Table 2. HRs and 95% CIs for the association between NSAID and HT use before diagnosis and colorectal cancer mortality, stratified by tumour sub-site.

	All tumour sub-sites combined				Proximal tumours		Distal/rectal tumours
Medication exposure	Cases at risk a	Colorectal cancer deaths	HR	95% CI	HR	95% CI	HR	95% CI
NSAID use b
Never	444	81	1.00	Referent	1.00	Referent	1.00	Referent
Ever	466	62	0.88	0.62–1.24	0.55	0.32–0.92	1.32	0.83–2.10
Aspirinc	340	50	1.14	0.77–1.69	0.75	0.41–1.37	1.64	0.96–2.78
Ibuprofen	182	19	0.51	0.26–1.00	0.09	0.01–0.69	0.94	0.45–1.97
HT use d
Never	344	57	1.00	Referent	1.00	Referent	1.00	Referent
Ever	465	68	0.95	0.66–1.35	0.90	0.53–1.54	0.95	0.57–1.56
Oestrogen only	429	60	1.04	0.69–1.55	1.17	0.65–2.08	0.91	0.51–1.63
Oestrogens and progestin	206	26	0.92	0.43–1.97	0.72	0.17–3.16	1.19	0.46–3.09

Abbreviations : CI=confidence interval; HR=hazard ratio; HT=hormone therapy; NSAID=non-steroidal anti-inflammatory drug.

Models in table above are adjusted for age at diagnosis, body mass index, smoking status, family history of colorectal cancer, history of preventive screening and stage of disease at diagnosis.

^aReported number of cases at risk and number of colorectal cancer deaths are out of the total number of Caucasian women with non-advanced disease (n=933) for each exposure. All HT models are restricted to Caucasian women at least 50 years of age with non-advanced disease (n=818). Note that not every case answered the medication questions in the interview.

^bAdditionally adjusted for HT use (never, ever).

^cResults by type are examined as ever use, adjusted for use of other medication types. For example, results reported for aspirin are generated from a model comparing ever users of aspirin to never users, including ‘ever use’ of ibuprofen in the model.

^dAdditionally adjusted for NSAID use (never, ever).