Feedback Intervention to Reduce Routine Electrocardiogram Use
Feedback Intervention to Reduce Routine Electrocardiogram Use
Background: Although physicians frequently order non-essential diagnostic tests, including screening tests such as electrocardiograms (ECGs), effective strategies for achieving evidence-based test-ordering are not proven. Our objective was to evaluate the impact of a feedback intervention designed to reduce the rate of screening ECG use and its variation.
Methods: A non-randomized pre-post intervention trial assessed the ordering of ECGs among primary care providers affiliated with Massachusetts General Hospital. Among outpatients visiting providers, those with cardiac diagnoses or symptoms were excluded, as were providers with <120 annual visits. Data were available on 117 providers, 105,682 patients, and 511,328 patient visits. During a 9-month intervention, providers received periodic case-mix-adjusted peer-comparisons of their screening ECG use. Two computerized billing systems tracked baseline (December 1996 to March 1998), intervention (April 1998 to December 1998), and post-intervention (January 1999 to September 1999) ECG use. Our outcome measures were: 1) the likelihood of ECG ordering at office visits and 2) provider practice variation, indicated by coefficient of variation and interquartile range.
Results: ECGs were ordered in 4.6% of visits before the intervention. Provider variation in case-mix adjusted ECG ordering was substantial (coefficient of variation, 101.6%; interquartile range, 3.2% [1.5%-4.7%]). ECG ordering averaged 3.5% during the intervention and 2.8% post-intervention (P < .001). Variation in ECG ordering narrowed considerably (coefficient of variation, 92.5%; interquartile range, 2.0% [1.0%-3.0%]) after the intervention. Multivariate analysis confirmed a persistent impact of the intervention.
Conclusions: Feedback to primary care providers successfully reduced ECG use and its variation. This approach may be applicable to other physician behaviors that remain contrary to evidence, but are not questioned because peer comparisons are unavailable.
Electrocardiograms (ECGs) are the oldest, most widely available, and most frequently used cardiac test. National estimates suggest that annually 20 million ECGs are ordered by U.S. office-based physicians at 2.6% of all U.S. office visits. On the basis of 1999 Medicare allowable charges, outpatient ECGs directly account for more than $600 million in annual health care expenditures.
Despite the usefulness of ECGs in evaluating sentinel cardiac symptoms and monitoring patients with cardiac disease, much doubt surrounds the value of ECGs as a routine screening tool. Recommendations against screening ECGs have been issued by the U. S. Preventive Services Task Force (USPSTF), the Canadian Task Force on the Periodic Examination, the American College of Cardiology/American Heart Association, and the American College of Physicians (for ages 18-65 years). The USPSTF, for example, states that insufficient evidence exists to recommend for or against screening in high-risk individuals and that recommendations against routine screening may be made on other grounds for persons who are not at high risk of developing symptomatic coronary artery disease. Cited in these negative recommendations are the low sensitivity of ECGs, the modest specificity of ECGs, and the inconvenience and risk of follow-up procedures, and the direct and indirect cost of ECGs and follow-up procedures. Altering the use of ECGs to conform to these guidelines may be difficult because physicians may perceive routine ECGs to be valuable clinically. Other barriers to change may include medicolegal concerns, testing induced by preoperative evaluation, patient request for services, and loss of income from office-based ECG testing.
Interventions aimed at influencing physician behavior toward established evidence-based guidelines may improve health care quality and, possibly, reduce health care expenditures. Among behavioral models that can inform efforts to alter physician behavior are the Predisposing, Reinforcing, and Enabling Constructs in Ecosystem Diagnosis and Evaluation (PRECEDE) model and the Knowledge, Attitudes, and Practices model. The PRECEDE model identifies predisposing, enabling, and reinforcing factors as the principal components that facilitate behavioral changes. The Knowledge, Attitude, and Practices model identifies a hierarchy of barriers that must be sequentially overcome for behavior change to occur. These 2 models suggest that successful interventions must go beyond educating physicians about guidelines and use alternative methods of motivation. Possible strategies include physician education, dissemination of guidelines, economic incentives, computerized and manual reminder systems, provision of feedback, and patient education. Previous studies indicate modest success in interventions aimed at changing physician behavior. However, those studies that used multiple strategies for confronting specific barriers, including feedback to physicians, were more likely to be successful. Even for effective interventions, positive changes in physician behavior may be short-lived after the withdrawal of the intervention.
After observing relatively frequent and quite variable screening ECG use, and on the basis of the previous studies of physician behavior change, a feedback intervention was developed to modify ECG ordering by physicians. Key hypotheses were that the intervention would diminish variation in ECG ordering rates and reduce overall ECG use.
