Kale MS, Bishop TF, Federman AD, Keyhani S. Trends in the Overuse of Ambulatory Health Care Services in the United States. JAMA Intern Med. 2013;173(2):142-148. doi:10.1001/2013.jamainternmed.1022
Author Affiliations: Division of General Internal Medicine, Department of Medicine, Mt Sinai School of Medicine (Drs Kale and Federman), and Division of Outcomes and Effectiveness, Departments of Public Health and Medicine, Weill Cornell Medical College (Dr Bishop), New York, New York; and Division of General Internal Medicine, Department of Medicine, University of California (Dr Keyhani), and San Francisco Veterans Affairs Hospital (Dr Keyhani), San Francisco, California.
Background Given the rising costs of health care, policymakers are increasingly interested in identifying the inefficiencies in our health care system. The objective of this study was to determine whether the overuse and misuse of health care services in the ambulatory setting has decreased in the past decade.
Methods Cross-sectional analysis of the 1999 and 2009 National Ambulatory Medical Care Survey and the outpatient department component of the National Hospital Ambulatory Medical Care Survey, which are nationally representative annual surveys of visits to non–federally funded ambulatory care practices. We applied 22 quality indicators using a combination of current quality measures and guideline recommendations. The main outcome measures were the rates of underuse, overuse, and misuse and their 95% CIs.
Results We observed a statistically significant improvement in 6 of 9 underuse quality indicators. There was an improvement in the use of antithrombotic therapy for atrial fibrillation; the use of aspirin, β-blockers, and statins in coronary artery disease; the use of β-blockers in congestive heart failure; and the use of statins in diabetes mellitus. We observed an improvement in only 2 of 11 overuse quality indicators, 1 indicator became worse, and 8 did not change. There was a statistically significant decrease in the overuse of cervical cancer screening in visits for women older than 65 years and in the overuse of antibiotics in asthma exacerbations. However, there was an increase in the overuse of prostate cancer screening in men older than 74 years. Of the 2 misuse indicators, there was a decrease in the proportion of patients with a urinary tract infection who were prescribed an inappropriate antibiotic.
Conclusions We found significant improvement in the delivery of underused care but more limited changes in the reduction of inappropriate care. With the high cost of health care, these results are concerning.
Given the rising costs of health care, policymakers are increasingly interested in identifying the inefficiencies in our health care system.1 In an analysis of the estimated $700 billion that is wasted annually in our health care system, overuse, or the delivery of services for which the risks exceed the benefits, has been identified as a significant component, equaling roughly $280 billion.2 Interest in overuse has started to gain traction, notably by physician leaders. Several national physician groups have tackled the overuse of screening and diagnostic testing, identifying many common scenarios in which services have low value and high cost.3,4 Research has confirmed that overuse is widespread and occurs across multiple specialties.5,6
Assessments of the current state of our health care system typically examine 1 of 3 interrelated dimensions of quality: structure (the characteristics of the resources of the health care system), process (interactions between clinicians and patients), and outcomes (changes in patients' health status).7 Evaluations of process measures dominate quality improvement because they are activities that clinicians control most directly. Process measures can be further categorized into overuse, underuse, and misuse. Overuse represents the delivery of health care for which the risks outweigh the benefits (eg, use of an antibiotic to treat viral respiratory syndromes); underuse represents the failure to deliver health care for which the benefits outweigh the risks (eg, use of an aspirin in patients with coronary disease); and misuse is the delivery of the wrong care (eg, use of an antibiotic other than nitrofurantoin, trimethoprim-sulfamethoxazole, or quinolone for the treatment of an uncomplicated urinary tract infection).
Recent studies8 have demonstrated an improvement in the underuse of needed medical services; however, it is unclear whether the rates of misuse and overuse have also decreased over time. Understanding the relationship of changes in underuse to overuse and misuse helps to characterize the state of our evolving health care system, particularly with respect to the quality of care delivered and the growing costs associated with care. In this study, we apply the quality framework of underuse, overuse, and misuse to a nationally representative sample of patients cared for in ambulatory care settings to determine whether the overuse and misuse of health care services have decreased in the past decade.
We performed a cross-sectional analysis using data from the 1998, 1999, 2008, and 2009 National Ambulatory Medical Care Survey (NAMCS) and the outpatient department component of the National Hospital Ambulatory Medical Care Survey (NHAMCS). The NAMCS and NHAMCS are nationally representative surveys conducted annually by the Centers for Disease Control and Prevention's National Center for Health Statistics. The NAMCS surveys patient visits to physicians in non–federally funded, non–hospital-based offices; the NHAMCS surveys patient visits to physicians in non–federally funded hospital outpatient departments. The visits sampled take place during a 1-week period that is randomly assigned for each practice (a 4-week sample period is used in NHAMCS). We pooled 1998 and 1999 data and 2008 and 2009 data to increase the sample sizes.
