Table 4. Overall effect estimates for prostate cancer and statin use according to study characteristics.
No. of studies | Pooled estimate | Tests of heterogeneity | pinteraction | Tests of publication bias | |||||
RR (95% CI) | p-value | Q value (d.f.) | p-value | I2 (%) | Begg's p | Egger's p | |||
All studies | 27 | 0.93 (0.87–0.99) | 0.03† | 145.30 (26) | <0.001 | 82 | 0.56∥ | 0.12 | |
Study design | 0.45§ | ||||||||
Cohort | 15 | 0.93 (0.87–1.01) | 0.09† | 88.60 (14) | <0.001 | 84 | 0.56∥ | 0.07 | |
Case-control | 12 | 0.87 (0.72–1.05) | 0.15 | 56.64 (11) | <0.001 | 81 | 0.31∥ | 0.09 | |
PSA testing | 0.76§ | ||||||||
Adjusted | 6 | 0.91 (0.81–1.02) | 0.13 | 14.87 (5) | 0.011 | 66 | >0.99∥ | 0.49 | |
Not adjusted | 21 | 0.93 (0.86–1.01) | 0.11 | 121.23 (20) | <0.001 | 83 | 0.53∥ | 0.04 | |
BMI and ALS | 0.24§ | ||||||||
Adjusted | 11 | 0.88 (0.78–0.99) | 0.04† | 44.23 (10) | <0.001 | 77 | 0.54∥ | 0.22 | |
Not adjusted | 16 | 0.96 (0.88–1.04) | 0.37 | 81.09 (15) | <0.001 | 81 | 0.56∥ | 0.12 | |
Bonovas et al. [24] analysis | 0.63§ | ||||||||
Before | 10 | 0.95 (0.82–1.11) | 0.58 | 26.62 (9) | 0.002 | 66 | 0.73∥ | 0.51 | |
After | 17 | 0.91 (0.84–0.99) | 0.03† | 118.62 (16) | <0.001 | 87 | 0.27∥ | 0.01 | |
Results for long-term statin use | 11 | 0.94 (0.84–1.05) | 0.31 | 28.80 (10) | 0.001 | 65 | 0.74§ | 0.28∥ | 0.17 |
Cohort studies | 7 | 0.91 (0.81–1.02) | 0.12 | 14.79 (6) | 0.02 | 59 | 0.14∥ | 0.02 | |
Case-control studies | 4 | 0.97 (0.64–1.48) | 0.92 | 9.75 (3) | 0.02 | 69 | 0.33∥ | 0.49 | |
Results for advanced prostate cancer | 7 | 0.80 (0.70–0.90)* | <0.001† | 8.98 (6) | 0.17‡ | 33 | 0.25§ | 0.77∥ | 0.90 |
Cohort studies | 5 | 0.85 (0.72–1.00)* | 0.04† | 7.75 (4) | 0.10‡ | 48 | 0.81∥ | 0.62 | |
Case-control studies | 2 | 0.74 (0.62–0.88)* | 0.001† | 0.01 (1) | 0.90‡ | - | - | - |
PSA, Prostate specific antigen; BMI, Body mass index; ALS, Adverse life style; RR, Relative risk; CI, Confidence interval; d.f., Degree of freedom.
Relative risk from fixed-effects model due to no heterogeneity among the studies;
P value representing significant inverse association between statin use and prostate cancer;
Statistically significant for homogeneity;
Test of interaction was not statistically significant;
Statistically significant for no publication bias.