Table 2.

Relative risks and 95% CIs from 12 univariate models estimating the effect of intervention by outcome measure

Outcome measuresNo. of observations (no. of patients)RS versus non-RSAspirin versus nonaspirin
Relative riska (95% CI)PRelative risk (95% CI)P
Total number of polyps in rectum and sigmoid215 (116)1.05 (0.73–1.49)0.800.77 (0.54–1.10)0.16
Crypt width95 (58)1.03 (0.93–1.13)0.600.95 (−0.15 to 0.04)0.28
Crypt length95 (58)1.01 (0.93–1.09)0.861.08 (1.00–1.16)0.04
Mean total CCP95 (58)1.28 (0.94–1.73)0.121.37 (1.00–1.86)0.05
Mean number of MIB1-positive cellsb116 (78)1.02 (0.89–1.16)0.820.99 (0.87–1.14)0.93
Hemicrypt cellsb,c116 (78)−0.82 (−5.05 to 3.42)c0.713.67 (−0.55 to 7.90)c0.09

NOTE: All models adjusted for first result and time on intervention (years).

  • aEstimated as the exponential coefficient of the intervention effect in the random effects model.

  • bAdjusted for calendar time to account for possible effect of differences in storage time of these samples before analysis; MIB1 is the antibody used in staining for Ki67.

  • cVariable not logged, coefficient not exponentiated.