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

Gillett M, Royle P, Snaith A, et al. Non-Pharmacological Interventions to Reduce the Risk of Diabetes in People with Impaired Glucose Regulation: A Systematic Review and Economic Evaluation. Southampton (UK): NIHR Journals Library; 2012 Aug. (Health Technology Assessment, No. 16.33.)

Cover of Non-Pharmacological Interventions to Reduce the Risk of Diabetes in People with Impaired Glucose Regulation: A Systematic Review and Economic Evaluation

Non-Pharmacological Interventions to Reduce the Risk of Diabetes in People with Impaired Glucose Regulation: A Systematic Review and Economic Evaluation.

Show details

4Systematic review of clinical effectiveness

Research question

Are there effective non-pharmacological interventions that will reduce the progression to diabetes in those with IGT and IFG?

The following interventions, either alone or in combination, are considered:

  • weight loss
  • exercise
  • qualitative changes in diet.

Methods

The review adopted the methodological approach published by the NHS Centre for Reviews and Dissemination.178

Inclusion and exclusion criteria

Inclusion criteria

Intervention
  • Weight loss.
  • Exercise.
  • Qualitative changes in diet.

Alone or in combination.

Comparators
  • Standard treatment.
  • Non-intensive lifestyle treatment.
Population
  • People with IGT or IFG.
Study design
  • RCTs of at least 2 years' duration.
  • Systematic reviews of RCTs.
Outcomes
  • Progression to diabetes.
  • Weight loss.
  • Adverse events (AEs).
  • Changes in blood glucose.
  • Changes in diet and physical activity.
  • Changes in blood cholesterol.

Exclusion criteria

Population
  • People with diabetes.
Study design
  • RCTs with < 2 years' duration.
  • Study designs other than RCTs.

Search strategy

Electronic databases were searched for published systematic reviews, RCTs, economic evaluations and ongoing research up to September 2007. The databases searched were MEDLINE, EMBASE, The Cochrane Library, Science Citation Index, the National Research Register and the UKCRN. Appendix 2 shows the databases searched and the strategy in full. Updating searches were carried out in February 2011, mainly to identify more papers from the main studies. Auto-alerts on MEDLINE were run until September 2011. Selective updating searches in MEDLINE and EMBASE were carried out in January 2012, focusing on new cost-effectiveness analyses and recent reviews.

Abstracts returned by the search strategy were examined independently by two researchers (AS and NW) and screened for inclusion and exclusion. Disagreements were resolved by discussion and consultation with a third researcher (PR). Full texts of the identified studies were obtained. Two researchers (AS and MI) examined these independently.

Data extraction

Two reviewers (AS and MI) extracted data regarding study design and characteristics, details of the intervention, and patient characteristics and outcomes into a specially designed form. Differences in data extraction were resolved by discussion, referring back to the original paper and in consultation with PR and NW.

Quality assessment

To assess the quality of the RCTs, the following criteria were used:

  1. method and description of randomisation
  2. description of attrition/losses to follow-up
  3. specification of eligibility criteria
  4. blinding
  5. power calculation
  6. robustness of outcome measurements
  7. similarity of group participants at baseline
  8. data analysis.

Overall study quality was rated as follows: A (all quality criteria met), B (one or more of the quality criteria only partially met) or C (one or more criteria not met).

Internal validity:

  • sample size

    power calculation at design

  • selection bias

    explicit eligibility criteria

    proper randomisation and allocation concealment

    similarity of groups at baseline

  • performance bias

    similarity of treatment other than the intervention across groups

  • attrition bias and intention-to-treat (ITT) analysis

    all patients accounted for

    number of withdrawals specified and reasons described

    analysis undertaken on an ITT basis

  • detection bias:

    blinding

    objective outcome measures

    appropriate data analysis

    any potential conflict of interest was noted.

Results

Systematic reviews

Five good-quality systematic reviews were identified. These were ones that scored highly using the five quality criteria as used for the NHS Centre for Reviews and Dissemination Database of Abstracts of Reviews of Effects (DARE). Four of these met five quality criteria,38,95,179181 and the other met four and was uncertain on one.182 These reviews do not include all of the trials now available, nor do they include recent papers from some trials which were available.

The AHRQ published a review of the evidence on the diagnosis, prognosis and treatment of IGT and IFG in 2005, which included six trials of lifestyle interventions.38 The authors included studies of > 6 months' duration. As will be reported later, that seems too short because changes are often achieved with lifestyle interventions in the short term which do not persist. We preferred a minimum follow-up of 2 years. However, five of the six trials had a duration of ≥ 2 years. A meta-analysis of four trials of combined diet and exercise gave a RR of 0.54 (95% CI 0.42 to 0.70). The review concluded that there was good evidence that lifestyle interventions could reduce the risk of diabetes in people with IGT.

The Diabetes Australia Guideline covered a wider range of topics, but included consideration of the roles of physical activity and diet in reducing the risk of diabetes.95 The guideline review concluded that exercise can slow the progression from IGT to T2DM, and that reduction in dietary fat, especially saturated fat, also reduces the risk of diabetes. It also concluded that a combined diet and exercise programme was more effective than either alone.

A review by Gillies et al. (2007),179 published during the preparation of this review, examined the evidence for both pharmacological and lifestyle interventions.179 They noted that the trials were heterogeneous in the interventions, ethnicity, weight and age, but they carried out meta-analyses that gave the following HRs:

  • diet 0.67 (95% CI 0.49 to 0.92)
  • exercise 0.49 (95% CI 0.32 to 0.74)
  • combined diet and exercise 0.49 (95% CI 0.40 to 0.59)
  • all studies pooled 0.51.

The studies that carried most weight in their meta-analysis were the Finnish DPS183 and the US DPP,108 which will be described in detail later. The longer term results from the Finnish trial were not then available.

The Cochrane review by Norris et al. (2005)180 focused on weight loss or control. In four studies of at least 1 year's duration, mean weight loss was 2.8 kg, and in the two studies of 2 years' duration, it was 2.7 kg. The weight loss results from the DPP were larger (4.9 kg at 2 years) but were not included in the meta-analysis because of lack of data on distribution.

Yamaoka et al. (2005)182 reviewed trials on ‘long-term’ (but included studies of ≥ 6 months) non-pharmacological weight loss interventions in pre-diabetes. They concluded that the incidence of T2DM could be reduced by about 50%. They excluded the DPP study,108 but concluded that the meta-analysis of smaller studies matched the results from DPP.

Two Cochrane reviews have looked at separate elements of lifestyle change. Nield et al. (2008)184 set out to assess the effects of dietary advice for preventing T2DM but included only two trials, one being the Da Qing trial described later in this chapter. The other trial did not appear to be restricted to people with IGT. Nield et al. (2008)184 concluded that there was a lack of good data on prevention of T2DM by diet alone.

Orozco et al. (2008)185 assessed the effects of exercise alone or exercise and diet compared with standard advice. They concluded that the combination was effective, but not exercise alone.

Randomised trials

Nine RCTs comparing lifestyle intervention with standard care of IGT were identified: DPP,108 Kosaka et al. (2005),186 Liao et al. (2002),187 Mensink et al. (2003),188 Oldroyd et al. (2006),189 Da Qing,190 Indian DPP,173 Finnish DPS183 and Wein et al. (1999).191

Some trials gave rise to many papers, only some of which are cited below. Where there are multiple papers from a study, we cite the website from where, in most cases, details of the protocol, copies of the papers and slides can be downloaded.

The Diabetes Prevention Program

The DPP study repository website contains full details of the DPP, including a complete protocol, results, and list of publications.192

Description and quality of study

The DPP included 3234 participants with IGT, and compared intensive dietary and physical activity advice with standard advice.108 The sample size necessary to achieve 90% statistical power was estimated to be 2279 participants. This was based on two main assumptions, (1) an expected conversion rate to diabetes of 6.5 per 100 person-years among participants assigned to the standard lifestyle recommendations plus placebo, and (2) that for participants assigned to intensive lifestyle or metformin intervention groups, the diabetes development HR is reduced by ≥ 33%, i.e. to < 4.33 per 100 person-years. The primary outcome was development of diabetes by ADA criteria, using an OGTT.

Participants were recruited using a variety of methods including volunteering in response to advertisements, open screening and referral by health-care providers with the aim of recruiting ≥ 50% women, ≥ 50% ethnic minority and roughly 20% aged ≥ 65 years old. Participants were randomised (stratified by clinical centre) into three treatment groups:

  • intensive lifestyle intervention
  • standard advice plus metformin
  • standard advice plus placebo.

A fourth arm with randomisation to troglitazone was discontinued in 1998 once the toxicity of that drug was realised.

The treatment groups were similar at baseline with respect to age, sex, race, weight and BMI. Baseline physical activity was reported; one of the inclusion criteria for the DPP was being able to walk one-quarter of a mile in 10 minutes. Baseline leisure physical activity levels of the DPP participants using the modifiable activity questionnaire (MAQ) showed that women reported being less active than men (p < 0.0001), and older individuals (> 60 years of age) reported more leisure physical activity (p < 0.0001) than younger age groups.

Baseline diet in terms of total energy intake and fat consumption was reported for only the intensive lifestyle intervention group.193 Treatment regimens were reported in detail. Staff were provided with ongoing training and provision of intervention materials. Drug administration was double blind; however, if a diagnosis of diabetes was made then participants, investigators and primary care providers were unblinded to the diagnosis and measurements. It was not reported whether any concurrent medication (apart from metformin) was taken; however, participants were excluded at screening if they were using medications known to impair glucose tolerance. Patients were assessed over a mean follow-up of 2.8 years (range 1.8 to 4.6 years) using self-reporting of diet and physical activity and using objective methods for all other measurements. Study attrition was 8% by the end of the study (92.5% had attended a scheduled visit within 5 months of the close of the study). AEs were reported according to treatment group.

Participants

The DPP trial recruited 3234 overweight participants with IGT of ≥ 7.8 to < 11.1 mmol/l (2-hour plasma glucose detected by a 75 g OGTT) and a FPG of between 5.3 and 6.9 mmol/l.

Participants came from a range of ethnic groups: 54.7% were white, 19.9% were African American, 15.7% were Hispanic, 5.3% were American Indian and 4.4% were Asian. In total, 67.7% were female and 69.4% had a family history of diabetes. The minimum age was 25 years. The mean age (± SD) of the 3234 participants was 50.6 ± 10.7 years and the mean weight was 94.2 ± 20.3kg. The minimum BMI was 24 kg/m2. The majority of participants had a BMI of < 40 kg/m2; however, the BMI was ≥ 40 kg/m2 in 8% of men and 21% of women. The overall mean BMI was 34 kg/m2: 30.8% had BMIs of 30 kg/m2 to < 35 kg/m2 and 36.9% had BMIs of ≥ 35 kg/m2, so overall 67.7% had BMIs of ≥ 30 kg/m2. In men, 56.5% had BMIs of ≥ 30 kg/m2, and 73% of women had BMIs of ≥ 30 kg/m2.

A history of and/or treatment for hypertension was present in 27% of participants. More than 37% of men and 33% of women reported a history of and/or treatment for high cholesterol. The authors noted: ‘the DPP cohort includes individuals who are more overweight and hyperinsulinaemic and less hypertensive than the subjects in other studies’. As such, ‘DPP participants may be less susceptible to hypertension-related morbid events that may confound the secondary CVD outcomes attributed to IGT or hyperglycaemia per se’.192

One issue in interpreting trials is the representativeness of the recruits. Figure 5, below, shows the stages of recruitment to the DPP.

FIGURE 5. Screening and recruitment for the DPP.

FIGURE 5

Screening and recruitment for the DPP. Reprinted from Controlled Clinical Trials, volume 23, Rubin et al., authors, The Diabetes Prevention Program: recruitment methods and numbers. pp. 157–171. Copyright (2002) with permission from Elsevier. (more...)

About 80% of the 158,177 potential participants were eliminated between steps 1 and 2. About one-third gave no reason for exclusion, but for those for whom a reason was available it was found that the five primary reasons for stopping after step 1 included (1) choosing not to have an OGTT (18%); (2) being excluded because the finger-stick glucose reading was outwith the entry criteria (17%); (3) BMI (12%); (4) being excluded because taking medications, such as thiazide diuretics, was likely to confound assessment for diabetes (7%); and (5) being given a diagnosis of diabetes (6%).

There was another large reduction (26,266 individuals) at step 2, which involved administration of an OGTT to determine glucose tolerance criteria for eligibility. About two-thirds of these were people whose fasting or 2-hour OGTT results were not in the eligible range. Many people were excluded at this stage for more than one reason. Step 3 was a 3-week run-in period; 81% (3819) who started the run-in were eventually randomised to one of the study arms.

Most of the exclusions of the initial volunteers, therefore, were based on plasma glucose results. However, many people selected themselves out, and this should be borne in mind when considering the generalisability of the results.

