Chapter  166:  Diabetes Education for Children With Type 1 Diabetes Mellitus and Their Families

Literature Search

Search terms were adapted for the following electronic databases: MEDLINE^® Ovid, Ovid MEDLINE^® In-Process & Other Non-Indexed Citations, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews (CDSR), Database of Abstracts of Reviews of Effects (DARE), HealthSTAR, EMBASE, CINAHL^®, ERIC, PsycINFO^®, CINAHL Plus with Full Text (EBSCO), Science Citation Index Expanded^® and Social Sciences Citation Index^® (both via ISI Web of Knowledge^SM), PubMed^®, LILACS (Latin American and Caribbean Health Science Literature), Proquest^® Dissertations & Theses, CRISP (Computer Retrieval of Information on Scientific Projects), National Library of Medicine (NLM) Gateway, OCLC ProceedingsFirst and PapersFirst, and trial registries such as The National Research Register, ClinicalTrials.gov, and Current Controlled Trials. We also searched websites of relevant professional associations, and reference lists of relevant reviews and included studies. Only English-language studies were included.

Study Selection

Two reviewers independently screened titles and abstracts to determine if an article met general inclusion criteria. The full text of all articles identified as “include” or “unclear” was retrieved for formal review.

Using a priori inclusion criteria, two reviewers independently assessed each full text article using a standard form. Disagreements were resolved by consensus or third-party adjudication.

Quality Assessment

Two reviewers independently assessed the methodological quality of included studies. The Jadad Scale and Schulz criteria for allocation concealment were used to assess the methodological quality of randomized and nonrandomized controlled clinical trials. In a post hoc assessment, trials were given credit if outcome assessors were blinded to outcomes. The Thomas Quality Assessment Tool for Quantitative Studies was used to assess studies of other designs. In addition, the funding source was recorded.

Data Extraction

Data were extracted by one reviewer using standardized forms and checked for accuracy and completeness by a second reviewer. Extracted data included inclusion/exclusion criteria, and the characteristics of participants, interventions, and outcomes. Disagreements were resolved by consensus or third-party adjudication.

Data Analysis

Due to extreme heterogeneity in study designs, interventions, populations, and outcomes, no meta-analyses were performed. Interventions were grouped into nine broad categories: general diabetes education (21 studies), interventions based on cognitive behavioral therapy (24 studies), family therapy (9 studies), skills training (7 studies), and programs delivered at diabetes camps (17 studies), as well as psychoeducation (2 studies) and a physical activity program (1 study). Results are presented first by outcome (e.g. HbA1c), then by intervention, and then by population subgroup (general population, children with newly diagnosed diabetes, or children with poor metabolic control).

Included Studies

As a result of the search, 12,756 citations were identified. One hundred articles were included in the review, representing 80 unique primary studies.⁵–⁸⁴ The number of enrolled participants in the studies ranged from 11 to 332 (median = 50 [IQR = 30 to 89]). The mean age of study participants ranged from 2.7 to 16 years.

Outcomes

HbA1c. 52 studies assessed the effectiveness of diabetes education programs in controlling HbA1c levels (Table 1). Overall, the methodological quality of studies was low, with three RCTs assessed as high quality and nine studies assessed as moderate quality. Thirty-one studies assessed the general population of children with diabetes. Nine studies reported that HbA1c levels decreased significantly following the interventions. The successful interventions fell into the categories of family therapy (n = 3), cognitive behavioral therapy (n = 4), general diabetes education (n = 1), and skills training (n = 1). The one moderate quality RCT examined the addition of a coping skills intervention to intensive diabetes management. Six studies found that HbA1c levels improved significantly for both the intervention and control groups; however, differences between groups were not statistically significant. The remaining 16 studies reported that diabetes education had no significant effect on improving HbA1c levels, thus in the majority of the studies there was no evidence for increase in effectiveness of the educational interventions over the education provided in standard care.

Eight studies assessed diabetes education delivered to children with newly diagnosed diabetes and their families (Table 1). Three studies assessed general diabetes education or family therapy programs that were delivered in an ambulatory setting vs. education delivered during hospitalization. One study reported that the decrease in HbA1c levels was significantly greater for the home-based group vs. the inpatient group, whereas the other two found no significant differences in HbA1c levels between groups. The results of the remaining studies were inconsistent. Three studies reported that cognitive behavioral therapy and skills training interventions resulted in lower HbA1c levels in the intervention group, one study reported that both groups improved, and one study found no difference between the two groups.

Thirteen studies examined HbA1 in children with poorly controlled diabetes (Table 1). Two large, high quality RCTs examining general diabetes education and family therapy reported that the intervention had no significant impact on HbA1c levels. The results of the remaining studies were inconsistent. Four studies reported that HbA1c levels improved following the intervention. Two studies reported an improvement in HbA1c in both intervention and control groups, but there was no significant difference between groups. The remaining five studies reported that diabetes education had no significant effect on HbA1c levels. Due to the heterogeneity across studies and the general low methodological quality, it is difficult to conclude which interventions may have an effect over and above standard care.

Health service utilization. Eleven studies assessed impact of diabetes education on health service utilization (Table 1). Measures of health service utilization included length of stay and hospital or ED admissions for diabetes- and non-diabetes-related complications. Overall, the methodological quality was moderate with one RCT assessed as high quality, six rated as moderate, and four rated as low quality.

The four studies assessing this outcome in the general patient population provide some evidence that diabetes education has an impact on utilization (Table 1). However, there is considerable heterogeneity across studies making it difficult to generalize the results.

Three of the four studies that examined children with newly diagnosed diabetes reported some reduction in health services utilization following the intervention. The results of the three studies that targeted children with poor metabolic control were inconsistent. The high quality RCT reported increased hospital admission rates for both intervention and control groups (i.e., no improvement), while the two low quality trials reported significantly fewer hospital admissions in the intervention group compared to the control group.

Complications. Fifteen studies examined the effect of diabetes education in controlling complications (Table 1). Most studies reported on the incidence of severe hypoglycemia; six reported on the incidence of DKA. Six of the studies were of moderate quality and the other nine were assessed as low quality.

Of the 10 studies that examined the general population of children with diabetes, six found that the intervention had a significant effect. The effective interventions included general diabetes education (n = 2), cognitive behavioral therapy (n = 2), and skills training (n = 2).The remaining studies either showed that there was no improvement or that there was no significant difference between groups for this outcome.

Of three studies assessing this outcome in children with newly diagnosed diabetes, two reported a significant improvement in the intervention group. The third study comparing inpatient vs. outpatient delivery of education found no difference between groups.

The remaining 2 studies assessed children with poor metabolic control and the results both favored the interventions (general diabetes education and diabetes camp).

Knowledge. Thirty studies assessed the effectiveness of diabetes education in improving knowledge. Most studies (n = 20) were assessed as low quality, however, 10 were evaluated as moderate and one RCT as high quality (Table 1). Twenty-four studies examined the general population of children with diabetes. Interventions included general diabetes education (n = 8), cognitive behavioral therapy (n = 6), diabetes camps (n = 7) and skills training (n = 2). Two before-after studies of moderate quality studies reported gains in knowledge after a cognitive behavioral therapy and general diabetes education intervention; however, three other moderate quality before-after studies did not find a significant change following the intervention. The results of the remaining studies were also inconsistent: five reported statistically significant increases in knowledge; two reported found the interventions to be effective in subgroups of the study population; six reported knowledge gains that were not statistically significant or were not retained over the long-term.

Three moderate quality studies compared inpatient vs. ambulatory education in children with newly diagnosed diabetes. All three studies reported increases in knowledge levels for both groups, but differences between groups were not statistically significant. It should be noted that the interventions and outcome measures among studies were different from one another, so results may not be generalizable.

Three studies assessed knowledge among children with poor metabolic control. One high quality RCT did not find a significant change in knowledge for either group. One before-after study reported significant increases in knowledge following a diabetes camp, and one CCT reported significantly higher knowledge levels over the short-term but these gains were not sustained over the long-term.

Skills. Nine studies assessed the effect of diabetes education on the development of diabetes management skills, including self-monitoring of blood glucose (SMBG), nutrition and diet-related skills, and urine testing (Table 1). In the general population of children with diabetes, four studies (2 cognitive behavioral therapy⁴⁸ and 2 camp interventions) found gains in these skills, while four reported no significant change in skills following the intervention.

In the population of children with poor metabolic control, one RCT conducted at diabetes camp found no significant difference between intervention and control groups in SMBG.

Self-management/adherence. Twenty-one studies assessed self management and regimen adherence (Table 1). Of these, fourteen focused on the general population of children with diabetes. One moderate quality and seven low quality studies reported a significant improvement in self-management in the intervention group. Successful interventions included general diabetes education (n = 2), cognitive behavioral therapy (n = 3) and family therapy (n = 2). The remaining studies did not show a significant change.

One study assessed this outcome among children who were newly diagnosed with diabetes and found no significant change. Five studies focused on children with poor metabolic control. One RCT found a significant effect on self-management in blood testing and adherence; the remaining studies found no significant change.

Psychosocial outcomes. Thirty-nine studies examined one or more psychosocial outcomes, including family or social relationships, family or social support, social skills, coping, self-perception, self-efficacy, stress, depression and anxiety (Table 1). Diabetes education was effective in improving several psychosocial outcomes; however, the methodological quality of the studies was generally low and there was considerable heterogeneity across interventions, time points, and measures used.

Quality of life. Two studies found that cognitive behavioral therapy and general diabetes training intervention improved quality of life in the intervention group (Table 1). The remaining two studies that examined this outcome reported that the education interventions had no significant effect.

School performance. There was limited evidence relating to this outcome (Table 1). In the two studies assessing this outcome, diabetes education programs did not have a significant effect on school absence or sick days.

Future Research Opportunities

Recommendations for further research include:

• Appropriately powered RCTs are needed to assess cognitive behavioral therapy, family therapy, motivational interviewing and frequency of contact with health care professionals on HbA1c and short-term complications, especially in the population of children with poorly controlled diabetes

• RCTs should specify the components of education that is part of standard care and followup. A survey of standard diabetes education programs so that researchers are aware of the diversity of standard care would be a useful addition to the literature.

• Additional research is needed in the area of medical nutrition therapy education.

• Future studies assessing the effect of educational strategies on quality of life should use a standard, validated outcome measure.

• Five- to ten-year followup will help to assess the effect of education on long-term complications of diabetes.

• RCTs and CCTs should blind outcome assessors to the intervention and should report allocation concealment.

• Additional research is needed to examine the aspects of education that improve outcomes associated with intensive diabetes management.

Acute Complications of Type 1 Diabetes

Type 1 diabetes and its treatment has two major acute complications: DKA and hypoglycemia. DKA is a metabolic state resulting from acute hyperglycemia. DKA has a mortality rate of 0.5 percent, mostly due to cerebral edema, the most frequent diabetes-related cause of death. DKA is most common at presentation, occurring in an average of 40 percent (15 to 83 percent in population studies) of children presenting with diabetes. In established diabetes, the rate is 1 to 8 percent per year. Risk factors include infection, insulin omission, and equipment malfunction. DKA is treated with immediate hospitalization, insulin replacement, and rehydration.

Hypoglycemia is a complication of insulin treatment. Symptoms caused by a fall in blood glucose include shakiness and emotional instability. In severe cases, there may be seizures or unconsciousness. There has been concern about possible brain dysfunction due to prolonged or repeated hypoglycemic episodes; however, there is limited evidence of permanent cognitive sequelae and they are considered minor.^88,89 The prevalence rates for this complication vary due to potential under-reporting of minor episodes. Studies looking at the prevalence of severe hypoglycemia in children and adolescents report a range of 4 to 86 episodes per 100 patient years.^3,90 Hypoglycemia is most frequent at night, is most serious in younger children due to their relatively higher rate of glucose utilization, and is a possible cause of death. It can be avoided with education about symptoms, careful meal planning, and nighttime glucose monitoring.

Chronic Complications

Chronic complications associated with type 1 diabetes include microvascular complications such as retinopathy, nephropathy, and neuropathy, and macrovascular complications. Macrovascular complications include circulatory and cardiovascular events such as stroke and myocardial infarction, which are rare in children and adolescents; however, risk factors such as hypertension, smoking and dislipidemia should be managed. Chronic complications have been linked to poor glycemic control and the duration of the disease.⁹¹ Many chronic complications are rare in childhood, but management of diabetes in childhood has implications for later development of complications. Donaghue et al.⁹² found that although the survival-free period of retinopathy and microalbinuria was significantly longer for those diagnosed before 5 years of age compared with those diagnosed later, the risk of clinical retinopathy increased by 28 percent for every prepubertal year of duration and by 36 percent for every post-pubertal year of duration. However, there has been a declining incidence of some of the longterm complications over recent decades,^93,94 likely due to improvements in diabetes management.