Both surveys use a multistage stratified probability sampling design that allows for the generation of national estimates on the patient-visit level. Information collected in both surveys includes the visit characteristics, diagnoses, medications, and services ordered.
We developed our quality indicators using a combination of current performance measures and guideline recommendations (Table 1). Each indicator was chosen because it pertained to outpatient quality of care and could be reliably calculated using information in the NAMCS and NHAMCS in the study years. We identified 22 measures, which we organized into 1 of 3 categories: underuse, overuse, or misuse (Table 1) of health care services.9- 31 Although in some cases we applied guideline recommendations that were published after 1999 or after 2009, this approach allowed for comparative assessments of the quality of care over time and is consistent with previous examinations of quality using NAMCS and NHAMCS.8
For each indicator, we identified the eligible population (denominator) using a combination of variables: patient's reason for the visit, the diagnosis (classified using the International Classification of Diseases, Ninth Revision, Clinical Modification diagnostic codes), and the diagnosis check-boxes (owing to changes in survey design, this variable was not used to identify eligible visits in the 1998 and 1999 NAMCS or NHAMCS). We excluded patient visits on the basis of clinical contraindications. For example, we were interested in examining the extent to which patients with atrial fibrillation are prescribed anticoagulation, on the basis of a quality measure developed by the American College of Cardiology, American Heart Association, and American Medical Association Physician Consortium for Performance Improvement.9 We constructed the denominator for this quality indicator by identifying all visits by patients with documented atrial fibrillation who did not have a contraindication to anticoagulation, such as a diagnosis of gastrointestinal bleeding. We then measured the proportion of visits in which the patients were prescribed anticoagulation. Medications were identified using a combination of the medication codes developed by the National Center for Health Statistics and the Multum Lexicon Plus database.32
For each measure, we calculated the weighted proportion of eligible visits in which the patient received recommended care or, in the case of our overuse and misuse measures, the weighted proportion of eligible visits in which the patient received nonrecommended care. We then used the χ2 test to compare differences in these weighted proportions between 1998-1999 and 2008-2009.
We took into account the sampling weights and sample design variables available in NAMCS and NHAMCS to generate these weighted, nationally representative estimates. The reliability of the estimates is in accordance with the standards specified by the National Center for Health Statistics, and quality indicators were not included if they had less than 30 unweighted cases in each cell.33 We generated 95% CIs using Stata statistical software, version 11.0 (StataCorp).
In our study sample, there were 79 083 and 102 980 unweighted visits by adult patients at least 18 years of age in 1998-1999 and 2008-2009, respectively (Table 2). Compared with visits made in 1998-1999, visits in 2008-2009 were by slightly older patients (mean age, 54.2 years in 2008-2009 vs 50.9 years in 1998-1999; P < .001) and by more patients insured through Medicare (26.2% vs 22.7%; P = .03). Otherwise, the 1998-1999 and 2008-2009 study samples were similar with respect to patient sex, race, ethnicity, reason for visit, and practice region.
In our analysis of underuse measures, we observed a statistically significant improvement in 6 of 9 quality indicators (Table 3). In the 10-year interval under consideration, there was an improvement in the use of antithrombotic therapy for atrial fibrillation (45.9% to 71.9%; P < .01). There was also an improvement in the use of aspirin (28.4% to 64.5%; P < .01), β-blockers (28.1% to 55.2%; P < .01), and statins (26.8% to 58.6%; P < .01) in coronary artery disease. There were also improvements in the use of β-blockers in congestive heart failure (20.6% to 59.7%; P < .01) and the use of statins in diabetes mellitus (12.1% to 36.2%; P < .01). We did not find statistically significant differences in the remaining underuse quality indicators: the use of angiotensin-converting enzyme inhibitors in congestive heart failure, the use of antiplatelets in stroke, and the pharmacologic treatment of osteoporosis.
