Presentation on theme: "Exploring efficiency and quality of care among hospitals of the US Veteran Health Administration and Germany Jonas Schreyögg, PhD Commonwealth Harkness."— Presentation transcript:
Exploring efficiency and quality of care among hospitals of the US Veteran Health Administration and Germany Jonas Schreyögg, PhD Commonwealth Harkness Fellow, Center for Health Policy, Stanford University, USA Harkness Fellowships, Final Reporting Seminar, Orlando, June 5-6
Hospital Spending per Inpatient Acute Care Day in 2004 adjusted for PPP Background Source: OECD Health Data 2006 -> we want to know more!
Research approach Measuring efficiency and quality between hospitals with individual level data -> costs and mortality per patient To adequately control for case-mix we select three major episodes of care: Acute Myocardial Infarction (AMI), hip replacement, appendectomy Advantages of AMI over other care episodes: Requires immediate medical attention -> no patient selection High incidence and leading cause of death in the elderly Quality of care provided by hospitals can substantially avoid mortality Reflects level of technology utilization
Objective 1) to explain variation in costs as a measure of efficiency and 2) to explain variation in hospital mortality as a measure of clinical quality between hospitals of the Veteran Health Administration and Germany
Compression effect: Most German DRGs pay the same amount per day -> very similar in Medicare Why cant we just take reimbursement rates? Schreyögg J, Tiemann O, Busse R (2006) Cost accounting to determine prices: how well do prices reflect costs in the German DRG-system? Health Care Management Science 9 (3): 269-280
Data Data for year 2005 from two settings were chosen which have a very similar data structure and follow a similar cost accounting approach 1) Hospitals of the US Veteran Health Administration (VHA) –standardized accounting system –subject to extensive periodic audits –maintained and continuously improved since the mid 90ties –we used data from administrative files/EHR –data from 135 hospitals 2) German hospitals –obtained as part of the national cost data study (263 hospitals) maintained by InEK (Institute for the Calculation of Hospitals reimbursement) since 2004 –participation of hospitals requires adherence to cost accounting standards –to increase reliability of cost data we only included hospitals which participated 2004 and 2005 –InEK databases is not accessible for researchers -> each hospital was approached –60 hospitals met our sample definition criteria -> 19 delivered us their data
Methodology 1. Multilevel regression is performed -> What determines costs and mortality in both systems? 2. Imposing VHA prediction functions on German sample and imposing German prediction functions on the VHA sample -> How would VHA hospitals perform with German patients and vica versa? 3. Applying a propensity matching score approach to match VA and German patients -> How do hospitals from both systems perform with the same patient co-morbidities?
Modeling strategy Co-morbidities Hospital characteristics Treatment strategies (Use of Technology) Costs per case/ Mortality German sample Definiton of common set of variables Lower level (Unit: cases) Estimation of seperate cost/mortality multilevel functions for German and VHA hospitals Co-morbidities Hospital characteristics Treatment strategies (Use of Technology) Costs per case/ Mortality VHA sample Higher level (Unit: hospitals) Lower level (Unit: cases) Higher level (Unit: hospitals)
Sample characteristics (1)
Sample characteristics (2)
Median costs by cost category Factor 3.5 (40% due to wages/ 60% due to nursing ratio) Factor 4.6 (higher admin. per bed ratio 2.0-VA/ 0.12-Ger, higher admin. wages -> more documentation, bureaucracy etc. )
Regressions with dependent "ln_costs per case"
Regressions with dependent hospital mortality"
How do VHA hospitals perform with German patients and vice versa?
How do hospitals from both systems perform with the same patient co-morbidities?
How do AMI costs compare to Medicare and other European countries? w/o physician costs and drugs Comparison to Medicare hospitals (results from other studies): costs for Medicare tend to be higher/ increase with complexity of procedure Medicare hospitals perform more procedures Mortality in Medicare hospitals is lower Source: data for other European countries based on findings of EU HealthBasket Project; data for Medicare based on estimates for Medicare part A from MEDPAR Inpatient Hospital National Data for Fiscal Year 2005, Centers for Medicare and Medicaid Services. Hospital costs per AMI case adjusted for PPP
The analysis demonstrates the potential of health system comparison with micro-level data German hospitals are more efficient and hospital mortality is lower -> VHA: higher nursing ratio, higher wages, use of more complex technologies, higher overheads, higher mortality -> Germany: higher LOS and higher relative marginal costs of technology -> it is not either the prices or the utilization which make the US health care system the most expensive health care system in the world it is a combination of both Next steps: further decomposition of costs into input prices and utilization Inclusion of more US subsystems or countries into the analysis Conclusion
Implications for Germany: vertical integration can improve public health planning abilities central purchasing lowers prices for technology and drugs substantially -> but is the VHA purchasing model transferable? Implications for the VHA or US in general: Utilization of more advanced technology does not necessarily lead to lower mortality Bureaucracy has to be reduced in order to cut down overheads (documentation requirements due to liability issues etc.) Can staffing ratios be reduced in order to increase efficiency? General Implications: Financial incentives are obviously not the major driver for utilization Quality and availability of data in health care should be improved -> Each country should maintain an accessible representative micro data panel to improve cross-country learning possibilities Policy Implications
Generalizability of German results
Definition of variables Dependent variables: Costs per case in PPP-US$: costs incurred during hospitalization for treatment after AMI Hospital mortality: if a patient admitted with AMI dies during hospitalization Explanatory variables: Hospital characteristics: beds, urban status, teaching status, no. AMI cases, nursing ratio Co-morbidities: –Ontario AMI prediction rules and Charlson Co-morbidity Index were used –ICD-9 for VHA and ICD-10 for Germany – crosswalk tables used Cardiac cath. w/o. PCI Cardiac cath. w. PTCA Cardiac cath. w. bare-metal stent Cardiac cath. w. drug-eluting stent PTCA w/o. cardiac cath. Bare-metal stent w/o. cardiac cath. Drug-eluting stent w/o. cardiac cath. If cardiac catheter was performed during same stay If cardiac catheter was performed before the stay Treatment strategies: –treatment strategies identified for cost function