Eight studies (31,284 participants) were included in the review: 1 RCT, 2 post hoc subgroup analyses from RCTs, 1 prospective cohort study and 4 retrospective cohort studies.
All studies were considered to be of acceptable quality, with scores ranging from 13 (41%) to 22 (69%); the median score was 16 (50%).
Insulin (4 studies, 19,205 participants).
Studies only reported on the outcome of all-cause mortality and were too heterogeneous to pool. Three studies found increased risks of all-cause mortality, cardiovascular morbidity and mortality, and/or hospital admission with insulin compared with diet, sulfonylurea, metformin and thiazolidinediones. One study reported no difference in all-cause mortality for insulin-treated patients compared with those using sulfonylurea metformin, thiazolidinediones, non-sulfonylurea secretagogues and alpha-glucosidase inhibitors.
Metformin (3 studies, 17,501 participants).
One study found no difference in mortality between patients using metformin and those taking sulfonylurea, non-sulfonylurea secretagogues, alpha-glucosidase inhibitors and insulin. Two studies reported reduced mortality with metformin, alone or in combination with sulfonylurea, compared with sulfonylurea, non-sulfonylurea secretagogues, alpha-glucosidase inhibitors and insulin. Studies were too heterogeneous to pool for this outcome. All-cause hospital admission at 1 year was also reduced (OR 0.85, 95% CI: 0.76, 0.95; 2 studies; I-squared 20.9%; p=0.26).
Thiazolidinediones (4 studies, 22,476 participants).
In a pooled analysis, thiazolidinediones were associated with reduced mortality compared with sulfonylurea, non-sulfonylurea secretagogues, alpha-glucosidase inhibitors and insulin (OR 0.83,
95% CI: 0.71, 0.97; 4 studies), although there was moderate statistical heterogeneity (I-squared 52%; p=0.10). Hospital admission for heart failure was, however, increased with thiazolidinediones (OR 1.13, 95% CI: 1.04, 1.22; 4 studies; I-squared 0%). One study reported no effect on all-cause hospital
Sulfonylurea (2 studies, 18,042 participants).
One study found no difference in risk of mortality for sulfonylurea compared with non-sulfonylurea secretagogues, alpha-glucosidase inhibitors, metformin, thiazolidinediones and insulin. The other study found that sulfonylurea monotherapy was associated with a higher risk of mortality than metformin or a