Use of Preoperative Testing and Physicians’ Response to Professional Society Guidance

Importance  The value of routine preoperative testing before most surgical procedures is widely considered to be low. To improve the quality of preoperative care and reduce waste, 2 professional societies released guidance on use of routine preoperative testing in 2002, but researchers and policymakers remain concerned about the health and cost burden of low-value care in the preoperative setting.

Objective  To examine the long-term national effect of the 2002 professional guidance from the American College of Cardiology/American Heart Association and the American Society of Anesthesiologists on physicians’ use of routine preoperative testing.

Design, Setting, and Participants  Retrospective analysis of nationally representative data from the National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey to examine adults in the United States who were evaluated during preoperative visits from January 1, 1997, through December 31, 2010. A quasiexperimental, difference-in-difference (DID) approach evaluated whether the publication of professional guidance in 2002 was associated with changes in preoperative testing patterns, adjusting for temporal trends in routine testing, as captured by testing patterns in general medical examinations.

Main Outcomes and Measures  Physician orders for outpatient plain radiography, hematocrit, urinalysis, electrocardiogram, and cardiac stress testing.

Results  During the 14-year period, the average annual number of preoperative visits in the United States increased from 6.8 million in 1997-1999 to 9.8 million in 2002-2004 and 14.3 million in 2008-2010. After accounting for temporal trends in routine testing, we found no statistically significant overall changes in the use of plain radiography (11.3% in 1997-2002 to 9.9% in 2003-2010; DID, −1.0 per 100 visits; 95% CI, −4.1 to 2.2), hematocrit (9.4% in 1997-2002 to 4.1% in 2003-2010; DID, 1.2 per 100 visits; 95% CI, −2.2 to 4.7), urinalysis (12.2% in 1997-2002 to 8.9% in 2003-2010; DID, 2.7 per 100 visits; 95% CI, −1.7 to 7.1), or cardiac stress testing (1.0% in 1997-2002 to 2.0% in 2003-2010; DID, 0.7 per 100 visits; 95% CI, −0.1 to 1.5) after the publication of professional guidance. However, the rate of electrocardiogram testing fell (19.4% in 1997-2002 to 14.3% in 2003-2010; DID, −6.7 per 100 visits; 95% CI, −10.6 to −2.7) in the period after the publication of guidance.

Conclusions and Relevance  The release of the 2002 guidance on routine preoperative testing was associated with a reduced incidence of routine electrocardiogram testing but not of plain radiography, hematocrit, urinalysis, or cardiac stress testing. Because routine preoperative testing is generally considered to provide low incremental value, more concerted efforts to understand physician behavior and remove barriers to guideline adherence may improve health care quality and reduce costs.

The value of routine preoperative testing before most elective surgical procedures is widely considered to be low.^1-4 The national cost of this testing may be considerable, with 30 million Americans undergoing surgery annually and 60% of those patients undergoing ambulatory procedures.⁴ In recognition of these challenges and broader concerns about value, several major physician-education initiatives were undertaken to more appropriately guide medical decision making, improve the quality of health care that physicians provide, and reduce the incidence of unnecessary testing. Three in particular—the American Board of Internal Medicine’s “Medical Professionalism in the New Millennium: A Physician Charter,” which helped catalyze the Choosing Wisely campaign⁵; the American College of Cardiology/American Heart Association (ACC/AHA) update for Perioperative Cardiovascular Evaluation for Noncardiac Surgery^6,7; and the American Society of Anesthesiologists’ (ASA) practice advisory guidance for preanesthesia evaluation⁸—were concurrently disseminated in 2002.