Background: Although physicians frequently order non-essential diagnostic tests, including screening tests such as electrocardiograms (ECGs), effective strategies for achieving evidence-based test-ordering are not proven. Our objective was to evaluate the impact of a feedback intervention designed to reduce the rate of screening ECG use and its variation.
Methods: A non-randomized pre-post intervention trial assessed the ordering of ECGs among primary care providers affiliated with Massachusetts General Hospital. Among outpatients visiting providers, those with cardiac diagnoses or symptoms were excluded, as were providers with <120 annual visits. Data were available on 117 providers, 105,682 patients, and 511,328 patient visits. During a 9-month intervention, providers received periodic case-mix-adjusted peer-comparisons of their screening ECG use. Two computerized billing systems tracked baseline (December 1996 to March 1998), intervention (April 1998 to December 1998), and post-intervention (January 1999 to September 1999) ECG use. Our outcome measures were: 1) the likelihood of ECG ordering at office visits and 2) provider practice variation, indicated by coefficient of variation and interquartile range.
Results: ECGs were ordered in 4.6% of visits before the intervention. Provider variation in case-mix adjusted ECG ordering was substantial (coefficient of variation, 101.6%; interquartile range, 3.2% [1.5%-4.7%]). ECG ordering averaged 3.5% during the intervention and 2.8% post-intervention (P < .001). Variation in ECG ordering narrowed considerably (coefficient of variation, 92.5%; interquartile range, 2.0% [1.0%-3.0%]) after the intervention. Multivariate analysis confirmed a persistent impact of the intervention.
Conclusions: Feedback to primary care providers successfully reduced ECG use and its variation. This approach may be applicable to other physician behaviors that remain contrary to evidence, but are not questioned because peer comparisons are unavailable.
Electrocardiograms (ECGs) are the oldest, most widely available, and most frequently used cardiac test. National estimates suggest that annually 20 million ECGs are ordered by U.S. office-based physicians at 2.6% of all U.S. office visits. On the basis of 1999 Medicare allowable charges, outpatient ECGs directly account for more than $600 million in annual health care expenditures.
Despite the usefulness of ECGs in evaluating sentinel cardiac symptoms and monitoring patients with cardiac disease, much doubt surrounds the value of ECGs as a routine screening tool. Recommendations against screening ECGs have been issued by the U. S. Preventive Services Task Force (USPSTF), the Canadian Task Force on the Periodic Examination, the American College of Cardiology/American Heart Association, and the American College of Physicians (for ages 18-65 years). The USPSTF, for example, states that insufficient evidence exists to recommend for or against screening in high-risk individuals and that recommendations against routine screening may be made on other grounds for persons who are not at high risk of developing symptomatic coronary artery disease. Cited in these negative recommendations are the low sensitivity of ECGs, the modest specificity of ECGs, and the inconvenience and risk of follow-up procedures, and the direct and indirect cost of ECGs and follow-up procedures. Altering the use of ECGs to conform to these guidelines may be difficult because physicians may perceive routine ECGs to be valuable clinically. Other barriers to change may include medicolegal concerns, testing induced by preoperative evaluation, patient request for services, and loss of income from office-based ECG testing.
Interventions aimed at influencing physician behavior toward established evidence-based guidelines may improve health care quality and, possibly, reduce health care expenditures. Among behavioral models that can inform efforts to alter physician behavior are the Predisposing, Reinforcing, and Enabling Constructs in Ecosystem Diagnosis and Evaluation (PRECEDE) model and the Knowledge, Attitudes, and Practices model. The PRECEDE model identifies predisposing, enabling, and reinforcing factors as the principal components that facilitate behavioral changes. The Knowledge, Attitude, and Practices model identifies a hierarchy of barriers that must be sequentially overcome for behavior change to occur. These 2 models suggest that successful interventions must go beyond educating physicians about guidelines and use alternative methods of motivation. Possible strategies include physician education, dissemination of guidelines, economic incentives, computerized and manual reminder systems, provision of feedback, and patient education. Previous studies indicate modest success in interventions aimed at changing physician behavior. However, those studies that used multiple strategies for confronting specific barriers, including feedback to physicians, were more likely to be successful. Even for effective interventions, positive changes in physician behavior may be short-lived after the withdrawal of the intervention.
After observing relatively frequent and quite variable screening ECG use, and on the basis of the previous studies of physician behavior change, a feedback intervention was developed to modify ECG ordering by physicians. Key hypotheses were that the intervention would diminish variation in ECG ordering rates and reduce overall ECG use.
