep disturbances in a large multicentre randomised trial.³⁸⁷ No significant difference between groups in daytime pain or global assessment of efficacy was recorded. However, another study showed no improvement in the McGill Questionnaire or on the visual analogue scale for pain to five weeks (Ib).⁴⁸³

In contrast a study of mexiletine compared to placebo for 26 weeks found that the Five Item Symptom Scale Score was improved in all but one patient during treatment with mexiletine, but in only two patients during the placebo phase.³⁴⁵ Mexiletine significantly improved pain, dysaesthesia and paraesthesia. During treatment with mexiletine the visual analogue score fell significantly. Three patients had mild side effects when treated with mexiletine, including nausea, hiccough and tremor (Ib).

Tramadol: A medium-scaled prospective randomised trial of tramadol at up to 200 mg/day found that by day 14 people in the tramadol group had less pain than patients in the placebo group.²⁰¹ This benefit lasted through to the end of the trial at day 42. They also scored better on outcomes of physical and social functioning. No statistically significant differences between treatments were noted for current health perception, psychological distress, overall role functioning and the two overall sleep problem indexes and sleep subscales. The most common adverse events in the tramadol group were nausea (23%), constipation (22%), headache (17%) and somnolence (12%). Nine patients treated with tramadol and one treated with placebo discontinued due to adverse events. The most common adverse events leading to discon- tinuation of tramadol were nausea and dyspepsia. However, this study recruited only people with Type 2 diabetes (Ib).

Management during acute arterial events

91.

Optimal insulin therapy, which can be achieved by the use of intravenous insulin and glucose, should be provided to all adults with Type 1 diabetes with threatened or actual myocardial infarction or stroke. Critical care and emergency departments should have a protocol for such management.