Intervention

In the DPP trial, the intensive lifestyle intervention group was assigned a goal of achieving and maintaining at least 7% weight loss through low-calorie, low-fat diet and moderate-intensity physical activity (at least 150 minutes/week). Participants were encouraged to achieve the weight loss (through reduction in dietary fat intake to < 25% of calories) and exercise goals within the first 24 weeks. Sixteen individual (one-to-one) sessions with a case manager (‘lifestyle coach’) within this time covered general information about diet and exercise and behaviour strategies, such as self-monitoring, goal-setting, stimulus control, problem-solving and relapse prevention training (i.e. resource intensive). During maintenance, group courses were offered every 3 months on topics related to exercise, weight loss or behaviour issues. Two comparator groups were randomised to standard lifestyle advice with either placebo (tablets twice daily) or metformin (850 mg twice daily). The standard lifestyle advice (given to all participants including the intensive lifestyle intervention group) consisted of written information and a 20- to 30-minute individual advice session recommending 5–10% weight loss and 30 minutes of physical activity 5 days a week. In addition, participants were advised to avoid excessive alcohol intake and stop smoking. Participants were reviewed annually. The average duration of intervention and follow-up was 2.8 years (range 1.8 to 4.6 years).

Results

Primary outcomes

Progression to diabetes The DPP assessed the incidence of diabetes by annual OGTT testing or semi-annual FPG testing over 4 years. In addition, testing was prompted if symptoms arose that were suggestive of diabetes. Results were expressed in four ways: (1) the number of cases per 100 person-years; (2) per cent reduction in incidence; (3) estimated cumulative incidence at 3 years; and (4) number needed to treat (NNT) one patient with diabetes.

The number of cases per 100 person-years was significantly lower in the lifestyle intervention and metformin groups than in the placebo group (4.8, 7.8 and 11.0 in the lifestyle, metformin and placebo groups, respectively) equating to 58% (95% CI 48% to 66%) and 31% (95% CI 17% to 43%) lower incidence in the lifestyle and metformin groups, respectively, compared with the placebo group. At 3 years the estimated cumulative incidence of diabetes was 14.4%, 21.7% and 28.9% in the lifestyle, metformin and placebo groups, respectively (so they were quite a high-risk group).

The NNT to prevent one case of diabetes during a 3-year period was estimated to be 6.9 for the lifestyle intervention and 13.9 for metformin. The authors stated:

the incidence of diabetes in the placebo group was higher than expected perhaps owing to greater frequency of glucose testing or selection of persons at higher risk.108

Subgroup analysis (note: the study had inadequate power for this analysis) found that treatment effects did not differ significantly according to sex, race or ethnic group; however, the effect of the lifestyle intervention was significantly greater among participants with lower baseline glucose concentrations 2 hours after glucose load. Similarly, the effect of the lifestyle intervention over metformin was greater in older participants and those with lower BMIs. The authors stated:

The racial and ethnic-group differences in incidence of diabetes were perhaps reduced by the selection of participants who were overweight, and had elevated fasting and post-load glucose concentrations which are three of the strongest risk factors for diabetes.108

After adjustment for weight change in the lifestyle group, no independent effects of increased physical activity or decreased per cent fat on diabetes risk were found.193

Several different measures of body size at baseline were predictive of the subsequent development of diabetes.83 When analysed at 3.2 years, large WC at baseline was a better predictor of risk for developing diabetes in both sexes than other measures in the placebo and lifestyle groups. The HR for WC was 1.29 (p < 0.01) and 1.53 (p < 0.01) for women in the placebo and lifestyle groups, respectively, and 1.43 and 1.49 for men (adjusted for age and self-reported race/ethnicity) relative to smaller waists. A graded increase in the risk of developing diabetes was seen as the tertile of WC increased in both lifestyle and placebo groups.

The age of participants had an impact on progression to diabetes.194 Diabetes incidence rates did not vary by age in the placebo group (11, 10.8 and 10.3 cases per 100 person-years in young, middle-aged and older groups, respectively, p = 0.71). In contrast, intensive lifestyle intervention was more effective with increasing age (6.3, 4.9 and 3.3 cases per 100 person-years, in the 25- to 44-year, 45- to 59-year and 60- to 85-year age groups, respectively, p = 0.007). Those in the oldest age group lost more weight and were more physically active.

Diabetic retinopathy was detected in 12.6% of the participants who developed diabetes during the DPP compared with 7.9% of those who did not (p = 0.03). The only characteristics reported to be different between those who developed retinopathy and those who did not were HbA1c level (6.4% ± 0.55% vs 6.2% ± 0.63%; p < 0.05 – as reported in the paper, although the difference of only 0.2% seems trivial) and systolic blood pressure (SBP) (128 mmHg vs 125 mmHg; p < 0.05).195

Regression to normal glucose values The DPP measured FPG and HbA1c levels in all participants every 6 months for up to 4 years. Both FPG and HbA1c levels changed significantly over the time of the study. In the first year, the FPG of the lifestyle and metformin groups decreased; however, in subsequent years the FPG values increased and returned to baseline levels by 2.5 years; further increases were observed thereafter up to 4 years (significant difference between groups; p < 0.001). In the placebo group, FPG values increased at each time point from baseline up to 4 years (significant difference between groups 0.5–3 years; p < 0.001). HbA1c values showed a similar initial decrease in the lifestyle and metformin groups; however, both groups showed an increase thereafter with the metformin group lying between lifestyle and placebo values, as shown in Figure 6. At the start, the HbA1c level was 5.9%. At 4 years, the means were, approximately, placebo 6.1%, lifestyle 6.0% and metformin 6.0%.

FIGURE 6. Mean change in HbA1c level vs years from randomisation in the DPP.

FIGURE 6

Mean change in HbA1c level vs years from randomisation in the DPP. Reprinted with permission from the Massachusetts Medical Society. The Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes with lifestyle intervention (more...)

The placebo group saw a constant increase in values from baseline to 4 years. The percentage of participants with normal glucose concentrations was greater at all time points up to 4 years in the lifestyle intervention group compared with both metformin and placebo groups. At 4 years, the percentages with normal fasting glucose were lifestyle 54.1%, metformin 45.1% and placebo 43.8%. The percentages with normal 2-hour glucose were lifestyle 37.8%, metformin 27.9% and placebo 24.2%. The percentages with both normal fasting and 2-hour glucose levels were lifestyle 29.6%, metformin 19.7% and placebo 17.2%. There were no differences in the percentages reverting to NGT among the age groups.164

The 10-year results showed continuing but reduced benefit, with the cumulative incidence of diabetes reduced by 34% in the lifestyle group (and by 18% in the metformin group). This was despite all groups being offered a modified form of the original intervention (in groups rather than 1 : 1) after the end of the trial, and despite the metformin group being allowed to continue on the drug.

Adverse events Gastrointestinal symptoms were less frequent in the lifestyle group than in the placebo group (12.9/100 person-years vs 30.7/100 person-years, p < 0.02), whereas the rate of GI symptoms was significantly higher in the metformin group than in the placebo group (77.8/100 person-years vs 30.7/100, p < 0.02). The rate of musculoskeletal symptoms was significantly higher in the lifestyle group (24.1/100 person-years) than in both the metformin and placebo groups (20.0 and 21.1/100 person-years, respectively). No significant differences between groups were seen in hospitalisation or deaths.

Secondary outcomes

Weight loss Data on weight change were available for 3085 participants at 6 months, 3064 participants at 1 year, 2887 at 2 years and 1510 at 3 years. Participants assigned to the lifestyle intervention had significantly greater weight loss than participants assigned to receive metformin or placebo. Over 4 years the average weight loss was 5.6 kg, 2.1 kg and 0.1 kg in the lifestyle, metformin and placebo groups, respectively (p < 0.001).

Weight loss, reduction in WC and percentage of participants who achieved the 7% weight loss goal all increased with age.194 The percentages of participants who achieved the weight loss goal (7%) of body weight were 25–44 years = 33%, 45–59 years = 39% and 60–85 years = 55%. Participants aged 60–85 years had the most weight loss.

At the 10-year follow up, the lifestyle group had regained most of the weight, going from a nadir of a mean 7 kg reduction at 1 year back up to a mean weight loss of 2 kg.

What makes the difference? One of the papers from the DPP explored the relative contributions of the changes in weight and exercise achieved by studying the intensive lifestyle group.193 Dietary and exercise data were collected by questionnaires. The number of participants with ≥ 3-year or longer follow-up was only 638 because recruitment went on until May 1999 and the data were collected at end of July 2001.

Hamman et al. (2006)193 reported that the mean weight loss at 3 years was 4.1 kg. Physical activity increased at each year end compared with baseline. The percentage of calorie intake from fat fell from 34% at baseline to 28% at the end of year 1. A total of 153 participants developed diabetes – a rate of 5 per 100 person-years. When all of the changes were examined in a multivariate Cox model, weight loss was the dominant factor. Even small amounts of weight loss helped – on average there was a 16% reduction in the risk of diabetes per kilogram of weight loss. Weight loss reduced diabetes incidence similarly across all race/ethnicity groups, for both sexes, for all ages, and for several levels of physical activity, regardless of the level of initial obesity.

Hamman et al. (2006)193 then looked at the effects of meeting the various goals at year 1, dividing participants into eight subgroups. The subgroup that met all of the goals had the lowest risk of diabetes, with an 89% reduction in risk (HR 0.11, 95% CI 0.05 to 0.24; p < 0.0001) compared with the group meeting none of the goals. Weight loss was responsible for much of the reduction, but there was halving of the incidence of diabetes in those who met the physical activity goal (150 minutes of moderate-intensity activity per week) as shown in the comparison of subgroups 3 and 4 compared with 1 and 2. The reduction was 46% after adjusting for baseline variables and 44% (HR 0.56; 95% CI 0.36 to 0.89; p = 0.012) after adjusting for weight loss over follow-up. There is of course an interaction, in that increased physical activity helped to sustain weight loss. However, among 495 recruits who did not achieve the 12-month weight loss target (7% or 6.6 kg on average), the incidence of diabetes was reduced by 44% in those who achieved the physical activity target. Note that this group did lose some weight (mean 2.6 kg) but adjustment for this had little impact on the effect of physical activity.

The results may also indicate that at least ∼150 minutes per week of moderate activity are required before an effect on diabetes risk is achieved.

Adherence

Dietary intake DPP examined the adherence of participants to intervention. Diet adherence was assessed at 1 year only and showed that the lifestyle group had a significantly greater reduction in daily energy intake over the first year than metformin and placebo groups (−450, −296 and −249 kcal, respectively; p < 0.001). This was associated with a significantly greater change in the average intake of fat from baseline to 1 year (−6.6%, −0.8% and −0.8% of total calories, respectively; p < 0.01). Data for longer time points were not reported. There was no significant difference in reduction of reported caloric intake among different age groups, although older (60–85 years) recruits reduced calorie intake less (by 10% at 12 months) than younger recruits (25–44 years; 18%).194

Physical activity Compared with baseline, all study groups had increased their physical activity at 1, 2, 3 and 4 years. However, participants in the lifestyle intervention group had significantly greater change in physical activity at all time points than both the metformin and placebo groups, the physical activity of which increased only slightly over baseline values.108 The authors stated:

After adjustment for weight change no independent effect of increased physical activity or decreased per cent fat on diabetes risk was found.193

Paradoxically, those participants who met the physical activity goal of 150 minutes/week of moderate activity had a 44% reduction in diabetes risk, independent of weight loss.

In a representative sample of DPP lifestyle participants (n = 274; 94% of the final 293 lifestyle participants randomised), characteristics that correlated with high levels of baseline, 1-year and end-of-study physical activity were (1) being a man and (2) having lower BMI and lower perceived stress, depression and anxiety scores at baseline. Higher baseline BMI and being a woman correlated with lower baseline, 1-year and end-of-study physical activity levels, with women having significantly higher BMIs and higher levels of depression, anxiety and perceived stress than men.196

Older age was an independent predictor of achieving the goal of 150 minutes of physical activity at 1 year and 2 years. Lifestyle participants aged > 60 years achieved greater minutes of physical activity and greater per cent weight loss and greater risk reductions for developing diabetes (71% risk reduction compared with 48% risk reduction in persons aged 25–44 years). Higher levels of baseline physical activity correlated with greater readiness to change physical activity levels (p < 0.0001) and lower levels of perceived stress (p = 0.009), depression (p < 0.003), and anxiety (p = 0.03) at baseline, 1-year and end-of study levels.196

Perhaps time is a factor, with those who are older having more time post retirement. The older age groups had the highest MET-hours of activity and did better with lifestyle change than younger groups, although it should be noted that they started leaner and with better insulin sensitivity.194

Metabolic syndrome Recruitment to DPP was on the basis of IGT, but a later paper by Orchard et al. (2005)197 reported the prevalence of metabolic syndrome at baseline and the effects of the interventions.

The metabolic syndrome was defined by the criteria from the National Cholesterol Education Program's Adult Treatment Panel III, with three or more of the following:

  • WC of > 102 cm in men and > 88 cm in women
  • serum triglyceride level of at least 1.7 mmol/l
  • HDL-C level of < 1.03 mmol/l in men and < 1.3 mmol/l in women
  • blood pressure of ≥ 130/85 mmHg
  • FPG level of 6.2 mmol/l.

