Penrod JD (James J. Peters VAMC/Mount Sinai School of Medicine), Deb P
(Hunter College), Dellenbaugh C
(James J. Peters VAMC), Burgess JF
(VA Boston Healthcare System), Zhu C
(James J. Peters VAMC/Mount Sinai School of Medicine), Christiansen CL
(Boston University), Luhrs C
(VA New York Harbor Healthcare System), Livote EE
(James J. Peters VAMC), Cortez TB
(VA New York Harbor Healthcare System), Morrison Rs
(Mount Sinai School of Medicine)
Objectives:
In 2008, VA launched an $80 million dollar initiative to expand palliative care programs. Palliative care consultation teams (PCCTs) aim to relieve suffering and improve quality of life for patients with advanced illness and their families. Several studies have shown cost savings with PCCTs, both in the VA and private sector, but none have taken selection of palliative care on unobservable factors into account. We estimated selection-corrected VA hospital costs for veterans with advanced disease receiving palliative versus usual care in five VA medical centers during FY2005 and FY06.
Methods:
Using a sample of 6595 hospital stays for advanced disease for 606 palliative care and 3321 usual care patients, we used generalized linear models (GLM) with selection-corrections to estimate hospital daily total direct, pharmacy, nursing, laboratory, and radiological costs, and a probit model with selection-correction to estimate the probability of ICU admission. The key independent variable was a binary indicator for whether the patient had received palliative care consultation. Predictors included: patient age, race, marital status, principal diagnosis, inpatient (LOS), comorbidities, indicator for death, and enrollment category. We corrected for potential selection bias using nonlinear instrumental variables (IV) estimation (using simulated likelihood methods in the GLM models and a bivariate probit for ICU admissions) with patients’ attending MD as the identifying instrument for selection into palliative care.
Results:
Hospital total direct, pharmacy, nursing, laboratory, and radiological costs were $464, $51, $182, $49 and $11 lower per day (p < 0.01 in all cases), respectively, for patients receiving palliative compared to usual care. All cost results have been corrected for potential selection bias. Palliative care patients were 43.7 percentage points (p < 0.001) less likely to be admitted to ICU during the hospitalization than usual care patients.
Implications:
Costs of hospital care were significantly lower per day for patients with advanced disease seen by PCCTs compared to those receiving usual care. Less frequent use of ICU by palliative care patients is a major source of cost differences.
Impacts:
Given previous findings of better patient and family outcomes with palliative care, our results suggest both a cost and quality incentive for VA medical centers to develop and expand palliative care programs.