Abstract
Calcium and, to a lesser extent, dairy products are consistently modestly inversely associated with colorectal cancer (CRC). Dairy products may contain components other than calcium and fat, such as insulin-like growth factor-1, that may affect CRC risk. In the prospective Iowa Women's Health Study, calcium, dairy product, and vitamin D intakes were assessed using a semiquantitative food frequency questionnaire. To investigate dairy products independent of their calcium components, we estimated residuals from linear regression models of their associations with dietary calcium. Of the 35,221 55-69-year-old cancer-free women at baseline in 1986, 1,731 developed CRC during follow-up through 2012. For those in the highest relative to the lowest intake quintiles, the adjusted hazards ratios and 95% confidence intervals from multivariable Cox proportional hazards regression models for overall and distal CRC were 0.81 (0.67-0.98; P(trend) = 0.004) and 0.59 (0.44-0.80; P(trend) = 0.003), respectively, for total calcium; and 0.79 (0.66-0.94; P(trend) = 0.01) and 0.69 (0.53-0.90; P(trend) = 0.003) for total dairy products, respectively. The various dairy product residuals were not associated with CRC. These results support that, among women, calcium and dairy products may be inversely associated with CRC-perhaps primarily distal CRC-but suggest that the non-calcium, non-fat component of dairy products may not be associated with CRC.