. 2012 Nov 14;7(11):e48991. doi: 10.1371/journal.pone.0048991

Table 2. Overall cancer risk at doses in line with NDA, restricting the analysis to trials during at least 52 weeks (n = 13, continuity correction used: 0.01, I² = 0).

Model	Per protocol analysis Odds ratio [95% confidence interval]	Modified intention-to-treat analysis Odds ratio [95% confidence interval]
Random effect (DerSimonian and Laird)	1.04 [0.58–1.87]	1.39 [0.77–2.49]
Fixed effect (Mantel-Haenszel) ^*	1.22 [0.71–2.11]	1.65 [0.96–2.85]
Fixed effect (Peto) ^*	1.22 [0.72–2.08]	1.60 [0.97–2.64]

Restricted to trials with at least one event, n = 10.

Table 2. Overall cancer risk at doses in line with NDA, restricting the analysis to trials during at least 52 weeks (n = 13, continuity correction used: 0.01, I2 = 0).