Self-reported depression among cancer survivors in Medicare beneficiaries: A statistical method comparison between complex survey design and simple random sample design
Depression had negative effect on cancer treatment and survivorship. This study evaluated self-reported depression prevalence among elder cancer survivors (age 65+). 2012 Medicare Beneficiary Survey (MCBS) data was used to address this question. The beneficiaries were classified into two groups: cancer survivors and non-cancer beneficiaries. We discovered the prevalence of self-reported depression among cancer survivors was 64.48% and 61.71% among non-cancer beneficiaries, and the difference was significant (P=0.048). In multiple logistic regression, the odds of self-reported depression in elder cancer survivors was 1.235 (P=0.0004) times higher than the odds in elder adults without cancer. Because of such high prevalence among cancer survivors, appropriate medical management and social support would be needed to address this issue. Also this study conducted a statistical methodology comparison between 'Complex sampling design and considering cluster/stratum/weight', 'Complex sampling design and considering weight only' and 'Simple random sample design'. For complex survey data, 'Complex sampling design and considering cluster/stratum/weight' was the best available method and was set as reference. 'Complex sampling design and considering weight only' had the same point estimates but narrower confidence intervals. And 'Simple random sample design' was biased in both point estimates and confidence intervals.
Zhao, Bo, "Self-reported depression among cancer survivors in Medicare beneficiaries: A statistical method comparison between complex survey design and simple random sample design" (2016). Texas Medical Center Dissertations (via ProQuest). AAI10126740.