Re: Insulin-Like Growth Factor 1 and Prostate Cancer Risk: a Population-Based, Case-Control Study

Extract

Wolk et al. ( 1 ) present an important contribution to the evidence supporting a relation between insulin-like growth factor 1 (IGF-1) and the risk of prostate cancer. However, their decision to form the reference group by combining the subjects in the two lowest quartiles of IGF-1 concentration merits further consideration. Based on information in the report, I have reconstructed the case and control distribution for each quartile (see “Notes” below).

Wolk et al. formed the combined reference group because “the odds ratio [OR]. . . did not differ for the two lowest quartiles” and because the combination “increased the precision of the OR estimates in the two higher quartiles.” Although the case-control ratio in the second quartile (Q2) is not statistically significantly different from the case- control ratio in Q1, these ratios are certainly not the same (OR 4 0.69; 95% confidence interval [CI] 4 0.39- 1.23). The Q2 ratio differs from Q1 more than the case-control ratio in Q4 differs from Q1 (OR41.23; 95% CI4 0.72-2.09). Furthermore, while it is true that the combination increased the precision of the estimates of effect in Q3 and Q4 by 13%—the standard errors fell from 0.27 to 0.23 and 0.24, respectively— the combination had a larger impact on the point estimates of effect. The estimate of excess relative risk in Q3 increased by 60% (from 0.46 to 0.73) and in Q4 by 100% (from 0.23 to 0.46).