AcademyHealth Annual Research Meeting – Innovative Payment Mechanisms in Maryland Hospitals The Effect of Global Budgets on Hospital Utilization and Outpatient Quality in Maryland Nicolae Done Department of Health Policy and Management June 27, 2016 *I am grateful to the JHSPH Sommer Scholars Program for funding support for my doctoral studies. Based on joint work with Bradley Herring, Susan Hutfless, and Tim Xu.
Context: Payment Incentives Affect Hospital Behavior Under a DRG-based system, hospitals receive payment per case As long as payment > marginal cost, more admissions leads to more “profit” Incentive to drive up admissions Disincentive to provide high-quality preventive care Facilitated by information asymmetries and regulatory inability to fully monitor quality 2
Maryland Introduced Global Budgets for Hospitals in Two Waves In 2010 – Total Patient Revenue (TPR) pilot program Essentially for rural hospitals In 2014 – Global Budget revenue – state-wide expansion Tweaks to the TPR to accommodate urban hospitals with overlapping service areas Each hospital is assigned a Primary Service Area (PSA) and Secondary Service Area (SSA) Hospital is at risk for expenditures over the approved revenue but keeps any surplus One-time adjustments to next- year’s budget Market share changes monitored closely by HSCRC (Hospital Services Cost Review Commission), which regulates hospital rates for all payers and all hospitals in the state 3
Key Features of the Global Budget Program Base population Base year = FY2010 Residents living in hospital’s PSA and SSA Regulated services Inpatient and outpatient At the hospital campus Adjustments Payer mix, variation in prices RESULT: APPROVED COMBINED TOTAL REVENUE 4
Research Questions Did the TPR program decrease hospital utilization among residents in rural Maryland? 1. Inpatient 2. Outpatient A. Total admissions rate B. Total visits rate B. Total inpatient days rate C. ED visit rates C. Preventable admission rates (PQIs) a) Total a) b) Preventable Chronic b) Acute c) Primary Care Treatable D. Readmissions rate ED visits were categorized using the Billings criteria (Billings et al. 2000) 5
Study Design Difference-in-differences regression approach to adjust for time-varying confounders and control for unobserved heterogeneity and common trends Poisson model specification using counts as the outcome and population estimates as the population at risk 1. TPR only – Rural Hospitals, 2008-2013 (excluding long-term global budgets hospitals) ZIP codes served by the TPR hospitals (N=147) with ZIP codes assigned to a control A. group of rural, non-participating hospitals (N=62) TPR hospitals compared to all non-participating hospitals in the state B. (another N=263 ZIP codes) 6
Main Empirical Model Specification Y ~ Poisson( , ) ict ict exp{ T X } ic t it it ct i t Y Outcome for ZIP code at time i t it Expected counts it verdispersion parameter O estimate of p rogr am effe ct D D X ZIP code level time-varying characteristi cs it ZIP code fixed effects i Co unty-lev el time-varying characteristics ct year fixed effects t 7
Data Discharge Abstracts Database from HSCRC: universe of inpatient discharge abstracts to non-federal acute care hospitals in Maryland Detailed information on date of admission, diagnoses (ICD-9), performed procedures, patient demographics, ZIP code, etc. Linked to ZIP-code level data from Claritas Demographic Reports and the Census Bureau Supplemented with county-level data from the Area Resource File Calendar Years 2008-2013 8
Characteristics of ZIP Codes in the Service Areas of TPR and Rural Control Hospitals Pre-pilot (2008-2009) 2010-2013 TPR Rural Ctrl Urban TPR Rural Ctrl Urban Average population 5,235 9,088 16,004 5,262 9,453 16,534 Percent female 49.8 50.1 50.7 50.5 50.8 51.0 Median age 40.0 37.5 37.8 42.8 40.9 39.7 Percent non-white 10.3 16.2 31.8 10.1 21.3 34.6 Median household income ($) 56,112 80,060 75,362 66,284 95,895 87,834 Percent unemployed 4.3 3.0 4.0 6.8 5.0 6.4 Percent at least college degree 17.6 26.5 33.6 19.8 29.5 34.4 Percent uninsured 13.5 10.4 12.5 12.4 9.4 11.9 Physicians per 1,000 pop. 1.5 1.5 3.2 1.5 1.4 3.2 PCPs per 1,000 pop. 0.2 0.2 0.2 0.2 0.2 0.2 Specialists per 1,000 pop. 1.3 1.3 3.0 1.3 1.2 3.0 Number of FQHCs 1.9 0.5 4.6 2.0 0.7 5.5
