Thirteen studies (n=1,276) were included in the review: 10 double-blind RCTs (n=987) and 3 randomised case-controlled studies (n=289). In total, 231 patients were treated with pioglitazone and 485 with rosiglitazone.
Study quality ranged from 1 to 5 points on the Jadad scale: one study (a case-controlled study) scored 1 point, five (including 2 case-controlled studies) scored 2 points, two scored 3 points, three scored 4 points and two scored 5 points. Four studies did not report information on patient drop-outs.
No evidence of publication bias was observed in either the funnel plot or when using Egger's regression method.
The overall mean change in serum CRP level was -0.82 (95% CI: -1.15, -0.49) in favour of TZD in comparison with control. However, significant heterogeneity was associated with this result (chi-squared 1,120.60, d.f.=13, P<0.00001; I-squared 98.8%) (13 studies, n=1,371; 14 comparisons since 1 study included 2 comparisons). Repeating the analysis and removing one study at a time did not significantly alter the results, suggesting that no one study had undue influence on the overall findings.