How Do Older Adults Consider Age, Life Expectancy, Quality of Life, and Physician Recommendations When Making Cancer Screening Decisions? Results from a National Survey Using a Discrete Choice Experiment

Abstract

Background. Older adults with limited life expectancy frequently receive cancer screening, although on average, harms outweigh benefits. We examined the influence of life expectancy on older adults' cancer screening decisions relative to three other factors. Methods. Adults aged 65+ years (N = 1272) were recruited from a national online survey panel. Using a discrete choice experiment, we systematically varied a hypothetical patient's life expectancy, age, quality of life, and physician's recommendation and asked whether the participant would choose screening. Participants were randomized to questions about colonoscopy or prostate-specific antigen/mammography screenings. Logistic regression produced preference weights that quantified the relative influence of the 4 factors on screening decisions. Results. 879 older adults completed the survey, 660 of whom varied their screening choices in response to the 4 factors we tested. The age of the hypothetical patient had the largest influence on choosing screening: the effect of age being 65 versus 85 years had a preference weight of 2.44 (95% confidence interval [CI]: 2.22, 2.65). Life expectancy (10 versus 1 year) had the second largest influence (preference weight: 1.64, CI: 1.41, 1.87). Physician recommendation (screen versus do not screen) and quality of life (good versus poor) were less influential, with preference weights of 0.90 (CI: 0.72, 1.08) and 0.68 (CI: 0.52, 0.83), respectively. Conclusions. While clinical practice guidelines increasingly use life expectancy in addition to age to guide screening decisions, we find that age is the most influential factor, independent of life expectancy, quality of life, and physician recommendation, in older adults' cancer screening choices. Strategies to reduce overscreening should consider the importance patients give to continuing screening at younger ages, even when life expectancy is limited.

Keywords: cancer screening; decision making; discrete choice experiment; older adults; patient preferences.

Figure 1

An example of a choice task presented in the survey. *Male participants were presented with a hypothetical patient, Mr. Jones. Female participants were presented with a hypothetical patient Ms. Jones. **Participants were randomized to receive questions about mammography, PSA, or colonoscopy.

Figure 2

Preference weights for different levels of life expectancy, age, quality of life, and physician recommendation in participants’ cancer screening decisions (n=660).^{a,b a} The bars represent the preference weights for each attribute level. The brackets around the preference weights indicate the 95% confidence intervals for the estimates. Within each attribute the vertical distance between bars indicates the relative change in preference weight when changing from one level of the attribute to another level of the same attribute. ^b Choice analysis only included participants whose screening decisions varied in response to different combinations of the attribute levels.