Trends in Inpatient Utilization in Areas Served by TPR And Rural Control Hospitals, 2008-2015 Source: Author’s calculations obtained by applying the AHRQ PQI algorithms to HSCRC discharge abstract data and dividing by population estimates 10 from the Claritas Demographic Reports
Trends in Inpatient Utilization in Areas Served by TPR And Rural Control Hospitals, 2008-2015 Source: Author’s calculations obtained by applying the AHRQ PQI algorithms to HSCRC discharge abstract data and dividing by population estimates 11 from the Claritas Demographic Reports
Trends in Inpatient Utilization in Areas Served by TPR And Rural Control Hospitals, 2008-2015 Source: Author’s calculations obtained by applying the AHRQ PQI algorithms to HSCRC discharge abstract data and dividing by population estimates 12 from the Claritas Demographic Reports
Trends in Inpatient Utilization in Areas Served by TPR And Rural Control Hospitals, 2008-2015 Source: Author’s calculations obtained by applying the AHRQ PQI algorithms to HSCRC discharge abstract data and dividing by population estimates 13 from the Claritas Demographic Reports
Trends in Inpatient Utilization in Areas Served by TPR And Rural Control Hospitals, 2008-2015 Source: Author’s calculations based on the HSCRC discharge abstract data and and population 14 estimates from the Claritas Demographic Reports
Effects of TPR 2010 Program on Inpatient Utilization (2008-2013) Control = Rural Hospitals ONLY Control = All Maryland Hospitals Unadjusted Adjusted % Unadjusted Adjusted % 95% CI* 95% CI* % Δ Δ * % Δ Δ * Admission rate -3.5 -4.4 (-5.5 to -3.3) -2.0 -2.4 (-3.0 to -1.7) Inpatient days rate -6.0 -7.9 (-8.4 to -7.3) -2.0 -4.0 (-4.4 to -3.7) Readmissions rate 4.3 6.5 (3.7 to 9.3) -4.7 -4.0 (-5.7 to -2.3) Overall composite (PQI 90) -3.5 -4.8 (-8.1 to -1.4) -3.6 -6.4 (-8.3 to -4.4) Acute composite (PQI 91) 7.8 3.8 (-1.7 to 9.3) 4.6 -1.0 (-4.2 to 2.2) Chronic composite (PQI 92) -10.3 -10.1 (-14.4 to -5.9) -8.7 -9.8 (-12.3 to -7.3) “Non - preventable” adm. -3.3 -4.1 (-5.3 to -2.9) -1.7 -1.8 (-2.5 to -1.1) Non-deferrable adm. 2.6 3.6 (-2.3 to 9.5) 0.1 2.5 (-1.1 to 6.0) Admissions through ED -3.0 -4.2 (-5.7 to -2.7) 0.6 -0.1 (-0.9 to 0.8) *Obtained from Poisson regression models controlling for ZCTA fixed effects and year fixed effects, and adjusted for ZCTA demographics (percent female, median age, percent non-white, median household income, percent with at least a college degree, percent unemployment) and county characteristics (percent uninsured, percent enrolled in Medicare Advantage, physicians per capita, primary care physicians per capita, specialists per capita, and number of Federally Qualified Health Centers). 15
Trends in Outpatient Utilization in Areas Served by TPR And Rural Control Hospitals, 2008-2015 Source: Author’s based on the HSCRC discharge abstract data and population estimates from the 16 Claritas Demographic Reports
Trends in Outpatient Utilization in Areas Served by TPR And Rural Control Hospitals, 2008-2015 Source: Author’s based on the HSCRC discharge abstract data and population estimates from the 17 Claritas Demographic Reports
Trends in Outpatient Utilization in Areas Served by TPR And Rural Control Hospitals, 2008-2015 Source: Author’s calculations obtained by applying the Billings criteria to HSCRC discharge abstract data and dividing by population estimates from the 18 Claritas Demographic Reports
Trends in Outpatient Utilization in Areas Served by TPR And Rural Control Hospitals, 2008-2015 Source: Author’s calculations obtained by applying the Billings criteria to HSCRC discharge abstract data and dividing by population estimates from the 19 Claritas Demographic Reports
Effects of TPR 2010 Program on Outpatient Utilization (2008-2013) Control = Rural Hospitals ONLY Control = All Maryland Hospitals Unadjusted Adjusted Unadjusted Adjusted % 95% CI* 95% CI* % Δ % Δ * % Δ Δ * Outpatient visit rate -12.9 -13.7 (-14.2 to -13.3) -10.7 -11.8 (-12.1 to -11.6) ED visit rate 3.8 0.0 (-0.7 to 0.7) 5.0 3.8 (3.4 to 4.2) Emergent/PC treatable 5.1 0.7 (-0.9 to 2.3) 6.7 4.8 (3.9 to 5.7) Emergent/Preventable 6.4 3.8 (0.8 to 6.8) 9.6 8.4 (6.7 to 10.1) 20
Recommend
More recommend
