Xuxi Zhang

Results of subgroup analysis based on intervention provider Based on intervention provider, we divided the studies into three groups, PLI, PPI and CHWI (Fig. 2, Part C). In the PLI group 13, 17, 18, 20 ‐ 24, 28 ‐ 31 , we found heterogeneity reduced ( I 2 = 26.7%) but the difference in HbA 1c outcomes between IG and CG was smaller and no longer statistically significant ( ‐ 0.08%, 95% CI  ‐ 0.18 to  ‐ 0.03, or  ‐ 0.9mmol/mol, P = 0.141). In the PPI group, the difference in HbA 1c outcomes of  ‐ 0.49% (95% CI  ‐ 0.86 to  ‐ 0.12) or  ‐ 5.4mmol/mol ( P = 0.009) was much larger and without heterogeneity ( I 2 = 0%). There were, however, only two studies 26, 27 in this group. In the CHWI group 14 ‐ 16, 19, 25, 32 , there was larger difference in HbA 1c outcomes of  ‐ 0.35% (95% CI  ‐ 0.54 to  ‐ 0.16) or  ‐ 3.8 mmol/mol ( P < 0.001) but the heterogeneity increased substantially ( I 2 = 62.4%). Additionally, there was statistical significant difference among these three groups ( P = 0.009). Results of subgroup analysis based on intervention type The subgroup analysis also examined intervention type (TDI, SGI, CCOI, CCRI and HVI, see Fig. 2, Part D). Results indicated nonsignificant differences in HbA 1c outcomes between IG and CG in both the TDI group 17, 18, 21, 29 ( ‐ 0.12%, 95% CI  ‐ 0.31 to 0.07, or  ‐ 1.3mmol/mol, P = 0.202) and the SGI group 19, 22, 23, 27 ( ‐ 0.13%, 95% CI  ‐ 0.29 to 0.04, or  ‐ 1.4mmol/mol, P = 0.134) with substantial heterogeneity respectively ( I 2 = 74.2%; I 2 = 63.3%, respectively). In the CCRI group 13, 20, 24 ‐ 26, 31, 32 , there was no heterogeneity ( I 2 = 0%) and the pooled results showed larger statistical significant difference in HbA 1c outcomes of  ‐ 0.24% (95% CI  ‐ 0.43 to  ‐ 0.06), or  ‐ 2.6mmol/mol ( P = 0.011). Combining results of HVI group 14 ‐ 16 , there was also a larger difference in HbA 1c outcomes of  ‐ 0.30% (95% CI  ‐ 0.53 to  ‐ 0.07), or  ‐ 3.3mmol/mol ( P = 0.011) but the heterogeneity increased substantially ( I 2 = 79.9%). In the CCOI group 28, 30 , the difference in HbA 1c outcomes between IG and CG was 0.01% (95% CI  ‐ 0.23 to 0.25), or 0.1mmol/mol ( P = 0.93). These five groups did not differ significantly from each other ( P = 0.356). Results of peer support intervention at different follow ‐ up intervals We also combined all the results of 28 follow ‐ up intervals of 20 studies in meta ‐ analysis (Fig. 2, Part E). The overall pooled results of 28 follow ‐ up intervals indicated statistically significant difference in HbA 1c outcomes between IG and CG of  ‐ 0.19% (95% CI  ‐ 0.27 to  ‐ 0.11), or  ‐ 2.1mmol/mol ( P <0.001) with acceptable heterogeneity ( I 2 = 40.7%). We divided the 28 follow ‐ up intervals into four groups (follow ‐ up during the intervention, immediately after the intervention, >1& ≤ 6 months after the end of intervention, >6 months after the end of intervention) for subgroup analysis. For HbA 1c measured immediately after the intervention 13 ‐ 15, 17 ‐ 27, 29, 31, 32 , the heterogeneity increased ( I 2 = 50.9%) and the pooled results indicated larger difference in HbA 1c outcomes between IG and CG of  ‐ 0.21% (95% CI  ‐ 0.31 to  ‐ 0.11), or  ‐ 2.3mmol/mol ( P < 0.001). For HbA 1c measured during the intervention 15, 20, 32 , we found larger difference in HbA 1c outcomes of  ‐ 0.37% (95% CI  ‐ 0.59 to  ‐ 0.15), or  ‐ 4.0mmol/mol ( P = 0.001) without heterogeneity ( I 2 = 0%). For HbA 1c measured at >1& ≤ 6 months after the end of intervention 13, 15, 16, 24, 28 , the heterogeneity was reduced ( I 2 = 33.5%) but the difference in HbA 1c outcomes of  ‐ 0.06% (95% CI  ‐ 0.26 to 0.15), or  ‐ 0.7mmol/mol ( P = 0.585) was not 6 143 Peer support among adults with type 2 diabetes

RkJQdWJsaXNoZXIy ODAyMDc0