While the American Board of Internal Medicine charter focused on guiding principles that are important in the practice of medicine, such as improving the quality and cost-effectiveness of health care, the ACC/AHA and ASA guidance documents made more specific recommendations about appropriate testing in the preoperative setting.⁵ However, despite these efforts, many researchers and policymakers remain concerned that a substantial disparity persists between practice guidelines and clinical care patterns.^2,9 Evidence supporting their concerns includes the rise in cardiac stress testing among patients who are enrolled in Medicare before elective surgery,² the wide use of laboratory blood testing in the preoperative setting,¹ and uncertainty among physicians about the potential adverse consequences—such as delayed or canceled surgery—of performing fewer tests.¹⁰

While single-site studies^2,9 have reported poor rates of adherence to guidelines for preoperative testing and larger studies^1,2 suggest that some preoperative tests are overused and of low value for specific surgical procedures, the long-term national effect of the 2002 initiatives on informing preoperative testing practices across diverse tests and surgery types is unknown. This fact is particularly important because routine preoperative testing is common and costly,^1,2 and the ACC/AHA and ASA guidance documents—which were both updated in recent years—discourage this practice.^11,12 In this study, we use nationally representative data from the National Ambulatory and Hospital Medical Care Surveys to examine preoperative testing patterns between January 1, 1997, and December 31, 2010, before and after the release of the 2002 guidance for preoperative testing.

We used a quasiexperimental, difference-in-difference (DID; time × group interaction) approach to examine the long-term effect of professional guidance in 2002 on preoperative testing patterns. The DID approach measures changes in an outcome associated with a policy change, after accounting for secular trends in that outcome, as reflected in a control group that is not exposed to the policy change.¹³ In this study, we chose general medical examinations, which primarily comprise routine annual examinations or checkups, as in our control group, because testing patterns in these visits likely capture temporal trends in the use of routine testing.^14,15 This study was exempt from institutional review board approval by New York University School of Medicine.

We analyzed data collected in the National Ambulatory Medical Care Survey (NAMCS) and National Hospital Ambulatory Medical Care Survey (NHAMCS) from January 1, 1997, through December 31, 2010.¹⁶ The National Center for Health Statistics and the Centers for Disease Control and Prevention conduct the NAMCS and NHAMCS annually on a nationally representative sample of visits to office-based physicians, hospital-based outpatient clinics, and emergency departments in the United States. For the NAMCS, each physician is randomly assigned to a 1-week reporting period during which a systematic random sample of visits are surveyed. Data are collected on patients' symptoms and demographic characteristics, physicians' diagnoses, medications ordered or provided, and medical services provided. For the NHAMCS, a systematic random sample of patient visits in selected noninstitutional general and short-stay hospitals are surveyed during a randomly assigned 4-week reporting period. The data collected on patient and physician characteristics are comparable with those collected in the NAMCS. From January 1, 1993, through December 31, 2010, the physician and hospital or outpatient clinic response rates in the NAMCS and NHAMCS ranged from 58.3% to 73.0% and 80.4% to 95.0%, respectively, and item nonresponse rates were generally 5.0% or lower in both surveys.

The NAMCS and NHAMCS record up to 3 reasons for each visit (1 “most important” and 2 “other” reasons) and 3 diagnoses related to the visit (1 “primary” and 2 “other” diagnoses); these are centrally coded using the National Center for Health Statistics Reason for Visit Classification and the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), respectively. Adult patients (≥18 years) who saw physicians for preoperative evaluations or general medical examinations comprised our study population. Preoperative visits were identified using Reason for Visit Classification code 4200.0 or ICD-9-CM diagnosis codes V72.81 through V72.84, and general medical examinations were identified using Reason for Visit Classification code 3100.0.

The 2002 ACC/AHA guidance, which aimed to “identify the most appropriate testing and treatment strategies…and avoid unnecessary testing”^6(p543) in patients undergoing noncardiac surgery, primarily focused on the use of cardiac stress testing and invasive coronary angiography, although other preoperative tests and management approaches were reviewed. In particular, this guidance suggested that routine electrocardiograms (ECGs) generally be used only in patients with recent chest pain or ischemic equivalents or those who undergo higher-risk surgery.⁶ The 2002 ASA advisory, which informs preanesthesia care in general and preoperative testing in particular, addressed a broad range of preoperative tests, emphasized the importance of assessing the comparative effectiveness of preoperative testing strategies, and specifically recommended against routine testing, defined as “a test ordered in the absence of a specific clinical indication or purpose.”^8(p488) Therefore, their advisory is relevant to a broad range of physicians who provide preoperative care. Additional details about the context of the 2002 ACC/AHA and ASA guidance documents, along with the 2007 ACC/AHA guidance update, are provided in eAppendix 1 and eTable 1 in the Supplement. Based on the tests reviewed by these professional societies and data availability in the NAMCS and NHAMCS, we considered orders for plain radiography, hematocrit, urinalysis, ECG, or cardiac stress testing as our primary outcome measures. Data on hematocrit studies were only available for 1997-2004; these studies, therefore, were included only in subanalyses. We identified these tests using ICD-9-CM procedure codes or with survey fields that directly captured test referrals.¹⁷

