Aspirin Use After Diagnosis Improves Survival in Older Adults With Colon Cancer

A Retrospective Cohort Study

Marlies S. Reimers, MD; Esther Bastiaannet, MSc; Myrthe P. P. van Herk-Sukel, PhD; Valery E. P. Lemmens, PhD; Colette B. M. van den Broek, MD; Cornelis J. H. van de Velde, MD, PhD; Anton J. M. de Craen, PhD; Gerrit Jan Liefers, MD, PhD


J Am Geriatr Soc. 2012;60(12):2232-2236. 

In This Article



The central patient database of PHARMO (PHARMO Institute for Drug Outcomes Research, Utrecht, the Netherlands), which links to more than 10 databases using different medical record linkage algorithms, was recently combined with data from the Eindhoven Cancer Registry (ECR).[15] From the PHARMO database, prescriptions of low-dose aspirin (80 mg) were selected and linked to individuals diagnosed with colorectal cancer who were registered in the ECR between 1998 and 2007. Four thousand four hundred eighty-one individuals with colorectal cancer were included in this database. A subgroup analysis was performed of this previously published cohort comprising specifically individuals aged 70 and older diagnosed with colon cancer who used aspirin only after diagnosis (users) or never used aspirin (nonusers) (n = 536).[2] The date of prescription and date of diagnosis were compared to assess whether the aspirin was prescribed before or after the diagnosis.


Vital status of participants was established directly from the participant's medical record or through linkage of cancer registry data with the municipal population registries, which record information on the vital status of their inhabitants. Follow-up started 30 days from diagnosis of colorectal cancer (T0), because information concerning prescriptions in the hospital was unavailable. Follow-up was until the last contact date or death. Users were defined as participants who had at least one prescription for aspirin for at least 14 days; participants who were prescribed aspirin for less than 14 days were defined as nonusers. Time-dependent survival analyses were used to assess survival. Participants were defined as nonusers from T0 to first use and as users from first use to the end of follow-up. Poisson regression survival models were used to study the effect of aspirin on overall survival. In multivariate proportional hazards regression analysis, adjustments were made for sex, age (continuous), stage (pathological stage or clinical stage if pathological stage was unknown), adjuvant chemotherapy (yes/no), comorbidity (yes/no), surgery (yes/no), grade, location of tumor, and year of diagnosis. Finally, stratified analyses were performed for type of comorbidity, chemotherapy, grade, stage, surgery, and location of tumor.