We observed an improvement in only 2 of 11 overuse quality indicators, 1 indicator became worse, and 8 did not change. There was a statistically significant decrease in the overuse of cervical cancer screening in visits for women older than 65 years (3.1% to 2.2%; P = .02) and in the overuse of antibiotics for asthma exacerbations (22.3% to 6.8%; P < .01). Rates of urinalysis testing at general medical examinations also decreased, although the difference was of borderline significance (39.9% vs 25.3%; P = .05). However, there was an increase in the overuse of prostate cancer screening in men older than 74 years (3.5% to 5.7%; P = .03). There were no changes in the remaining 7 overuse measures: complete blood count and electrocardiogram testing in general medical examinations, use of antibiotics for upper respiratory tract infections and acute bronchitis, mammography for women 75 years or older, imaging in acute back pain, and chest x-ray in general medical examinations.
Of the 2 misuse indicators, there was 1 significant improvement. The proportion of patients with a urinary tract infection who were prescribed an inappropriate antibiotic decreased from 24.9% to 2.7% (P < .01). There was no change in the proportion of elderly patients who were prescribed inappropriate medications. Adjusting for insurance status to account for potential differences in access to care did not change our results.
In our examination of ambulatory health care services over 10 years, we found an improvement in 6 of 9 measures of underuse but only 3 of 13 measures of inappropriate care (both overuse and misuse). Our findings of the continued delivery of inappropriate care, such as the use of prostate-specific antigen testing in older men and cervical cancer screening in older women, are consistent with other studies34,35 that demonstrate the persistence of inappropriate care. Our results also suggest that there has been little change in the delivery of inappropriate ambulatory care in the past decade.
Given the questionable sustainability of the current trajectory of health care costs, our findings uniquely inform the discussion of strategies to improve the quality of health care, particularly as solutions are analyzed with an eye on their affordability and financial impact. We found considerable room for improvement in most of our overuse measures, a space in which the dual goals of high quality and reduced costs can be met, and demonstrated that attention to underuse and overuse has been uneven. The United States has a higher total expenditure on health relative to its gross domestic product compared with all other countries.36 Although there is continuing debate about what constitutes a reasonable cost of health care, there is growing momentum in delivering higher quality care that costs less. Reducing inappropriate care where patients clearly do not benefit and for which there may be added risk is certainly part of this stated goal.37
There are several possible explanations for our findings; however, among the most likely is that targeting and reducing inappropriate care has not been a real focus of the quality of care movement. In the past 2 decades, there has been substantial growth in methods to measure quality in health care. These quality measures have developed alongside the growing understanding that medicine can and should be delivered on the basis of evidence. Using a combination of information from clinical trials and observational studies, panels of expert physicians have created clinical practice guidelines, a repository of which is maintained by the Agency for Healthcare Research and Quality.38 The creation of clinical practice guidelines has informed the development of metrics to assess the quality of our health care system.39 Although quality assessments are dominated by process-based measures, these have mostly taken the form of underuse measures. And despite the acknowledgment that overuse contributes to waste and inefficiency in our health care system, it is not routinely measured in quality assessments. In light of the abundance of literature and practice guidelines related to underuse, our finding that the overuse of ambulatory care may have changed little during the past 10 years is not entirely unexpected. Reducing inappropriate care will require the same attention to guideline development and performance measurement that was directed at reducing the underuse of needed therapies.
Developing guidelines and performance measures to reduce inappropriate care may be easier said than done. Many methodologic, political, and cultural challenges have impeded progress in these areas. There are 2 main methodologic challenges to creating quality measures that address the delivery of inappropriate care.40 First, overuse, unlike underuse, is not easily studied within publicly reported databases or within hospital claims data. For example, if a patient has an acute myocardial infarction, all that may be needed to determine whether a patient appropriately received an aspirin is the discharge diagnosis, inpatient medication list, and discharge medications. The second challenge is the difficulty in creating guidelines and measures around overuse of many types of health care services. Determining if a patient inappropriately received a procedure requires a much more detailed set of clinical criteria than what is required for assessments of underuse. Although there are methods for assessing the appropriate use of services, such as the RAND Appropriateness Method, they are typically time-consuming and expensive processes.41 For example, creating appropriateness criteria using the RAND Appropriateness Method for the appropriate insertion of tympanostomy tubes requires not only a systematic review of the literature but also assembling an expert panel composed of physicians from multiple specialties, such as pediatricians and pediatric otolaryngologists grading an exhaustive and mutually exclusive list of clinical factors, such as the presence of hearing or language delay.42 Some specialty organizations (eg, American College of Cardiology/American Heart Association43) have developed appropriateness criteria around a number of procedures and diagnostic tests. However, the method has not been widely implemented to develop a robust set of guidelines across a large spectrum of services.
