The Relationship between the Use of Electronic Health Records and Quality of Care in U.S. Hospitals Catherine M. DesRoches DrPH Academy Health June 28, 2009 Introduction • Funding: Office of the National Coordinator for Health Information Technology, the Robert Wood Johnson Foundation • Study Team: Institute for Health Policy, MGH; George Washington University; Harvard School of Public Health; American Hospital Association Research Question • Is the adoption of an EHR system or the key clinical functionalities that comprise an EHR associated with higher quality care? – Nearly 400 studies examining the effect of specific electronic functionalities on quality. – Most of these studies come from a small set of pioneering hospitals. – Little evidence beyond these institutions. Methods Data Sources • 2008 HIT Survey of Acute Care Hospitals in the US • 2005 AHA Annual Survey • 2008 Hospital Quality Alliance Database • 2006 Medicare Provider Analysis and Review Database 2008 HIT Survey of Acute Care Hospitals in the United States • Survey of all 4,840 acute-care general medical/surgical hospitals in the US. • Field period: March – September 2008 • Response rate: 63.1% (3,054 hospitals) • EHR definition: created with an expert consensus panel – Comprehensive EHR: 24 clinical functionalities implemented across all major clinical units – Basic EHR: 10 clinical functionalities implemented in at least one major clinical unit • Analytic sample: 2,952 non-federal acute care general hospitals. Other Data Sources 2005 American Hospital Association Annual Survey 2007 Hospital Quality Alliance database • Performance scores for 4,470 acute care hospitals based on patients seen during the 2007 calendar year. – Created summary scores for AMI, CHF, Pneumonia, Surgical Infection Protection. • 30-day Risk Standardized Mortality for AMI, CHF, Pneumonia, Surgical Infection Protection. 2006 Medicare Provider Analysis and Review database (MEDPAR) • 30-day risk adjusted readmission rates for AMI, CHF, Pneumonia, Surgical Infection Protection. EHR Adoption Among Hospitals in the US Percent of US hospitals with a comprehensive or basic EHR 50% Comprehensive System: 1.5% 40% Basic System: 7.8% 30% 18.5% 20% 21.1% 9.8% 10% 8.0% 6.1% 10.5% 7.4% 0% Small Medium Hospital Size Large Teaching Nonteaching Teaching Status Yes No Member of a System HQA Summary Scores and EHR Availability Comprehensive or basic 100% 95.9% 95.0%* 86.9% 85.9% 91.2% 90.5% No EHR 87.5% 85.4%* 90.9% 89.4%* 75% 50% 25% 0% AMI CHF Pneumonia Surgical Infection prevention SUMMARY SCORES *Difference is significant at p < 0.01 Summary scores adjusted by hospital size, region, teaching, urban, CICU, and percent Medicaid. Overall HQA Summary Scores and Clinical Decision Support - Guidelines Comprehensive or basic 100% 95.7% 94.8% * 87.5% 85.3%* 91.1% 90.4%* No EHR 86.2% 85.4% 90.1% 89.9%* 75% 50% 25% 0% AMI CHF Pneumonia Surgical Infection prevention Overall SUMMARY SCORES *Difference is significant at p < 0.01 Summary scores adjusted by hospital size, region, teaching, urban, CICU, and percent Medicaid. HQA Summary Scores and Clinical Decision Support – Clinical Reminders Comprehensive or basic 100% 95.5% 94.8% * 86.9% 85.4% * 91.0% 90.4% * No EHR 85.9% 85.5% 89.9% 89.3% ** 75% 50% 25% 0% AMI *Difference is significant at p < 0.01. **Difference is significant at p < 0.05. CHF Pneumonia Surgical Infection prevention Overall SUMMARY SCORES Summary scores adjusted by hospital size, region, teaching, urban, CICU, and percent Medicaid. Risk Adjusted 30-day Mortality and EHR Availability Comprehensive or basic No EHR 25 20 16.1 16.1 15 11.1 11.1 11.4 11.4 10 5 0 AMI CHF Pneumonia Mortality rates are risk adjusted and adjusted for hospital size, region, teaching, urban, CICU, and percent Medicaid patients. Risk Adjusted 30-day Mortality and Computerized Order Entry for Medications Comprehensive or basic No EHR 25 20 16.1 16.0 15 11.1 11.1 11.3 11.5 * 10 5 0 AMI CHF Pneumonia *Difference is significant at p =0.02. Mortality rates are risk adjusted and adjusted for hospital size, region, teaching, urban, CICU, and percent Medicaid patients. 30-Day Readmissions Rates and EHR Availability Comprehensive or basic No EHR 40 30 25.8 24.9 25.1 24.4 19.0 20 20.2 10 0 AMI CHF Pneumonia Readmission rates are risk adjusted and adjusted for hospital size, region, teaching, urban, CICU, and percent Medicaid patients. 30-Day Readmissions Rates and Electronic Discharge Summaries Comprehensive or basic No EHR 40 29.6 * 30 25.0 26.2 ** 24.4 19.6 20 20.9 ** 10 0 AMI CHF Pneumonia *Difference is significant at p =.0.02. **Difference is significant at p < 0.01. Readmission rates are risk adjusted and adjusted for hospital size, region, teaching, urban, CICU, and percent Medicaid patients. Limitations • Response rate: Non-responders were different from responders. • We could not measure use of HIT, only availability. • Limited detail on where the systems were implemented. • Cross-sectional data • Limited set of quality metrics Conclusions • Modest but generally consistent relationships between EHR adoption and performance on standard quality process measures and readmission rates. • This study moves us beyond the “efficacy” argument for EHRs by assessing effectiveness. • Study suggests that even beyond pioneering institutions, functionalities such as clinical decision support seem to be associated with better performance on process measures. • Overall, findings are sobering: HIT adoption is likely to be an important part of improving the efficiency and efficacy of our health care system BUT just implementing systems will not have a dramatic effect on care.