The metabolic syndrome was present in 53% (n = 1711) of the 3234 participants at baseline, with little variation by age. After 3 years, the prevalence of the metabolic syndrome increased from 55% at baseline to 61% in the placebo group (p = 0.003), did not change (54–55%) in the metformin group (p > 0.2) and decreased from 51% to 43% (p < 0.001) in the lifestyle group. The decrease in the lifestyle group correlated most strongly with decreases in WC and blood pressure.

Fujimoto et al. (2007)198 examined the relationship between changes in body fat and progression to diabetes by 1 year. At baseline, recruits were mainly obese, with mean BMI of 32.1 kg/m2 in men and 33.0 kg/m2 in women. Visceral and subcutaneous fat was measured by CT, as well as by standard measurements such as BMI. The lifestyle group had big reductions in both visceral fat (reduced at L2–3 by 24% in men and 18% in women at 1 year) and subcutaneous fat (again at L2–3, reduced by 16% in men and 11% in women). Progression to diabetes was associated with fat changes differently in the arms of the study. In the metformin group, the reduced diabetes risk was independent of body fat changes. In the placebo group, only the subcutaneous fat changes correlated with diabetes risk, and then only in men. In the lifestyle group, all of the fat variables correlated with diabetes reduction in men. In women, weight, BMI and WC were significant predictors but the association with visceral fat did not quite reach statistical significance.

After the end of the randomised trial, all three groups (lifestyle, metformin and placebo) were offered the lifestyle intervention, albeit in groups rather than the original individualised provision. At the 10-year follow-up, the original lifestyle group had regained about 5 kg of their original (by 12 months) 7 kg weight loss, so that by 10 years their weight was little different from the metformin group, who lost about 2.5 kg.199 However, the incidence of diabetes by 10 years remained lower in the former lifestyle (by 34%) group and metformin (by 18%) group than in the former placebo group. The onset of T2DM was delayed by about 4 years with lifestyle and by about 2 years by metformin. One feature of note was that not only was the best effect of lifestyle seen in the 60- to 85-year age group, but also that this group had no significant response to metformin.199

Perreault et al. (2009)200 noted that in some diabetes prevention trials, glucose tolerance regressed to normal, and used DPP data to examine the factors that were associated with this. The factors included baseline evidence indicating a milder condition (lower baseline FPG and 2-hour plasma glucose) and those indicating response to the intervention, especially greater weight loss. As reported by Hamman et al. (2006)193 for every 1 kg weight loss there was a 16% reduction in diabetes risk, and it was the most important predictor of regression.193 However, Perreault et al. (2009)200 noted that intensive lifestyle interventions appeared to have other components that were independent of weight loss, most probably physical activity.

In another paper, Perreault et al. (2008)201 noted that meeting the goals of the lifestyle intervention was a strong predictor of reduction in progression to diabetes. However, men in the lifestyle arm met more goals than women but had the same progression to diabetes. Perreault et al. (2009)200 explain this on the basis that men were at higher risk from baseline.

The DPP also examined changes in cardiovascular risk factors over time, and found that among those who progressed to diabetes, there were rises in blood pressure and triglycerides, and a fall in HDL-C.202 These changes were statistically significant, although quite small. Those who regressed from IGT to NGT showed reductions in blood pressure and triglycerides, and improvements in HDL-C levels. These improvements in HDL-C levels were more marked in the lifestyle group than in the metformin one, and LDL cholesterol (LDL-C) also fell in the lifestyle group. Goldberg et al. estimated that in the lifestyle group these changes should bring about a 10–13% reduction in heart disease.

The effect of metformin has been examined in several of the DPP papers. In the main study report it was noted that metformin reduced the risk of diabetes by 31%.108 However, the effects varied among groups. Metformin had little effect in the oldest age group, whereas lifestyle change was more effective with increasing age.194 Crandall et al. (2006)194 suggest that this may be related to the oldest age group (60–85 years) being leaner, and having less insulin resistance, at baseline. Metformin may work by reducing insulin resistance and the older group will have less to gain. Another DPP paper reported that metformin was effective mainly in those with BMIs of > 35 kg/m2. In this BMI subgroup, metformin was as effective as lifestyle.83

Some of the benefits of metformin were the result of weight loss rather than improvements in insulin sensitivity.203 Lachin et al. (2007)203 estimated that 64% of the metformin benefit was mediated through weight loss. Interestingly, they note that the effect of weight loss achieved with metformin appears to be less than the same weight lost by lifestyle change.

Most DPP papers report prevention of diabetes, but the 10-year follow-up199 also reported that diabetes was delayed, with the time at which 40% of the (high-risk) groups progressed to diabetes being 4 years later with lifestyle and 2 years later with metformin compared with placebo.

Ackermann et al. (2009)204 estimated the direct utility of weight loss, after adjusting for a range of possible confounding factors. They used the SF-6D utility measure, and estimated that there was a gain in utility of 0.007 (on a scale of 0.29–1.00) for every 5 kg of weight loss. Although this fell below the minimum effect size that was deemed to be clinically important (0.04), they considered that the change could improve cost-effectiveness if sustained for years.

In summary, DPP found that intensive diet and exercise intervention in those with IGT significantly reduced the incidence of diabetes by 58% compared with standard lifestyle advice. The reduction was sustained, with a cumulative incidence of diabetes at 6 years of 23% in the intervention group, compared with 38% in the control groups. At 10 years, the incidence of diabetes was 34% lower in the lifestyle group and 18% lower in the metformin group compared with the control group. This was despite the lifestyle group regaining much of the 7 kg weight lost in the first year.

The Finnish Diabetes Prevention Study

Data were obtained from the main study by Tuomilehto et al. (2001),183 with supplementary information cited where appropriate from additional DPS papers.205219

Description and quality of trial

This RCT of 523 participants with IGT in Finland compared intensive lifestyle intervention with a control group. Participants were recruited by screening high-risk individuals (e.g. obese subjects or first-degree relatives of those with T2DM), who were identified through previous epidemiological surveys and advertisements. Those with IGT (according to WHO 1985 criteria) on two separate OGTTs were recruited. Additional inclusion criteria were age 40–64 years and BMI of > 25 kg/m2. Power calculation estimated that, in order to detect a 35% reduction in incidence of diabetes with 80% power at 5% significance level, 3252 person-years were required, i.e. 650 subjects to be followed for 5 years or 542 subjects to be followed for 6 years. A total of 523 participants were randomised into two groups: a diet-plus-exercise group and a control group.

Treatment groups were not significantly different at baseline apart from two measurements: SBP and per cent total energy consumed as saturated fat (see Participants, below). Treatment regimens were reported in detail. Use of concurrent medication was reported. Participants were assessed over 6 years (mean follow-up 3.2 years) using subjective self-reporting of diet and physical activity, and using objective measures for all other outcomes. Forty subjects (8%) withdrew from the study: 23 in the intervention group and 17 in the control group. Of these, nine could not be contacted, three withdrew because of severe illness, one died and 27 withdrew for personal reasons. AEs were not reported. No conflict of interest was reported.

Participants

Tuomilehto et al. (2001)183 recruited 523 Finnish men and women with IGT (as defined by WHO 1985 criteria: FPG of < 7.8 mmol/l and/or 75 g OGTT 2-hour ≥ 7.8 mmol/l and < 11.1 mmol/l). Exclusion criteria were: previous diagnosis of diabetes mellitus [other than gestational diabetes mellitus (GDM)]; persons involved regularly in a vigorous exercise programme; subjects receiving treatment to lower blood pressure other than routine dietary and health advice; persons with any chronic disease making a 6-year survival improbable; other medical characteristics that were likely to interfere with participation in the study; and subjects with clinical conditions, such as thyroid and liver diseases, which could interfere with glucose metabolism. The male–female ratio was 33 : 67 and mean age (years ± SD) was 55 ± 7 years. Mean BMI was 31.4 ± 4.6 kg/m2 in the intervention group and 31.0 ± 4.5 kg/m2 in the control group, and mean weight was 86.7 ± 14.0 kg versus 85.5 ± 14.4 kg, respectively. Sixty-six per cent of participants in the intervention group had first-degree relatives with diabetes compared with 61% in the control group. Five per cent of the intervention group were taking cholesterol-lowering drugs compared with 6% of the control group; 30% of the intervention group were taking antihypertensive drugs compared with 31% of the control group. Subjects receiving treatment to lower blood glucose were excluded. Baseline characteristics were similar across groups apart from (1) SBP (± SD) at baseline was slightly, but significantly, higher in the diet-plus-exercise group than in the control group (140 ± 18 vs 136 ± 17; p = 0.03 between groups) and (2) proportion (% ± SD) of total energy consumed as saturated fat was significantly higher in the control group [17 ± 4.3 percentage total energy (E%)] than the intervention group (16.2 ± 4.0 E%; p = 0.019 between groups).

A further paper209 reported the prevalence of the metabolic syndrome, using a modified WHO definition. In the DPS cohort, 78% of the men and 72% of the women had the metabolic syndrome. The mean BMI was higher in women (32 kg/m2) than in men (30 kg/m2) but WHR was higher in men. Men had slightly higher diastolic blood pressure (DBP) and FPG, but lower HDL-C. Obesity was seen in 96.5% of men and 66.3% of women, hypertension in 62.9% of men and 60.9% of women, and dyslipidaemia in 51.2% of men and 48.6% and women.

Intervention

Participants were randomised to two groups: (1) intensive lifestyle (diet-plus-exercise) group and (2) control group. Participants randomised to the diet-plus-exercise intervention group (n = 265) received 15 individual sessions with a nutritionist over 3 years. Participants were advised to consume a diet with: 50% of daily calories from carbohydrates; < 30% fat; < 10% from saturated fat; < 20% from mono- and polyunsaturated fat; < 300 mg/day cholesterol; ≥ 15 g/1000 kcal fibre and approximately 1 g protein/kg ideal body weight/day. Additional topics covered included diabetes risk factors, problem-solving and physical activity. Attendances at further group sessions, expert lectures, low-fat cooking lessons and visits to local supermarkets was encouraged. Participants were also encouraged to lose weight at a rate of 0.5–1 kg/week towards a goal of BMI of < 25 kg/m2 or ≥ 5% weight reduction. (Note: in 48 participants very low-calorie diets were considered after 6 months.) Participants had their level of physical activity assessed individually and were advised to increase their overall level of physical activity making use of supervised (and individualised) exercise sessions (endurance training, resistance training and voluntary group walking). The control group were given general information at baseline about lifestyle and diabetes risk. They were advised to adjust their diet to reduce BMI to < 25 kg/m2, consume < 30% energy as fat, reduce alcohol intake and stop smoking. Counselling was done individually or in one group session (but not individualised, i.e. not tailored to the individual) with annual follow-up visits.

The cost of the intervention diet was, if anything, slightly less than their usual diet had been.216 Dietary fibre and fat contents were strong predictors of success, even after adjustment for weight loss.213

Results: primary outcomes

Progression to diabetes Tuomilehto et al. (2001)183 assessed the incidence of T2DM (WHO 1985 criteria) by annual oral glucose tolerance testing for up to 6 years (mean follow-up 3.2 years). Results were expressed in a number of ways. First, the cumulative incidence of diabetes (years 1–6) was significantly lower (58%) in the intervention group than in the control group [year 1, 5 (2%) in intervention group vs 16 (6%) in control group; year 2, 15 (6%) in intervention group vs 37 (14%) in control group; year 3, 22 (9%) in intervention group vs 51 (20%) in control group; year 4, 27 (11%) in intervention group vs 53 (23%) in control group; year 6, 27 (10%) in intervention group vs 59 (23%) in control group; RR = 0.4; p < 0.001 between groups]. The absolute incidence of diabetes (number of cases per 1000 person-years) was 32 cases per 1000 person-years in the intervention group compared with 78 per 1000 person-years in the control group. Tuomilehto et al. (2001)183 concluded that 22 subjects with IGT must be treated for 1 year (or five subjects for 5 years) to prevent one case of diabetes.

In a later publication, in which mean follow-up was 4.1 years, 114 of the participants (intervention and control group combined) had been diagnosed with diabetes.214 Those who developed diabetes were more obese at baseline and had higher fasting and 2-hour plasma glucose values. Study authors noted that ‘those subjects who developed diabetes during the first 2 years of the trial did not have a 3 year examination’ [also noted in the 4-year publication by Uusitupa et al. (2003)218]; this may have affected the apparent effectiveness of the intervention.

Results: secondary outcomes

Weight loss Tuomilehto et al. (2001)183 assessed weight loss annually using standard methods. Results were presented as mean change in body weight (kilograms and per cent) from baseline to year 1 and year 2. The study found that both groups had lost weight at both time points but the intervention group had significantly greater weight loss than the control group (4.2 kg vs 0.8 kg at 1 year; p < 0.0001; and 3.5 vs 0.8 kg gain at 2 years; p < 0.001). By 3 years the weight reductions were 3.5 kg in the intervention group and 0.9 kg in the control groups.215

BMI was also assessed annually and showed a similar significantly greater improvement in the intervention group (−1.3 kg/m2) compared with the control group (−0.3 kg/m2) (p < 0.0001 for year 1 and year 3 between groups).

