Modelling Trajectories of Care for Persons with Terminal Illness

Objective:  Heart failure is one of the most common reasons for hospitalization in patients over the age of 65.  Heart failure is also a common cause of death, and the proportion of Medicare beneficiaries with heart failure selecting the Medicare hospice benefit has increased substantially over the last decade.  Adherence to appropriate treatments (e.g., medications) can help avoid hospitalizations and maximize quality of life.  For example, the “Get With The Guidelines-Heart Failure Registry” reported that over 10% of heart failure hospitalizations during 2005-2007 were due to medication or dietary non-adherence.   Although provider prescription rates for evidence-based therapies have improved over time, especially during hospitalization for heart failure, overall non-adherence to heart failure medications has been reported to be between 11 and 48%. Furthermore, as the disease progresses, palliative care approaches may also help reduce hospitalizations and improve quality of life.  While understanding the appropriate point for initiation of palliative approaches is challenging, characterization of the trajectories of care for heart failure patients according to whether the patients are adherent to medications may help identify phases where initiation of palliative care discussions could be beneficial.

Methods:  This analysis will use data from the Atherosclerosis Risk in Communities (ARIC) Study to evaluate trajectories of service use for heart failure patients separately for patients who are adherent versus non-adherent to medications.  ARIC is an ongoing prospective epidemiologic study that enrolled a cohort of 15,792 participants aged 45-64 years in 1987-1989 from four communities.  From 2005-2010, medical records were abstracted for 2,575 hospitalizations for heart failure for 1,079 cohort members, and these data were combined with Medicare claims to track service use.  Approximately 60% of these cohort members were enrolled in Medicare Part D; Part D claims are being used to construct measures of “proportion days covered” for each month for heart failure medications.  The analysis will use a person-month file to characterize the trajectories of service use (e.g., inpatient stays including hospitalizations as well as observation stays, medication use, inpatient and total Medicare reimbursements, and hospice days) following incident heart failure hospitalization up to death.  Survival will also be an important outcome of interest, though since heart failure ultimately leads to death for most patients, alternative measures such as “hospitalization-free survival days” or “institutional day-free survival days” will be explored. 

Analytic Challenges and Expected Results: The ARIC Study provides measures of socio-economic status and living arrangement that are not available through Medicare claims alone; these measures can be used to reduce the problem of omitted variable bias. However, bias from selection is an inherent problem in identifying the effects of treatments such as medications.  The limited geographic areas of the ARIC study likely preclude the use of variation in provider prescription patterns to identify medication effects.  Regardless, characterizing the patterns of care over the full course of an important disease can help identify ranges of potential benefits such as reductions in hospital use and likely associated improvements in quality of life.