Seven observational studies were included in the review. Three studies used nested case control designs, three used non-nested case control designs and one used a cohort study design. Sample sizes ranged from 40 to over 1,233 cases of C. difficile. Study quality was rated as being high for three studies, moderate for two studies and low for two studies.
Compared with no antibiotic exposure, a statistically significantly increased risk of C. difficile was shown with the use of any antibiotics (OR 3.55, 95% CI, 2.56 to 4.94; five studies; Ι²=90.6%). Where antibiotics were stratified according to their classes, all but one of them were associated with a statistically significant increase in risk of C. difficile. The largest effect sizes were observed with clindamycin (OR 16.80, 95% CI 7.48 to 37.76; three studies; Ι²=66.7%), fluoroquinolones (OR 5.50, 95% CI 4.26 to 7.11; five studies; Ι² =10.9%) and CMCs (OR 5.68, 95% CI 2.12 to 15.23; five studies; Ι²=93.8%). Smaller effect sizes were observed with macrolides (OR 2.65, 95% CI 1.92 to 3.64; four studies; Ι²=48%), sulphonamides and trimethoprim (OR 1.81, 95% CI 1.34 to 2.43; four studies; Ι²=0%) and penicillins (OR 2.71, 95% CI 1.75 to 4.21; five studies; Ι²=76.8%). No statistically significant difference in C. difficile risk was observed in the comparison between tetracyclines and non-exposure to antibiotics (OR 0.92, 95% CI 0.61 to 1.40; three studies; Ι²=0%). No evidence of publication bias was found.
The two studies with no antibiotic-free controls demonstrated no statistically significant differences in C. difficile risk between those exposed to penicillin and those exposed to tetracyclines, sulphonamides and trimethoprim, or macrolides. Statistically significant increases in C. difficile risk were observed with exposure to CMCs or fluoroquinolones compared with penicillin exposure. Neither study reported any C. difficile cases associated with exposure to clindamycin.
Further results were fully reported in the review.