Table 3

Incident diabetes and additional glycaemia outcomes

Author, publication dateNumber of studies included in synthesesOutcomes
Incident diabetes
Ashra, 201511 studiesIntervention associated with lower rate of progression to diabetes: meta-analysis IRR=0.74 (95% CI 0.58 to 0.93)
Baker, 20117 studiesIntervention associated with significantly lower incidence of T2DM in all studies reviewed. RR reduction ranged from 29% to 75%.
Balk, 201516 studiesIntervention associated with lower rate of incident diabetes: summary RR=0.59 (95% CI 0.51 to 0.66)
Gillett, 20125 systematic reviews; 9 RCTsAuthors observed that lifestyle interventions were associated with lower rates of progression to diabetes in most studies and concluded that some diabetes can be prevented or delayed by lifestyle interventions, with larger, longer term trials (the DPS, DPP and Da Qing study) providing the best evidence. Some evidence that this intervention effect was temporary was also noted, as well as the DPS suggestion that adherence to lifestyle change may be an important mediator of impact on diabetes risk.
Glechner, 20155 studiesMeta-analysis outcomes suggest intervention non-significantly associated with lower risk of progression to T2DM at 1 year: RR=0.60 (95% CI 0.35 to 1.05; 4 studies). Evidence of significant intervention effect on progression to diabetes at 3 years: RR=0.63 (0.51 to 0.79; 5 studies).
Hopper, 20114 studiesIn meta-analysis, intervention associated with lower rate of progression to T2DM: RR=0.52 (95% CI 0.46 to 0.58).
Merlotti, Morabito and Pontiroli, 201411 studiesIntervention associated with lower rate of progression to diabetes: meta-analysis OR=0.43 (95% CI 0.35 to 0.52).
Merlotti, Morabito, Ceriani and Pontiroli, 20144 studiesIntervention associated with lower rates of progression to diabetes: meta-analysis OR=0.44 (95% CI 0.36 to 0.52).
Modesti, 20168 studiesIntervention associated with lower rates of progression to diabetes: meta-analysis OR=0.55 (95% CI 0.44 to 0.70).
Norris, 20055 studiesIntervention associated with significantly lower cumulative incidence of diabetes in three of the five trials reviewed (RR reductions=58% (95% CI 48 to 66), 51% and 58%). Trials in which effect observed involved intensive, sustained, multicomponent interventions.
Orozco, 20088 studiesIntervention associated with lower rates of progression to diabetes: meta-analysis RR=0.63 (95% CI 0.49 to 0.79). Similar results when largest study (DPP; weight=26%) excluded: RR=0.69 (0.55 to 0.87).
Santaguida, 20055 studiesSignificantly lower rates of progression to diabetes or higher rates of reversion to normal glucose tolerance, observed in intervention vs control arms of 4/5 trials. (NB. Number of studies considering progression to diabetes specifically not described). ARR for progression to diabetes ranged from 1.6% to 7.1%. RR reduction for progression to diabetes in intervention, cf. control scenario=31–55%. Observed NNT for 1 year to avoid one case of diabetes=14.2 to 62.5. In meta-analysis, RR of progression to diabetes in intervention, cf. control arms=0.54 (95% CI 0.42 to 0.70).
Selph, 20156 studiesIntervention associated with lower risk of progression to diabetes: meta-analysis RR=0.55 (95% CI 0.43 to 0.70). Similar results when Da Qing study (23-year follow-up) excluded: RR=0.53 (0.44 to 0.63).
Shellenberg, 20137 studiesIntervention associated with lower risk of progression to diabetes at 1 year (meta-analysis RR=0.35, 95% CI 0.14 to 0.85; four studies), 4 years (RR=0.56, 0.48 to 0.64; 2 studies), 6 years (RR=0.47, 0.34 to 0.65; 3 studies), and 10 years (RR=0.80, 0.74 to 0.88; one study). Da Qing study not included in meta-analysis, but noted that intervention associated with lower rates of progression to diabetes at 6 and 20 years, in this study.
Stevens, 201516 studiesIn network meta-analysis (incorporating 16 lifestyle vs placebo/standard care studies) lifestyle intervention associated with lower risk of progression to diabetes: HR=0.65 (95% CI 0.56 to 0.74).
Yoon, 20137 studiesT2DM incidence ranged from 3% to 46% in the intervention groups, cf. 9.3% to 67.7% in the control groups. Significantly lower T2DM incidence associated with intervention observed in 5/7 studies (RR reduction 28.5% to 64.7%). In the sixth study, lower diabetes incidence observed in the intervention vs control scenario at 1-year follow-up, but not at 3-years or 5-years. In the seventh study, intervention effect was observed only in the per protocol analysis (ie, no effect observed in intention to treat analysis).
Glycaemic control
