Evaluation of Appropriate Use of Preoperative Echocardiography before Major Abdominal Surgery: A Retrospective Cohort Study

Data were obtained from the Truven Health MarketScan Commercial and Medicare Supplemental Databases from 2004 through 2017. These databases contain de-identified records of health services for more than 250 million patients with primary or Medicare supplemental health insurance coverage through employer-based fee-for-service, point-of-service, or capitated plans. The University of Chicago Institutional Review Board (Chicago, Illinois) considered this study exempt from review as there are no patient identifiers in the MarketScan databases (IRB19-1891).

Our analysis included records of patients with an International Classification of Diseases, Ninth Revision–Clinical Modification or International Classification of Diseases, Tenth Revision–Clinical Modification principal procedure code for a major abdominal surgical procedure (fig. 1). We focused on major abdominal procedures as they are a high-risk procedure for major adverse cardiac events as per the Revised Cardiac Risk Index.²  We did not include intrathoracic and suprainguinal vascular procedures as there may be procedure-specific indications that warranted a resting echocardiogram. Procedures included open or laparoscopic small bowel resections, large bowel resections, gastric resections, pancreatic resections, cystectomies, and nephrectomies from 2005 to 2017 (Supplemental Digital Content table 1, https://links.lww.com/ALN/C688). Subjects were excluded from the analysis if (1) they were under 18 yr old (n = 7,327), (2) the surgery was emergent (identified by correlation with an emergency room service claim from the same date as the surgery; n = 83,227), (3) the surgery was not the first procedure of interest that the patient underwent during the study period (n = 9,476), (4) the subject was not enrolled in an insurance plan for 1 yr before the surgery (n = 132,939), and (5) there was no drug benefit data available in the MarketScan database for the patient (n = 85).

The primary outcome measure was the presence of a “rarely appropriate” resting echocardiogram performed within 60 days before a major abdominal surgery. A 60-day period was chosen as echocardiograms performed within this time period are likely preoperative tests consistent with previous analyses.^9–11  Outpatient resting transthoracic echocardiograms were identified using Current Procedural Terminology codes (Supplemental Digital Content table 2, https://links.lww.com/ALN/C688). The indications for the preoperative echocardiograms were determined based on the International Classification of Diseases, Ninth Revision–Clinical Modification or International Classification of Diseases, Tenth Revision–Clinical Modification principal or secondary diagnosis code associated with the echocardiography claim. Indications were categorized as “appropriate” if they were clearly specified as “appropriate” in the Appropriate Use Criteria for Echocardiography, such as a resting echocardiogram ordered for dyspnea or acute heart failure (e.g., International Classification of Diseases, Ninth Revision–Clinical Modification code 786.05 or 428.x, respectively; Supplemental Digital Content table 3, https://links.lww.com/ALN/C688). Indications were categorized as “rarely appropriate” if they were specified as “rarely appropriate” in the Appropriate Use Criteria for Echocardiography, such as a resting echocardiogram ordered for chronic heart failure (e.g., International Classification of Diseases, Ninth Revision–Clinical Modification code 428.22) without any additional new symptom identified in a patient with an echocardiogram performed in the previous year, or a resting echocardiogram ordered for coronary atherosclerosis (International Classification of Diseases, Ninth Revision–Clinical Modification code 414.0x). No studies in our cohort were found to be “may be appropriate” as there are very few overall indications that are classified as “may be appropriate” in the Appropriate Use Criteria for Echocardiography, and they could not be correlated to International Classification of Diseases, Ninth Revision–Clinical Modification/International Classification of Diseases, Tenth Revision–Clinical Modification codes. Echocardiograms were considered “unclassifiable” if the indication did not allow appropriateness classification according to the Appropriate Use Criteria for Echocardiography (e.g., heart disease, unspecified). While all the indications for the echocardiograms were based on the 2011 Appropriate Use Criteria for Echocardiography document, the recommended updated classification terminology from the 2013 methodology update was applied. This update did not change the classification of tests from the 2011 document, but rather adjusted the terminology used to describe these classifications in order to more closely reflect their application in practice.⁵  In particular, “inappropriate” and “uncertain” indications from the 2011 document are now referred to as “rarely appropriate” and “may be appropriate,” respectively, in the 2013 document.