At the 7-year follow-up, the control group had lost 0.7 kg since baseline, and the intervention group 3.1 kg, and the modelling showed that weight loss was the main predictor of success. Weight loss did not vary by level of education when level of education was divided into tertiles.220

Blood pressure (1-year data from Tuomilehto et al. (2001);183 Uusitupa et al. (2000)219 based on ‘incomplete database’183) Blood pressure levels improved significantly more in the intervention group than in the control group. SBP fell 5 mmHg in the intervention group and did not change in the control group when measured at 2 years (p = 0.0005). DBP fell 5 mmHg in the intervention group and 3 mmHg in the control group, p = 0.0125 (derived from the AHRQ review38).

Adherence

Dietary intake Tuomilehto et al. (2001)183 examined the adherence of participants to intervention. Diet adherence was assessed subjectively over the first year by examining self-reported 3-day food records filled in by participants and completed four times during the year. Data were presented as percentage of groups achieving specific dietary goals. From baseline to year 1, compared with the control group, a significantly greater (p = 0.001) proportion of participants in the intervention group achieved separate goals of fat intake < 30% total energy (47% intervention group vs 26% control group), saturated fat intake < 10% total energy intake (26% vs 11%) and fibre intake > 15 g/1000 kcal (25% vs 12%). Change in dietary intake from baseline to 1 year and 3 years was reported in detail in Lindstrom et al. (2003).215 Compared with the control group, the intervention group showed a significant decrease in total energy consumed (kcal/day), E% as fat, saturated fat (E%) and monounsaturated fat (E%) from baseline to year 1 and year 3. Similarly, the intervention group showed a significant increase from baseline at 1 year and 3 years in E% of carbohydrates, fibre density (g/1000 kcal) and intake of both water-soluble and insoluble fibre (g/1000 kcal) compared with control groups. No significant differences between groups were seen in alcohol or polyunsaturated fat consumption.

The association between dietary macronutrient composition and change in body weight and WC and diabetes risk was assessed using DPS data.213

During a mean follow-up of 4.1 years, weight loss was related to an increase in fibre [p-value (p) for trend = 0.001] and decrease in fat (p for trend = 0.018) and energy density (p for trend = 0.001). Reduced diabetes risk was associated with higher fibre density (p for trend = 0.01) and lower fat intake (p for trend = 0.004), after adjusting for group assignment, sex, age, baseline weight, baseline 2-hour glucose, physical activity and baseline intake. This result did not change significantly when further adjusted for weight change.

Therefore, it would seem that a high-fibre, low-fat diet can result in sustained weight reduction, and it can significantly decrease the risk of progression to diabetes, even independently of weight loss.

The authors doubted the accuracy of some of the self-reported dietary changes:

  • The energy intakes calculated from the food records revealed that under-reporting had taken place.
  • Overweight and obese people are known to be even more prone to dietary under-reporting than normal-weight people.
  • Individuals who succeeded in weight reduction were possibly more likely to report consuming the recommended diet.
Adherence

Physical activity Adherence to physical activity guidance was assessed at baseline and at every annual visit using the Kuopio Ischaemic Heart Disease Risk Factor Study 12-month LTPA quantitative questionnaire. The duration (minutes per week) of total physical activity and moderate- and high-intensity LTPA were calculated. Lindstrom et al. (2003)215 reported changes from baseline to year 1 and year 3: 86% of the intervention group compared with 71% of the control group (p = 0.001) achieved the exercise goal at year 1.215 These showed that compared with the control group (n = 250), the intervention group (n = 256) showed no significant difference in total LTPA at either time point but a significant increase in the minute/week spent doing moderate-to-vigorous LTPA: baseline to year 1, +49 minutes in intervention group compared with +14 minutes in control group (p = 0.0073); baseline to year 3, +61 minutes in intervention group compared with +6 minutes in control group (p < 0.0057).

Compared with the control group, significantly fewer individuals in the intervention group were classified as sedentary at year 1 (14% intervention group vs 30% control group; p < 0.001) and year 3 (17% intervention group vs 29% control group; p = 0.028). The study authors noted that:

Frequency, duration, and intensity of leisure time and lifestyle physical activity during the preceding 12 months was estimated by the participants at each annual visit. It was not straightforward and may have been incomplete due to difficulty recollecting.215

Laaksonen et al. (2005)212 provide further analysis of physical activity data from an extended follow-up to a mean of 4.1 years, compared with the original trial end of 3.2 years. During this time, the randomised allocation and intervention was maintained.210

Questionnaires were completed each year. Moderate to vigorous exercise was defined as ≥ 3.5 METs. The intervention group reported an increase of about 48 minutes per week in this level of exercise; the control group reported little change. The main difference was in what was classed as ‘strenuous structured’ physical activity (other than walking).

There was a dose–response relationship. Those who increased exercise the most were > 60% less likely to develop diabetes, although this difference was slightly reduced after adjusting for weight loss – the RR in the highest tertile of activity then became 0.51 compared with the lowest tertile. Participants whose increase in walking for exercise was in the upper third were 59% less likely to develop diabetes than those whose change was in the lower third, independent of other factors.

The most common form of LTPA was walking and the second most common was cycling. Non-leisure activity included gardening and shovelling snow. The conclusion is that increasing physical activity may substantially reduce the incidence of T2DM in high-risk individuals. A key message from a public health standpoint would be that at least 2.5 hours/week of walking for exercise during follow-up seemed to decrease the risk of diabetes by 63–69%, largely independent of dietary factors and BMI.

Fibrinolysis One question for this review is whether or not interventions to reduce progression to diabetes would also reduce the increased risk of CVD in people with IGT. A substudy of the Finnish DPS183 examined changes in fibrinolytic activity. A reduction in fibrinolytic activity is thought to increase CVD. Hamalainen et al. (2005)206 reported that in five centres improvement in fibrinolytic activity was seen by 1 year [31% reduction in plasminogen activator inhibitor-1 (PAI-1)].206 In one centre (Turku) it was measured again at 3 years and was shown to persist. The factor that most explained the improvement was weight loss. Hence, it appears that cardiovascular risk is also reduced by lifestyle change.

Post-intervention follow-up after discontinuation of the intervention After a median of 4 years of active intervention period, participants who were still free of diabetes were further followed up for a median of 3 years, with median total follow-up of 7 years. No further intervention was provided but the participants were seen annually by the study nurse (which could be regarded as an intervention).214

Participants were divided into six groups according to how many lifestyle goals were achieved, so that group 5 achieved all and group 0 achieved none. Table 3 shows the incidence of diabetes for each group, expressed as a ratio to group 0.

TABLE 3. Success scores and HRs for diabetes in the DPS.

TABLE 3

Success scores and HRs for diabetes in the DPS.

In terms of incidence per 100 person-years, group 0 had a rate of 8.4 and group 4/5 a rate of 2.0.

Weight loss from baseline was strongly associated with the other goals, as is shown by the weight loss in each group (Table 4).

TABLE 4. Success scores and 3-year weight reduction in the DPS.

TABLE 4

Success scores and 3-year weight reduction in the DPS.

Lindstrom et al. (2006)214 also examined the incidence of diabetes in the post-intervention years, and showed that it remained 36% lower because most of the intervention group maintained their lifestyle changes. However, the difference between the incidence in the intervention and control groups narrowed with longer follow-up, from a RR reduction of 58% at the study end at 3 years to a 43% reduction at 7 years.

The results vary according to whether ITT analysis is used, because not all of the intervention group complied. If only those who achieved high success scores are considered then reduction in diabetes is much greater. Weight reduction appeared to be the most important factor.

Baseline factors and diabetes risk A further study212 examined whether baseline factors predicted response. This could be important if there was scope for selecting those most likely to respond. They looked at the effect of gender, age, BMI, 2-hour plasma glucose and the Finnish Diabetes Risk Score (FINDRISC). HRs for diabetes in the intervention group compared with the control group were calculated. The effect of the intervention increased with age (HRs for tertiles of age were 0.78, 0.49 and 0.36; p for interaction = 0.013) and also for tertiles of baseline FINDRISC (HRs 1.19, 0.77 and 0.25). Gender, BMI and 2-hour plasma glucose were not significant predictors.

These results may imply that intervention would be more cost-effective in older people, and those with a high baseline risk.

A follow-up paper208 on the metabolic syndrome reported a significant reduction in the prevalence of the metabolic syndrome in the intervention group, compared with the control subjects, with ORs of 0.62 for metabolic syndrome and 0.48 for the prevalence of abdominal obesity. The authors comment that these changes should reduce the risk of CVD as well as diabetes.

The cost-effectiveness of the DPS intervention was addressed by Lindgren et al. (2007).221 They applied the findings to a population-based cohort from Stockholm, using a Markov model in which people start in IGT and are at risk of developing diabetes or vascular disease. Therefore the model can assess the effects of the intervention on cardiovascular risk, not just diabetes. The cost assumptions reflected the intensity of the DPS, with, for example, seven meetings with the dietitian in year 1, and four times per year later, plus group circuit training and an annual physician visit. The high cost of the intervention is offset by reduced costs, but it is not clear from the paper where these come from or what proportions of the savings are from reducing diabetes, or reducing CVD. Intervention is said to yield a gain of 0.2 quality-adjusted life-years (QALYs), in a cohort of 397 people.

One way of improving the cost-effectiveness of the intervention would be to target it at people at higher risk. In the DPS,212 the NNT to prevent one case of diabetes was 7.7. However, among those with a baseline FINDRISC score of < 15, the NNT was 25, whereas among those with a baseline FINDRISC score of ≥ 15 the NNT was only 3.6.

Disappointingly, the 10-year follow-up did not show any difference in CVD between the groups, nor any change in overall mortality. However, as Uusitupa et al. (2009)222 report, this may be as a result of low power, arising from very low event rates in the DPS volunteers. A comparison with a population-based cohort, the FINDRISC group, showed that the DPS groups (combined) had a relative mortality risk of only 0.3 (95% CI 0.17 to 0.54), probably related to lower cardiovascular risk scores at baseline.

In summary, lifestyle intervention significantly reduced the incidence of diabetes in Finnish subjects with IGT, with the 7-year follow-up showing a fall in the incidence of diabetes from 7.4% to 4.3%, a reduction of 42%.

The Da Qing Study

Description and study quality

This trial was conducted in the Chinese industrial city of Da Qing, and compared three lifestyle interventions and a control group.52,190,223 Five hundred and seventy-seven participants with IGT were recruited; no power calculation was reported. Participants were recruited from health-care clinic patients screened for T2DM and IGT. Clinics (rather than participants) were randomised (authors performed analysis that showed overall outcomes would probably not be altered as a result) to one of four groups: diet only, exercise only, diet plus exercise or control. Of the 577 participants randomised, 530 completed the study and had baseline values reported. Treatment groups were similar at baseline and treatment regimens were described in detail. All physicians, nurses and technicians involved in the study attended annual 2-day training sessions to receive standardised instructions on diet and exercise interventions. Exclusion criteria were not specified.

Participants were assessed over 6 years using subjective self-reporting of diet and physical activity and using objective measures for all other outcomes. The 6-year analysis included 530/577 (92%) participants: seven refused follow-up, 29 left Da Qing and 11 died during the course of the study. There were no deaths in the exercise-only group, three deaths in the control group (pneumonia, n = 1; cirrhosis, n = 2), three deaths in the diet-only group (cancer, n = 2; septicaemia, n = 1) and five deaths in the diet-plus-exercise group (stroke, n = 1; cancer, n = 2; accident, n = 1; Crohn's disease, n = 1). AEs were not reported. No conflict of interest was reported.

Participants

Pan et al. (1993)52 recruited 577 men and women with IGT (as defined by WHO 1985 criteria; FPG of < 7.8 mmol/l and 75 g OGTT 2-hour ≥ 7.8 mmol/l and < 11.1 mmol/l). Participants with IGT were recruited from a total of 110,660 patients screened (87% of target population) for diabetes and IGT at 33 health clinics across the city. Participants were aged > 25 years. Baseline characteristics were not reported for the randomised population (n = 577) but were reported for those participants (n = 530) who completed 6 years of follow-up. The mean age (± SD) of the participants who completed 6 years of follow-up was 45.0 ± 9.1. Fifty-three per cent of participants were male; the male–female ratio varied between groups, although not significantly, with the diet-only group having slightly more women and with a male predominance in the other three groups.190 Mean BMI was 25.8 ± 3.8 kg/m2.

Intervention

Participants allocated to the diet-alone group (n = 130 at 6 years) received individual instruction (frequency not reported) and small group counselling from physicians weekly for the first month, monthly for 3 months and quarterly every year for the remainder of the study. Participants were given advice according to their baseline BMI. Those with a BMI of < 25 kg/m2 were prescribed a diet containing 25–30 kcal/kg body weight, 55–56% carbohydrate, 10–15% protein and 25–30% fat. They were also encouraged to eat vegetables, control alcohol intake and reduce intake of simple sugars. Those with BMI ≥ 25 kg/m2 were advised to reduce their calorie intake with the aim of losing weight at a rate of 0.5–1.0 kg/month until they had achieved a BMI of 23 kg/m2.

