Figure 3.

Conditional logistic regression for purchases prior to diagnosis (Dx). Cases (n=153) were matched with controls according to age and store card use (HSR1 only, HSR2 only, or HSR1+HSR2) at a 2:1 ratio with replacement (n=306 controls in matched analysis, 77 unique individuals). Participants with a previous diagnosis of cancer were excluded. Where possible, the same 24 months of purchases from controls were aligned to each matched case leading to their diagnosis (–24 months to –1 month on the X-axis); otherwise, the last 24 months of purchases were used for controls. Conditional logistic regression of matched case-control sets adjusting for oral contraceptive pill use and household number for combined sum of pain and indigestion medication (A), pain medication only (B), or indigestion medication only (C). HSR: high street retailer; OR: odds ratio.