Diabetes Management and Education

Type 1 diabetes is managed by a combination of insulin replacement and balancing of diet and exercise in order to maintain glycemic control and prevent the occurrence of complications. Glycemic control, which is directly linked to complication rates,⁹¹ is monitored by the measurement of glycosylated hemoglobin (HbA1c), which reflects the mean blood glucose level over the previous 2 to 3 months. Lowering HbA1c has been associated with a reduction of microvascular and neuropathic complications of diabetes.³

It is generally accepted that in order to effectively manage diabetes, education about components of management such as blood glucose monitoring, insulin replacement, diet, exercise, and problem solving strategies must be delivered to the patient and family. Education seems necessary both at diagnosis, where there is usually no knowledge base and patient and family are given the basic skills for controlling the disease,⁴ and throughout the patient's lifetime, with ongoing attention to self-management skills, screening and prevention of complications, and new developments in these areas. Since management of diabetes requires lifestyle changes, most clinicians feel it is important for education to be delivered to the whole family. The following report attempts to determine whether there is evidence to support the general belief that diabetes education is necessary and/or beneficial.

Intensive Diabetes Management and DCCT

The Diabetes Control and Complications Trial (DCCT) demonstrated that intensive therapy was highly beneficial in decreasing the incidence of complications in type 1 diabetes.⁹⁸ Intensive therapy included the administration of insulin by injection or pump 3 or more times daily. The dosage was adjusted in accordance with the results of self-monitoring of blood glucose at least 4 times per day, dietary intake, and anticipated exercise. Specific blood glucose concentration goals were set and patients visited the study center each month and were in telephone contact to review and adjust regimens. This was in contrast to conventional treatment, which had one or two daily injections of insulin, daily self-monitoring of urine or blood glucose, education about diet and exercise, and did not usually include daily adjustments in insulin dosage. Goals included absence of symptoms and maintenance of normal growth and development, but not specific blood glucose targets.

Followup demonstrated that the benefit of intensive therapy in decreasing complications was maintained. This trial established a new philosophy of treatment in using a multidisciplinary team approach and using adjustments based on data monitored by the patient. It is of interest to know if families can be educated to use this intensive style of management outside the trial setting.

Educational Interventions

Educational interventions can take many forms. Didactic education, computer games, board games, cognitive behavioral therapy, and telephone calls are some of the possible methods of delivery. The education may be directed at the patient alone, parent alone, the whole family, and even peers. Education delivered at diagnosis is different than the education on self-management that occurs throughout the patient's lifetime. At diagnosis, the skills needed to manage the disease are first introduced. Later, education may be needed to adjust to the ongoing challenges of developmental changes with a chronic disease, and to keep apprised of new treatments. The aim of the educational interventions may be, among other things, to improve metabolic control, reduce complications, gain skills in self-management, or improve quality of life.

Reviews of educational and psychosocial interventions for adults with diabetes have been conducted and have shown beneficial effects.⁴ Those reviews that have examined programs targeted to meet the particular needs of children have primarily focused on the adolescent population.^4,99 The systematic review by Grey et al. (2000)²¹ appears to be the only one that has assessed diabetes education for both children and adolescents.²¹ The review by Gage et al. (2004)⁴ found that most interventions could be categorized into programs focusing on knowledge/skills, psychosocial issues, and behavior/self-management. They found that there were modest improvements across outcomes such as behavior and metabolic control but that there was little evidence regarding their long-term effectiveness. They also reported that hospital inpatient education at diagnosis was not significantly more effective than home based education and suggested that education may be most beneficial in patients whose metabolic control is poor. The review on psychological interventions by Winkley et al. (2006)¹⁰⁰ found that psychological treatments such as supportive or counseling therapy, cognitive behavior therapy, psychoanalytically informed therapies, and family systems therapy improved glycemic control in children and adolescents with diabetes but had no effect in adults. The review by Grey et al.,²¹ which examined education in children specifically, concluded that educational interventions were useful in improving diabetes knowledge, but they were not consistently useful in improving metabolic control. They also reported that psychosocial interventions such as coping skills training helped adolescents to improve adjustment and metabolic control, and that family interventions may be helpful in reducing parent-child conflict about diabetes management.

The Key Questions

The key questions for this Task Order are:

• 1

  What is the evidence that diabetes education on day-to-day management of diabetes improves metabolic control (as determined by glycosylated hemoglobin⁷⁶ [HbA1c] values, numbers of diabetes-related hospitalizations, frequency of diabetic ketoacidosis [DKA] and numbers of episodes of hypoglycemia)?

• 2

  What is the evidence that medical nutrition therapy education in day-to-day management of diabetes improves HbA1c values and results in less variability in blood glucose levels?

• 3

  What is the evidence that diabetes education results in improved long-term management of diabetes, including better adherence to recommendations made in clinic and decreased hospitalizations and emergency room visits for diabetes-related complications?

• 4

  What is the evidence that diabetes education programs improve knowledge about diabetes management?

  • a

    What is the evidence that this knowledge increases the child's self-confidence in his or her ability to handle the disease and has a positive impact on the child's quality of life (QOL) and other psychosocial issues (e.g., school absences, school performance, adherence to a medical regimen)?

  • b

    What is the evidence that this knowledge improves long-term metabolic control (i.e., decreases or prevents diabetes-related complications), as shown in the Diabetes Control and Complications Trial (DCCT) (as measured by retinal, renal, cardiovascular and neurological evaluations), in children of families who receive these diabetes education or medical nutrition therapy program services compared to children of families who do not receive these services?

• 5

  What is the evidence that training in intensive diabetes management (consistent with DCCT, including blood glucose monitoring at least 4 times a day, 3 or more daily insulin injections or use of an insulin pump and education on when and how to adjust insulin doses) conducted in the practitioner setting yields:

  • a

    Improved metabolic control, (as determined by HbA1c values, numbers of diabetes-related hospitalizations, frequency of DKA and numbers of episodes of hypoglycemia)?

  • b

    A decrease in or prevention of diabetes-related complications (as measured by retinal, renal, cardiovascular and neurological evaluations), as demonstrated by DCCT?

Randomized Controlled Trials

Overall, the methodological quality of the 42 RCTs was low (median Jadad¹⁰¹ score = 2/5; IQR, 1 to 2). The Jadad score was 3 for 4 trials,^30,55,57,83, 2 for 31 trials^5,6,10,16–^18,21,24–^26,31–^{33,36,38,43,46,47,56,60,72,8213,15,22,23,58,59,70,71,81} and 1 for 7 trials^{8,9,49,52,63,79,80} (Table 3). All were described as randomized; however, the descriptions of randomization varied. Five were adequately randomized;^30,32,55–^57,83 the rest were unclear. Two trials reported clear concealment of allocation^30,83 and the remaining trials were unclear. No trials were described as double-blind (patient and outcome assessor blinded). Withdrawals and dropouts were clearly described in 33 trials; in 9 studies, withdrawals and dropouts were not described.^{8,9,32,49,52,56,63,79,80} The post hoc assessment of blinding revealed that 11 RCTs had included an attempt to blind outcome assessors.^{17,18,21,23,26,52,56,56,59,71,72,80}

The most common source of funding was government (n = 24, 57 percent of RCTs). Three trials (7 percent) received funding from industry.^9,25,43 Other sources of funding included foundations (n = 10, 24 percent), internal (n = 7, 17 percent), and private (n = 4, 10 percent). For nine trials (21 percent), the funding is not known.^{13,15,26,32,36,49,79}–⁸¹ Several trials received funding from multiple sources (Table 3).^{5,9,16,17,21,33,38,52,71}

Controlled Clinical Trials

The methodological quality of the 11 CCTs was low (median modified Jadad¹⁰¹ score = 1/3; IQR, 0 to 1) (Table 4). Seven CCTs (64 percent) obtained only 1 point for the individual components of the Jadad scale, all for the description of withdrawals and dropouts. The remaining four trials did not score any points.^42,45,51,84

The CCTs received funding from government sources (n = 2), foundations (n = 2), internal (n = 2), and private sources (n = 2) (Table 4). One trial received funding from multiple sources.⁵¹ For six studies (55 percent), the funding source is not known.^{7,42,45,48,62,84}

Before and After Studies

The quality of reporting of the 20 before-and-after studies was evaluated using individual components of the Thomas tool.¹⁰⁵ Overall, the methodological quality of the before-and-after studies was weak (Table 5 and Table F3, Appendix F).^* For 10 (50 percent) studies participants were unlikely to be representative of the target population and were rated as weak on the selection bias component. The remaining studies were rated as moderate. Control of potential confounders (age, sex, duration of diabetes, socioeconomic status) was rated as weak for 13 (65 percent) studies. None of the studies reported the blinding of outcome assessors to the intervention. Data collection methods were rated as strong for 12 (60 percent) studies. For the remaining eight, the reliability and validity of the data collection tools was not reported. The risk of attrition bias was moderate with 14 (70 percent) studies rated as strong or moderate on this component. Data analysis was appropriate in most studies. None of the studies provided a sample size calculation. Nine studies reported a statistically significant difference between groups for the primary outcome, five studies found no significant differences, and six studies did not report whether group differences were statistically significant. Overall, the intervention integrity was moderate. For 19 (95 percent) studies, over 80 percent of participants received the allocated intervention. Methods to assess the consistency of the intervention were included in eight (40 percent) studies; for the remainder they were not reported. For most studies (n = 14, 70 percent), it was unlikely that contamination occurred or that there was a cointervention. For the remaining six, there was insufficient information to answer this question. Finally, funding source was disclosed in 70 percent (n = 14) of the before-and-after studies (Table 5).

Cohort Studies

The quality of reporting of the seven cohort studies was evaluated with individual components of the Thomas tool.¹⁰⁵ (Table 5 and Table F3, Appendix F).^* Overall, the methodological quality was moderate. Most studies did a reasonable job of protecting against selection bias with five of seven (71 percent) choosing study groups that were at least somewhat representative of the target population. The risk of attrition bias was low; all but one study included a description of the number of withdrawals and the followup rates were unlikely to introduce differences between the comparison groups. Data collection methods were assessed as valid and reliable for six studies (86 percent). Control of potential confounders (age, sex, duration of diabetes, socioeconomic status) either in the design or analysis was rated as weak for six (86 percent) studies and as moderate in one study (14 percent). No studies reported that the ascertainment of outcome exposure was blinded. For all studies data analysis was considered appropriate. None of the studies provided a sample size calculation; however, four studies (57 percent) reported a statistically significant difference between groups for the primary outcome. Intervention integrity was good. In six studies (86 percent), over 80 percent of participants received the allocated intervention and the risk of contamination or cointervention was low. Finally, three of the cohort studies (43 percent) reported the source of funding (Table 5).

General Diabetes Education—HbA1c

Description of studies. We identified 12 studies (7 RCTs,^{9,15,18,24,32,55,79} 2 cohorts,^39,69 3 uncontrolled before-and-after^61,75,77) that assessed the effectiveness of general diabetes education on HbA1c. Studies were conducted in the United States,^9,32 Europe,^{24,61,75,77,79} Canada,^15,18,39 and Australia.^55,69 The median year of publication was 2003 (IQR, 1995 to 2005).

The number of participants enrolled in the studies ranged from 28 to 146 (median = 65 [IQR 35 to 109]). The mean age of participants ranged from 10 to 15.8 years (n = 9 studies). For 3 studies, the mean age was less than 12 years.^18,55,77 Two studies examined education interventions delivered to children and their families at the time of diagnosis.^18,69 Two studies focused on children who demonstrated poor metabolic control.^15,55

Most interventions were targeted to children and their parents or to the entire family;^{9,15,18,32,39,55,61,69,75,77,79} one was delivered to the child.²⁴ The settings for the interventions were described as a hospital inpatient unit,^61,77 outpatient clinic,^24,79 or home.⁹ For 5 studies, the setting was mixed (e.g., diabetes center and home).^{18,32,39,55,69} The setting was not clearly described in 2 studies.^15,75

Five studies did not measure HbA1c beyond the completion of the education program.^{9,24,32,55,79} In the 7 studies that reported post-intervention assessments,^{15,18,39,61,69,75,77} the median followup period was 12 months and ranged from 3.5 to 36 months.

Results

General population of children with diabetes. Howe et al.³² randomly assigned 75 patients to one of three treatment groups: standard care (routine quarterly clinic visits), standard care plus one education session on basic diabetes management skills, and standard care plus the education session plus weekly telephone calls to review management techniques. At 6 months, mean HbA1c levels dropped slightly from baseline (0.3 percent, 0.4 percent, and 0.5 percent, respectively). The differences were not statistically significant; however, they do represent a clinical improvement for the education plus telephone call group. The trial was ended early due to lack of enrolment. The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

The retrospective cohort study (n = 28) by Lawson et al.³⁹ compared intensive, individualized education to group education delivered to patients. The content for both interventions was the same. Both groups showed a statistically significant decrease in mean HbA1c at 3 months; however, over the next 12 months the HbA1c levels increased for both groups. For the individualized education cohort, the mean HbA1c remained significantly lower than at initiation of the program (9.5±0.3 vs. 8.2±0.4 percent, p = 0.001); for the group education cohort, the mean HbA1c was not significantly different than baseline (8.2±0.4 vs. 8.1±0.3 percent). However, the baseline levels for the group education cohort were significantly lower than for the individual education group (p = 0.02). The methodological quality of this study was rated as weak using the Thomas instrument.

