Among 418,458 women in our dataset, 68,641 were at least 66 years of age and had undergone bilateral mammography between April 2001 and June 2002. Of these patients, 46,080 fulfilled eligibility criteria, including 1,090 (2.4%) cases and 44,990 (97.6%) controls. Patient characteristics are presented in Table . Patients diagnosed with breast cancer were slightly older than controls. Compared to controls, cases more commonly had a history of diagnostic and therapeutic breast procedures and a history of benign neoplasm of the breast or other breast diseases in the three years prior to the index date. Cases used estrogen replacement therapy less frequently than controls and were less likely than controls to have seen a gynecologist in the year prior to the index date. Twenty one cases (1.9%) and 101 controls (0.2%) died during the 6-month period following the index date.
Patient characteristics at the index date and mortality in the following 6 months
Exposure to the drugs of interest
Frequency of exposure to the medications of interest are indicated in Table (≥ 90 days of ns-NSAIDs/COX-2 inhibitors, aspirin, acetaminophen). Controls were more likely than cases to be exposed to ≥ 90 days of ns-NSAIDs and to COX-2 inhibitors. Patients exposed to at least 90 days of either ns-NSAIDs or COX-2 inhibitors during the year prior to the index date filled on average (± standard deviation) 8.0 ± 4.4 prescriptions for these drugs during that year with a median number of 7 prescriptions. Ninety percent of all aspirin prescriptions were for 325 mg/day or less and 25% of them were for ≤ 80 mg/day, the median dose (interquartile range) for all aspirin prescriptions were 312.5 (80, 312.5).
Exposures to ≥ 90 days of COX-2 inhibitors, ns-NSAIDs, aspirin and acetaminophen in the year prior to the index date: Crude and adjusted odds ratios.
Patient characteristics associated with breast cancer
Logistic regression models showed that cases were older (odds ratio, 95% confidence interval for 75–84 years and ≥ 85 years compared to 66–74 years: 1.90, 1.66–2.18 and 3.68, 2.69–5.04, respectively) and more likely to have received combined estrogen therapy 1.35 (1.06, 1.70) in the year prior to the index date. Cases were also more likely to have a history of diagnostic/therapeutic breast procedures in the three years prior to the index date (2.60, 1.59–4.26), benign breast neoplasm diagnosed in the year prior to the index date (1.40, 1.12–1.75), and/or other breast disease in the three years prior to the index date (1.87, 1.57–2.23). Cases were less likely to have used estrogen replacement therapy in the year prior to the index date (0.71, 0.61–0.84). Cases were also less likely to have seen a gynecologist in the year prior to the index date (0.68, 0.57–0.81) and to have had mammography in the years 2–3 prior to the index date (0.29, 0.26–0.33) (data not shown).
Table shows the crude and adjusted odds ratios of the association between exposure to the drugs of interest and breast cancer. Small differences between the crude and adjusted odds ratios were observed. We investigated these differences and found that they resulted from the adjustment for other exposure categories. In fact, exposure categories were not mutually exclusive. For example, some patients exposed to at least 90 days of ns-NSAIDs and/or COX-2 inhibitors were also exposed to at least 90 days of aspirin. When only exposure categories were entered in the model, the effect of each was adjusted for the effects of the others. None of the other patient characteristics listed above was a confounder or effect modifier for exposure to ns-NSAIDs and/or COX-2 inhibitor. However, age was a confounder for exposure to aspirin.
As shown in Table , exposure to at least 90 days of an ns-NSAID and/or COX-2 inhibitor (0.75, 0.64–0.89) was inversely associated with the risk of breast cancer as was exposure to at least 90 days of aspirin at an average daily dose > 100 mg (0.75, 0.64 – 0.89). In contrast, exposure to at least 90 days of aspirin at an average daily dose ≤ 100 mg and exposure to at least 90 days of acetaminophen were not found to be associated with breast cancer. Exposures to at least 90 days of COX-2 inhibitors (0.81, 0.68–0.97) or ns-NSAIDs (0.65, 0.43–0.99), assessed separately, were associated with a lower risk of breast cancer. Exposure to at least 90 days of celecoxib (0.88, 0.70–1.10) or rofecoxib (0.80, 0.63–1.02) showed trends toward an association with a lower risk of breast cancer.
Consistent with the results of the primary analysis, exposure to ≥ 90 days of ns-NSAIDs and/or COX-2 inhibitors was also associated with a lower risk of breast cancer in a subgroup analysis of patients who had mammography in years 2–3 prior to the index date (0.73, 0.55–0.93). Again, in this subgroup, exposure to ≥ 90 days of acetaminophen (0.86, 0.59–1.31) and exposure to ≥ 90 days of aspirin at an average daily dose ≤ 100 mg (0.97, 0.68–1.39) were not associated with breast cancer. In contrast with the findings of the primary analysis, exposure to ≥ 90 days of aspirin at an average daily dose >100 mg (0.93, 0.72–1.18) did not seem to be associated with breast cancer in this subgroup. It is difficult to conclude if this difference was due to chance given the multiple modeling conducted in this study or to differences in the characteristics of the women in the two analyses. In particular, in the primary analysis, only 25% (54/211) of the cases who received aspirin for 90 days or more at an average daily dose >100 mg also received hormone replacement therapy compared to 42% (45/93) of women in the secondary analysis. Thus, there may have been better surveillance of the women in the secondary analysis. Of note, the effect of aspirin at an average daily dose >100 mg in the subgroup analysis differed between women who were not receiving hormone replacement therapy (0.83, 0.60, 1.15) and those who did (1.08, 0.74, 1.56).
The results of the main analysis did not change substantially when we reassessed the effect of exposure to ≥ 90 days of ns-NSAIDs and/or COX-2 inhibitors (0.75, 0.63–0.89), aspirin at an average daily dose ≤ 100 mg (0.90, 0.71, 1.15) aspirin at an average daily dose > 100 mg (0.75, 0.63, 0.89) and acetaminophen (0.89, 0.69, 1.15) excluding the prescriptions filled in the month prior to the index date.