To further account for patient and clinical characteristics that may be associated with testing, we extracted information on patient age, sex, race or ethnicity, insurance (private, Medicare, Medicaid, self-pay or no charge, and other or unknown), US census region (Northeast, Midwest, South, and West), urban or rural setting, smoking status, and important comorbidities known to increase the risk of adverse events in the postoperative period (hypertension, coronary artery disease, diabetes mellitus, chronic kidney disease, and chronic obstructive pulmonary disease).⁶ Race and ethnicity were determined, per NAMCS and NHAMCS guidance, according to the office or clinic’s “usual practice, based on your knowledge of the patient, or from information in the medical record.”^17(p94) We categorized patients as non-Hispanic white, non-Hispanic black, Hispanic, other race, and unknown race or ethnicity when information on ethnicity was missing. Comorbidities were identified using visit diagnoses and reasons for the visit.¹⁸

All analyses accounted for the complex sampling design of the NAMCS and NHAMCS.¹⁹ We conducted descriptive data analysis and used simple logistic regressions, including ordinal and multinomial models, to compare the characteristics of patients in preoperative and general medical examinations. We also estimated simple and multivariate DID logistic regressions and estimated the predicted probability of testing to examine the effect of the 2002 professional society guidelines on the use of each type of test during preoperative visits. Our models were generally implemented as follows:

in which Testing, Preoperative visit, and Postguidelines were indicator variables and Postguidelines specifically captures the period 2003-2010. The coefficient on the interaction term between Preoperative visit and Postguidelines estimates the effect of interest. Specifically, this coefficient captures the difference in routine testing rates between preoperative visits and general medical examinations in the period before the release of the professional guidance in 2002 vs the difference after the release of this guidance; it therefore represents the independent relationship between the 2002 guidance and preoperative testing rates. We tested for a difference in pre-2002 testing trends between preoperative visits and general medical examinations by estimating simple logistic regressions that were limited to 1997-2002 and included an interaction variable between time and our indicator for preoperative visits. The coefficient for this variable was not significant for any of the routine tests we examined.

Multivariable logistic regression models also adjusted for patients’ clinical risk factors and demographic characteristics, insurance, geographic region, and urban or rural setting. To account for secular trends in testing, we also included a continuous variable for the year. In 2007, the ACC/AHA preoperative testing guidance was updated to further narrow recommendations for preoperative ECG and cardiac stress testing,¹² a development that could erroneously inflate any effects we attributed to the 2002 guidance alone. In light of this development, we conducted a sensitivity analysis, limiting the time frame to 2007 instead of 2010. We also conducted a sensitivity analysis in which we used ICD-9-CM procedure codes for cardiovascular and pulmonary procedures (see eAppendix 3, eTable 9, and eTable 10 in the Supplement for details) to identify patients who may have been referred for higher-risk procedures. Analyses were performed using Stata, version 12 (StataCorp LP).

During the 14-year period, the average annual number of preoperative visits in the United States increased from 6.8 million in 1997-1999 to 9.8 million in 2002-2004 and 14.3 million in 2008-2010. Compared with patients who had general medical examinations, patients who had preoperative visits were more likely to be female, less likely to be black vs white, more likely to have Medicare or no insurance vs private insurance, and more likely to live in the South or West vs the Northeast (Table 1). Patients who had preoperative visits were also less likely to have a diagnosis at the visit for hypertension, coronary artery disease, diabetes, or chronic obstructive pulmonary disease but more likely to have a diagnosis at the visit for chronic kidney disease.