Povlsen et al.⁶¹ assessed the effect of an intervention targeted at 37 families from ethnic minority groups in Denmark. The intervention included adapted educational material and guidelines and re-education that focused on increasing knowledge and self-care. The authors reported a significant difference in mean HbA1c levels immediately after the 12-month intervention (9.2±1.4 vs. 8.6±1.0 percent, p = 0.01); however, this improvement disappeared at the 6-month followup. The methodological quality of this study was rated as weak using the Thomas instrument.

In an uncontrolled before-and-after study, von Sengbusch et al.⁷⁷ assessed the impact of a mobile diabetes education service in a group of 107 children. Overall HbA1c levels for the whole cohort did not change significantly. The researchers conducted a post hoc subgroup analysis and found that among children with poor metabolic control (defined as HbA1c >8.0 percent), there was improved metabolic control from baseline to 6-month followup (9.4±0.9 vs. 8.6±1.5 percent, p<0.01). The methodological quality of this study was rated as moderate using the Thomas instrument.

Brown et al.⁹ randomly assigned 59 children and adolescents to a group that played a video game featuring characters with diabetes who manage their diabetes by monitoring blood glucose, taking insulin injections, and choosing foods, or a group that played a pinball video game with no information on diabetes. At the end of the 6-month study period, both groups exhibited higher levels of HbA1c (i.e., poorer metabolic control). There was no statistically significant difference between groups. The methodological quality of this study was rated as low (1/5 on the Jadad score; unclear allocation concealment).

In the RCT by Hackett et al.,²⁴ three cohorts of families received educational packages that were delivered over 8 months. The content was the same for all groups and included a nutrition therapy education component; one group (cohort 1) received a second reinforcement package for an additional 8 months. The comparison group did not receive the education package. Overall, the education program did not have an impact on HbA1c levels. The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Wadham et al.⁷⁹ conducted an RCT (n = 67) to evaluate a 4-session family centered, structured education program for adolescents and parents. The program focused on skill-based sessions and teamwork and communication between parents and adolescents. The control group received routine clinical care. Preliminary results at 6 months did not show any significant change in HbA1c levels for either group. The methodological quality of this study was rated as low (1/5 on the Jadad score; unclear allocation concealment).

In an uncontrolled before-and-after study Verrotti et al.⁷⁵ studied 30 adolescents who attended nine education sessions on general diabetes management. At the 12-month followup, mean baseline levels of HbA1c had decreased significantly (11.8±2.8 vs. 10.0±2.7 percent, p = 0.019). The methodological quality of this study was rated as moderate using the Thomas instrument.

Children with newly diagnosed diabetes. Dougherty et al.¹⁸ randomly assigned 63 newly diagnosed patients to a 24-month home-based diabetes education program or to traditional hospitalization and outpatient followup. Treatment differences between the groups consisted of duration of initial hospital stay, timing of initial teaching, and the nature and extent of subsequent nursing followup. Both groups exhibited significantly lower mean HbA1c values at the 3-month followup. These differences were still present at 36 months, 1 year after the intervention ended (10.5±1.3 vs. 6.4±1.4 percent for the home-based group and 10.0±1.3 vs. 7.1±1.3 percent for the hospital-based group). The difference between the two groups at 36 months was statistically significant. The methodological quality of this study was rated as moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

In the prospective cohort study by Srivivasan et al.⁶⁹ a group of 61 newly diagnosed patients attended a diabetes day care program (DDCP) at which they received “survival skills” diabetes education. The pre-DDCP cohort comprised 49 patients who were admitted to hospital for 4 to 7 days for a detailed education program. Neither group demonstrated improvement in HbA1c levels at 3, 6 or 12-month followup. The methodological quality of this study was rated as moderate using the Thomas instrument.

Children with poorly controlled diabetes. Nunn et al.⁵⁵ randomly assigned 123 patients to receive scheduled telephone calls from a pediatric diabetes educator or standard care (i.e., routine clinic visits but no telephone calls). Over the course of the 8-month study HbA1c levels increased in both groups (i.e., poorer metabolic control). There was no statistically significant difference between groups. The methodological quality of this study was rated as high (3/5 on the Jadad score; unclear allocation concealment).

In the RCT by Coupland¹⁵ adolescents and their families participated in a family-based intervention to improve adherence (n = 15); the comparison group (n = 14) comprised adolescents who were taught stress management techniques. At 6-month followup, both the intervention and comparison groups demonstrated significant improvements in mean HbA1c (11.6±1.41 vs. 9.8±2.23 percent and 11.3±1.22 to 10.5±1.57 percent, respectively). There was no statistically significant difference between groups. The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Cognitive Behavioral Therapy—HbA1c

Description of studies. Nineteen studies (14 RCTs,^{5,6,8,13,16,17,21,22,25,31,33,47,72,80} 4 CCTs,^7,14,19,76 1 cohort⁵⁴) examined various interventions that used cognitive behavioral therapy techniques. All but three studies were conducted in the United States; two were conducted in Europe,^54,76 and one was conducted in Australia.¹⁴ The median year of publication was 1992 (IQR, 1989 to 1999) and ranged from 1983 to 2003.

The number of participants enrolled in the studies ranged from 13 to 139 (median = 31 [IQR 20 to 69]). The mean age of participants ranged from 2.7 to 15.4 years (14 studies). For five studies, the mean age was less than 12 years.^{17,19,22,47,72} Two studies examined education interventions delivered to children and their families at the time of diagnosis.^17,19 Six studies focused on children who demonstrated poor metabolic control.^{8,13,14,16,25,76}

In seven studies the interventions were delivered to children;^{7,8,13,14,21,25,31} in two studies they were delivered to parents.^19,33 In the remaining studies parents and children or the entire family were targeted.^{5,6,16,17,22,47,54,72,76,80} The settings for the interventions were described as a diabetes center,^5,7,47 hospital inpatient unit,^6,13,25 outpatient clinic,^8,16,72,80 home,³¹ or summer school.³³ In four studies, the setting was mixed (e.g., diabetes center and home).^14,17,21,54 Two studies did not report the setting of the intervention.^22,76 All but three studies^6,19,31 conducted post-intervention assessments for HbA1c. The median followup period was 4 months and ranged from 1 month to 14 months.

Results

General population of children with diabetes. Anderson et al.⁶ randomized 70 adolescents and parents to a group that received standard care (routine clinic visits) or a group that received standard care plus a problem solving intervention that focused on self-monitoring of blood glucose (SMBG) around meal planning. At completion of the 18-month program, the intervention group exhibited a mean decrease of 0.37 percent in HbA1c levels compared with a mean increase of 0.62 percent in the standard care group. This difference was statistically significant (p = 0.04). The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

In a subsequent study, Anderson et al.⁵ assessed the effectiveness of an intervention for families that focused on teamwork and shared parent-teen responsibility for diabetes tasks. Eighty-nine families were randomly assigned to the intervention group or to one of two comparison groups: standard care (routine clinical care from the diabetes team) or standard care plus didactic diabetes education. At the end of the 12-month program, there was no significant difference in mean HbA1c among the three groups. While the intervention group demonstrated improvement in HbA1c levels at the 12-month followup, there was no significant difference in change scores between the groups (-0.20±1.1 percent for the teamwork group vs. 0.11±1.1 percent for the combined comparison groups). The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Grey et al.²¹ randomly assigned 77 adolescents to receive intensive diabetes management (IDM) as described in the DCCT or IDM plus a behavioral program of coping skills training intervention. At the 12-month followup, both groups reported significant decreases in HbA1c levels. The change from baseline measures of HbA1c for the intervention group was greater than the control group and the difference was statistically significant (9.1±1.5 vs. 7.5±1.1 percent and 9.2±1.4 vs. 8.5±1.4 percent, respectively; p<0.001). The methodological quality of this study was rated as moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

In the study by Kaplan et al.,³³ 21 patients were randomly assigned to a group that received social skills training or a group that received lectures on diabetes. At the 4-month followup, the social skills group reported improved HbA1c levels compared with the comparison group (12.6±2.4 vs. 11.72 percent [SD not reported] and 13.5±1.6 vs. 14.42 percent [SD not reported]; p<0.05). The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

In a prospective study with historical controls, Nordfeldt and Ludvigsson⁵⁴ examined the effect of multiple dose insulin therapy combined with problem-based training and psychosocial support. The prospective cohort comprised 248 children admitted to hospital in 1994 and 1995. The comparison group (156 patients admitted in 1980-81) received standard clinical care. The researchers reported that the annual mean levels of HbA1c were significantly lower in the cohorts that received intensive diabetes training compared with the historical cohort that did not receive the same level of education and support (6.9±1.3 percent [1994], 7.1±1.1 percent [1995] and 7.4±1.2 [1980-81]; p = 0.004). The methodological quality of this study was rated as weak using the Thomas instrument.

In the study by McNabb et al.⁴⁷ 24 children were randomly assigned to either a 6-week self-management education program or to receive standard care (routine clinic visits). At the 6-week followup, HbA1c levels had decreased from baseline in both groups; however, the difference was not statistically significant (10.5±2.9 vs. 9.6±1.8 percent and 12.9±3.8 vs. 12.5±3.4 percent, respectively). The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Barglow et al.⁷ conducted a CCT (n = 42) to assess the effectiveness of a 4-month intensive multicomponent treatment and education program. The comparison group received standard care (routine clinic visits). At the 4-month followup both groups reported lower levels of HbA1c (change from baseline = -2.45±2.3 percent and -0.85±3.5 percent, respectively); however, the difference was not statistically significant. The methodological quality of this study was rated as low (1/3 on the modified Jadad score).

In the RCT by Gross et al.²² 14 children and parents were randomly assigned to either behavior modification training or to a group that included discussion and role-playing. Both the intervention and comparison groups exhibited improvement in metabolic control at the 6-month followup (13.0 vs. 12.5 percent and 13.4 vs. 11.0 percent, respectively). There was no statistically significant difference between the groups at followup. The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Szumowski⁷² randomly assigned 27 young children to an 8-week behavioral intervention that included information on diabetes management plus instruction and practice in the application of behavioral principles and goal setting to reinforce children's regimen adherence, or to the comparison group that received information on diabetes management but no additional instruction. At the 3-month followup there were no statistically significant changes from baseline HbA1c levels for either group. The methodological quality of this study was rated as moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

Horan et al.³¹ conducted an RCT (n = 20) that compared goal setting and problem solving using dynamic computer-assisted teaching modules vs. conventional education using an education booklet. At the end of the 15-week program, there were no significant changes in HbA1c levels in either group. The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

In the RCT by Webb,⁸⁰ 45 families were assigned to a group that received intensive collaborative goal setting training or a group that used a goal setting worksheet with guidance from a therapist. At the 3-month followup neither group showed a significant change from baseline HbAc1 levels. The methodological quality of this study was rated as low (1/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

Children with newly diagnosed diabetes. In the CCT by Golden et al.¹⁹ 11 families received an integrated diabetes education and psychosocial support program that was delivered in a motel-like setting. The comparison group (n = 8) received the same training 15 months after diagnosis. Overall, the children whose families received immediate education achieved lower levels of HbA1c at all time points during the study up to 24 months after diagnosis (p<0.05). Baseline levels of HbA1c were not reported; therefore, we could not assess whether there was a significant change from baseline for either group. The methodological quality of this study was rated as low (1/3 on the modified Jadad score).

Delamater et al.¹⁷ conducted an RCT (n = 36) that assessed the impact of a family-based self-management training (SMT) program conducted over the 6 months following diagnosis of diabetes. The description of the intervention included a medical nutrition therapy education component. One comparison group received standard care (regular outpatient contact with the health care team); a second comparison group received standard care plus supportive counseling. HbA1c measures taken at 12 and 24 months after diagnosis showed that all three groups improved their metabolic control. The SMT group had significantly lower HbA1c levels than the standard care patients at both 1 and 2 years after diagnosis (10.4±3.1 vs. 8.2±1.5 percent and 12.3±2.5 vs. 9.8±2.4 percent, respectively at 2 years). The SMT group also had lower levels than the supportive care group, but this difference was not statistically significant (10.4±3.1 vs. 8.2±1.5 percent and 11.1±2.4 vs. 9.1±1.7 percent, respectively at 2 years). The methodological quality of this study was rated as moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

Children with poorly controlled diabetes. In a CCT by Viner et al.,⁷⁶ a 6-week motivational and solution-focused therapy group program was delivered to 21 adolescents and their parents. The control group comprised 20 adolescents who did not receive any intervention. At the 12-month followup, the intervention group demonstrated statistically significant decreases in HbA1c levels compared with the control group (10.2±1.37 vs. 8.9±1.37 percent and 10.0±1.34 vs. 9.9±2.24 percent, respectively). The methodological quality of this study was rated as low (1/3 on the modified Jadad score).