If records of patients included multiple resting echocardiograms, only the claim for the test performed on the date closest to the surgery was analyzed. If records included multiple claims for the same resting echocardiogram with disparate principal diagnosis codes, the claim used for analysis was determined using the following criteria: (1) if there were claims for both physician and facility fees, the claim for the physician fee was analyzed; (2) if claims were paid on different dates, the claim with the earlier paid date was analyzed; and (3) if claims differed only in principal diagnosis, then the claim with the most specific diagnosis code or the code most closely aligned with Appropriate Use Criteria for Echocardiography was analyzed.

Secondary outcomes were major adverse cardiac events that included a diagnosis of myocardial infarction, cardiac arrest, or new heart failure during the inpatient claim associated with the major abdominal surgery. Outcomes were identified using International Classification of Diseases, Ninth Revision–Clinical Modification and International Classification of Diseases, Tenth Revision–Clinical Modification codes and can be found in the Supplemental Digital Content table 4 (https://links.lww.com/ALN/C688). Outcomes were stratified based on the appropriateness classification of the echocardiogram as outlined in the Appropriate Use Criteria for Echocardiography.

Patient characteristics analyzed were age, sex, geographical region, comorbid conditions, and Elixhauser score. Elixhauser comorbidities can be assessed as a single numeric score that summarizes disease burden and is adequately discriminative for in-hospital mortality.¹²  Consistent with previous analyses of MarketScan, comorbidity diagnoses were included for cohort subjects if the International Classification of Diseases, Ninth Revision–Clinical Modification and International Classification of Diseases, Tenth Revision–Clinical Modification codes were present at least once in inpatient or outpatient claims from the year before the procedure.¹¹  Diagnosis codes used to classify comorbidities can be found in the Supplemental Digital Content table 5 (https://links.lww.com/ALN/C688).

Records of patients who underwent nonpreoperative resting echocardiography within 1 yr (i.e., between 61 days and 365 days) of the procedure were also identified, as the Appropriate Use Criteria for Echocardiography specify that a repeat resting echocardiogram is unnecessary within 1 yr of a previous study in the absence of new symptoms or a change in clinical status.⁴  This guidance for repeating a resting echocardiogram applies for patients undergoing preoperative evaluation who have had an echocardiogram in the last year.

Other cardiac tests, including exercise or pharmacologic stress tests, myocardial nuclear imaging, stress magnetic resonance imaging, and transesophageal echocardiograms, performed between 61 and 365 days before the procedure were also identified via Current Procedural Terminology codes in outpatient claims (Supplemental Digital Content table 2, https://links.lww.com/ALN/C688). These tests can influence the decision to pursue a preoperative echocardiogram and were included in the analysis.

Data were complete except for region of service (n = 1,576). Apart from this, there were no missing data for any of the variables that were used for the analysis.

The data analysis was completed using SAS software, Version 9.4 (SAS Institute Inc., USA). A data analysis and statistical plan was written after the data were accessed. As this is a retrospective cohort study, the sample size was based on the available data, and no statistical power calculation was conducted. The Cochrane–Armitage test was used to identify any trends in the frequency of “rarely appropriate” echocardiograms throughout the study period. Chi-square and unpaired t tests were used to compare patient characteristics between subjects who underwent preoperative resting echocardiography and those who did not, and chi-square and one-way ANOVA tests were used to compare patient characteristics between subjects who had “appropriate,” “rarely appropriate,” or “unclassifiable” indications for resting echocardiography and outcomes.

We conducted three sensitivity analyses. The first compared the incidence of “appropriate” resting echocardiograms between the preoperative period and outside of the preoperative period to determine if there was a higher frequency of “rarely appropriate” resting echocardiograms performed in the preoperative period. We applied the Appropriate Use Criteria for Echocardiography in the same manner as resting echocardiograms performed more than 60 days before the operative procedure and compared the appropriateness using a chi-square test. Second, we evaluated the incidence of resting echocardiograms over the 12-month period before surgery in the entire cohort to confirm that resting echocardiograms performed 60 days before the procedure were in response to the upcoming surgical procedure. The third sensitivity analysis evaluated comorbidity burden and cardiac outcomes of patients who underwent a repeat echocardiogram in the 60 days before surgery to evaluate if patients who underwent a repeat echocardiogram had a higher comorbidity burden and higher frequency of major adverse cardiac events. Significance for these analyses was set at P ≤ 0.01 with Bonferroni correction for multiple comparisons and two-tailed hypothesis testing where appropriate.

