Does Hospital Price Competition Influence Nurse Staffing and Quality of Care? Julie Sochalski, PhD1 R. Tamara Konetzka, PhD2 Jingsan Zhu, MBA1 Joanne Spetz, PhD3 Kevin Volpp, MD, PhD1,4 Introduction • Over past 20 years hospitals shift from competing on quality/amenities competing on price. • Evidence that price competition rate of increase in hospital costs, profits efficiencies or lower quality? • Examine impact of price competition on one feature associated with hospital quality – nurse staffing. Academy Health June, 2005 3 1 University of Pennsylvania University of California at San Francisco 2 University of Chicago Philadelphia VA Medical Center 4 Nurse Staffing – Patient Outcomes Relationship • CrossCross-sectional studies over 3 decades show higher nurse staffing associated with reduced mortality. • Recent longitudinal study found increases in RN staffing linked to lower mortality, with diminishing returns. • Most studies rely on hospitalhospital-wide measure of nurse staffing which may obscure relationship. In summary: • Substantial gaps in understanding of nurse staffing— staffing— quality relationship. • Rely on crude staffing measures to explore relationship. • Lack information on current hospital responses to price competition. Hospital Responses to Price Competition • Hospital personnel increased from 1980s to early 1990s. • RNs increased commensurate with volume and CMI while other nursing personnel declined. • Spetz (1999) found HMO penetration was not associated with RN staffing through early 1990s. Research Questions • Are changes in nurse staffing levels associated with patient outcomes? • What hospital and market features are associated with staffing changes and thereby outcomes? • 1999 – California passes AB 394 to establish minimum nurse staffing ratios. 1 Data Study Design • California’s Office of Statewide Health Planning and Development (OSHPD) discharge data from 1991. • California acute care hospitals, 19911991-2001 • Three AHRQ inpatient quality indicators: • OSHPD annual disclosure (financial) data 1991-2001 • State death certificates 1991-2001. – 3030-day mortality for AMI, stroke, and hip fracture • Sample: – Hospitals: n = 421 short-term acute hospitals (non-federal, non-Kaiser) – Patients: • AMI: n = 352,536 (15.5%) • Stroke: Stroke n = 592,651 (14.1%) • Hip fracture: n = 276,628 (5.3%) Control Variables Key Study Variables • Nurse staffing – RN, LVN, Nurse Aide – Nursing productive hours per patient day – Acute medicalmedical-surgical units • Market factors – HMO penetration for hospital market area (fixed radius) – High vs. low competition market areas • Age • Gender • Race • Ethnicity • Expected source of payment: Medicare, Medicaid, uninsured, private • Elixhauser comorbidities • Hospital casecase-mix index • Year dummies 19911991-2001 (1991 is reference) • Hospital fixed effects controls for timetimeinvariant hospital and market factors Model Generalized linear model with hospitalhospital-level fixed effects + time fixed effects Model 1 Pr( Death) pht = β 0 h + β v Staffing ht + β wYeart + β1HospitalCMI ht + β x PatientSeverity pht + β y PatientDemographics pht + β z PaymentSource pht + ε pht Model 2 RNhppd ht = β 0 h + β v Staffing ht + β wYeart + β1 HospitalCMI ht + β 2 MCPht + β 3 MCP * COMPht + β 4WageIndexht + β x PatientSeverityht + β y PatientDemographicsht + β z PaymentSourceht + ε ht Hospital Summary Statistics No. of hospitals: Avg. # beds: Urban: Teaching: Ownership: Non-profit: Government: For-profit: Avg. CMI 421 192 88% 18.7% 52.7% 20.7% 26.6% 1.114 2 Change in CMCM-adjusted RN medicalmedical-surgical hours per patient day, 19911991-2001 Effects of nurse staffing on 3030-day mortality 20.0% 15.0% 10.0% 5.0% 0.0% 1992 1993 1994 1995 AMI Stroke Hip Fracture RN -0.004* (0.001) -0.002* (0.001) 0.002 (0.001) LVN -0.003 (0.002) 0.0004 (0.001) 0.0008 (0.001) Aide 0.001 (0.001) -0.0002 (0.0007) -0.0001 (0.0007) RN*baseline 0.0004† (0.0002) -0.0001 (0.0001) -0.0004 (0.0003) Model: 75th 1996 1997 1998 1999 2000 2001 -5.0% 25th -10.0% * p < .05 † p < .1 -15.0% Effects of price competition on nurse staffing Caveats/Limitations HMO Penetration 2.479** (0.696) • Changes over time in DRGs, DRGs, coding, zip codes (but smoothed/corrected to the extent possible) HMO Penetration*HHI -3.192* (1.245) • Limited to California – generalizable to other states? * p < .01 • Limited to mortality– mortality– generalizable to other quality measures? ** p < .001 • Are there thresholds to staffingstaffing-quality relationship? Conclusions • Extent to which changes in RN staffing levels are associated with lower mortality varies by condition. • Increasing managed care penetration is associated with higher RN staffing except in most competitive markets. • Limiting the number of patients per nurse may improve quality outcomes. 3