Determinants of functional decline and survival in elderly people following cancer diagnosis: the need to consider geriatric factors

Abstract

Background Growing incidence of cancer in aging population is an epidemiologic reality and the disparate prognosis in elderly needs to be addressed. Studies on these aspects are limited and usually overlook geriatric-specific factors. The aim of this work was to study determinants of prognosis in elderly after cancer diagnosis. Methods Subjects aged ≥ 65 years from elderly cohorts (PAQUID, 3 C, AMI), alive on January 1st 2005 and with a validated cancer diagnosis recorded in cancer registries of Gironde, a French department, from January 1st 2005 through to December 31st 2014 were included. Functional decline (FD) and overall survival (OS) were the primary outcomes. Functional status was measured using the Instrumental Activities of Daily Living (IADL) and the basic Activities of Daily Living (ADL) scales. FD was measured between cancer pre- and post-diagnosis cohort visits. FD was considered as occurrence of ADL limitation (severe disability) and occurrence of ADL and/or IADL limitation (mild disability). As probability of FD accelerates as death approaches, a composite outcome of either occurrence of ADL limitation or death was also considered. Survival time was measured from the date of diagnosis until patient's death or 31st of July 2017, whichever came first. Age at diagnosis, sex, living alone, education, diagnosis stage, treatment, smoking status, polypharmacy, depressive symptomatology, cognitive impairment or dementia and general practitioners (GP) per 100,000 inhabitants were the variables studied. For FD determinants, subjects with disability at baseline (severe or mild depending on analysis) were excluded. Logistic regression models were performed and adjusted on cohort study and delay between pre- and post-diagnosis visits or death. For determinants of OS, Cox models were fitted and adjusted on cohort study and age at diagnosis. To account for different levels, we applied random effects with clustering at the level of the number of GPs. Only multivariate models were reported as variances were non-significant. Analyses were case-completed. Results A total of 486 subjects were included in the study. The median age at cancer diagnosis was 83 years. Overall, 55% were male, 51% had an education higher than primary school, 62% did not live alone, 59% took 6 or less daily drugs and 87% did not present either cognitive impairment or dementia. Over 25% of subjects were diagnosed at advanced cancer stage and 77% received cancer treatment. At the post-diagnosis visit, 258 (53%) subjects were seen, 43 (9%) were lost to follow-up and 185 (38%) were dead. OS after cancer diagnosis was 77%, 66% and 41% at six months, 1 and 2 years, respectively. Regarding FD, in the final model, older age (OR = 18.3; 95% CI = 3.7–90.9), presence of cognitive impairment or dementia at the pre-diagnosis visit (OR = 8.3%; 95% CI = 2.6–27.0) and an advanced stage of cancer at diagnosis (OR = 4.7; 95% CI = 1.3–16.7) were associated with a significantly higher risk of severe disability. In the mild disability analysis, subjects with a higher education than primary school (OR = 0.4; 95% CI = 0.2–0.9) were at lower risk, while older age (OR = 3.3; 95% CI = 1.3–8.7) and those taking 6 or more daily drugs (OR = 2.3 95% CI = 1.0–5.2) were at higher risk. Regarding OS, current or former smoker patients (HR = 1.44 95% CI = 1.06–1.95), presenting low (HR = 1.63 95% CI = 1.16–2.29), moderate (HR = 2.62 95% CI = 1.80–3.83) or high disability (HR = 3.85 95% CI = 2.19–6.75), diagnosed at advanced stage (HR = 3.97 95% CI = 3.00–5.25) and not receiving treatment (HR = 1.98 95% CI = 1.44–2.71) had higher risk of death, while women had lower risk (HR = 0.66 95% CI = 0.48–0.90). Conclusion In addition to classical determinants of prognosis in cancer, we demonstrated the impact of cognitive impairment on FD and that of disability on OS. It appears essential to consider geriatric factors in outcomes studies on the elderly cancer population.