Participants allocated to the exercise-alone group (n = 141 at 6 years) received counselling sessions weekly for the first month, monthly for the next two months and quarterly for the remainder of the study. They were advised to increase the amount of physical activity undertaken by at least one unit/day (two units/day if possible for those < 50 years of age with no evidence of CVD or arthritis); one unit equated to mild exercise for 30 minutes, moderate exercise for 20 minutes, strenuous exercise for 10 minutes or very strenuous exercise for 5 minutes. Participants in the diet-plus-exercise group (n = 126 at 6 years) received instructions and counselling for both diet and exercise interventions similar to both regimens described above. Participants in the control group (n = 133 at 6 years) received general information and instructions about diabetes and IGT, diet and increased physical activity. No individual instruction or group counselling was offered.

Results: primary outcomes

Progression to diabetes Pan et al. (1997)190 assessed the incidence of diabetes (WHO 1985 criteria: FPG of ≥ 7.8 mmol and/or 2-hour glucose ≥ 11.1 mmol/l) by oral glucose tolerance testing at 2 years, 4 years and 6 years. Results were expressed as the cumulative number (%) of participants with diabetes at 6 years and incidence per 100 person-years. They found that at 6 years the incidence of diabetes was significantly lower (p < 0.05) in each of the three intervention groups than in the control group (68%). The lowest incidence was reported in the exercise-alone group (41%) followed by the diet-alone group (44%) and the diet-plus-exercise group (46%).

Incidence per 100 person-years Incidence was 10.0 (95% CI 7.5 to 12.5) in the diet-alone group compared with 8.3 (95% CI 6.4 to 10.3) in the exercise-alone group vs 9.6 (95% CI 7.2 to 12.0) in the diet-plus-exercise group compared with 15.7 (95% CI 12.7 to 18.7) in the control group. The influence of type of intervention and baseline characteristics on development of diabetes was assessed using a proportional hazards model; overall reduction in incidence of diabetes of 33% in the diet-only group (p < 0.03), 47% in the exercise-only group (p < 0.0005) and 38% in the diet-plus-exercise group (p < 0.005). Only modest changes in incidence were seen after adjustment for baseline factors.

Subgroup analysis of those participants with BMIs of < 25 kg/m2 (n = 208) compared with those with BMIs of > 25 kg/m2 (n = 322) showed that the incidence rates of diabetes in the control group of overweight participants were higher than those in the control group of lean participants (17.2 vs 13.3/100 person-years, p < 0.05). Li et al. (2002)223 stratified groups according to insulin resistance and β-cell insulin secretion levels at baseline and analysis showed that both were significantly associated (p < 0.05 and p = 0.01, respectively) with development of diabetes at 6-year follow-up; increasing insulin resistance and decreasing β-cell insulin secretion levels at baseline were associated with greater incidence of diabetes. Similarly, BMI and 2-hour plasma glucose levels at baseline were also significantly positively related to the development of diabetes (p < 0.01 and p = 0.01, respectively).

The 20-year results showed mixed results.224 The follow-up 14 years after the 6-year intervention showed a 43% reduction in progression to diabetes, but no difference in cardiovascular events or all-cause mortality. Li et al. (2002)223 comment that the Da Qing study did not have statistical power for such events, but the RR for first CVD events was 0.98, suggesting that there was no difference.

Most people progressed to diabetes: 80% in the intervention group, 93% in the control groups.

Results: secondary outcomes

Weight loss Pan et al. (1997)190 assessed weight loss at 3-month intervals using standard methods. Results are presented as weight change (kg) from baseline to 6 years in those with and without diabetes.190 It found that in those participants without diabetes only those in the diet-plus-exercise group had mean weight loss (−1.77 kg) compared with a mean gain of 0.93 kg, 0.71 kg and 0.27 kg in the diet-alone, exercise-alone and control groups, respectively. In those who had developed diabetes, participants in each group had mean weight loss of 2.43 kg in the diet-alone group, 1.93 kg in the exercise-alone group, 3.33 kg in diet-plus-exercise group and 1.55 kg in the control group. No further analysis was shown.

Gong et al. (2011)225 reported retinopathy rates at 20-year follow-up. There was a 47% reduction in severe retinopathy (9% in the intervention group, 16% in the control group). No reduction in renal failure was seen, but numbers of end points were very small.

Adherence

Dietary intake Pan et al. (1997)190 examined the adherence of participants to diet intake. It was not entirely clear from the publications how diet adherence was assessed (unlike other studies no mention was made of food diaries); however, compliance with the intervention regimen was discussed with nurses and clinic staff at 3-month intervals using interviews and forms and at 2-year intervals physicians in Beijing recorded diet changes. These showed that there was no significant difference between groups in estimated total calorie intake, per cent carbohydrates, per cent protein, per cent fat or amount of alcohol consumed (g/day). Study authors noted that dietary changes and assessments were carried out by interviewers who were not masked as to the intervention.

Physical activity As with assessment of dietary adherence, it is not clear how physical activity was recorded; however, compliance with the intervention regimen was discussed with nurses and clinic staff at 3-month intervals using interviews and forms and at 2-year intervals physicians in Beijing exercise changes. Change in physical activity (units/day) from baseline to 6 years was reported and showed that, compared with the control group, average units per day of exercise were slightly increased at 6 years in the diet-plus-exercise group, whereas little change or a decrease occurred in other groups; however, it should be noted that average units per day of exercise were significantly higher at baseline in the exercise and in the diet-plus-exercise groups.

In summary, lifestyle interventions (either diet, exercise or both) in a population with IGT led to a significant decrease in the incidence of diabetes at 6 years' follow-up from 68% to 41–46%. Both increased insulin resistance and decreased β-cell function at baseline were predictors of greater incidence of diabetes. Longer-term follow-up showed continuing reduction of progression to diabetes, but no benefit in terms of CVD or all-cause mortality.

The Indian Diabetes Prevention Programme

Description and quality of trial

This trial in Chennai, India, compared three intervention groups (lifestyle interventions plus or minus pharmacological intervention and pharmacological only) with a control group.173,226228

Participants were recruited by screening a middle-class population (n = 10,839) working in service organisations and their families, who were identified by workplace announcements and circulars. Those with IGT (according to WHO 1999 criteria) on two separate OGTTs were recruited. Power calculation was not reported. A total of 531 participants were randomised into four groups: (1) diet plus exercise; (2) metformin; (3) diet plus exercise plus metformin; and (4) control group. Treatment groups were not significantly different at baseline apart from slightly greater prevalence of family history of diabetes in the diet-plus-exercise-plus-metformin group (p = 0.031). Treatment regimens were reported in detail. Use of concurrent medication was not reported and no exclusion criteria were reported. Participants were assessed over 3 years (median follow-up 30 months) using subjective self-reporting of diet and physical activity and using objective measures for all other outcomes. Twenty-nine participants (5%) did not complete the study; three in the control group (one death and two lost to follow-up), 13 in the diet-plus-exercise group (one death, five not willing and seven lost to follow-up), five in the metformin group (two not willing and three lost to follow-up) and eight in the diet-plus-exercise-plus-metformin group (one death, two not willing and five lost to follow-up). AEs (cardiovascular, gastrointestinal and deaths) were reported. No conflict of interest was reported.

Participants

Ramachandran et al. (2006)173 recruited 531 Indian men and women with IGT (as defined by WHO 1999 criteria: FPG of < 7.0 mmol/l and/or 75 g OGTT 2-hour ≥ 7.8 mmol/l and < 11.0 mmol/l). Participants with IGT were recruited from a total population of 10,839 working in service organisations and their families. Participants were to have no diabetes, no major illness, aged 35–55 years. The age distribution of participants was as follows: 35–39 years (15.4%), 40–44 years (24.5%), 45–49 years (29.2%) and 50–55 years (30.9%). Seventy-nine per cent (429/531) of participants were male, 49.5% had a family history of diabetes, 21.7% were smokers and 31.8% had hypertension. Mean BMI was between 25.6 ± 3.7 kg/m2 and 26.3 ± 3.7 kg/m2 in the four groups. The authors noted:

The Indian population have a young age of onset of diabetes, relatively lower BMI, with high rates of insulin resistance and lower thresholds for risk factors for diabetes.173

In addition:

The Indian study cohort consisted of a middle-class working population, many of whom were already physically active and were on a diet similar to that prescribed.173

Intervention

Participants were randomised to four groups: (1) diet plus exercise [lifestyle modification (LSM)]; (2) metformin; (3) diet plus exercise plus metformin (LSM + metformin); and (4) control. Participants randomised to LSM intervention group (n = 133) and the group randomised to LSM + metformin (n = 129) received monthly telephone calls and individual sessions every 6 months for lifestyle advice. Participants were encouraged to improve their diet (reduce total calories, refined carbohydrates and fats, avoidance of sugar and inclusion of fibre-rich food). Participants had their level of physical activity assessed and advice given accordingly; those who were involved in physical labour or who had to walk or cycle > 30 minutes/day or were performing exercises regularly were asked to continue their activity. Those participants engaged in sedentary or light physical activity were advised to walk briskly for at least 30 minutes per day. Participants randomised to metformin (n = 133) or LSM + metformin (n = 129) were given 250 mg twice daily. (Note: first 50 patients received doses titrated up to 500 mg twice daily but the dose was lowered owing to high incidence of symptoms suggestive of hypoglycaemia.) The control group regimen was not described.

Results: Primary outcomes

Progression to diabetes Ramachandran et al. (2006)173 assessed the incidence of T2DM (WHO criteria) by oral glucose tolerance testing at 6, 12, 18, 24, 30 and 36 months. Results were expressed in a number of ways: as the cumulative incidence of diabetes at 3 years, absolute and RR reduction and the NNT for 3 years. At 3 years, compared with the control group, the incidence of diabetes was significantly lower (p < 0.03) in each of the three intervention groups (55% control vs 39.3% LSM vs 39.5% LSM + metformin vs 40.5% metformin). Absolute risk reductions for the three intervention groups were 15.7%, 14.5% and 15.5% for the LSM, metformin and LSM + metformin groups, respectively. Similarly, the RR reductions (NNT) were 28.5% (6.4), 26.4% (6.9) and 28.2% (6.5), respectively.

The relationship between development of diabetes and other variables was examined using a Cox's proportional hazards model. This found that, in addition to the three intervention regimens, the following variables also significantly influenced the development of diabetes: baseline 2-hour plasma glucose, fasting insulin and 2-hour insulin. Factors shown to have no influence on the development of diabetes were age, sex, family history of diabetes, BMI, WC, FPG, hypertension and smoking. Presumably these factors became non-contributory after the glucose and insulin levels were factored in.

In all of the groups, progression to diabetes was commoner in the subgroups that had both IGT and IFG than in those with only IGT, but these difference were not statistically significant, perhaps because of the relatively small numbers in the IFG + IGT groups.229

Cardiovascular events Eleven cardiovascular events were reported: two in the control group (one person died following surgery for CVA), four in the LSM group and five in the LSM-plus-metformin group.

Results: secondary outcomes

Weight loss Weight loss was assessed annually using standard methods. Results were presented as mean change in body weight (kg) from baseline to 12, 24 and 30 or 36 months. The study found that, of the four groups, only the weight of the control group increased significantly at each time point compared with baseline (p < 0.01). Weight change was not significant in either the metformin group or the metformin + LSM group; however, in the LSM group a significant increase in weight was seen at 24 months relative to baseline values (p = 0.035). BMI did not change much in the intervention group (25.5 kg/m2 at baseline, 25.7 kg/m2 at follow-up) but increased a little more in the control group (26.0–26.4 kg/m2). So the reduction in progression to diabetes occurred without weight loss.230

Adherence

Dietary intake Ramachandran et al. (2006)173 examined the adherence of participants to intervention. Diet adherence was assessed subjectively every 6 months up to 3 years by examining self-reported weekly food records filled in by participants and calculating average ‘adherence scores’ every 6 months. These showed that both the LSM group and the LSM-plus-metformin group improved their adherence over the study. The diet adherence of the LSM group improved from 62.5% to 81.6%, whereas the LSM + metformin group improved from 62% to 81.9%.

Physical activity Adherence to physical activity guidance was assessed subjectively by weekly self-report physical activity and by averaging the ‘adherence scores’ from 6 months to 3 years. There was an increase in the percentage physical activity adherence from 41.7% to an average of 58.8% in the LSM group and an increase from 45.9% to 62.9% in the LSM-plus-metformin group.

Metabolic syndrome Ramachandran et al. (2006)173 also determined the prevalence of the metabolic syndrome, as defined by the WHO criteria, in their trial participants.227 It was found in 46% (95% CI 42% to 52%) overall, but was commoner in women (62%; 95% CI 52% to 71%) than men (42%; 95% CI 37% to 47%). The presence of metabolic syndrome did not increase the incidence of diabetes.

