The Centers for Medicare and Medicaid Services has decided to limit payments to hospitals most likely to have patients readmitted. However, the data suggest that readmissions are best correlated with patient-specific issues.
With significant impact on patients’ quality of life and healthcare spending, hospital readmissions are a popular target for policy makers to influence the behaviors of health care providers. One recent Medical Care Research and Review study examines readmissions in the context of congestive heart failure using data from the Healthcare Cost & Utilization Project (HCUP). In contrast to most published research based only upon Medicare data, HCUP includes information on all payers within 14 states.
The authors’ primary outcome measure was all-cause readmission rates within 30 days of the index hospitalization. They used multilevel hierarchical logistic modeling to progressively account for the cost of index admission and covariates at patient, hospital, and community levels. Patient-level covariates included sociodemographics, payer status, and functional severity-of-illness. Hospital-level covariates included size, ownership, teaching status, and location. Census data on wealth and educational attainment served as community-level covariates.
The study yielded two significant findings. First, discharge against medical advice was associated with the highest the odds of readmission (odds ratio = 2.0, p-value < 0.001). This finding was persistent (odds ratio = 1.9, p-value <0.001) even in a fully adjusted model that controlled for patient comorbidities, community sociodemographic factors, hospital demographics, and cost.
The second factor strongly associated with readmission was Medicaid payer status (relative to private insurance). In the patient-level model, these Medicaid beneficiaries were 70 percent more likely (odds ratio = 1.7, p-value <0.001) to be readmitted to the hospital. However, in the fully adjusted model, this decreased to just 30 percent more likely (odds ratio = 1.3, p-value <0.001).
Even though Medicare patients have a small but statistically significant increased likelihood of readmissions, Medicaid patients represented the only payer demographic to substantially be more likely to suffer readmission.
Interestingly enough, hospital and community characteristics – number of beds, ownership, teaching status, metropolitan area versus rural, income, and education – appeared unrelated to the risk-adjusted odds of readmission. The only other variable with a magnitude of effect greater than insurance status (besides leaving against medical advice) was drug abuse, a patient-specific condition.
This research is especially timely in light of the Affordable Care Act’s recently implemented provisions limiting reimbursement for “preventable” readmissions. The findings highlight one rather vexing circumstance for health care providers – discharges against medical advice (AMA) double the odds of readmission. While medical personnel are partly to blame for such discharges, modern notions of patient autonomy are generally inimical to the involuntarily confinement of patients.
Nonetheless, it may be possible to improve using crisis intervention teams akin to those employed in cases of preventable medical harm. Prior reports from patient focus groups suggest that differing expectations underlie the problem of leaving AMA. Theoretically, crisis teams might be better able to manage divergent patient and provider expectations.
With funding cuts already underway for “preventable” readmissions, addressing the knowledge deficit regarding the underlying mechanisms of readmissions is of paramount importance. This study illuminates several areas for immediate intervention (leaving AMA, drug abuse, and nuances of Medicaid).
Andrew Gonzalez, MD, JD, MPH