In a three-arm RCT (n = 37), Cigrang¹³ investigated the effects of a coping skills program delivered to adolescents with a history of poor metabolic control. There were two comparison groups: conventional diabetes education and standard care (routine clinic visits). At the 3-month followup, HbA1c levels decreased for all three groups. The change from baseline for each group was clinically significant (1.06±1.31 percent, 0.94±2.25 percent, 0.92±1.74 percent, respectively); however, there were no statistically significant differences among the three groups. It is likely that the study did not have sufficient power to detect a difference among the groups. The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Couper et al.¹⁴ evaluated the effects of monthly home visits by a nurse educator plus weekly telephone calls that focused on goal-setting. Thirty-seven adolescents received routine care plus the intervention; 32 received routine clinical care only. At the 6-month followup, the intervention group reported statistically significant lower mean levels of HbA1c compared with the comparison group (11.1±1.3 vs. 9.7±1.6 percent and 10.5±1.6 vs. 10.3±2.2 percent, respectively [p = 0.0001]). The difference between HbA1c levels at baseline and the 18-month assessment in the intervention group were clinically, but not statistically, significant (11.1±1.3 vs. 10.0±1.5 percent [0.06]^1190.06). The methodological quality of this study was rated as low (1/3 on the modified Jadad score; unclear allocation concealment).

Delamater et al¹⁶ randomized 13 adolescents to receive a 2-month family-based behavior therapy program or to standard outpatient care. The description of this intervention included details of a medical nutrition therapy component. There were no significant changes in HbA1c levels in either group at the 4-month followup (11.5±2.1 vs. 11.0±2.4 percent and 10.4±0.8 vs. 10.3±1.5 percent, respectively). The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

In the RCT (n = 19) by Boardway et al.⁸ a 3-month stress management training program for adolescents was compared to standard outpatient care. The intervention was not effective in reducing HbA1c levels. For the intervention group HbA1c increased (i.e., poorer metabolic control) at the 6-month followup; there was no change in HbA1c levels for the control group (13.98±2.41 vs. 16.4±2.41 percent and 15.75±3.52 vs. 15.69±2.76 percent, respectively). The methodological quality of this study was rated as low (1/5 on the Jadad score; unclear allocation concealment).

Hains et al.²⁵ randomly assigned 15 adolescents to a cognitive behavioral stress training program or to a control group that did not receive any intervention. At 1-month followup there were no significant changes in HbA1c levels for either group. The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Family Therapy—HbA1c

Description of studies. Seven studies (6 RCTs,^{36,38,63,70,82,83} 1 uncontrolled before-and-after²⁸) assessed interventions that focused on family dynamics. Six studies were conducted in the United States; one took place in Europe.⁷⁰ The median year of publication was 2000 and ranged from 1996 to 2007.

The number of participants enrolled in the studies ranged from 18 to 119 (median = 38 [IQR 19 to 105]). The mean age of participants ranged from 6.0 to 14.6 years. In three studies, the mean age was less than 12 years.^28,36,70 Two studies examined education interventions delivered to children and their families at the time of diagnosis.^36,70 Two studies focused on children who demonstrated poor metabolic control.^28,83

The interventions were targeted at the entire family in six studies; in one study the intervention was delivered to the parents only.³⁶ The settings for the interventions were described as a diabetes center,⁸² hospital inpatient unit,⁶³ the doctor's office,⁸³ or home.^28,36 For two studies, the setting was mixed (e.g., diabetes center and home).^38,70 Two studies reported HbA1c levels immediately following the intervention.^38,83 Four studies conducted post-intervention assessments for HbA1c.^28,63,70,82 The median followup period was 6 months and ranged from 3 to 24 months.

Results

General population of children with diabetes. Wysocki et al.⁸² randomized 104 families to one of three groups: standard care plus behavioral family systems therapy (BFST), standard care plus family education and support, or standard care (physician directed clinical care). At the completion of the 6-month intervention, all groups showed improvement in mean HbA1c levels. However, only the BFST group maintained their improved HbA1c levels at 12-month followup, while both comparison groups reverted toward their baseline levels (BFST group: 9.6±1.6 percent at baseline vs. 8.8±1.5 percent at 12 months). The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Satin et al.⁶³ randomly assigned 32 adolescents and their parents to one of three groups. One group received multifamily training on family teamwork (MF). For the second group, families received the same content as group 1 plus simulation activities for parents (MF+S). The control group received no education intervention. At the 6-month followup, the MF+S group demonstrated a statistically significant improvement in HbA1c levels compared to the control group (WMD -1.09 percent; 95% CI, -2.11, -0.07). Similarly the MF group also showed improvement in HbA1c levels; however, the results were not statistically significant (WMD -0.31 percent; 95% CI, -1.40, 0.78). The methodological quality of this study was rated as low (1/5 on the Jadad score; unclear allocation concealment).

Laffel et al.³⁸ randomly assigned 105 families to a family-focused teamwork intervention (TW) or to standard care (i.e., routine multidisciplinary clinical care). At the end of the 12-month program, the TW group showed improved HbA1c levels (8.4±1.3 percent at baseline vs. 8.2±1.1 percent) compared to the control group (8.3±1.0 percent at baseline vs. 8.7±1.5 percent). In a multivariable regression analysis that controlled for age, duration of diabetes, and diabetes management, the change in HbA1c level after 1 year was 0.5 percent better in the TW group than in the control group (R² = 0.17, p = 0.04). The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Children with newly diagnosed diabetes. The RCT by Sundelin et al.⁷⁰ compared conventional inpatient education (n = 19 families) to a multidisciplinary program for family-oriented crisis intervention delivered in an outpatient setting (n = 19 families). There were no statistically significant differences in HbA1c levels between the two groups up to 5 years following diagnosis; both groups demonstrated improved levels in HbA1c levels compared to baseline measures. The mean baseline measure for the intervention group was 9.6±0.14 percent and at 5 years, it was 7.6±15 percent; for the control group, baseline HbA1c level was 9.8±0.74 percent and at 5 years it was 7.2±1.5 percent. The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Kennedy-Iwai³⁶ randomly assigned 19 families to receive standard care (standard clinical treatment and diabetes education) or standard care plus a couple communication program delivered to parents. There was no improvement in HbA1c levels at the post-treatment assessment for children in either group. The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Children with poorly controlled diabetes. Wysocki et al.⁸³ randomly assigned 119 children to 3 groups: standard care (physician directed clinical therapy), education and support (standard care plus 10 sessions of a diabetes support group), and BFST (standard care plus family problem-solving and communication training and individualized treatment plan). At the end of the 3-month intervention none of the groups demonstrated any significant change from their baseline HbA1c levels. The methodological quality of this study was rated as high (3/5 on the Jadad score; adequate allocation concealment).

In an uncontrolled before-and-after study, Harris et al.²⁸ reported 6-month followup data for 18 adolescents and their families who enrolled in a BFST program. HbA1c levels were unchanged from baseline levels. The methodological quality of this study was rated as weak using the Thomas instrument.

Skills Training—HbA1c

Description of studies. Six studies (3 RCTs,^43,49,57 2 cohorts,^40,53 1 uncontrolled before-and-after¹¹) examined various interventions that focused on skills training. Studies were conducted in the United States,¹¹ Canada,^49,57 Europe,^43,53 and Thailand.⁴⁰ The median year of publication was 2001 and ranged from 1983 to 2003.

The number of participants enrolled in the studies ranged from 32 to 130 (median = 51 [IQR 39 to 56]). The mean age of participants ranged from 7.1 to 14.4 years. For three studies, the mean age was less than 12 years.^11,40,49 Two studies examined education interventions delivered to children and their families at the time of diagnosis.^40,49 Two studies focused on children who demonstrated poor metabolic control.^43,57

Most interventions were targeted at children and their parents or the entire family; in one study the intervention was delivered to the child.⁵⁷ The settings for the interventions were described as a diabetes center,⁴⁹ hospital inpatient unit,⁴⁰ or home.⁵³ For three studies, the setting was mixed (e.g., diabetes center and home).^11,43,57 All studies conducted post-intervention assessments for HbA1c. The median followup period for the studies was 18 months and ranged from 6 to 43.5 months.

Results

General population of children with diabetes. Nordfeldt and Ludvigsson⁵³ evaluated the effect of self-study material on diabetes education aimed at self-management skills and the prevention of hypoglycemia. Over a 3-year period (1997 to 1999) brochures and videos were distributed to approximately 130 patients and families each year. HbA1c levels were compared to the mean levels for 1996 (prior to the distribution of the brochures). Mean HbA1c levels in 1997 were statistically significantly lower than those reported in 1996 (6.5±1.1 vs. 6.8±1.2 percent, respectively); similar results were reported comparing 1998 and 1996 (6.4±1.1 vs. 6.8±1.2 percent, respectively). The methodological quality of this study was rated as weak using the Thomas instrument.

In an uncontrolled before-and-after study by Caravalho and Saylor,¹¹ 56 children and their parents were taught insulin adjustment procedures and received group support and education to improve self-management. Among the 38 patients who were assessed at 6 months, there was no significant improvement in HbA1c levels from baseline to post-intervention (9.15±2.32 vs. 8.99±1.79 percent). The methodological quality of this study was rated as weak using the Thomas instrument.

Children with newly diagnosed diabetes. Likitmaskul et al.⁴⁰ conducted a nonconcurrent cohort study to compare the effect of a multidisciplinary team approach to diabetes education to conventional education that focused on insulin injection and how to control diet. The 28 children who received conventional education were diagnosed with diabetes prior to 1996 and served as a historical comparison group to the 24 children who were diagnosed after 1996. The children who received the multidisciplinary education program demonstrated significantly lower levels of HbA1c up to 3 years after diagnosis (12.4±2.7 vs. 13.6±5.4 percent, respectively [p = 0.03]). The methodological quality of this study was rated as moderate using the Thomas instrument.

In the study by Mitchell,⁴⁹ 32 children were randomly assigned to either the intervention or the control group. The intervention group received standard multidisciplinary education and support plus a booklet targeted at improving compliance with treatment. The control group received standard education and support. The intervention group showed a general trend to lower HbA1c values over the 24-month followup; however, the differences were not statistically significant except at 10 to 13 months post-diagnosis. The methodological quality of this study was rated as low (1/5 on the Jadad score; unclear allocation concealment). Both the intervention and control groups had substantial dropouts over the study period (47 and 53 percent, respectively).

Children with poorly controlled diabetes. Children with poorly controlled diabetes were targeted in the RCT by Mann et al.⁴³ Children in the intervention group (n = 19) received intensive diabetes education combined with regular SMBG. Children in the control group (n = 20) received intensive education only. At the end of the study, there was no significant change in HbA1c levels from baseline to 18 months in either group (14.1±1.3 vs. 14.3±1.9 percent for the intervention group; 12.7±2.0 vs. 12.8±2.4 percent for the control group). The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

In the RCT by Panagiotopoulos et al.⁵⁷ 25 children received weekly telephone calls and guidance with SMBG, while 25 children received standard care (i.e., routine clinical care). The authors reported that at 6 months, HbA1c levels decreased significantly from baseline in both the intervention and control groups (9.7±1.2 to 8.8±1.3 percent and 9.6±1.3 to 9.1±1.4 percent, respectively). However, the magnitude of change between the two groups was not statistically significant. In a post hoc subgroup analysis of children with HbA1c levels greater than 9.5 percent at baseline, the authors found that HbA1c levels decreased significantly in the intervention group. Six months after study completion, this subset continued to have improved HbA1c levels compared to the control group (from 10.5±1.0 to 9.3±0.9 vs. 10.6±1.0 to 10.4±1.5, respectively). The methodological quality of this study was rated as high (3/5 on the Jadad score; unclear allocation concealment).

Diabetes Camp—HbA1c

Description of studies. In five studies (1 RCT,⁴⁶ 2 CCTs,^62,84 2 uncontrolled before-and-after studies^34,35) the education intervention was delivered as part of a diabetes camp program. Four studies were conducted in the United States and one took place in Europe.³⁴ The median year of publication was 1997 and ranged from 1986 to 2005.

The number of participants enrolled in the studies ranged from 25 to 237 (median = 42 [IQR 33 to 56]). The mean age of participants ranged from 10.0 to 14.5 years. In two studies, the mean age was less than 12 years.^35,84 One study focused on children who demonstrated poor metabolic control.³⁴ All interventions were delivered to the children. All studies conducted post-intervention assessments for HbA1c; the median followup period was 3.5 months and ranged from 3 to 12 months.

Results

General population of children with diabetes. The specific interventions that were delivered during the diabetes camps were diverse. In the trial by Massouh et al.,⁴⁶ 34 adolescents were randomly assigned to a group that received a daily 1-hour teaching session about diabetes or to a group that received the lecture plus a social learning intervention (relationship skills) that involved role playing. Both groups demonstrated increases (i.e., poorer control) in HbA1c levels at 3.5 months following the end of camp (1.5 percent [p = 0.008] for the intervention group; 1.2 for the comparison group [p = 0.14]). The methodological quality of this study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

Remley⁶² conducted a CCT to compare a social cognitive theory-based program with a standard non-theory based camp education program. Eight 1-week camps across the United States were designated deliver the either the theory-based intervention or the standard program. At the 3-month followup, there was no significant change in HbA1c levels for either group. The attrition rate for both groups was high (40 percent). The methodological quality of this study was rated as low (1/3 on the modified Jadad score; unclear allocation concealment).