The study cohort consisted of 230,535 reports of patients, and characteristics of these patients are presented in table 1. Overall, 13,936 reports of patients (6%) included a preoperative resting echocardiogram in the 60 days before a major abdominal surgical procedure. Patients who underwent preoperative resting echocardiography were older, with a mean age of 63 yr (SD = 14 yr) as compared to 57 yr (SD = 14 yr) in the untested cohort (P < 0.001). Patients who received a preoperative resting echocardiogram had higher frequencies of multiple comorbidities, including insulin-dependent diabetes mellitus, stroke, chronic kidney disease, coronary artery disease, heart failure, and any cardiac valve pathology (P < 0.001). We did not identify a difference between the cohorts in frequency of nonpreoperative resting echocardiography (1,502 of 13,936, 11% vs. 24,284 of 216,599, 11%; P = 0.115) in the year before the procedure. The untested cohort did have a higher frequency of other nonpreoperative cardiac tests performed in the year before surgery (1,119 of 13,936, 8% vs. 18,983 of 216,599, 9%; P = 0.002); however, this difference is unlikely to be clinically significant.

Characteristics of patients who underwent preoperative resting echocardiography for “appropriate,” “rarely appropriate,” and “unclassifiable” indications are presented in table 2. Of all preoperative resting echocardiograms, 12,638 (91%) were classifiable by the Appropriate Use Criteria for Echocardiography, and 1,298 (9%) were unable to be classified. Of classifiable studies, 8,959 (71%) were classified as “appropriate,” and 3,679 (29%) were classified as “rarely appropriate.” There was no change in the frequency of “appropriate” versus “rarely appropriate” echocardiograms that were classifiable by the Appropriate Use Criteria or Echocardiography throughout the study period (Supplemental Digital Content table 6, https://links.lww.com/ALN/C688). Repeat echocardiograms (i.e., a preoperative resting echocardiogram when another resting echocardiogram was done within the year before the procedure) accounted for 11% (1,502 of 13,936) of all perioperative resting echocardiograms, and 52% (785 of 1,502) were classified as “rarely appropriate,” 36% (537 of 1,502) as “appropriate,” and 12% (180 of 1,502) as “unclassifiable.” Men had a higher proportion of “rarely appropriate” echocardiograms (2,031 of 6,308, 32% vs. 1,648 of 6,330, 26%; P < 0.001) as compared to women. Patients with coronary artery disease had the highest frequency (44%, 1,458 of 3,281) of “rarely appropriate” echocardiograms among all the cardiac comorbidities. The most common “rarely appropriate” echocardiogram indications in our study were surveillance of chronic ischemic heart disease (954 of 3,679, 26%), general preoperative examination (767 of 3,679, 21%), and hypertension (634 of 3,679, 17%).

The overall frequency of major adverse cardiac events that included myocardial infarction, cardiac arrest, and heart failure was 0.6% (1,317 of 230,535). The frequency among patients that did not have a preoperative echocardiogram was 0.5% (1,116 of 216,599) as compared to 1.4% (201 of 13,936) in patients who did receive a preoperative echocardiogram (P < 0.001). There was no difference in the frequency of major adverse cardiac events in patients with a “rarely appropriate” (1.8%, 66 of 3,679), “appropriate” (1.3%, 116 of 8,959), or “unclassifiable” resting echocardiogram (1.5%, 19 of 1,298; P = 0.086).

In our cohort, 11.2% (25,786 of 230,535) of patient records contained a claim for a resting echocardiograms performed before the preoperative period. Of those echocardiograms, 23,160 (90%) were classifiable by the Appropriate Use Criteria for Echocardiography, 17,213 (74%) were “appropriate,” and 5,947 (26%) were “rarely appropriate.” The proportion of resting echocardiograms that were classified as “rarely appropriate” in the preoperative setting was higher than those outside of the preoperative period in our cohort. Additionally, the frequency for all resting echocardiograms done in the year before surgery can be seen in Supplemental Digital Content figure 1 (https://links.lww.com/ALN/C688) and Supplemental Digital Content table 7 (https://links.lww.com/ALN/C688). Resting echocardiograms in the 2 months before surgery accounted for 47% (28,122 of 59,788) of the total volume of resting echocardiograms. We observed a sharp increase in the frequency of resting echocardiograms in the 2 months before major abdominal surgery, as illustrated by Supplemental Digital Content figure 1 (https://links.lww.com/ALN/C688).

