Oobjectives: To determine whether use of symptom relief drugs (e.g., antidepressants, anxiolytics, opioid analgesics, sleep aids) rises and use of two commonly prescribed classes of chronic medications (statins and osteoporosis drugs) falls with greater probability of death for older Medicare beneficiaries.

Design: Pooled cross-sectional study.

Setting: Non-institutionalised older US Medicare population in 2000 to 2005.

Participants: Community-dwelling Medicare beneficiaries aged 65 and older (N = 20,233).

Measurements: Use of medications measured according to dichotomous flags; intensity of use by annual medication fills.  Annual probability of death modeled using logistic regression and stratified into 7 groups with predicted probabilities of death that range from less than 5% to greater than 50%.  Prevalence of use and intensity (mean prescription fills per month) were computed for each class of medication.

Results: For symptom relief medications, there is relatively constant use with increasing probability of death, along with greater intensity of use.  For the two chronic medications, there was a monotonic decrease in use but at a relatively constant intensity.  Decline in statin use ranged from 34.4% in the lowest mortality stratum to 17.6% for those in the highest (P less than 0.001).  Use of osteoporosis drugs fell from 10.4% to 6.6% over the same range (P less than 0.001).

Conclusions: Greater intensity of use of symptom relief medications with increasing probability of death is consistent with hypothesised use.  The different profile for chronic medications suggests that the time to benefit is being considered regarding therapy initiation, which results in lower use.