Zorumski⁸⁴ investigated the effects of self-care training for 49 children. All children received basic self-care training from their physicians; 27 also attended a 1-week day camp that provided additional self-care instruction. At the 4-month followup, there was no significant change from baseline HbA1c levels for either group (9.93±2.9 vs. 10.49±2.7 percent [intervention group] and 10.85±2.14 vs. 10.47±2.8 percent [comparison group]). The methodological quality of this study was rated as low (0/3 on the modified Jadad score).

In an uncontrolled before-and-after study Kemp et al.³⁵ assessed the effectiveness of diabetes education and carefully monitored blood glucose control among 42 children who attended a 2-week summer camp. One year later, HbA1c levels had increased (i.e., poorer control) from the baseline measure (8.1±1.9 vs. 10.1±1.9 percent). The methodological quality of this study was rated as weak using the Thomas instrument.

Children with poorly controlled diabetes. In an uncontrolled before-and-after study (n = 25), Karaguzel et al.³⁴ examined the effect of a 1-week diabetes camp that incorporated intensive insulin treatment into a general diabetes program (n = 25). There was a statistically significant decrease in HbA1c levels from baseline to 12-months post-intervention (9.3±2.5 vs. 8.2±1.5 percent, p<0.05). The methodological quality of this study was rated as weak using the Thomas instrument.

Other Interventions—HbA1c

General population of children with diabetes. Campaigne et al.¹⁰ assessed the effects of physical training on HbA1c levels. Fourteen adolescents were randomly assigned to either a 12-week exercise program or to a control group that did not modify the usual exercise routine. At the end of the program, HbA1c levels remained unchanged in both groups. The methodological quality of the study was rated as low (2/5 on the Jadad score; unclear allocation concealment).

The RCT by Olmsted et al.⁵⁶ evaluated the effect of a 6-session intervention for eating disorders for young women with disturbed eating habits. Eighty-five patients were randomized to the psychoeducation program or to routine care at the diabetes center. At the 6-month followup neither group demonstrated any significant improvement in HbA1c levels. The methodological quality of the study was moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

Children with poorly controlled diabetes. In a CCT, Moran et al.⁵¹ compared two equivalent groups of 11 children with a history of poor metabolic control. One group received an intensive inpatient psychotherapy program; the comparison group comprised children who were admitted to hospital for medical treatment of diabetes-related complications. Both groups received diabetes education while in hospital. HbA1c levels were significantly improved at the 12-month followup for the intervention group (14.3 vs. 11.5 percent); there was no change in HbA1c levels for the comparison group (13.7 vs. 13.5 percent). The methodological quality of the study was low (0/3 on the modified Jadad score).

Summary of Results—HbA1c

Overall, we identified 52 studies that assessed the effectiveness of diabetes education programs in controlling HbA1c levels (Table 6). Of these, 30 studies (19 RCTs, 3 CCTs, 3 cohorts, 5 uncontrolled before-and-after) examined the general population of children with type 1 diabetes, 8 (5 RCTs, 1 CCT, 2 cohorts) focused on children with newly diagnosed diabetes, and 14 (9 RCTs, 3 CCTs, 2 uncontrolled before-and-after) considered children with poorly controlled diabetes. Due to substantial differences across studies in terms of population, interventions, comparison groups, duration of intervention, and followup time points, we were not able to pool the results of any of the studies.

General population of children with diabetes. Overall, 30 studies (19 RCTs,^{5,6,9,10,21,22,24,31}–^{33,38,46,47,56,63,72,79,80,82} 3 CCTs,^7,62,84 3 cohorts,^39,53,54 5 uncontrolled before-and-after^{11,35,61,75,77}) assessed the effectiveness of diabetes education in controlling HbA1c levels in the general population of children with diabetes (Table 6). Interventions that were assessed included general diabetes education,^{9,24,32,39,61,75,77,79} cognitive behavioral therapy,⁵–^{7,21,22,31,33,47,54,72,80} family therapy,^38,63,82 skills training,^11,53 diabetes camp,^35,46,62,84 physical training,¹⁰ and psychoeducation.⁵⁶ Overall the methodological quality of the studies was low. Three RCTs were rated as moderate (Grey et al.,²¹ Olmsted et al.,⁵⁶ and Szumowski⁷²) and two uncontrolled before-and-after studies (Verrotti,⁷⁵ von Sengbusch et al.⁷⁷) were assessed as moderate.

Results

The results of the studies rated as moderate quality were mixed. Grey et al.²¹ found that the intervention group that received intensive diabetes management plus coping skills training had significantly lower HbA1c levels than the group that received intensive management only. Szumowski⁷² found no change from baseline levels of HbA1c for the intervention group that received goal setting training or the control group. Similarly, Olmsted et al.⁵⁶ reported no difference in HbA1c levels among diabetic patients with eating disorders comparing a psychoeducation intervention with routine care. The uncontrolled before-and-after study by Verrotti et al.⁷⁵ found decreased levels of HbA1c among teenagers who received general diabetes education. In contrast, the uncontrolled before-and-after study by von Sengbusch et al.⁷⁷ assessed a mobile diabetes education program and found no change in HbA1c following the intervention.

The results of the remaining studies were inconsistent. Seven studies^{6,33,38,53,54,63,82} reported that HbA1c levels decreased significantly following the diabetes education intervention. Three of these studies assessed cognitive behavioral therapy,^6,33,54 three studies assessed family therapy programs^38,63,83 and one assessed skills training.⁵³ Six studies^{5,7,22,32,47,61} found that HbA1c levels improved significantly for both the intervention and control groups; however, the differences between groups were not statistically significant. Four of these studies assessed cognitive behavioral therapy^5,7,22,47 and two assessed general diabetes education.^32,61 The remaining studies⁹–^{11,24,31,35,46,62,79,80,84} reported that diabetes education had no significant effect on improving HbA1c levels.

Children With Newly Diagnosed Diabetes

Overall, 8 studies (5 RCTs,^{17,18,36,49,70} 1 CCT,¹⁹ 2 cohorts^40,69 assessed the effectiveness of diabetes education among children with newly diagnosed diabetes (Table 6). Interventions that were assessed were general diabetes education,^18,69 cognitive behavioral therapy,^17,19 family therapy,^36,70 skills training.^40,49 In general, the methodological quality was low; two RCTs (Delamater et al.¹⁷ and Dougherty et al.¹⁸) and the two cohort studies (Likitmaskul et al.⁴⁰ and Srinivasan et al.⁶⁹) were assessed as being of moderate quality.

Results. Three studies (Dougherty et al.,¹⁸ Sundelin et al.,⁷⁰ Srinivasan et al.⁶⁹ compared general diabetes education programs^18,69 or family therapy programs⁷⁰ that were delivered in an ambulatory setting vs. the same program to delivered during hospitalization. The findings of the studies were inconsistent (Table 6). Dougherty et al.¹⁸ reported that the decrease in HbA1c levels was significantly greater for the home-based group compared with the inpatient group. Sundelin et al.⁷⁰ and Srinivasan et al.⁶⁹ found no significant differences in HbA1c levels between groups.

The RCT by Delamater et al.¹⁷ found that self-management training group had significantly lower HbA1c levels compared with the control group. Similarly, the cohort study by Likitmaskul et al.⁴⁰) found that HbA1c levels were significantly lower in the cohort that received skills training. Mitchell⁴⁹ assessed skills training and reported that HbA1c levels improved significantly for both the intervention and control groups; however, the difference was not statistically significant. The sample size was small and there may not have been sufficient statistical power to detect a difference. The remaining study that assessed a family therapy program delivered to parents³⁶ reported that diabetes education had no significant effect on improving HbA1c levels in children with newly diagnosed diabetes.

Children With Poorly Controlled Diabetes

Overall, 14 studies (9 RCTs,^{8,13,15,16,25,43,55,57,83} 3 CCTs,^14,51,76 2 uncontrolled before-and-after^28,34) assessed the effectiveness of diabetes education among children with poorly controlled diabetes (Table 6). Interventions that were assessed included general diabetes education,^15,55 cognitive behavioral therapy,^{8,13,14,16,76} family therapy,^28,83 skills training,⁴³ diabetes camp,³⁴ and psychoeducation.⁵¹ In general, the methodological quality of the studies was low; three RCTs were assessed as being of high quality (Nunn et al.,⁵⁵ Panagiotopoulos et al.,⁵⁷Wysocki et al.⁸³).

Results. Two large high-quality RCTs that assessed general diabetes education (Nunn et al.,⁵⁵) and family therapy (Wysocki et al.⁸³) reported that diabetes education had no significant impact on HbA1c levels. The RCT by Panagiotopoulos et al.⁵⁷ found HbA1c levels decreased for both groups, but the difference was not significant.

The results of the remaining studies were inconsistent. Four studies^14,51,76 reported that HbA1c levels decreased significantly following the diabetes education intervention. Two RCTs^13,15 found that HbA1c levels improved significantly for both the intervention and control groups; however, the differences were not statistically significant. Both studies had small samples (less than 40 participants each) and may not have had sufficient statistical power to detect a difference. The remaining studies^{8,16,25,28,43} reported that diabetes education had no significant effect on improving HbA1c levels in this patient population.

General Diabetes Education—Health Services Utilization

Description of studies. We identified 6 studies (3 RCTs,^9,55,71 2 cohorts,^41,66 1 uncontrolled before-and-after⁷⁷) that assessed the effect of general diabetes education programs on health service utilization. Studies were conducted in the United States,^9,41,66,71 Australia⁵⁵ and Europe.⁷⁷ The median year of publication was 2001 and ranged from 1988 to 2006.

The number of participants in the studies ranged from 30 to 301 (median = 83 [IQR 32 to 146]). The mean age of participants ranged from 7.4 to 11.9 years (n = 5 studies). Two studies focused on children with newly diagnosed diabetes.^41,66 One study examined an education intervention delivered to children who demonstrated poor metabolic control.⁵⁵

Five of the intervention programs targeted children and their parents or the family; one was delivered to children only.⁹ The settings for interventions were described as a hospital inpatient unit,^41,77 diabetes center,⁷¹ home,⁹ or mixed (e.g., diabetes center and home).^55,66 Three studies reported post-intervention assessments of 1 month,⁶⁶ 6 months⁷⁷ and 24 months.⁷¹

Results

General population of children with diabetes. Svoren et al.⁷¹ randomly assigned 299 patients to one of three groups: a case manager or “care ambassador” whose role was to monitor clinic attendance and provide telephone outreach to families (CA), a care ambassador plus eight psychoeducational modules on diabetes care that were delivered during visits to the diabetes clinic (CA+), or standard care (routine clinical care). At the end of the 24-month program, the CA+ group had a significantly lower rate of hospitalizations than the combined comparison groups (8.9 per 100 patient-years vs. 15.3 per 100 patient-years, respectively [p = 0.04]). The rate of emergency department (ED) visits was also lower in the CA+ group than for the comparison groups (21.0 per 100 patient-years vs. 34.9 per 100 patient-years, respectively [p = 0.004]). The methodological quality of the study was moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

Brown et al.⁹ randomly assigned 59 children and adolescents to a group that played a video game featuring characters with diabetes or a group that played a pinball video game with no information on diabetes. At the end of the 6-month study period, the number of unscheduled urgent visits to the physician for diabetes-related problems declined in the intervention group (0.57 visits per child vs. 0.13 visits). For the comparison group, the number of visits increased (0.61 visits per child vs. 0.64). However, the difference in the change from baseline for the two groups was not statistically significant (p = 0.08). The methodological quality of the study was low (1/5 on the Jadad score; unclear allocation concealment).

In an uncontrolled before-and-after study, von Sengbusch et al.⁷⁷ assessed the impact of a mobile diabetes education service in a group of 107 children living in rural areas. At 24 months, the number of hospitalizations decreased significantly from 17 (16.2 percent) at baseline to 7 (6.8 percent). The methodological quality of the study was moderate based on the Thomas instrument.

Children with newly diagnosed diabetes. In a retrospective cohort study (n = 30), Lipman⁴¹ compared the effect of diabetes education delivered by a clinical nurse specialist compared to the same program delivered by staff nurses. The authors reported that the length of hospitalization was significantly shorter for the intervention group than for the comparison group. (6.2±1.8 days vs. 8.4±1.8 days,[p<0.01], respectively). The methodological quality of the study was moderate based on the Thomas instrument.

In a prospective cohort study, Siminerio et al.⁶⁶ examined differences between outpatient and inpatient programs for newly diagnosed children (n = 32). In the month following diagnosis, none of the children in either group reported any diabetes-related hospital or ED visits. The methodological quality of the study was moderate based on the Thomas instrument.

Children with poorly controlled diabetes. Nunn et al.⁵⁵ randomly assigned 123 patients to receive telephone calls from a pediatric diabetes educator or standard care (i.e., routine clinic visits but no telephone calls). Over the course of the 8-month study, both groups showed an increase in hospitalizations of 0.02 per year; however, the difference in the change from baseline between the groups was not statistically significant (p = 0.57). The methodological quality of the study was high (3/5 on the Jadad score; unclear allocation concealment).