Ashra, 201516 studiesNo significant impact of intervention, cf. control condition observed at 12–18 months: net FPG difference=−0.06 mmol/L (95% CI −0.11 to 0.00; 16 trials). Significant impact observed at >18 months: net FPG difference=−0.07 mmol/L (−0.13 to −0.02).
10 studiesNo significant impact of intervention, cf. control condition observed at 12–18 months (net 2h-OGT difference=−0.28 mmol/L, 95% CI −0.57 to 0.00; 10 trials), or >18 months (difference=−0.52 mmol/L, −1.05 to 0.01; 7 studies).
Balk, 20156 studiesIntervention associated with reversion to normoglycaemia: meta-analysis summary RR=1.53 (95% CI 1.26 to 1.71).
18 studiesAt follow-up closest to 1 year, summary net change in FPG associated with intervention vs control condition=−0.12 mmol/L (−0.20 to −0.05; 17 studies). Net change in 2h-OGT=−0.48 mmol/L (−0.86 to −0.17; 11 studies) and net change in HbA1c=−0.08% (−0.12 to −0.04; 8 studies).
Cardonna-Morrell, 20109 studiesAuthors concluded that the 1-year FPG outcomes across nine translational studies were in many cases similar to the DPP outcomes, but that effect size was too small to be clinically relevant.
4 RCTsNo observed significant net impact of intervention vs control condition on 12-month FPG (difference=−0.19 mmol/L; 95% CI −0.44 to 0.06; 3 trials) or 2h-OGT (0.04 mmol/L, −0.49 to 0.42; 2 RCTs).
Gillett, 20125 systematic reviews; 9 RCTsAuthors conclude that most studies suggest intervention associated with reversion to normal glucose tolerance
Glechner, 20153 studiesAt 1-year follow-up, intervention associated with significantly lower FPG (meta-analysis=−0.28 mmol/L; 95% CI −0.47 to −0.008), and 2h-OGT (−0.63 mmol/L, −1.08 to −0.18). Similar outcomes observed at 3-year follow-up (for FPG: −0.31 mmol/L; −0.48 to −0.15; for OGT: −0.68 mmol/L; 95% CI −1.03 to −0.34).
Gong, 20157 studiesMean net 2h-OGT difference associated with lifestyle vs control condition observed in meta-analysis=−0.65 mmol/L; 95% CI −1.35 to 0.05
Norris, 20056 studiesThe six studies that reported on HbA1c were considered not to be representative of all nine studies identified for review. Results therefore not pooled, but effect of intervention ranged from 0.0% to −0.3%.
Orozco, 20087 studiesMeta-analysis of 6/7* studies demonstrated significant impact of intervention on FPG: net difference, cf. control condition=−0.19 mmol/L (95% CI −0.32 to −0.05).
4 studiesMeta-analysis of 3/4* studies found no impact of intervention on 2h-OGT: net difference, cf. control condition=−0.23 mmol/L (−1.08 to 0.61).
Santaguida, 20055 studiesLifestyle intervention associated with significantly lower risk of progression to diabetes, or higher rate of reversion to normal glucose tolerance, in 4/5 trials reviewed (NB. number of studies considering glucose tolerance specifically not described.).
Shellenberg, 20137 studiesIntervention associated with significantly lower FPG at 0.5–4 years follow-up: summary mean difference=−0.28 mmol/L, 95% CI −0.33 to −0.23. Authors concluded that data post-4 years follow-up insufficient to draw conclusions.
5 studiesIntervention associated with significantly lower 2h-OGT at 1–4 years follow-up: summary mean difference=−0.54, −1.06 to −0.02. Again, data at post 4 years follow-up considered insufficient to draw conclusions.
3 studiesNo significant difference in HbA1c observed between intervention and control groups at 1–3 years follow-up: summary mean difference=−0.10, −0.22 to −0.01.
Zheng, 201512 studiesIntervention associated with significantly lower FPG, cf. control condition: mean difference=−0.22 mmol/L (95% CI −0.25 to −0.18; 9 studies). Also noted that the intervention effect increased with intervention duration, with the effect among the subgroup receiving the longest interventions (≥2 years duration) demonstrating the highest subtotal effect: mean difference for this subgroup=−0.24 mmol/L (−0.43 to −0.05; 12 studies).
  • Synthesis outcomes related to glycaemia are listed for each review, as relevant, alongside the number of primary studies drawn on in the associated syntheses. Italicised entries are those assigned AMSTAR scores <8, excluded from sensitivity analyses.

  • *Da Qing study not included in either meta-analysis due to cluster randomisation.

  • ARR, absolute risk reduction; DPP, Diabetes Prevention Programme; DPS, Diabetes Prevention Study; FPG, fasting plasma glucose; HbA1c, glycated haemoglobin; IRR, incidence rate ratio; NNT, number needed to treat; RCT, randomised controlled trial; RR, relative risk; T2DM, type 2 diabetes mellitus; 2h-OGT, 2-hours oral glucose tolerance.