In summary, lifestyle intervention (either alone or in combination with metformin) significantly reduced the incidence of diabetes in Asian Indians with IGT, from 55% to 40%. No added benefit was seen when combining them.

Kosaka et al.186

Description and quality of study

This RCT compared a diet and exercise intervention with standard advice.186

Japanese male subjects with IGT were randomised (number not reported; however, n = 458 at 1 year and 4.7–5.5% dropped out during first year) and assessed over a 4-year period using glucose, anthropometry, blood pressure and lipid measurements. Inclusion and exclusion criteria were explicit; however, a priori sample size calculation and randomisation method was not reported. Subjects were randomised into two treatment groups: in the first group (diet and exercise n = 102 at 1 year) participants were recommended to reduce their weight at a rate of 0.5–1 kg/month to achieve their target BMI (22 kg/m2) through a combination of intensive individually-tailored dietary and physical activity advice. In the second group (control n = 356 at 1 year) participants received standard advice to improve diet, increase physical activity and lose weight.

Lifestyle and control groups were similar at baseline in terms of age, sex, BMI, blood pressure and cholesterol levels. Baseline physical activity was not reported; however, 15% of subjects in the lifestyle intervention group were already performing the required amount of exercise and were recommended to maintain this level. The percentage of subjects in the control group already performing exercise was not reported. Concurrent medication and mean follow-up were not reported.

Participants

Kosaka et al. (2005)186 recruited Japanese males with IGT (as defined by WHO 1980 criteria; FPG of < 7.8 mmol/l and/or 100 g OGTT 2-hour of ≥ 8.8 mmol/l and < 13.27 mmol/l) randomly selected from population of government workers undergoing health screening. Men were excluded if they had a previous history of diabetes, diagnosed or suspected malignant neoplasms, diagnosed or suspected disease of the liver, pancreas, endocrine organs, or kidney, ischaemic heart disease or cerebrovascular disease or history of such disease. Baseline characteristics were not reported for the overall population but were given separately for the two groups who completed 1 year. The mean age of the participants was not reported; however, approximately 87% of participants in both groups were between the ages of 40 and 60 years. All participants were Japanese males and approximately 42% in both groups had first-degree relatives with diabetes. Mean BMI was approximately 24 mg/kg2 in both groups (so participants in Japan were leaner than those in the American and Finnish studies) and mean weight was not reported. Baseline characteristics were not significantly different between groups.

Intervention

Participants allocated to the intensive lifestyle intervention group (n = 102 at 1 year) were given advice according to their baseline BMI. Those with BMI of ≥ 22 kg/m2 were informed of their desirable body weight to achieve a BMI of 22 kg/m2. They were advised to reduce their weight at a rate of 0.5–1 kg per month until their target weight was achieved. Those with a BMI of < 22 kg/m2 were advised to maintain their present weight. To achieve the body weight objective participants were given individually tailored advice every 3–4 months at each hospital visit. They were advised to reduce amount of food, increase consumption of vegetables, reduce intake of fat and alcohol, and take 30–40 minutes of moderate exercise per day. Participants in the control group (n = 356 at 1 year) were also given advice based on their baseline BMI but this was given every 6 months; those with a BMI of > 24 kg/m2 were advised to lose weight by eating smaller meals and increasing their physical activity, whereas those with a BMI of < 24 kg/m2 were told to avoid weight gain by dieting and exercise.

Results: primary outcomes

Progression to diabetes The progression to diabetes (FPG of ≥ 7.8 mmol/l in consecutive tests within 2-week time period) was assessed by testing FPG every 2–3 months over 4 years. Results were expressed as the cumulative incidence of diabetes during the 4-year follow-up. The study found that the cumulative incidence of diabetes was significantly lower in the intensive lifestyle intervention compared with the control group (3.0% vs 9.3% respectively; p < 0.043) equating to a reduction of 67.4% in the intensive lifestyle group.

The relationship between progression to diabetes and weight change was evaluated in the control group alone. It was found that when participants in the control group were subdivided according to change in weight, the progression to diabetes was significantly less in those who had lost weight than in those whose weight had increased. Cumulative incidence of diabetes was 4.3% in those whose weight had decreased by > 1.0 kg (n = 126) compared with 10.6% [p = not significant (NS)] in those with no weight change (< ± 1 kg) (n = 173) and 14.7% (p = 0.006) in those who had gained ≥ 1.0 kg (n = 57).

Regression to normal glucose levels Kosaka et al. (2005)186 also assessed regression of participants from IGT to NGT (FPG of < 7.8 mmol/l and/or 100 g OGTT 2-hour cut-off ≤ 8.8 mmol/l). After 4 years, 53.8% in the intervention group compared with 33.9% in the control group had improved from IGT to non-IGT. When the control group was stratified according to weight change, it was seen that the rate of improvement was significantly greater in those whose weight had decreased relative to those whose weight was unchanged or increased: 47.6% in those whose weight had decreased > 1 kg (n = 126) compared with 12.5% in those whose weight had increased by ≥ 1.0 kg (n = 57) and 30.1% in those whose weight was unchanged (< ± 1kg) (n = 173 kg).

Results: secondary outcomes

Weight loss Participants assigned to the lifestyle intervention had significantly greater weight loss than did those in the control group. Over 4 years the average weight loss was 2.18 ± 1.63 kg and 0.39 ± 1.42 kg in the intervention and control groups, respectively (p < 0.001). In the intervention group, mean body weight had decreased by 2.5 kg at 1 year and tended to increase slightly thereafter; however, it remained significantly lower than baseline at the end of 4 years. In contrast, the control group had mean weight loss of 0.57 kg (measured from graph) at 1 year with weight increasing slightly every year thereafter.

In summary, Kosaka et al. (2005)186 found that intensive diet and exercise intervention in those with IGT significantly reduced the incidence of diabetes by 67.4% compared with standard lifestyle intervention (3.0% vs 9.3%; p = 0.043) and increased the rate of regression from IGT to NGT by 58.7% compared with standard lifestyle intervention (53.8% vs 33.9%; p < 0.001).

Liao et al.187

Description and quality of study

The RCT undertaken by Liao et al. (2002)187 in 74 Japanese American participants with IGT compared an intensive diet and aerobic exercise intervention with a standard diet and stretching exercise intervention.187

Participants were recruited after telephone screening of 340 Japanese individuals; of 102 diagnosed with IGT, 74 were randomised (28 were not randomised because of medical information or declining to participate) and baseline data were presented for 64 who completed 6 months. No power calculation was reported. Participants were randomised into two treatment groups: an intensive-diet-and-aerobic-exercise group compared with a standard-diet-and-stretching-exercise control group. Treatment groups were not significantly different at baseline when sex was taken into consideration (see Participants, below). Treatment regimens were reported in detail. It was not reported whether any concurrent medication was taken; however, participants were excluded at screening if they were using lipid-lowering drugs.

Patients were assessed over 24 months using subjective self-reporting of diet and physical activity and using objective methods for all other measurements. Fifty-eight participants completed the study (n = 29 in both groups); three participants did not take part because of poor venous access; one had an abnormal treadmill test; nine dropped out; two developed diabetes at 6 months; and one developed diabetes at 12 months. Sixty-four participants completed 6 months and their results were included from baseline to 24 months. (Note: it is not clear why an additional six patients have contributed data to the end of the study.) AEs were not reported. No conflict of interest was reported.

Participants

Liao et al. (2002)187 recruited 74 Japanese American participants with IGT (as defined by WHO 1998 criteria; FPG of < 7.0 mmol/l and/or 75 g OGTT 2-hour ≥ 7.8 mmol/l and < 11.1 mmol/l) selected after telephone screening of 340 individuals of full Japanese ancestry. Subjects were excluded if they had a history of significant coronary artery disease; valvular heart disease; hypertension (blood pressure > 160/90 mmHg); arthritis; pulmonary, neurological/psychiatric disease or dementia, which hindered ability to participate; unusual dietary restrictions (e.g. strict vegetarians); current use of lipid-lowering drugs; or tobacco use. Participants were also excluded if laboratory tests showed evidence of liver or kidney disease or anaemia (haematocrit: < 38% for men, < 36% for women) or if triglyceride levels were > 300 mg/dl.

Baseline characteristics were not reported for the overall population but were given separately for the two groups that completed 6 months. The mean age of the participants in the intervention and control groups was 55.8 ± 1.8 years and 52.2 ± 1.8 years, respectively. Mean BMI was 25.6 ± 0.8 kg/m2 and 26.6 ± 0.8 kg/m2 in the intervention and control groups, respectively, and mean weight was 66.1 ± 2.9 kg and 69.7 ± 2.6 kg. Baseline characteristics were not significantly different between groups; the intervention group had more women and hence a significantly lower amount of intra-abdominal fat (p = 0.038) and waist girth (p = 0.04). The differences were not significant after adjusting for sex. It was noted in the discussion that many of the study participants were already consuming an American Heart Association (AHA) step 1 diet at baseline (low in saturated fat) and that in this respect they may not be representative of all Japanese Americans.

Intervention

Participants randomised to the intensive lifestyle intervention group (n = 36) received 6 months' supervised endurance exercise on a treadmill for 1 hour three times a week with a goal to be exercising at 70% of heart rate reserve by 3 months. In addition, they were prescribed an isocaloric AHA step 2 diet comprising < 30% of total calories as fat (< 7% as saturated fat), 55% as carbohydrate, the balance as protein, and < 200 mg of cholesterol daily. Participants randomised to the control group (n = 38) received supervised stretching exercises for 1 hour three times per week and were prescribed an isocaloric AHA step 1 diet comprising 30% of total calories as fat (10% as saturated fat), 50% as carbohydrate, 20% as protein, and < 300 mg cholesterol daily. All participants were instructed to continue their diet and exercise unsupervised for an additional 18 months and were reminded of this at 12 months.

Results: primary outcomes

Progression to diabetes Liao et al. (2002)187 assessed the incidence of diabetes by oral glucose tolerance testing at 6, 12 and 24 months. It should be noted that this study was not designed to demonstrate prevention of diabetes. Its objective was to determine whether a lifestyle intervention would improve adiposity and body fat distribution in Japanese Americans with IGT. Results were expressed as the incidence of diabetes at 12 months, and found that one participant in the intervention group had developed diabetes at 12 months compared with two participants in the control group. No analysis of these results was undertaken.

Regression to normal glucose levels Liao et al. (2002)187 also assessed regression of participants from IGT to NGT. Results were expressed as the proportion of participants showing NGT at least once during their 24 months of follow-up and they found that this was significantly greater in the intervention group than in the control group (67% vs 30%, respectively; p = 0.01).

Results: Secondary outcomes

Weight loss Participants assigned to the intensive lifestyle intervention had significantly greater weight loss than participants assigned to the control group. Weight was measured at 6 months and 24 months using a standard method. At 6 months the intervention group had mean weight change from baseline (± SD) of −2.7 ± 0.4 kg compared with −0.9 ± 0.3 kg in the control group (p = 0.0003 adjusted for sex and baseline values). At 24 months, the difference between groups, although significant, was not as marked; intervention group weight change −1.8 ± 0.5 kg compared with 0.7 ± 0.6 kg in the control group (p = 0.0043).

Change in BMI BMI was also measured at 6 months and 24 months. Results showed significant differences between groups at both time points; at 6 months change in mean BMI (± SD) from baseline in the intervention group was −1.1 ± 0.2 kg/m2 compared with −0.4 ± 0.1 kg/m2 in the control group (p = 0.0003 adjusted for sex and baseline values). Similarly, at 24 months the intervention group maintained the decrease in BMI below baseline values at −0.7 ± 0.2 kg/m2 compared with an increase from baseline of 0.2 ± 0.2 kg/m2 in the control group (p = 0.0022).

Adherence

Dietary intake Liao et al. (2002)187 examined the adherence of participants to intervention. Diet adherence was assessed at 3, 6, 9, 12 and 24 months using 3-day food records submitted by participants (subjective outcome) and these were compared with the recommended intake according to allocated diet (either AHA step 2 for intervention group or AHA step 1 for control group). Results showed that, on average, at the five time points both the intensive lifestyle intervention group and the control group met their dietary goals.

The intervention group were assigned dietary goals of < 30% calories from fat and 7% from saturated fat. On average over the five time points they consumed 22–23.3% of calories from fat and 5.8–6.6% of calories from saturated fat. However, not all participants in this group achieved the dietary goals; between 79% and 88% of participants consumed < 30% of calories from fat, 55–70% consumed < 7% of calories as saturated fat and 66–97% consumed < 200 mg of cholesterol. Similarly, the control group achieved their respective dietary goals of approximately 30% of calories as fat (24.6 to 29.7% over the five time points) and < 10% of calories as saturated fat (7.1% to 8.5% over the five time points). Between 59 and 79% of participants consumed < 30% of calories from fat, 77–88% consumed < 10% calories as saturated fat and 74–89% consumed < 300 mg of cholesterol.

