needleman

The Influence of Health Policy and Market Factors on Quality in the Hospital Safety Net : 

The Influence of Health Policy and Market Factors on Quality in the Hospital Safety Net Jack Needleman Department of Health Services UCLA School of Public Health AcademyHealth Annual Research Conference San Diego, CA June, 2004

Coauthors: 

Coauthors Gloria J. Bazzoli, Virginia Commonwealth University Richard C. Lindrooth, Medical College of South Carolina Ray Kang, Health Research and Education Trust Romana Hasnain-Wynia, Heath Research and Education Trust Funding Robert Wood Johnson Foundation Program on Health Care Financing and Organization

Session Outline: 

Session Outline Background and Research Questions Methods Overall approach Measures Sample Results Conclusions

Background: 

Background US depends on safety net hospitals to provide care to substantial numbers of individuals In many communities, care concentrated Safety net can be split between core and voluntary Safety net generally more heavily dependent on Medicare, Medicaid and special public funding BBA (1997) may have restricted funding disproportionately for safety net hospitals Lower involvement with private patients may also make safety net more vulnerable to private cutting

Background and Research Questions: 

Background and Research Questions One way to respond to financial pressures is to cut expenses in ways that adversely affect quality Staffing Investment in equipment and technology Questions How did quality in safety net hospitals and non-safety net compare at baseline? How did quality in highly vulnerable hospitals and less vulnerable hospitals compare at baseline? How has relative quality changed over time?

Methods: Overall Approach: 

Methods: Overall Approach Define core and voluntary safety net hospitals Define high vulnerability to BBA Construct quality measures Regress quality measures on safety net and vulnerability measures interacted with year for 1996-2000

Definition of Safety Net: 

Definition of Safety Net Use 1995 data on actual hospital uncompensated care provision reported to the American Hospital Association Level within hospital top decile Market share twice overall market share Core: both Voluntary: one of two

Definition of High Vulnerability: 

Definition of High Vulnerability Use Feder, Hadley, Zuckerman Fiscal Pressure Index Sensitive to both Medicare share and Medicare payment level Upper quartile defined as high

Measures: 

Measures In-hospital death Subset of AHRQ Patient Safety Indicators 14 measures Present data on Failure to Rescue Overall ratio of actual to expected Based on geometric mean of actual to expected for hospital Weighted by share of expected in overall sample Expected based on logit using demographics, payer, source of admission, risk in DRG (some)

Measures: 

Measures Failure to rescue Decubitus ulcer Pneumothorax Infection due to care Accidental puncture or laceration Birth Trauma OB Trauma in Vag Delivery with instruments OB Trauma in Vag Delivery without instruments OB Trauma in C-Section Post-op hemorrhage or hematoma Post-op respiratory failure Post-op Deep Vein Thrombosis/ Pulmonary Embolism Post-op Sepsis Post-op Wound Dehiscence

Methods: 

Methods Bivariate analysis of ratio of actual to expected Regression of actual to expected count of deaths and failure to rescue Negative binomial regression with expected as exposure variable, safety net, high and other variables as independent variables Incidence Risk Ratios Overall ratio of actual to expected OLS regression of ratio on safety net, high, and other variables OLS Coefficients (additive change)

Data Sources: 

Data Sources American Hospital Association Annual Survey Safety net FPI Hospital control variables: technology, beds, teaching (COTH), ownership, Medicare market share Interstudy: HMO market share Other Medicare case mix index, population of metro area, population living in poverty Discharge data State discharge data sets for six states California, Florida, New Jersey, New York, Washington, Wisconsin

Results: Bivariate Analysis: 

Results: Bivariate Analysis

Regression of Outcome on Safety Net StatusIRR/Coef Relative to Non-SN in each year: 

Regression of Outcome on Safety Net Status IRR/Coef Relative to Non-SN in each year

Regression of Outcome on High FPI StatusIRR/Coef Relative to Non-High in each year: 

Regression of Outcome on High FPI Status IRR/Coef Relative to Non-High in each year

Regression of Outcome on SN and High FPI Status (1 of 2): 

Regression of Outcome on SN and High FPI Status (1 of 2)

Regression of Outcome on SN and High FPI Status (2 of 2): 

Regression of Outcome on SN and High FPI Status (2 of 2)

Conclusions: 

Conclusions Some evidence that core safety net quality lower than non-safety net No evidence that high FPI hospitals had lower base quality No evidence that between 1996 and 2000 disparity in quality grew for high FPI, core SN, or core SN high-FPI hospitals

Limitations of Study: 

Limitations of Study Limited number of states Limited quality measures AHRQ PSIs tend to produce low counts Limited time

Next steps: 

Next steps Expand number of quality measures Test sensitivity to alternative definitions of safety net hospitals