Factors Associated with Advance Care Plans and End-Of-Life Care Choices Among Elderly Americans: An Analysis of Health and Retirement Study Data

Introduction Advance care plans (ACP) allow people to plan for their end-of-life care before they become incompetent to make their treatment decisions. The umbrella term Advance Care Plans (ACP) includes the three most commonly used end-of-life care plans: Advance Care Planning Discussions (ACP discussions) and two advanced directives: living will and Durable Power of Attorney for Health Care (DPAHC). The ACP discussions are the verbal discussions about end-of-life plans, whereas the advance directives are written documents. ACPs are distinct nuances of end-of-life care planning. ACP discussions address a wide array of end-of-life care issues, including terminal care, funeral, burial and the place of death, etc. A living will outlines specific end-of-life care choices and elicits yes or no responses. The choices pertain to use of artificial respiration; artificial feeding and hydration; dialysis; or antibiotics; etc. A DPAHC, appoints a proxy to make treatment decisions on behalf of the incompetent patient at a terminal stage of life. While previous studies have used ACPs as distinct outcomes, in real life the ACPs exist in combinations. People who undertake ACP discussions are more likely to complete advance directives. More than 25 states have combined directives forms. Therefore, it is imperative to evaluate the factors associated with the combinations of ACPs: No ACP; ACP discussions only; a directive (a living will or DPAHC); a directive and ACP discussions; both directives (a living will and DPAHC); and all ACPs (a living will, DPAHC and ACP discussions). Among the factors associated with ACPs, health status has shown an inconsistent association. Some studies have shown that poor health is associated with higher ACP uptake rates, whereas others have noted no association. The possible reasons for inconsistent association include 1) examining the association without controlling for the change in health status and other health factors — prior research shows health status and change in health are closely related in influencing the uptake of ACPs and the end-of-life care choices 2) use of each ACP as a separate outcome instead of using them in combinations. Therefore, our first study attempted to clarify the association between health status, change in health status and interaction between the two measures with the combinations of ACPs. Our second study determined the factors associated with end-of-life care choices. Prior concerning the association between health status and end-of-life care choices have used prospect theory. However, previous research has used convenience samples and end-of-life care scenarios. We tested the prospect theory using a representative population-based sample and using the choices that people make considering their own health status and possible end-of-life circumstances. Methods We used the Health and Retirement Study (HRS) panel data from 1992-2014 and the HRS exit interview data from 2002-2014. The HRS captures health and retirement characteristics of a representative sample of Americans over 50 years using biennial panel surveys since 1992. It also conducts one-time post-death interviews with the next-of-kin of HRS decedents in the survey waves following their death. The post-death surveys collect information about medical care expenditures and use; advance care planning and end-of-life care choices and distribution of assets towards end-of-life. We used the Analytics Software and Solutions SAS version 9.4 to examine the association between health status and ACPs, we used a multinomial regression model. The combinations of ACPs were used as the study outcome. To study the association between health status and choices, a separate logistic regression model was used for each choice — limit care in certain situations, comfort care and all care possible. Results In study 1, self-reported health was not associated with any category of ACP combinations. However, change in health status was associated with ACPs — “worse or somewhat worse” change in health status since the last survey wave was associated with a higher uptake of “two directives” and “all ACPs”, compared with “much or somewhat better or the same”. The number of health conditions and a history of cancer were also associated with “all ACPs”. In study 2, we did not find association between self-reports of health and its change with the two care-limiting choices, including “limit care in certain situations” and “comfort care”. However, change in health status was associated with the “all care possible” option — a decline in health status since the last wave was associated with a higher likelihood of “all care possible” choice than improvement or no change in health status since the last wave. Among other health factors, a psychiatric illness was associated a higher uptake of “all care possible” and a lower uptake of “comfort care”. The decedents with a history of stroke chose less “limit care in certain situations” option. Recommendations We recommend further research on the factors associated with the combinations of ACPs. Future research should also use the combinations to determine the effects of ACPs on the cost and quality of end-of-life care