Objective improvement in physical activity Physical fitness was assessed at 6 months (end of supervised exercise) and 24 months by measuring the change in VO2max. At the end of the 6-month supervised period, the intervention group achieved a 3.3 ± 0.8 ml/kg/minute improvement in VO2max (p < 0.0001) compared with the control group (−0.6 ± 0.6 ml/kg/minute). At 24 months, the intervention group still had a significantly greater improvement over baseline (2.6 ± 0.7 ml/kg/minute) compared with the control group (−0.7 ± 0.5 ml/kg/minute, p = 0.0002). The percentage of participants showing a change in VO2max of ≥ 1.5 ml/kg/minute (selected arbitrarily as indicative of physical fitness) was 51.6% in the intervention group compared with 15.6% in the control group at the end of the 6-month supervision period (p = 0.006), and 59.3% and 13.8%, respectively, at 24 months (p = 0.001).

In summary, Liao et al. (2002)187 found that AHA step 2 diet and endurance exercise intervention in those with IGT significantly improved BMI and weight loss up to 24 months compared with AHA step 1 diet and stretching exercises intervention. A significantly greater percentage of patients in the intensive lifestyle intervention (AHA step 2 and endurance exercise) (67%) achieved NGT at least once during 24 months than in the control group (30%).

Maastricht (SLIM) Study: Mensink et al.188

Description and quality of study

This trial188,231234 compared intensive diet and exercise intervention with standard advice about lifestyle. Participants were recruited from an existing cohort in the Maastricht area; subjects with high risk of glucose intolerance were invited to undergo OGTT. The authors stated:

Participation rate was low, approximately 50%, as subjects were selected from an ongoing monitoring project for health and disease and therefore may be suffering from ‘research-fatigue’ possibly leading to selection bias.233

It should be noted that classification was based on single OGTT only (the risk of diabetes is higher in those with IGT classified using two OGTTs). Power calculations indicated that two groups of 50–60 participants would be sufficient to detect a 1-mmol/l difference in 2-hour glucose between groups following 2 years of intervention. A total of 114 participants were randomised into two groups: intensive lifestyle intervention compared with control (standard lifestyle advice). Treatment groups were not significantly different at baseline. Treatment regimens were reported in detail. Use of concurrent medication was not reported; however, subjects taking medication known to interfere with glucose intolerance (e.g. steroids) were excluded. Patients were assessed over 2 years using subjective self-reporting of diet and physical activity and using objective measures for all other outcomes. One hundred and two participants were included in the 1-year analysis and 88 at 2 years. Twenty-six participants dropped out of the study: 12 at 1 year and another 14 at 2 years (no differences in baseline values between participants and dropouts). AEs were not reported. No conflict of interest was reported.

Participants

Mensink et al. (2003)188 recruited 114 participants with IGT (as defined by WHO 1999 criteria; FPG of < 7.8 mmol/l and/or 75 g OGTT 2-hour ≥ 7.8 mmol/l and < 12.5 mmol/l) selected from a cohort at high risk of glucose intolerance. Subjects were excluded if they had previously diagnosed diabetes mellitus (other than GDM); mean 2-hour blood glucose > 12.5 mmol/l; medication known to interfere with glucose tolerance (e.g. chronic steroid use); participation in regular vigorous exercise or an intensive weight reduction programme during the year before the start of the study; any chronic disease that made participation in a lifestyle intervention programme impossible, or had an improbable 5-year survival. Baseline characteristics were not reported for the overall population but were given separately for the two groups following randomisation. The mean ages [± standard error (SE)] of the participants in the intervention and control groups were 55.6 ± 0.9 years and 57.8 ± 1.0 years, respectively. Mean BMIs were 29.8 ± 0.5 kg/m2 and 29.3 ± 0.4 kg/m2 in the intervention and control groups, respectively, and mean weights were 86 ± 1.9 kg and 83.7 ± 1.5 kg. In the intervention group, 25.5% had a family history of diabetes compared with 35.5% of those in the control group. Details of ethnicity were not reported.

Intervention

Participants randomised to intensive lifestyle intervention group (n = 55) were encouraged to undertake ≥ 30 minutes of moderate physical activity every day for at least 5 days/week. A supervised exercise programme (including aerobic exercise and resistance training) was encouraged. In addition, specific dietary recommendations were issued (based on Dutch Guidelines): carbohydrate intake ≥ 55% total energy; fat 30–35% total energy; saturated fatty acids < 10%; cholesterol < 33 mg/MJ; protein 10–15% total energy; dietary fibre 3 g/MJ. Participants were given the goal of losing 5–7% of their body weight. Follow-up visits took place at 4–6 weeks, 3 months and every 3 months thereafter. Participants in the control group were given oral and written information about healthy diet, weight loss and increased physical activity. No individual advice, programmes or follow-up visits were offered to the control group.

Results: primary outcomes

Regression to NGT The number of participants with regression to NGT was assessed by annual OGTT. At 2 years significantly more participants in the intervention group (50%) than in the control group (29%) had NGT (p < 0.05).

At 3 years, the cumulative incidence of diabetes among those who completed the study was 18% in the intervention group and 38% in the control group, a RR of 0.42 (0.18 to 0.96).235 However, an ITT analysis gave an incidence of only 32% in the control group, a RR of 0.52 (0.25 to 1.10).

Results: secondary outcomes

Weight loss Weight loss was assessed using standard methods at 1 year (n = 102) and 2 years (n = 88 participants who completed the study). Participants assigned to the intensive lifestyle intervention had significantly greater weight loss than participants assigned to the control group. At 1 year the intervention group had mean weight change from baseline (± SE) of −2.7 ± 0.5 kg compared with −0.2 ± 0.5 kg in the control group (p < 0.01). At 2 years, a significant difference between groups was maintained; intervention group weight change −2.4 ± 0.7 kg compared with +0.1 ± 0.5 kg in the control group (p < 0.01).

The paper by Roumen et al. (2008)235 that reported the 3-year results gives slightly different figures for weight loss, reporting that in the intervention group, weight loss at 12 months was 2.77 kg. The 3-year weight loss was only 1.08 kg, so the gap between intervention and control groups had narrowed, although it was still just statistically significant, partly because the control group had gained a little more.

Body mass index Change in mean BMI from baseline was significantly different between groups at both 1 year (n = 102) and 2 years (n = 88). At 1 year in the intervention group mean BMI changed by −0.9 ± 0.2 kg/m2 compared with 0.0 ± 0.2 kg/m2 in the control group (p < 0.001). This difference was maintained at 2 years; intervention group mean BMI decreased by 0.8 ± 0.2 kg/m2 compared with 0.0 ± 0.2 kg/m2 in the control group (p < 0.01).

Total cholesterol and LDL cholesterol No significant difference in LDL or cholesterol levels between groups was reported.

Adherence

Dietary intake Diet adherence was assessed subjectively at 1 year (n = 102) and 2 years (n = 88) by examining 3-day food records filled in annually by participants and checked by a dietitian. These showed that the lifestyle group had a significantly greater reduction in daily energy intake over the first year, but not the second year, compared with the control group (baseline to year 1: −1.2 MJ/day with intervention vs −0.3 MJ/day with control; p = 0.02; baseline to year 2: −0.9 MJ/day vs −0.3 MJ/day; p = NS); this was associated with a significantly greater change in the carbohydrate intake as percentage of total calories from baseline to both 1 year and 2 years (baseline to year 1: +4.7% with intervention vs +0.7% with control; p < 0.02; baseline to year 2: +5.5% vs +0.8%; p < 0.01). Similarly, the intervention group had significantly reduced their intake of fat as percentage of total calories from baseline to both 1 year and 2 years (baseline to year 1: −5.0% with intervention vs −1.0% with control; p = 0.01; baseline to year 2: −4.8% vs −0.3%; p < 0.01).

Physical activity Physical fitness was assessed objectively at 1 year (n = 102) and 2 years (n = 88) by measuring VO2max. The lifestyle group had significantly improved their VO2max compared with the control group from baseline to both 1 year and 2 years. Change in VO2max (± SE) from baseline to year 1 was +0.10 l/min ± 0.03 in the intervention group compared with +0.00 l/min ± 0.03 in the control group (p < 0.05). Similarly, from baseline to year 2 change in VO2max (± SE) was +0.09 l/min ± 0.04 in the intervention group compared with −0.03 ± 0.04 l/min in the control group (p < 0.05).

In summary, Mensink et al. (2003)188 found that intensive lifestyle intervention in those with IGT significantly improved BMI and weight loss up to 24 months compared with standard advice about lifestyle. A significantly greater percentage (50% vs 29%) of patients in the intensive lifestyle intervention group achieved NGT during 24 months.

The Newcastle trials

Description and quality of study

The first study by Oldroyd et al. (2001,236 2006189) was a relatively small trial in 78 men and women with IGT. It compared an intensive diet and physical activity counselling intervention with a control group who were offered no dietary or physical activity advice. Power calculation found that a sample size of 100 participants was required to have 90% probability of detecting a 0.6-mmol/l difference in mean FPG and a 20% difference in the proportion with glucose intolerance; therefore, the study was underpowered with only 78 participants recruited, and results for only 69 participants at 6 months, 62 at 12 months and 54 at 2 years. Participants were selected from existing studies, hospital databases and general practitioner (GP) surgeries. Participants were selected from individuals diagnosed with IGT on two consecutive OGTTs, and randomised using a random number table to two treatment groups; intensive lifestyle intervention and usual care (no lifestyle advice). Treatment groups were not similar at baseline; the intervention group had twice as many females as the control group, mean resting pulse was lower in the control group compared with intervention group (p = 0.011) and significantly more control participants reported engaging in regular physical activity (p = 0.017) compared with the intervention group. The interventions were described in detail. It was not reported whether any concurrent medication was taken. Participants were assessed over 2 years using subjective self-reporting of diet and physical activity and using objective methods for all other measurements. Sixty-nine per cent (54/78) of participants completed the study (n = 24 in the control group and n = 30 in the intervention group); 24 patients were lost to follow-up. Of these, 14 participants (intervention, n = 5; control, n = 9) withdrew from the study owing to family problems, work commitments or ill health. Nine participants (intervention, n = 3; control, n = 6) failed to attend 2-year follow-up and one participant died after a stroke between 12 months and 2 years. AEs were not reported. No conflict of interest was reported.

Participants

Oldroyd et al. (2006,189 2001236) recruited 78 men and women with IGT. Inclusion criteria were age between 24 and 75 years old and European origin. Participants were excluded if they were pregnant, on therapeutic diets or unable to undertake moderate physical activity. Baseline characteristics were not reported for the randomised population (n = 78) but were reported for those participants (n = 69) who completed 6 months of follow-up. The mean age of the participants who completed 6 months of follow-up in the intervention and control group was 58.2 years (range 41–75 years) and 57.5 years (range 41–73 years). In the intervention group, 10/32 (31%) participants were females compared with 20/37 (54%) in the control group (p = NS). Mean weight ± SD was 85.5 ± 14.2 kg and 85.3 ± 17.9 kg, respectively.

Intervention

Participants randomised to the intensive lifestyle intervention group (n = 39) received 12 appointments over 24 months for regular motivational counselling and written information from a dietitian and physiotherapist. Participants were encouraged to improve their diet (regular meals, more fruit and vegetables, reduce fat and sugar intake and eat adequate fibre) with a goal of reducing their BMI to < 25 kg/m2 in those who were overweight or obese. Specific dietary goals included fat intake ≤ 30% of total energy; polyunsaturated to saturated fat ratio of ≥ 1.0; 50% of energy from carbohydrate and dietary fibre intake of ≥ 20 g per 4.2 MJ. In addition, participants were given an individually tailored physical activity plan designed to enable 20–30 minutes of aerobic exercise at least once a week. Participants in the control group received no dietary or physical activity advice for the duration of the study.

Results: primary outcomes

Progression to diabetes Oldroyd et al. (2001,236 2006,189) assessed the incidence of diabetes by oral glucose tolerance testing at 6, 12 and 24 months. At 24 months there was no difference in the percentage of participants in the intervention group 7/32 (22%) and control group 8/37 (22%) who developed diabetes over the 24-month trial.

Regression to normal glucose levels More participants from the intervention than the control group reverted to NGT (FPG < 7.8 mmol/l) at 12 months' and 24 months' follow-up (22% vs 17% at 12 months; 20% vs 13% at 24 months). Significance level was not reported.

Results: secondary outcomes

Weight loss Weight loss was assessed at 6 months (n = 69), 12 months (n = 62) and 2 years (n = 54 participants who completed the study). At all time points, participants assigned to the intensive lifestyle intervention had significantly greater weight loss than the control group. Mean weight loss from baseline (± SD) at 6 months was −1.1 kg in the intervention group compared with +0.54 kg in the control group (p = 0.010), at 12 months it was −1.1 kg vs +1.5 kg (p = 0.001) and at 24 months it was −1.8 kg vs +1.5 kg (p = 0.008).

LDL cholesterol and total cholesterol There were no significant differences in LDL or cholesterol levels between groups at 6, 12 or 24 months.