There has also been no formal effort to develop and promote the use of standardized overuse measures even though there are some simple measures of overuse that could be easily implemented and studied. For example, there is good evidence that screening for prostate cancer in the very elderly and infirm is not beneficial44 and yet it continues to be performed at alarming rates.35 Despite being easily measured, this practice has not been evaluated as a potential performance measure or adopted by the Healthcare Effectiveness Data and Information Set.
There are political and cultural challenges to addressing overuse as well, namely, resistance to limiting access to health care services. For example, the Agency for Healthcare Research and Quality (formerly the Agency for Health Care Policy and Research), whose charge includes the creation of practice guidelines, was nearly dissolved when it recommended against the use of surgery in the initial management of low-back pain.45 The government has not since taken up the mantle of addressing inappropriate care through the creation of practice guidelines. The unwillingness of our society to address overuse to achieve both high quality and affordability reflects the pervasive fear of rationing46 and the interests of industry stakeholders.
The reactions by physicians to limiting inappropriate care have been mixed. For example, when the United States Preventive Services Task Force recently updated and published its draft recommendations discouraging the use of prostate-specific antigen screening in asymptomatic men, it received strong words of rebuke from the American Urologic Association.47 However, not all physicians are opposed to limiting care with unclear benefits. Recently, the National Physicians Alliance through its Good Stewardship project launched a campaign to limit inappropriate ambulatory care and proposed a set of 5 overused practices in the fields of internal medicine, family medicine, and pediatrics.3,5 The American Board of Internal Medicine Foundation followed suit and launched the “ Choosing Wisely” campaign, in which they coordinated with 9 physician specialty organizations, including cardiologists, radiologists, and oncologists, to identify tests or procedures that are commonly used and not always appropriate.48 The campaign is notable for the collaboration with many procedure-oriented physician groups who are valuing high-quality care over financial gains. These initiatives may be foreshadowing a change in practice culture that may be necessary to begin the hard work of addressing the delivery of inappropriate care in the US health care system.
There are limitations to the conclusions of our study. First, the number of underuse, overuse, and misuse measures available in the NAMCS and NHAMCS data sets is limited; thus, our study presents just part of the picture of appropriateness of care in ambulatory care settings, and we cannot conclude with statistical confidence that misuse occurred with greater frequency than underuse. Second, some of the observed differences may not be statistically significant because of insufficient statistical power. Third, we were limited by the availability of data in the NAMCS database. For example, the NAMCS documents only 6 to 8 medications per visit and it is possible that appropriate or inappropriate medications were not documented for some visits, leading to overestimation or underestimation. Fourth, we may have underestimated the receipt of some services because NAMCS documents the services provided at only 1 visit per year. However, the trends remain informative because biases arising from a once-a-year assessment apply equally to all years of study. Last, we were unable to examine explanatory mechanisms for underuse or overuse, such as physician rationale and decision making. Understanding the root cause of overuse will require looking beyond most public use data sets or claims-based data.
In our examination of ambulatory care in the United States, we found an improvement in most of the underuse measures but limited changes in the delivery of inappropriate care. Reducing health care costs and improving the quality of care in the United States can be achieved by reducing overuse and misuse of health care services, but it will require making uncomfortable decisions that patients, physicians, and policymakers have been historically unwilling to make. Developing clinical practice guidelines that define when care should not be delivered and performance measures to address inappropriate care are critical steps to advance the mission of increasing the value and efficiency of health care delivery.
Correspondence: Minal S. Kale, MD, Division of General Internal Medicine, Department of Medicine, Mt Sinai School of Medicine, One Gustave L. Levy Place, PO Box 1087, New York, NY 10029 (firstname.lastname@example.org).
Accepted for Publication: August 19, 2012.
Published Online: December 24, 2012. doi:10.1001/2013.jamainternmed.1022
Author Contributions:Study concept and design: Kale and Keyhani. Acquisition of data: Kale. Analysis and interpretation of data: Kale, Bishop, and Federman. Drafting of the manuscript: Kale. Critical revision of the manuscript for important intellectual content: Bishop, Federman, and Keyhani. Statistical analysis: Kale and Federman. Administrative, technical, and material support: Bishop. Study supervision: Keyhani.
Conflict of Interest Disclosures: Dr Kale reports that she is supported by the Mt Sinai Primary Care Research Fellowship, funded by the Health Resources and Services Administration through the Ruth K. Kirchstein National Research Service Award. Dr Bishop reports that she is partially funded as a Weill Cornell Medical College Laitman Fellow. Dr Keyhani reports that she is funded by a Veterans Administration Health Services Research and Development Service Career Development Award.