Article type
Year
Abstract
Background: Cumulative meta-analysis may increase the risk of type 1 and 2 errors due to sparse data and repetitive testing on accumulating data. To limit risk of random errors, meta-analyses may be analysed with trial sequential analysis (TSA) using heterogeneity-adjusted required information size to determine when convincing evidence is reached.
Methods: Meta-analyses of the effects of targeting intensive versus conventional glycaemic control on mortality, non-fatal myocardial infarction, and severe hypoglycaemia in patients with type 2 diabetes were conducted. TSA was applied to limit the overall type 1 error to 5% and type 2 error to 20%. We used a heterogeneity-adjusted required information size calculated for a 10% relative risk reduction (RRR) of mortality and of non-fatal myocardial infarction and a 30% RRR of severe hypoglycaemia corresponding to numbers needed to treat (harm) of 50-100.
Results: The relative risk reduction of all-cause mortality and non-fatal myocardial infarction were 1.02, 95% CI 0.91 to 1.13; I$2=30$% (Figure 1) and 0.85, 95% CI 0.76 to 0.95; I2=0% (Figure 2). TSA showed that a 10% RRR or more can be rejected for all-cause mortality (Figure 3). TSA showed lack of firm evidence for a benefit of targeting intensive glycemic control for non-fatal myocardial infarction. Only 27,958 participants (44%) of the heterogeneity-adjusted required information size of 63,446 have been accrued to detect a 10% RRR of non-fatal myocardial infarction (Figure 4). The RR of serious hypoglycaemia was 2.18, 95% Cl 1.47 to 3.23; I2=73% (Figure 5). TSA showed firm evidence for at least a 30% increase of severe hypoglycaemia with intensive glycaemic control (Figure 6).
Conclusion: Intensive glycaemic control does not seem to reduce all-cause mortality, and if so then less than 10%. A 10% RRR of non-fatal myocardial infarction could not be confirmed. TSA confirmed that intensive glycaemic control increases the RR of severe hypoglycaemia with30%.
Methods: Meta-analyses of the effects of targeting intensive versus conventional glycaemic control on mortality, non-fatal myocardial infarction, and severe hypoglycaemia in patients with type 2 diabetes were conducted. TSA was applied to limit the overall type 1 error to 5% and type 2 error to 20%. We used a heterogeneity-adjusted required information size calculated for a 10% relative risk reduction (RRR) of mortality and of non-fatal myocardial infarction and a 30% RRR of severe hypoglycaemia corresponding to numbers needed to treat (harm) of 50-100.
Results: The relative risk reduction of all-cause mortality and non-fatal myocardial infarction were 1.02, 95% CI 0.91 to 1.13; I$2=30$% (Figure 1) and 0.85, 95% CI 0.76 to 0.95; I2=0% (Figure 2). TSA showed that a 10% RRR or more can be rejected for all-cause mortality (Figure 3). TSA showed lack of firm evidence for a benefit of targeting intensive glycemic control for non-fatal myocardial infarction. Only 27,958 participants (44%) of the heterogeneity-adjusted required information size of 63,446 have been accrued to detect a 10% RRR of non-fatal myocardial infarction (Figure 4). The RR of serious hypoglycaemia was 2.18, 95% Cl 1.47 to 3.23; I2=73% (Figure 5). TSA showed firm evidence for at least a 30% increase of severe hypoglycaemia with intensive glycaemic control (Figure 6).
Conclusion: Intensive glycaemic control does not seem to reduce all-cause mortality, and if so then less than 10%. A 10% RRR of non-fatal myocardial infarction could not be confirmed. TSA confirmed that intensive glycaemic control increases the RR of severe hypoglycaemia with30%.
Images