Cognitive Behavioral Therapy—Health Services Utilization

Children with newly diagnosed diabetes. One CCT by Golden et al.¹⁹ examined the effect of an intervention that used cognitive behavioral therapy techniques. Eleven families received an integrated diabetes education and psychosocial support program that was delivered in a motel-like setting. The comparison group (n = 8) received the same training 15 months after diagnosis. Over 24 months following diagnosis, the intervention group experienced 1 hospitalization for hypoglycemia compared with 11 hospitalizations for the comparison group. The methodological quality of the study was low (1/3 on the modified Jadad score).

Skills Training—Health Services Utilization

Children with newly diagnosed diabetes. Likitmaskul et al.⁴⁰ conducted a nonconcurrent cohort study to compare the effect of a multidisciplinary team approach to diabetes education to conventional education that focused on insulin injection and how to control diet. The 28 children who received conventional education were diagnosed with diabetes prior to 1996 and served as a historical comparison group to the 24 children who were diagnosed after 1996. Based on a review of medical records, the hospital length of stay after initial diagnosis was significantly shorter for children who received the multidisciplinary education program than for the comparison group (17.6±9.5 days vs. 36.0±46.5 days, respectively). The methodological quality was rated as moderate using the Thomas instrument.

Children with poorly controlled diabetes. In the RCT by Mann et al.⁴³ children in the intervention group (n = 19) received intensive diabetes education combined with regular SMBG; children in the control group (n = 20) received intensive education only. Over the 18-month followup, the group that received education only had 4 diabetes-related hospital admissions; the SMBG group reported no admissions (p< 0.04). The methodological quality of the study was low (2/5 on the Jadad score; unclear concealment of allocation).

Diabetes Camp—Health Services Utilization

General population of children with diabetes. In an uncontrolled before-and-after study, Koontz³⁷ examined the effect of a general diabetes education program delivered at a 1-week camp (n = 112). The mean age of the children was 5.0 years. At the 3-month followup, there were no hospitalizations reported for the 29 children who responded to the followup questionnaire. Three children reported physician visits for reasons other than their regular checkup. No baseline data were provided, therefore we do not know if this was different from their pre-camp experience. The methodological quality of the study was rated as moderate based on the Thomas instrument.

Other Interventions—Health Services Utilization

Children with poorly controlled diabetes. In the CCT by Moran et al.⁵¹ a group of 11 children received an intensive inpatient psychotherapy program; the comparison group comprised 11 children who were admitted to hospital for medical treatment of diabetes-related complications. Both groups received diabetes education while in hospital. At the 12-month followup there was no significant difference in the number of overall hospitalizations (1.0±1.4 admissions per child vs. 0.9±0.9 [comparison group]). When the number of diabetes-related hospital admissions was compared, the intervention group had significantly fewer admissions than the comparison group (0.3±0.5 admissions per child vs. 0.9±0.9 [p<0.05], respectively). The methodological quality of the study was low (0/3 on the modified Jadad score).

Summary of Results—Health Services Utilization

We identified 11 studies that assessed the impact of diabetes education on health services utilization (Table 7). Of these, four studies (2 RCTs, 2 uncontrolled before-and-after) examined the general population of children with type 1 diabetes, four (1 CCT, 3 cohort) focused on children with newly diagnosed diabetes, and three (2 RCTs, 1 CCT) considered children with poorly controlled diabetes. Health services utilization was measured in a variety of ways: length of stay, hospital or ED admissions for diabetes- and non-diabetes-related complications.

General population of children with diabetes. Four studies (2 RCTs,^9,71 2 uncontrolled before-and-after^37,77) assessed the effect of diabetes education on health services utilization in the general population of children with diabetes (Table 7). Three studies assessed general diabetes education programs^9,71,77 and one assessed a diabetes camp program.³⁷ In general, the methodological quality of the studies was moderate; only one RCT⁹ was rated as being of low quality.

Results. There is some evidence that diabetes education has an impact on health services utilization in this patient population. The RCT by Svoren et al.⁷¹ reported hospital and ED admissions were significantly reduced in the intervention group that were supported by a case manager plus psychoeducation training. The uncontrolled before-and-after study by von Sengbusch et al.⁷⁷ reported statistically significant reductions in hospital admissions following the intervention. The study by Koontz³⁷ did not report baseline information; however, there were no hospital or ED admissions 3 months following the diabetes camp. The RCT by Brown et al.⁹ reported a non-significant reduction in physician visits.

Children with newly diagnosed diabetes. Four studies (1 CCT,¹⁹ 3 cohorts^40,41,66) assessed the effect of diabetes education on health service utilization among children with newly diagnosed diabetes (Table 7). Overall, the methodological quality of the studies was moderate, with all but one¹⁹ rated as moderate quality. Two cohort studies assessed a skills training program (Likitmaskul et al.⁴⁰) and a general diabetes education program (Lipman et al.⁴¹); both found that the length of stay for initial hospitalization was shorter for the intervention groups than the comparison groups. The cohort study by Siminerio et al.⁶⁶ compared inpatient vs. ambulatory delivery of general diabetes education and reported that neither study group had any diabetes-related hospital or ED admissions. The RCT by Golden et al.¹⁹ reported significantly fewer hospital admissions for the intervention group.

Children with poorly controlled diabetes. The large high-quality RCT by Nunn et al,⁵⁵ reported increased rates of hospital admission for both the intervention group that received general diabetes education plus telephone followup and the control group (i.e., the intervention was not effective) (Table 7). In contrast, the two studies of low methodological quality^43,51 reported that their intervention groups had significantly fewer hospital admissions than the control groups.

General Diabetes Education—Complications

Description of studies. We identified 6 studies (2 RCTs,^18,71 1 retrospective cohort,³⁹ 3 uncontrolled before-and-after^61,75,77) that assessed the effect of general diabetes education programs on short-term complications. Studies were conducted in the United States,⁷¹ Canada,^18,39 and Europe.^61,75,77 The median year of publication was 2001 and ranged from 1993 to 2006.

The number of participants in the studies ranged from 28 to 301 (median = 50 [IQR 30 to 107]). The mean age of participants ranged from 9.8 to 15.8 years (n = 5 studies). For three studies, the mean age was less than 12 years.^18,71,77 One study examined education intervention delivered to children and their families at the time of diagnosis.¹⁸

All interventions were targeted to children and their parents or to the family. The settings for interventions were described as a hospital inpatient unit,^61,77 diabetes center,⁷¹ or mixed (e.g., diabetes center and home).^18,39 The setting was not clearly described in one study.⁷⁵ All six studies reported post-intervention assessments; the median followup period was 12 months and ranged from 6 to 36 months.

Results

General population of children with diabetes. Svoren et al.⁷¹ randomly assigned 299 patients to one of three groups: a case manager or “care ambassador” whose role was to monitor clinic attendance and provide telephone outreach to families (CA), a care ambassador plus eight psychoeducational modules on diabetes care that were delivered during visits to the diabetes clinic (CA+), or standard care (routine clinical care). At the end of the 24-month program, the CA+ group had significantly reduced rates of severe hypoglycemic events compared with either the CA or standard care group (0.45 events per person-year vs. 0.56 events or 0.65 events, respectively [p = 0.02]). The methodological quality of the study was moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

The retrospective cohort study (n = 28) by Lawson et al.³⁹ compared intensive, individualized education to group education delivered to patients. The content for both interventions was the same. After 12 months, two children (12 percent) in the intervention group experienced severe hypoglycemic reactions; there were no severe reactions in the comparison group. The difference was not statistically significant (p = 0.13). There were no episodes of DKA in either group. The methodological quality of the study was weak based on the Thomas instrument.

Povlsen et al.⁶¹ assessed the effect of an intervention targeted on 37 families from ethnic minority groups in Denmark. The intervention included adapted educational material and guidelines and re-education that focused on increasing knowledge and self-care. The authors observed that the number of severe hypoglycemia episodes increased from 3 to 10; however, the increase was not statistically significant. There was one DKA event during the intervention. The methodological quality of the study was weak based on the Thomas instrument.

In an uncontrolled before-and-after study Verrotti et al.⁷⁵ studied 30 adolescents who attended nine education sessions on general diabetes management. At the 12-month followup, the number of severe hypoglycemia episodes decreased significantly from 2.9±2.2 events to 1.1±1.3 events (p = 0.002). The methodological quality of the study was moderate based on the Thomas instrument.

In an uncontrolled before-and-after study, von Sengbusch et al.⁷⁷ assessed the impact of a mobile diabetes education service in a group of 107 children. At the 6-month followup the number of episodes of severe hypoglycemia did not change significantly from baseline measures (0.23 events per 100 patient-years to 0.21 events). The methodological quality of the study was moderate based on the Thomas instrument.

Children with newly diagnosed diabetes. Dougherty et al.¹⁸ randomly assigned 63 newly diagnosed patients to a 24-month home-based diabetes education program or to traditional hospitalization and outpatient followup. Treatment differences between the groups included duration of initial hospital stay, timing of initial teaching, and the nature and extent of subsequent nursing followup. Over the 24-month followup period, the authors reported no statistically significant differences between the groups in the number or rate of diabetes-related adverse events (severe hypoglycemia, hyperglycemia and ketosis, DKA, chronic hyperglycemia). The rate in the home-based group was 0.34 events per patient compared with 0.26 in the hospital-based group (RR = 1.45 [95% CI: 0.59 to 3.6]). The methodological quality of the study was moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

Cognitive Behavioral Therapy—Complications

Description of studies. We identified four studies (2 RCTs,^21,72 1 CCT,¹⁹ 1 prospective cohort⁵⁴) that examined the effect of interventions that used cognitive behavioral therapy techniques. Studies were conducted in the United States^19,21,72 and Europe.⁵⁴ The median year of publication was 1995 and ranged from 1985 to 2000.

The median number of participants in the studies was 52 and ranged from 27 to 139. The mean age of participants ranged from 2.7 to 14.7 years. For two studies, the mean age was less than 12 years.^19,72 One study examined education intervention delivered to children and their families at the time of diagnosis.¹⁹

Interventions were delivered to children,²¹ parents,¹⁹ or to both parents and children or the entire family.^54,72 The settings for interventions were described as an outpatient clinic,⁷² or mixed (e.g., diabetes center and home).^21,72 The setting was not clearly described in one study.¹⁹ All but one study¹⁹ reported post-intervention assessments; the followup periods were 3 months,⁷² 12 months,²¹ and 248 patient years.⁵⁴

Results

General population of children with diabetes. Grey et al.²¹ randomly assigned 77 adolescents to receive intensive diabetes management (IDM) as described in the DCCT or IDM plus a behavioral program of coping skills training intervention (IDM+). At the 12-month followup, the IDM+ group reported statistically lower rates for adverse events than the control group (severe hypoglycemia: 1.1 per patient year vs. 1.2 per patient year, respectively [p<0.001]; DKA: 0.02 vs. 0.06 events per patient year, respectively [p<0.001]). The authors observed that the incidence of hypoglycemia significantly decreased in females but not in males in the IDM+ group. Overall, the rate of complications was higher than that reported by the DCCT. The methodological quality of the study was moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

In a prospective study with historical controls, Nordfeldt and Ludvigsson⁵⁴ examined the effect of multiple dose insulin therapy combined with problem-based training and psychosocial support. The prospective cohort comprised 248 children admitted to hospital in 1994 and 1995. The comparison group (156 patients admitted in 1980-81) received standard clinical care. The annual incidence of severe hypoglycemia with unconsciousness was 0.17 events per patient-year for the intervention group compared with 0.23 events per patient-year for the 1980-81 cohort (p<0.05). The incidence of DKA requiring hospitalization during 1994-95 was 0.015 episodes per patient-year (not reported for the 1980-81 cohort). The methodological quality of the study was weak based on the Thomas instrument.

Szumowski⁷² randomly assigned 21 young children to an 8-week behavioral intervention that included information on diabetes management plus instruction and practice in the application of behavioral principles and goal setting to reinforce children's regimen adherence. The comparison group received information on diabetes management but no additional instruction. At the 3-month followup there was no significant difference in the number of hypoglycemic episodes experienced by either group, nor was the change from baseline significantly different for either group. The methodological quality of the study was moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

Children with newly diagnosed diabetes. In the CCT by Golden et al.¹⁹ 11 families received an integrated diabetes education and psychosocial support program that was delivered in a motel-like setting. The comparison group (n = 8) received the same training 15 months after diagnosis. At 12-month followup, the rate of severe hypoglycemia episodes reported by the intervention group was significantly lower than the comparison group (0.25 per patient vs. 1.1 per patient, respectively [p<0.01]). The methodological quality of the study was low (1/3 on the modified Jadad score).