Blood pressure Oldroyd et al. (2001,236 2006189) found changes in systolic BP and diastolic BP of borderline significance. Change in systolic BP at 6 months ± SD was −7.9 ± 16.7 mmHg in the intervention group vs −0.27 ± 14.3 mmHg in the control group (p = 0.05). Change in diastolic BP at 6 months was −2.9 ± 9.9 mmHg vs +1.9 ± 10.0 mmHg (p = 0.052). The authors noted:

We cannot exclude the possibility that greater familiarisation of intervention participants with the intervention team, which occurred during review appointments, compared with controls, contributed to the decrease we observed in blood pressure.236

Adherence

Dietary intake Diet adherence was assessed subjectively at 6 months (n = 69), 12 months (n = 62) and 2 years (n = 54) by examining 4-day food records filled in by participants and checked by a dietitian. These showed that, compared with the control group, the lifestyle group had a significantly greater reduction in total fat intake (g/day) at all time points (6 months: −13.6 ± 35.3 vs +3.7 ± 30.4; p = 0.037; 12 months −16.7 ± 26.5 vs −0.43 ± 33.5, p = 0.044; 24 months: −24.4 ± 24.5 vs −6.5 ± 30.9; p = 0.027). Changes in polyunsaturated/saturated fat (P/S) ratio and dietary fibre were not significantly different between groups at any time point. The number (%) of participants achieving dietary targets was assessed at baseline and 24 months. This showed that the number of participants in the lifestyle intervention group, but not the control group, who achieved the following nutritional targets was significantly increased from baseline to 24 months: ≤ 30% energy as fat; ≥ 50% total energy as carbohydrate and dietary fibre ≥ 20 g per 4.2 MJ energy (p < 0.02 for each outcome).

Physical activity Physical fitness was assessed at 6 months (n = 69), 12 months (n = 62) and 2 years (n = 54) by objective measurement of distance covered in shuttle walking test and change in resting pulse and subjective self-reported measures of participation in regular activity. Oldroyd et al. (2001,236 2006189) observed a significant increase in the percentage of participants at 6, 12 and 24 months who reported undertaking regular physical activity once a week sufficient to get their heart thumping: change from baseline in intervention and control group (6 months: +33.3% 95% CI 13 to 50 with intervention vs −3.1% with control 95% CI −14 to 8.5; p = 0.03; 12 months: +34.3 95% CI 16 to 49 vs +7.1 95% CI −8 to 21; p = 0.02; 2 years: +32.1 95% CI 12 to 48 vs −4.2 95% CI −23 to 14; p = 0.03). No change in resting pulse rate was observed at 6 and 12 months, but a small but significant decline from baseline in resting pulse rate was reported at 24 months (−4.4 ± 8.5 in the intervention group vs 1.2 ± 8.5 in the control group; p = 0.023); however, there was no significant difference at any time during follow-up for the mean distance walked in the shuttle test (data were not shown). It was noted by the authors that caution should be exercised in interpreting the results, as (1) medication that may affect pulse rate was not recorded and (2) self-reported physical activity and physical fitness are generally poorly correlated because of over-reporting of physical activity.

Adherence was also assessed by percentage attendance at review appointments (diet and exercise) of intervention participants. Twenty-four out of 39 (62%) participants attended all of the review appointments up to 6 months, and this increased to 28/39 (72%) at 12 months; however, by 24 months attendance lay at 36% with only 14/39 of the randomised participants attending the review appointments [average in the first 6 months was 80% (range 67–95%)].

In summary, Oldroyd et al. (2001,236 2006189) found that regular counselling on lifestyle over a 24-month period from a dietitian and physiotherapist resulted in significant improvements in weight. Cardiovascular risk factors were unchanged.

A later study from the same group also reported a reduction in diabetes incidence, by about half, from 33 per 1000 person-years in the intervention group, to 67 in the control group.237 This study came from the Newcastle centre of the European Diabetes Prevention Study (EDIPS), and had only 102 participants. The full EDIPS will have 750 participants.

Wein et al.191

Description and quality of study

This RCT compared an intensive lifestyle guidance intervention with initial lifestyle guidance in pregnant women with IGT over a median 4.25-year period (range 11.7–81.1 months) in an intervention group compared with median 4.0 years in a control group; p = 0.021.191 Inclusion criteria were IGT, English and non-English speakers, and no exclusion criteria were specified in the publication. A priori sample size calculation and randomisation method were not reported. Subjects were randomised into two treatment groups: intensive lifestyle group and control group. Both groups were similar at baseline. Concurrent medication was not reported. Seven patients were lost to follow-up (three in the intervention group and four in the control group). Adverse effects were not reported.

Participants

Wein et al. (1999)191 recruited 200 pregnant Australian females with IGT (as defined by WHO 1985 criteria; FPG < 7.8 mmol/l and 75 g OGTT 2-hour ≥ 7.8 mmol/l and < 11.1 mmol/l) selected from a long-term follow-up of women with GDM at a single hospital. Participants were regular attendees for follow-up testing. Baseline characteristics were similar at baseline. The mean age of the participants was 39.5 years in the intervention group and 37.8 years in the control group. Mean weight was 64.9 kg and 66.4 kg, respectively, and BMI was 25.2 kg/m2 and 25.6 kg/m2, respectively. Participant's country of birth was recorded and no significant difference between groups was found. In the intervention and control groups, 48% and 45%, respectively, were born in Australia or New Zealand, with smaller percentages from the Mediterranean/Middle East (20% and 21%, northern Europe (7% and 6%), South East Asia (20% and 23%) and the Indian subcontinent (5% and 5%). Parity was not significantly different between groups. The study authors noted the following: (1) participants may be a self-selected compliant group in that they elected to attend follow-up and (2) participants had already been exposed to counselling with respect to diet and exercise during their pregnancy.

Intervention

Participants allocated to the intervention group (intensive lifestyle guidance, n = 100) were given a standard diet advice sheet (‘Target on Healthy eating’ recommended by the state Health Department) and reminded of the need for regular exercise (brisk walking for 30 minutes, three times a week). Every 3 months, a dietitian contacted the participants by telephone. The dietitian's role was to answer questions regarding diet and encourage compliance with the diet and exercise recommendation. In comparison, the control group (n = 100) received the same standard advice sheet and exercise advice but had initial advice only and no additional contact with the dietitian.

Results: primary outcomes

Progression to diabetes The progression to diabetes was assessed by oral glucose tolerance testing annually. Results were expressed as the prevalence of diabetes, annual incidence of diabetes, the cumulative rates of decay to diabetes up to 5 years and the RR of diabetes.

The study found that the prevalence of diabetes over median follow-up of 51 months (longer follow-up in the intervention group) was not significantly different between groups [26/97 (26.8%) in intervention group vs 27/96 (28.1%) in the control group; p = 0.957]. (Note: percentage based on participants who completed study.) The annual incidence of diabetes was not significantly different between groups. Cox regression analysis of all participants showed that baseline BMI (p = 0.0007), fasting (p = 0.04) and 2-hour plasma glucose (p < 0.0001) levels were associated with an increased risk of diabetes. No significant association was found with change in BMI and age at start of study.

Regression to normal glucose levels Wein et al. (1999)191 also assessed regression of participants from IGT to NGT (no specific definition in publication). Results were expressed as n (%) participants with NGT status after a median 51 months' follow-up and showed that there was no significant difference between groups in the progression to NGT [43/97 (44.3%) in the intervention group vs 43/96 (44.8) in the control group]. (Note: percentage based on participants who completed study.)

Results: secondary outcomes

Weight loss No significant difference between groups was seen in BMI. Presumably that explains why there was no difference in diabetes.

Adherence

Dietary intake The adherence of participants to intervention was examined. Diet adherence was assessed using a questionnaire to record diet history at baseline and annually until final assessment. The fat, residue and sugar content of the diet were scored from ‘1’ to ‘3’ (representing ‘poor’ to ‘good’) and the total score calculated. Baseline and final assessment data were presented and both intervention and control groups showed an improvement in diet score (+0.64 and +0.56 points, respectively; p = 0.32); however, no significant difference between groups was found. Study authors noted that dietary advice may not have an impact without the addition of stronger reinforcement, such as periodic weighing.

Physical activity Adherence to physical activity guidance was assessed at baseline and annually to final assessment using a questionnaire to record exercise history. The amount of exercise was scored on a scale of 0–7, where ‘0’ was totally sedentary, ‘1’ mildly active, ‘2’ active without formal exercise, ‘3’ physical work or walking once a week, to ‘7’ athletic training. Results showed that there was no significant difference in exercise scores between groups at the initial visit (3.1 vs 2.9; p = 0.26) or at follow-up (3.2 vs 3.1; p = 0.43).

In summary, Wein et al. (1999)191 found that there were no significant differences, in progression to diabetes or regression to NGT, between intensive diet and lifestyle advice given 3-monthly compared with the same diet and lifestyle advice given once. However, no differences in weight and exercise were achieved between the groups. Progression to diabetes was associated with increased fasting and 2-hour plasma glucose at baseline at high baseline BMI levels.

Wing et al.238

This trial recruited individuals who were overweight or obese (30–100% of ideal body weight; mean weight at baseline about 98 kg, BMI 36 kg/m2) and had a family history of T2DM.238 About half had IGT, but most results are not given separately for that subgroup. However, the trial provides a useful illustration of the main problem with lifestyle interventions.

There were four arms of the trial: diet, exercise, both diet and exercise and control group. The intervention groups had weekly group meetings for 6 months then reduced to fortnightly for a further 6 months. The diet group started with a strict low-calorie diet aiming at 800–1000 calories daily, relaxed to 1200–1500 after 8 weeks. They also had weekly education sessions. The exercise group had the same frequency of meetings, with supervised weekly walks for about an hour. They were asked to take exercise such as brisk walking at least 5 days each week. The combined group had both interventions. The control group were given educational materials but had no meetings. In the second year, there were only two refresher courses, which were poorly attended: 36% of the diet group, 15% of the exercise group and 29% of the combined group participated. However, attendance at the final assessment visits at 2 years was good at 84%.

At 6 months, the diet group had lost 9 kg, the exercise group 2 kg, the combined group 10 kg and the control group 1.5 kg. Results for the IGT group were not given separately. Unfortunately, by 2 years, most of the weight had been regained in all groups, although the diet and combined group were still about 2 kg lighter than at baseline.

This illustrates a common problem: even among volunteers, improvements are not sustained once the intervention is stopped or reduced. Attendance dropped from 61% at meetings in the first 6 months to 27% thereafter.

Progression to diabetes is reported in the text of the article, but not separately for the arms. The presence of IGT at baseline increased the risk of diabetes, with 25% of those with IGT developing diabetes compared with 6% of those with NGT. Weight loss of 4.5 kg reduced the development of diabetes by 26% in those with IGT compared with those with NGT and no weight loss.

Similar findings were reported by Page et al. (1992)239 in a pilot study in 31 people with IGT. After a 6-month diet and exercise intervention, there were improvements in various parameters in the intervention group but no change in the eight control subjects. For example, total cholesterol fell from a mean of 5.2 to 4.5 mmol/l, and mean SBP by 6 mmHg. However, the gains did not persist once the intervention was stopped.

Overview of evidence base There is a good body of evidence that some T2DM can be prevented, with the best evidence coming from the larger longer-term trials such as DPS,183 DPP108 and Da Qing.190

Table 5 summarises the results. Combining studies into a meta-analysis was not done because of the differences in intervention, duration and recruits, which also means that the results in the table should not be compared between studies.

TABLE 5. Progression to diabetes and regression to NGT, large trials only.

TABLE 5

Progression to diabetes and regression to NGT, large trials only.

Most of the studies show that progression to diabetes can be reduced, and regression to NGT increased.

Adherence to the lifestyle measures was clearly a problem for some. The results come from groups among which compliance varied. Figure 2 from the DPS183 shows a strong inverse correlation between the proportions progressing to diabetes and the success scores in achieving the targets.

There was also a tendency for the benefits to be lost not long after the intervention ended, with, for example, regain of weight. The exception was the DPS,183 where benefit was largely maintained for 3 years after intervention ended. Perhaps a 4-year intervention can permanently improve lifestyle change, whereas short intervention does not.

However, as noted above, studies with the longest follow-up show disappointing results in terms of CVD.

The benefits of the lifestyle intervention were greatest in those with the highest compliance and who achieved more of the targets (such as weight loss and dietary change). For example, in the Finnish study,183 those who achieved four or five of the five targets had a risk of developing diabetes which was only 23% of those who achieved none.

However, even among the volunteers in the trials, many did not succeed and others succeeded in the short term (such as the first 6 months) but not in the longer term. The key to success is sustained lifestyle change, especially weight loss.

In conclusion, lifestyle measures can be highly effective in reducing progression to diabetes but adherence to lifestyle change is the most important factor.

© 2012, Crown Copyright.

Included under terms of UK Non-commercial Government License.

Bookshelf ID: NBK109430

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.5M)
  • Disable Glossary Links

Other titles in this collection

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...