Among records of patients that included a resting echocardiogram in the preoperative period and also an echocardiogram in the previous year, mean Elixhauser comorbidity scores were greater for patients who had an “unclassifiable” 13 (SD = 11) and “appropriate” 12 (SD = 10) resting echocardiogram as compared to a “rarely appropriate” 10 (SD = 10) resting echocardiogram (P < 0.001). The frequency of myocardial infarction, cardiac arrest, and heart failure associated with repeat echocardiograms overall was 2.7% (39 of 1,502). There was no difference in the frequency of complications in patients with a “rarely appropriate” (3.3%, 26 of 785), “appropriate” (1.9%, 10 of 537), or “unclassifiable” resting echocardiogram (1.5%, 3 of 180; P = 0.187).

In our retrospective cohort study, we found that while preoperative resting echocardiography was uncommon overall (6%), more than one in four echocardiograms (29%) that were classifiable by the Appropriate Use Criteria for Echocardiography were considered “rarely appropriate.” The American College of Cardiology and American Heart Association perioperative practice guidelines recommend a narrow set of conditions for preoperative resting echocardiography known to be associated with increased perioperative complications.²  An important and consistent theme throughout the practice guidelines is that cardiac testing should be rarely pursued in the absence of a clear indication independent of the impending surgery. The Appropriate Use Criteria for Echocardiography provide guidance to clinicians in the optimal use of resting echocardiography for all clinical settings and provide guidance to perioperative clinicians in appropriate use of resting echocardiography for conditions that may not directly apply to the perioperative period. The most frequent reason for “rarely appropriate” preoperative resting echocardiograms in our study was routine surveillance of known cardiovascular disease, such as chronic ischemic heart disease. Further, there was no difference in frequency of major adverse cardiac events among patients undergoing “rarely appropriate” as compared to “appropriate” or “unclassifiable” echocardiograms. These findings suggest that while overall preoperative resting echocardiography utilization is low, there still exists an opportunity to reduce testing that has not been shown to impact perioperative cardiac morbidity.

The high proportion of “rarely appropriate” echocardiograms may be secondary to the outpatient nature of the echocardiograms assessed in this study as well as ordering clinician characteristics. Outpatient studies are less likely to be associated with a change in clinical status and are known to have a higher frequency of “rarely appropriate” echocardiograms compared to inpatient exams. The frequency of “rarely appropriate” echocardiograms in other practice settings has been reported as high as 30%, which is similar to our findings.^13–15  In our sensitivity analysis of nonperioperative echocardiograms, we identified a high percentage of “rarely appropriate” echocardiograms in the outpatient setting, which suggests that potential overuse of cardiac imaging in the outpatient setting is not limited to the preoperative period. Ordering clinician characteristics may lead to a higher frequency of “rarely appropriate” echocardiograms in the preoperative setting. In a single academic center retrospective cohort of patients referred for outpatient transthoracic echocardiography, we demonstrated that anesthesia and surgical clinicians had the highest frequency of “rarely appropriate” echocardiograms (21% and 19%, respectively) as compared to cardiologists (3%) or internal medicine clinicians (10%).¹⁶  This may be due to a lack of familiarity with the Appropriate Use Criteria for Echocardiography among anesthesia and surgical clinicians.¹⁶  Anesthesiologists are required to address medical concerns during perioperative visits that extend beyond the upcoming procedure. As such, our specialty should consider incorporating the Appropriate Use Criteria for Echocardiography to guide preoperative cardiovascular imaging and develop interventions to reduce potential overuse. Our finding that a small number of diagnoses (e.g., surveillance of chronic ischemic heart disease) are associated with a large proportion of “rarely appropriate” echocardiograms may help to guide interventions that may be effective to decrease potential overuse of preoperative resting echocardiography.^13,17,18 

Chronic ischemic heart disease without change in clinical status accounted for just over a quarter of all the “rarely appropriate” echocardiograms in our study and may be an easy target to reduce overuse. In the only prospective trial across eight hospital systems to reduce “rarely appropriate” echocardiograms, the second most common reason for a “rarely appropriate” echocardiogram was routine surveillance of ventricular function with known coronary artery disease and no change in clinical status or cardiac examination.¹⁷  The intervention for the trial comprised an initial video lecture, a mobile application-based decision support tool, and monthly clinician feedback reports for cardiologists and primary care providers that summarized individual ordering behavior. The proportion of rarely appropriate echocardiograms decreased in the intervention group as compared to the control (8.8% vs. 10.1%; odds ratio, 0.75; 95% CI, 0.57 to 0.99; P = 0.039), and the decrease was sustained throughout the study period.