Skills Training—Complications

Three studies (1 RCT,⁵² 2 cohorts^40,53) examined the effect of skills training on short-term complications. The studies were conducted in Europe and were published in 2002^40,53 and 2005.⁵²

Results

General population of children with diabetes. In the RCT by Nordfeldt et al.,⁵² three groups of patients (n = 332; mean age = 5.3 years) were randomized to receive either videotapes and a brochure with information about practical skills for self-control and treatment (intervention group), a videotape and brochure with general diabetes information, or routine clinical care. Outcomes were assessed at 12 and 24 months following distribution of the education materials. At 12 months there was a reduction in the annual incidence of severe hypoglycemia in the intervention group; there was no significant change for either of the controls groups. This reduction was still present at 24 months; the incidence of severe hypoglycemia in the intervention group decreased from 42 percent to 25 percent (p = 0.023) in the intervention group. The methodological quality of the study was low (1/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

Nordfeldt and Ludvigsson⁵³ evaluated the effect of self-study material on diabetes education aimed at self-management skills and the prevention of hypoglycemia. Over a 3-year period (1997 to 1999) brochures and videos were distributed to the homes of approximately 130 patients and families each year. The mean age of the children was 4.6 years. The mean incidence of severe hypoglycemia with unconsciousness was lower during the 1997-1999 cohort compared to 1994-1996 (pre-intervention); however, the difference was not statistically significant (0.14 to 0.16 events per patient-year vs. 0.17 to 0.22 events per patient-year, respectively). The methodological quality of the study was weak based on the Thomas instrument.

Children with newly diagnosed diabetes. Likitmaskul et al.⁴⁰ conducted a nonconcurrent cohort study to compare the effect of a multidisciplinary team approach to diabetes education to conventional education that focused on insulin injection and how to control diet. The 28 children who received conventional education were diagnosed with diabetes prior to 1996 and served as an historical comparison group to the 24 children who were diagnosed after 1996. The children who received the multidisciplinary education program had a 4 percent readmission rate for recurrent DKA or hyperglycemia during the first year following diagnosis compared with 18 percent for the comparison group. This difference between groups continued up to 4 years post-diagnosis. The methodological quality of the study was moderate based on the Thomas instrument.

Diabetes Camp—Complications

Children with poorly controlled diabetes. In an uncontrolled before-and-after study, Karaguzel et al.³⁴ examined the effect of a 1-week diabetes camp that incorporated intensive insulin treatment into a general diabetes program (n = 25). The mean age of the children was 5.0 years. Over the 12-month followup, no severe hypoglycemic episodes were detected. As no baseline data were provided, we cannot determine if this was a change from pre-intervention. The methodological quality of the study was weak based on the Thomas instrument.

Other Interventions—Complications

General population of children with diabetes. Campaigne et al.¹⁰ assessed the effects of physical training on short-term complications. Fourteen adolescents were randomly assigned to either a 12-week exercise program or to a control group that did not modify the usual exercise routine. At the end of the program, there was no change in the occurrence of hypoglycemia in either group. The methodological quality of the study was low (2/5 on the Jadad score; unclear allocation concealment).

Summary of Results—Complications

Overall, we identified 15 studies that assessed the effectiveness of diabetes education in controlling complications. Of these, 11 (5 RCTs, 3 cohorts, 3 uncontrolled before-and-after) examined the general population of children with type 1 diabetes, 3 studies (1 RCT, 1 CCT, 1 cohort) focused on children with newly diagnosed diabetes, and 1 uncontrolled before-and-after study considered children with poorly controlled diabetes (Table 8). While one of the objectives of this evidence report was to assess the effect of diabetes education on both long- and short-term complications, all of the complications reported in the included studies were short-term. Most studies reported on the incidence of severe hypoglycemia. Six studies reported the incidence of DKA and one study measure chronic hyperglycemia.

General population of children with diabetes. Overall, 11 studies (5 RCTs,^{10,21,52,71,72} 3 cohorts,^39,53,54 3 uncontrolled before-and-after^61,75,77) assessed the effect of diabetes education on short-term complications among the general population of children with diabetes (Table 8). Three studies examined interventions based on cognitive behavioral therapy,^21,54,72 five studied general diabetes education interventions,^{39,61,71,75,77} 2 assessed skills training^52,53 and one assessed a physical training program.¹⁰ Overall the methodological quality was low; only two RCTs (Grey et al.,²¹ Szumowski⁷² and two uncontrolled before-and-after studies (Verrotti,⁷⁵ von Sengbusch et al.⁷⁷) were rated as being of moderate quality.

Results

Grey et al.²¹ reported that the event rate for severe hypoglycemic events decreased significantly following diabetes education. Verrotti et al.⁷⁵ also reported that the event rate for severe hypoglycemic events decreased significantly following diabetes education. In contrast Szumowski⁷² and von Sengbusch et al.⁷⁷ found that diabetes education had no significant impact on the rate for severe hypoglycemia.

The results of the remaining studies were inconsistent. Three studies^52,54,71 reported that the event rate for severe hypoglycemic events decreased significantly following diabetes education. Nordfeldt and Ludvigsson⁵³ found that the annual event rate for severe hypoglycemia decreased for both the skills training and comparison groups; however, the difference was not statistically significant. The remaining studies reported that diabetes education had no significant impact on short-term complications.^10,39,61

Children with newly diagnosed diabetes. Overall three studies (1 RCT,¹⁸ 1 CCT,¹⁹ 1 cohort⁴⁰) assessed the effect of diabetes education on short-term complications among children with newly diagnosed diabetes (Table 8). One study (Likitmaskul et al.⁴⁰), which was assessed as being of moderate quality, reported that the event rate for severe hypoglycemic events and the number of readmissions for DKA decreased significantly following a skills training intervention. Dougherty et al.¹⁸ reported that setting for delivery of general diabetes education (i.e., inpatient vs. ambulatory) had no significant impact on short-term complications. The low quality RCT by Golden et al.¹⁹ reported that the event rate for severe hypoglycemic events decreased significantly following a cognitive behavioral therapy intervention.

Children with poorly controlled diabetes. One uncontrolled before-and-after study assessed the effect of a diabetes camp³⁴ on short-term complications among children with poorly controlled diabetes (Table 8). No adverse events were reported by Karaguzel et al.³⁴ during followup.

General Diabetes Education—Knowledge

Description of studies. We identified 13 studies (5 RCTs,^{9,18,24,32,55} 2 cohorts,^66,69 4 uncontrolled before-and-after,^61,65,75,77 2 CCTs,^42,45) that assessed the effectiveness of general diabetes education on diabetes-related knowledge. The studies were conducted in the United States,^9,32,45,66 Europe,^{24,42,61,75,77} Canada,¹⁸ India,⁶⁵ and Australia.^55,69 The median year of publication was 1999 (IQR, 1993 to 2005).

The number of study participants ranged from 30 to 146 (median = 63 [IQR 37 to 107]). The mean age of participants ranged from 10.1 to 15.4 (n = 8 studies). The mean age was less than 12 years in six studies.^{18,24,42,55,66,77} Three studies assessed education interventions delivered to children and their families at the time of diagnosis.^18,66,69 One study examined children who demonstrated poor metabolic control.⁵⁵

All of the interventions were aimed at children and their parents or to the entire family. The interventions were delivered in a variety of settings including hospital inpatient units,^61,77 home,⁹ and an outpatient clinic.²⁴ In seven studies, the setting was mixed (e.g., hospital and home)^{18,32,45,55,65,66,69} and not reported in two studies.^42,75

There were four studies in which knowledge was not measured beyond completion of the education program.^9,24,32,55 In the eight studies that reported post-intervention assessments,^{18,42,61,65,66,69,75,77} ranged from 1 to 36 months.

Results

General population of children with diabetes. Brown et al.⁹ randomly assigned 59 children and adolescents to a group that played a video game with a focus on nutritional education featuring characters with diabetes or a group that played a pinball video game with no information on diabetes. At the end of the 6-month study period, the intervention group reported more gains in knowledge compared to the control group when assessed using the Diabetes Knowledge Test. The methodological quality of this study was low (1/5 on the Jadad score; unclear concealment of allocation).

In the RCT by Hackett et al.,²⁴ three cohorts of families received educational packages that were delivered over 8 months. The content was the same for all groups; one group (cohort 1) received a second reinforcement package for an additional 8 months. The comparison group did not receive the education package. The education program led to an increase in mean knowledge scores, as measured by a multiple choice test, across all cohorts. However, the knowledge was poorly retained by fathers. The methodological quality of this study was low (2/5 on the Jadad score; unclear concealment of allocation).

Howe et al.³² randomly assigned 75 patients to one of three treatment groups: : standard care (routine quarterly clinic visits), standard care plus one education session on basic diabetes management skills, and standard care, the education session plus weekly telephone calls to review management techniques. From results of the KNOW test (46 multiple choice questions) there were no statistically significant differences found in knowledge among the groups. The trial was ended early due to lack of enrolment. The methodological quality of this study was low (2/5 on the Jadad score; unclear concealment of allocation).

In the CCT by Lucey and Wing,⁴² a general education intervention was delivered to 49 families in the form of two 6-hour group sessions. In the followup period, which ranged from 2 to 44 weeks, the children in the intervention group performed better on a multiple choice questionnaire that tested knowledge than children in the comparison group. The methodological quality of this study was low (0/3 on the modified Jadad score).

Mason⁴⁵ examined a two to four player board game played by 93 groups of parents and children. The children would collect cards (tools) that could be used for continued control of diabetes. According to a questionnaire, there was no significant difference in knowledge between groups that played the game and those who did not play the game, though both groups significantly improved knowledge from baseline. The methodological quality of this study was low (0/3 on the modified Jadad score).

Povlsen et al.⁶¹ assessed the effect of an intervention targeted at 37 families from ethnic minority groups in Denmark. The intervention included adapted educational material and guidelines and re-education that focused on increasing knowledge and self-care. Using the results of a questionnaire, the authors reported a mean increase in knowledge from baseline; however, the differences between families were considerable at the 6-month followup. The methodological quality of this study was weak based on the Thomas instrument.

In the study by Shobhana et al.⁶⁵ a 1-hour didactic lecture on injection and monitoring skills and individualized diet counseling was delivered to 37 parent and child groups by a multidisciplinary team. A short questionnaire found that immediately post-intervention and at 3 months, there were significant increases in knowledge about diabetes, injections and hypoglycemia. At the 6-month followup, knowledge in all areas increased except with regard to self-monitoring. The methodological quality of this study was weak based on the Thomas instrument.

In an uncontrolled before-and-after study Verrotti et al.⁷⁵ studied 30 adolescents who attended nine education sessions on general diabetes management. At the 12-month followup, a 20-item multiple choice questionnaire showed that the participants had statistically higher knowledge scores compared to baseline values. The methodological quality of this study was moderate based on the Thomas instrument.

In an uncontrolled before-and-after study, von Sengbusch et al.⁷⁷ assessed the impact of a mobile diabetes education service in a group of 107 children. There was a statistical improvement in children's knowledge at the 6-month followup according to the results of an age-adapted questionnaire. The methodological quality of this study was moderate based on the Thomas instrument.

Children with newly diagnosed diabetes. Dougherty et al.¹⁸ randomly assigned 63 newly diagnosed patients to a 24-month home-based diabetes education program or to traditional hospitalization and outpatient followup. Treatment differences between the groups consisted of duration of initial hospital stay, timing of initial teaching, and the nature and extent of subsequent nursing followup. The Diabetes Knowledge Scale found no significant difference between the groups. The methodological quality of this study was moderate (2/5 on the Jadad score; unclear concealment of allocation; blinding of outcome assessors).

Siminerio et al.⁶⁶ conducted a prospective cohort study in which a group of 32 newly diagnosed children and their parents and peers received 10 to 12 hours of inpatient or outpatient education sessions over three days. The education addressed aspects of basic diabetes knowledge such as self-management skills, nutrition and exercise. The Test of Diabetes Knowledge found that there was no significant difference in knowledge between the inpatient and outpatient groups. The methodological quality of this study was moderate based on the Thomas instrument.

In the prospective cohort study by Srivivasan et al.⁶⁹ a group of 61 newly diagnosed patients attended a diabetes day care program (DDCP) at which they received “survival skills” diabetes education. The pre-DDCP cohort comprised 49 patients who were admitted to hospital for 4 to 7 days for a detailed education program. There was no significant difference between the groups' results on the Test of Diabetes Knowledge. The methodological quality of this study was moderate based on the Thomas instrument.

Children with poorly controlled diabetes. Nunn et al.⁵⁵ randomly assigned 123 patients to receive scheduled telephone calls from a pediatric diabetes educator or standard care (i.e., routine clinic visits but no telephone calls). The intervention was provided for five to eight months. Immediately after the intervention, authors measured knowledge using a modified test of diabetes knowledge and found there to be no significant differences between groups. The methodological quality of this study was high (3/5 on the Jadad score; unclear concealment of allocation).

Cognitive Behavioral Therapy—Knowledge

Description of studies. Seven studies (3 RCTs,^22,31,72 3 CCTs,^14,48,74 and 1 controlled before-and-after²⁰) examined various interventions that used cognitive behavioral therapy techniques to increase knowledge. All but two studies were conducted in the United States; 1 was conducted in Europe⁴⁸ and 1 was conducted in Australia.¹⁴ The median year of publication was 1993 (IQR, 1990 to 1998) and ranged from 1985 to 2001.

