End-of-Life Quality of Care in Nursing Homes: Rural-Urban Disparities

Differences in rural-urban practice styles and quality-of-care in nursing homes (NHs) were demonstrated in prior studies, but little is known about variations in end-of-life (EOL) quality. Our study examines this issue.

We constructed four risk-adjusted EOL quality measures (QMs) for long-term decedent residents: in-hospital-death; no hospice referral before death; pain; and dyspnea.

We used CY2005-2007 Minimum Data Set, Medicare beneficiary file, inpatient and hospice claims, Online Survey Certification and Reporting System and Rural-Urban Commuting Area Codes.

We estimated logistic regression models predicting the probability of each outcome conditional on decedents' risk factors (n=898,964 decedents). We then calculated QMs for each facility (n=14,611), as the difference between the actual and the expected risk-adjusted outcome rates.

To examine the association between EOL QMs and rurality, we fit multivariate linear regressions, with fixed state effects, for each QM (dependent variable). NH location – categorized as urban, large-town, small-town, isolated-rural – was the key variable of interest. Independent variables included: ownership, staffing, size, case mix, payer mix, etc.

Rural-urban disparities were found for all QMs. For example, for the in-hospital-death QM, we found that compared to urban NHs, facilities in large and small towns, and in isolated-rural areas, had worse (p<0.001) QMs (i.e. higher observed in-hospital-death rates than the risk-adjusted national average). QMs were significantly worse in small-town (p<.001) and in isolated-rural (p=0.039) NHs compared to large-town NHs.

We provide empirical evidence for rural-urban disparities in EOL quality of care. Identifying disparities is a necessary first step towards improving care for dying NH residents.