We compared unadjusted rates of testing in preoperative and general medical examinations for the duration of our study (Table 2). Patients who had preoperative visits were more likely to receive plain radiographs (10.3% vs 5.2%; odds ratio [OR], 2.11; 95% CI, 1.76-2.54; P < .001), ECG (16.0% vs 10.5%; OR, 1.63; 95% CI, 1.38-1.92; P < .001), and cardiac stress testing (1.7% vs 0.8%; OR, 2.11; 95% CI, 1.44-3.10; P < .001) compared with patients who had general medical examinations and were also less likely to undergo urinalysis (10.0% vs 21.6%; OR, 0.40; 95% CI, 0.34-0.48; P < .001) and hematocrit testing (7.8% vs 12.5%; OR, 0.59; 95% CI, 0.45-0.77; P < .001). Overall, patients who had preoperative visits were less likely to receive any test compared with patients who had general medical examinations (26.5% vs 29.4%; OR, 0.87; 95% CI, 0.76-0.99; P = .04).

Trends in testing during preoperative visits and general medical examinations are shown in the Figure. Overall rates of plain radiography and urinalysis fell over time across both visit types, as did rates of ECG in preoperative visits. However, rates of cardiac stress testing trended upward in preoperative visits.

Using DID regression models, we found that, in the years after the release of the professional guidance for preoperative testing, there was no significant change in the use of plain radiography (11.3% in 1997-2002 to 9.9% in 2003-2010; DID, −1.0 per 100 visits; 95% CI, −4.1 to 2.2), hematocrit (9.4% in 1997-2002 to 4.1% in 2003-2010; DID, 1.2 per 100 visits; 95% CI, −2.2 to 4.7), urinalysis (12.2% in 1997-2002 to 8.9% in 2003-2010; DID, 2.7 per 100 visits; 95% CI, −1.7 to 7.1), or cardiac stress testing (1.0% in 1997-2002 to 2.0% in 2003-2010; DID, 0.7 per 100 visits; 95% CI, −0.1 to 1.5) (Table 3). Publication of the preoperative guidance was associated with a decrease in the use of ECG testing, however, which decreased by −6.7 tests per 100 visits (95% CI, −10.6 to −2.7; P = .001 in adjusted model). In a sensitivity analysis, we limited the follow-up period to 2007, the year that the ACC/AHA updated its preoperative testing guidance. This change did not significantly affect our results (adjusted change in ECG testing, −6.1 tests per 100 visits; P = .006; no other testing changes were statistically significant), which are presented in eAppendix 2 and eTables 2 through 8 in the Supplement. In another sensitivity analysis, we adjusted for an indicator variable for patients who may have been referred for higher-risk cardiovascular and pulmonary procedures. Our results, presented in eAppendix 3 and eTable 10 in the Supplement, were not significantly affected.

We found that overall rates of routine testing in the United States among patients who were evaluated in preoperative visits declined across several, but not all, of the testing categories we examined. However, after accounting for temporal trends in routine testing patterns, we found little evidence that the release of the professional guidance in 2002 was associated with long-term changes in national preoperative testing rates. In particular, with the exception of ECG, the rates of preoperative evaluation with plain radiography, hemoglobin, urinalysis, and cardiac stress testing were not significantly affected by the release of the ACC/AHA and ASA guidance in the years following 2002. Our findings are consistent with the conclusions of previous studies that report overuse of testing in the preoperative setting, including studies of preoperative cardiac stress test use among patients with Medicare,² medical consultations for cataract surgery,³ and testing before hernia repair,¹ among other studies.^9,20 Although we found little evidence of an effect of the guidance, it is noteworthy that rates of preoperative testing have generally fallen over time. This change likely reflects strong secular trends in testing and may have been partially driven by temporal reductions in reimbursement.

The publication of national preoperative testing guidance in 2002 may have had a limited long-term effect on physician behavior for several reasons. Physicians may not have been aware of the guidance or may have believed that it was not applicable to their patients.²¹ Prior research on physician behavior and practice guidelines also suggests that financial incentives that link physician compensation with testing may also increase their use.²² The role of incentives might be most relevant for cardiac stress testing, a setting in which physicians are compensated for conducting tests and interpreting study results. In a recent national study of cardiac stress testing, we found that rates of imaging stress tests—which are reimbursed by Medicare and private insurers at higher rates than nonimaging cardiac stress tests—increased substantially during the past 2 decades.¹⁸