The number of participants enrolled in the studies ranged from 14 to 69 (median = 23 [IQR 20 to 32]). The mean age of participants ranged from 6.5 to 14.2 years (5 studies). For 2 studies, the mean age was less than 12 years.^22,72 One study examined education interventions delivered to children who had poor metabolic control.¹⁴

In three studies the interventions were delivered to children^14,22,74 and to parents as well as children in four studies.^20,31,48,72 The settings for the interventions were described as an outpatient clinic,^20,48,72 or home.³¹ For one study, the setting was mixed (e.g., diabetes center and home).¹⁴ Two studies did not report the setting of the intervention.^22,74 All but 2 studies^20,31 conducted post-intervention assessments for knowledge. The median followup period was 12 months and ranged from 3 to 13 months.

Results

General population of children with diabetes. In the RCT by Gross et al.²² 14 children and parents were randomly assigned to either behavior modification training or to a group that included discussion and role-playing. At 6-month follow up, the intervention group scored significantly higher on the Behavior Modification Test than the control group (p<0.01). The methodological quality of this study was low (2/5 on the Jadad score; unclear allocation concealment).

Szumowski⁷² randomly assigned 21 young children to an 8-week behavioral intervention that included information on diabetes management plus instruction and practice in the application of behavioral principles and goal setting to reinforce children's regimen adherence. The comparison group received information on diabetes management but no additional instruction. At the 3-month followup parents were assessed using the Knowledge of behavioral principles as applied to children (KBPAC) and the Test of Diabetes Knowledge while children's knowledge was assessed using the Children's Diabetes Quiz. The authors found that there was a significant group by time interaction (p<0.008) for both groups at 3 months. Both children and parents increased their knowledge scores from baseline to 3 months but statistical significance was not reported. The methodological quality of this study was moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

Horan et al.³¹ conducted an RCT (n = 20) that compared goal setting and problem solving using dynamic computer-assisted teaching modules vs. conventional education using an education booklet. At the end of the 15-week program, assessments of diabetes knowledge were made using multiple choice questions modified from the Test of Diabetes Knowledge. There was no significant difference between the groups with regards to their applied knowledge of diabetes. However more individuals in the intervention group showed improvement in their factual diabetes knowledge than individuals in the control group. The methodological quality of this study was low (2/5 on the Jadad score; unclear allocation concealment).

In a controlled clinical trial, Mendez and Belendez⁴⁸ delivered 12 sessions to children (n = 37) that provided them with audiovisual and print material and also allowed them to practice their skills. Their parents were taught to reinforce adherence behaviors rather than punish non-compliance. Immediately after the sessions ended, the intervention group had significantly higher results on the Diabetes Information Survey for Children (DISC) (p < 0.001) but at 13 months, the difference between the intervention and control group was not significant (p = 0.087). The methodological quality of this study was low (1/3 on the modified Jadad score).

Thomas-Dobersen et al.⁷⁴ conducted a controlled clinical trial in which 20 adolescents received 14 sessions over three months addressing various areas of diabetes management such as diet and knowledge of hypoglycemia. Youth Evaluation Scales (YES) found that at 1-year followup, the intervention group showed a statistically significant increase in knowledge compared to the control group (p<0.01). The methodological quality of this study was low (1/3 on the modified Jadad score).

Greco at al.²⁰ conducted an uncontrolled before-and-after study that aimed to implement a structured intervention for integrating peers into diabetes care in a healthy and adaptive manner. There were 23 pairs of patients and peers who were assessed using the Diabetes Education and Support Assessment (DESAT) tool. After the 4-week intervention, the children as well as their peers experienced a significant increase in knowledge compared to baseline (p<0.0001). The methodological quality of this study was moderate based on the Thomas instrument.

Children with poorly controlled diabetes. Couper et al.¹⁴ evaluated the effects of monthly home visits by a nurse educator plus weekly telephone calls that focused on goal-setting. Thirty-seven adolescents received routine care plus the regular intervention; 32 received routine clinical care only. At the 6-month followup, both parents and children in the intervention groups, assessed with the Diabetes Knowledge Assessment Scale, had significantly higher knowledge scores than those in the control group (p = 0.001). At 12 months, only the parents maintained this difference between the groups (p = 0.005). The methodological quality of this study was low (1/3 on the modified Jadad score).

Diabetes Camp—Knowledge

Description of studies. In eight studies (3 RCTs⁵⁸–⁶⁰, 1 CCT,⁶² and 4 uncontrolled before-and-after studies^12,27,34,35) the education intervention was delivered as part of a diabetes camp program. Seven studies were conducted in the United States; one took place in Europe.³⁴ The median year of publication was 1994 and ranged from 1983 to 2005.

The number of participants enrolled ranged from 25 to 237 (median = 76.5 [IQR 61 to 86]). The mean age of participants ranged from 10.0 to 14.83 years. For 1 study, the mean age was less than 12 years.³⁵ One study focused on children who demonstrated poor metabolic control.³⁴

In one study, the intervention was aimed at families³⁵ while the rest of the interventions were targeted only at children. Only three studies conducted assessments immediately after the intervention.^12,27,60 The remaining five studies conducted post-intervention assessments for knowledge; the median followup period was 8 months and ranged from 2 days to 12 months.

Results

General population of children with diabetes. In the uncontrolled before-and-after study by Christensen et al.¹² the campers attended two sessions on carbohydrate counting and two sessions on food portioning. They used flash cards, practiced reading labels, measuring utensils, used scales with real food and played a card game. The authors used laboratory values to assess the correlations between HbA1c and knowledge and found that the correlation was not statistically significant (p = 0.09). The methodological quality of this study was weak based on the Thomas instrument.

The campers in the controlled before-and-after study by Harkavy et al.²⁷ received informal and formal teaching sessions which covered diabetes etiology/pathology, self-management skills, effects of diabetes on stress and social issues. Immediately after the 2-week camp was over, campers completed a multiple choice questionnaire to assess their knowledge of etiology, pathophysiology and diabetes in general. The authors found that campers 12–13 years and 14–15 years of age showed a significant improvement in knowledge. There was no significant difference found in the 10–11 year olds. The methodological quality of this study was weak based on the Thomas instrument.

In an RCT by Pichert et al.,⁶⁰ 64 campers were randomly assigned to one of two groups: four 45-minute problem-solving sessions where they learned about metabolic management skills, diabetes and exercise guidelines and the social issues surrounding diabetes or the control group that received traditional instruction that did not emphasize problem solving. Immediately following the camp, the campers were asked to complete multiple choice and short answer tests to assess their factual knowledge of and ability to apply diabetes-related exercise guidelines. Repeated measures analysis of variance revealed that problem-solving groups gained more in factual knowledge and the ability to apply it to new problems than did the control group. The methodological quality of this study was low (2/5 on the Jadad score; unclear allocation concealment).

In another RCT by Pichert et al.,⁵⁸ 69 campers were randomized either to the intervention group comprising three 45-minute groups sessions using a problem-solving format devoted to nutrition-related skills and knowledge, or to the control group that received conventional direct instruction. When assessments were performed 2–4 or 5–6 days post-intervention, there was significant improvement in the knowledge in both groups when they were asked to complete a personal meal plan recall, but there was no significant difference between groups. The methodological quality of this study was low (2/5 on the Jadad score; unclear allocation concealment).

In the third RCT by Pichert et al.,⁵⁹ 81 campers were randomized to participate in two 45-minute sessions on sick day management that were taught via a problem-solving format, or to the control group that received conventional direct instruction. Knowledge was assessed using a Sick Day Knowledge Test and a Sick Day Problem Solving Test. At the 8-month followup, there was no significant difference between groups; however, the intervention group was better able to explain why the management guidelines applied to a hypothetical example. The methodological quality of this study was moderate (2/5 on the Jadad score; unclear allocation concealment; blinding of outcome assessors).

Remley⁶² conducted a CCT to compare a social cognitive theory-based program with a standard non-theory based camp education program. Eight 1-week camps across the United States were designated deliver either the theory-based intervention or the standard program. At the 3-month followup, there was no significant change in knowledge levels for either group. The attrition rate for both groups was high (40 percent). The methodological quality of this study was low (1/3 on the modified Jadad score).

In an uncontrolled before-and-after study, Kemp et al.³⁵ assessed the effectiveness of diabetes education and carefully monitored blood glucose control among 42 children who attended a 2-week summer camp. One year later, there was a significant improvement in knowledge when comparing pre to post intervention groups (p<0.01 vs. pre-camp). The methodological quality of this study was weak based on the Thomas instrument.

Children with poorly controlled diabetes. In an uncontrolled before-and-after study (n = 25), Karaguzel et al.³⁴ examined the effect of a 1-week diabetes camp that incorporated intensive insulin treatment into a general diabetes program which included dietary education. There was a statistically significant increase in knowledge about diabetes and diabetes nutrition at 6 and 12 month followup. The methodological quality of this study was weak based on the Thomas instrument.

Results

General population of children with diabetes. In the uncontrolled before-and-after study by Monaco et al.,⁵⁰ children were divided into two groups, ages 6–8 years and 9–11 years, to receive one session in which they were instructed in the use of injection site charts and injection site bears. The authors found that children in both age groups committed significantly fewer injection site identification errors when using the injection bears compared to using the chart. The methodological quality of this study was moderate based on the Thomas instrument.

In the uncontrolled before-and-after study by Templeton et al.,⁷³ adolescents attended one 90-minute session in which they were taught skills that would help them monitor their blood glucose. Immediately after the session they were asked to complete a five item true/false test. While the authors provided data on the percentages of questions correct, they did not comment on the statistical significance of results. The methodological quality of this study was weak based on the Thomas instrument.

Summary of Results—Knowledge

Overall, we identified 30 studies that assessed the effectiveness of diabetes education in improving knowledge (Table 9). Of these, 24 studies (9 RCTs, 5 CCTs, 10 uncontrolled before-and-after) examined the general population of children with type 1 diabetes, 3 studies (1 RCT, 2 cohorts focused on children with newly diagnosed diabetes, and 3 studies (1 RCT, 1 CCT, 1 uncontrolled before-and-after) considered children with poorly controlled diabetes.

General population of children with diabetes. Twenty-four studies (9 RCTs,^{9,22,24,31,32,58}–^60,72 5 CCTs,^{42,45,48,62,74} 10 uncontrolled before-and-after^{12,20,27,35,50,61,65,73,75,77}) assessed the effectiveness of diabetes education in improving diabetes-related knowledge among the general population of children with diabetes. Eight studies assessed general diabetes education,^{9,24,32,42,45,61,65,77} six examined interventions that used cognitive behavioral therapy techniques,^{20,22,31,48,72,74} 7 assessed interventions delivered at diabetes camps,^12,27,35,58–^60,62 and two assessed skills training.^50,73 Overall the methodological quality of the studies was low; only 2 RCTs^59,72 and 34 uncontrolled before-and-after studies^20,50,75,77 were rated as being of moderate quality.

Results. The results of the 5 moderate quality studies were inconsistent (Table 9). Sumowski⁷² reported increased knowledge levels in both the intervention and control groups, but the difference between groups was not statistically significant. Three uncontrolled before-and-after studies (Greco et al.,²⁰ Verrotti et al.,⁷⁵ and von Sengbusch et al.⁷⁷) reported gains in knowledge following a cognitive behavior therapy and general diabetes intervention, respectively. Monaco et al.⁵⁰ reported increases in knowledge levels, but the difference was not statistically significant. In the RCT by Pichert et al.⁵⁹ knowledge of nutrition and meal planning improved for both intervention and control groups, but the difference was not statistically significant.

The results of the remaining studies were also mixed. Two RCTs,^22,60 one CCT⁷⁴ and two uncontrolled before-and-after studies^35,65 reported statistically significant increases in diabetes-related knowledge following the education interventions. Two studies reported that their interventions were effective in subgroups of the study population.^24,27 Three trials and three uncontrolled before-and-after studies reported knowledge gains but the differences were not statistically significant^{9,42,58,61,73} or were not retained over the longer term.⁴⁸ In the remaining studies, the education interventions did not have significant effect on knowledge outcomes.^{12,31,32,45,62}

Children with newly diagnosed diabetes. Three moderate quality studies (1 RCT by Dougherty et al.,¹⁸ 2 cohorts by Siminerio et al.⁶⁶ and Srinivasan et al.⁶⁹) compared groups of children with newly diagnosed diabetes that received general diabetes education programs delivered in an ambulatory setting (e.g., home, outpatient clinic) vs. a hospital inpatient setting (Table 9). In all three studies, the knowledge levels increased for both intervention and comparisons groups; however, the differences were not statistically significant. The setting of the education program does not appear to have an effect on knowledge outcomes.

Children with poorly controlled diabetes. The large high-quality RCT by Nunn et al,⁵⁵ reported no significant differences in knowledge between the intervention group that received general diabetes education plus telephone followup and the control group that received diabetes education with no telephone followup (Table 9). The remaining studies that targeted patients with poorly controlled diabetes were of low quality. One uncontrolled before-and-after reported significant increases in knowledge following a diabetes camp;³⁴ one CCT reported significantly higher knowledge levels over the short-term, but these gains were not sustained over the long-term.¹⁴