DISCONTINUATION OF MAMMOGRAPHY FOLLOWING DECLINES IN LIFE EXPECTANCY

Purpose: The expected benefit of detection and treatment of early, asymptomatic cancer tumors is negatively related to remaining life expectancy. It makes no sense to treat early stage cancer in a patient who will die of another cause before the cancer becomes clinically apparent. Screening guidelines advise physicians and patients to consider remaining life expectancy, but are vague on the subject of when exactly to discontinue screening. We examine the impact of hospitalization, which represents a “negative health shock”, on discontinuation of mammography among previously screened women.

 

Method: Using Medicare claims and enrollment records, we identified women age 65+ continuously enrolled in fee-for-service Medicare who received a mammogram in 1997 (the index mammography) and were not previously diagnosed with breast cancer. We identified receipt of mammography using physician and outpatient claims. We identified hospitalizations in the 11 month window following the index mammography using inpatient claims. We used Cox survival models to estimate the impact of hospitalization on the time to receipt of a second mammography, controlling for age, race, and census tract income and education levels.

 

Results: The sample was comprised of 42,328 women, 87% of whom had a second mammography in the follow-up period (until the end of 2004). Women who were hospitalized for any cause were less likely to have a second mammography (78%, hazard ratio [HR] 0.73, p<0.01), as were women who had a heart attack (72%, HR 0.67, p<0.01), or stroke (68%, HR 0.61, p<0.01). However, the majority of women age 75+ who were hospitalized or experienced a heart attack or stroke were re-screened.

 

Conclusion: We find that new information about life expectancy influences screening decisions among some patients and physicians; they do not all blindly persist in past screening behavior following hospitalization. However, the fact that so many women ages 75+ continue to receive mammography after having experienced a heart attack or stroke suggests that overscreening and overdetection are common. Previous studies have examined the relationship between mammography and life expectancy cross-sectionally, but these may be biased by the fact that many women never undergo mammography in the first place. We limit our sample to previously screened women. To the best of our knowledge, this is the first study to examine discontinuation of mammography.