More broadly, prior studies of physician behavior have reported limited effects of professional guidelines on physician behavior.^21,23-29 Analyses of physician decision making have proposed myriad reasons for poor adherence, including lack of awareness of guidelines, unfamiliarity among physicians with the content of guidance, uncertainty about whether a guideline will improve patient outcomes, uncertainty in the setting of conflicting guidelines, and inertia of existing practice patterns.²¹ These factors may explain the absence of a robust response to the 2002 guidance on preoperative testing. In addition, some evidence suggests that physicians are more likely to follow guidelines that add, rather than eliminate, a test or procedure.^23,27 These behavioral patterns among physicians have broad implications for clinical health care quality and costs because professional societies routinely aim to improve quality of care and reduce inappropriate variation in practice patterns.²¹

In 2012, the ASA updated the 2002 practice advisory with new evidence but concluded that no change in their recommendations was necessary.¹¹ More recently, the ASA, as part of the Choosing Wisely campaign, surveyed practicing anesthesiologists to develop a “top 5” list of low-value preoperative practices. Consistent with earlier ASA guidance, their final list included routine laboratory studies and cardiac stress testing, with the latter being a major focus of ACC/AHA guidelines. Because cardiac stress tests, in particular, are costly and frequently result in downstream tests and procedures, more concerted efforts to understand physician behavior and remove barriers to guideline adherence may improve health care quality and reduce costs.¹⁸

Our study has several limitations. The NAMCS and NHAMCS provide only a limited amount of clinical information for each patient visit, and we were therefore unable to adjust for surgery type or identify patients who were referred for high-risk surgery. Because high-risk surgical procedures may be associated with higher rates of preoperative testing, we examined published national rates of cardiac and pulmonary procedures in the National Hospital Discharge Surveys to determine whether the distribution of these procedures changed during our study period. Our analysis demonstrated little difference in the distribution of these procedures in 2003-2010 (14.5% to 15.7% for cardiac procedures, and 2.3% to 2.7% for pulmonary procedures).^30-32 Another important limitation is that we were unable to separate the differential effects of the ACC/AHA and ASA guidance. In particular, the ASA guidance may have been less frequently accessed by internists, cardiologists, and other physicians who routinely perform preoperative evaluations. Another limitation is the relatively low incidence of some tests, such as cardiac stress tests, which may have limited our ability to detect the overall effects of the 2002 professional preoperative guidance. However, our overall estimates of test use meet the statistical reliability requirements issued by the National Center for Health Statistics. Because our study is cross-sectional, we also do not have information on patient outcomes. Another limitation of our analysis is that general medical examinations may not be an adequate control group for preoperative evaluations, although these visits are frequently also characterized by the routine use of tests that may be low in value. Finally, temporal trends in testing, reflected by the use of routine testing in general medical examinations, may have been influenced by other factors that differentially affected physician behavior in general medical visits compared with preoperative testing visits.^33,34

Physicians play a critical role in improving the quality of health care that patients receive and in reducing the use of low-value care. In the context of preoperative care, our findings suggest that professional guidance aimed at improving quality and reducing waste had little effect on physician practice. We add to the body of evidence that underscores the urgency of using evidence-based and novel approaches to remove barriers to adherence and influence physician behavior in the provision of high-quality health care.

Accepted for Publication: February 7, 2015.

Corresponding Author: Alana E. Sigmund, MD, NYU Langone Medical Center, 550 First Ave, Tisch Hospital, Room 1803, New York, NY 10016 (alana.cohen@nyumc.org).

Published Online: June 8, 2015. doi:10.1001/jamainternmed.2015.2081.

Author Contributions: Dr Sigmund had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: All authors.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: All authors.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Stevens, Ladapo.

Administrative, technical, or material support: Sigmund.

Study supervision: Ladapo.

Conflict of Interest Disclosures: Dr Ladapo reports serving as a consultant to CardioDx, Inc. No other disclosures were reported.

Funding/Support: This study was supported by grant K23 HL116787 from the National Heart, Lung, and Blood Institute (Dr Ladapo).

Role of the Funder/Sponsor: The National Heart, Lung, and Blood Institute had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Correction: This article was corrected on June 11, 2015, to fix a typographical error in the Abstract.