Sex bias may have played a role in the higher frequency of men receiving a “rarely appropriate” echocardiogram as men may be more likely to receive cardiovascular testing. Previous population-based studies that evaluated utilization of perioperative resting echocardiograms similarly identified a higher frequency of testing in men as compared to women.⁶  In a study of sex bias among cardiologists evaluating a simulated patient with suspected coronary artery disease, male patients were more consistently rated as likely to benefit from angiography as compared to female patients.¹⁹  Given the known bias toward men and cardiovascular disease, it appears highly plausible that sex bias played a role with the increased frequency of “rarely appropriate” preoperative resting echocardiograms in our analysis.

Our study is timely in light of the Centers for Medicare and Medicaid Services mandate starting in 2022 that will require appropriateness determinations according to the Appropriate Use Criteria for Echocardiography to be submitted at the time of study order for all “advanced imaging procedures” for Medicare beneficiaries.²⁰  Clinicians will be tracked through the program, and those considered outliers with a high frequency of “rarely appropriate” studies will be subject to more rigorous authorizing procedures and potential reimbursement denials. Initially, perioperative physicians will experience this program in relation to preoperative nuclear stress testing as resting echocardiography is not included in the first phase of the mandate. However, as the highest volume cardiac imaging test performed, resting echocardiography is expected to be included as the program expands.²¹  Further, perioperative testing is likely to be a particular focus and may be disproportionally affected by these programs. Our study addresses the knowledge gap of appropriate use criteria for resting echocardiography on a national level, and our finding of a high frequency of rarely appropriate studies suggests significant potential overuse. However, a full understanding of appropriate utilization of perioperative resting echocardiography also requires consideration of potential underuse when “appropriate” indications are present. In a study that used the National Inpatient Sample to identify “appropriate” use of inpatient resting echocardiography for certain conditions (e.g., acute myocardial infarction), only 8% of inpatient admissions received a resting echocardiogram.²²  Patients who received a resting echocardiogram had a lower inpatient mortality as compared to patients who did not when adjusting for patient and hospital characteristics. Given that there are a variety of appropriate indications unrelated to surgery that may first come to clinical attention in the perioperative setting (e.g., heart murmur), the potential for underuse of resting echocardiography is real. While our study did not address underuse, further study of this issue in the perioperative setting, and its association with patient outcomes, deserves study.

Our study has several limitations. First, the MarketScan databases only include patients with employer-sponsored health insurance or Medicare supplemental plans and may not be a representative sample of the U.S. population. This may explain why our frequency of preoperative echocardiograms was lower than other population-based studies that had higher mean ages for their cohorts. Thus, it remains unclear if our results are generalizable to a traditional Medicare population. MarketScan provides diagnosis and procedure codes for billing purposes, and these codes do not fully represent the scope of disease and may not capture clinical changes over time for each patient. Other unmeasured variables may have impacted the decision for preoperative echocardiography that we were unable to account for, and it is possible that the billing codes utilized did not appropriately reflect the clinical status of the patient. However, we used a robust approach to classify echocardiograms based on principal and secondary diagnoses that were directly matched to criteria found within the Appropriate Use Criteria for Echocardiography. Further, we did not identify any change in the frequency of “appropriate” echocardiograms across the transition from International Classification of Diseases, Ninth Revision–Clinical Modification to International Classification of Diseases, Tenth Revision–Clinical Modification coding that took place in 2015. In addition, our data collection focused on comorbidities that were documented in the year before surgery, which may not capture the full burden of comorbid disease for each patient. The Appropriate Use Criteria for Echocardiography rely much more heavily on consensus and expert opinion as compared to practice guidelines that rely heavily on a more evidence-based format.²³  However, the Appropriate Use Criteria for Echocardiography incorporate the preoperative cardiac evaluation practice guidelines with regards to their recommendations. Finally, the later years of the study cohort contained fewer patients as compared to earlier in the cohort, and this may have led to decreased precision in some of the later estimates of the frequency of “rarely appropriate” studies.

In conclusion, more than one in four classifiable resting preoperative echocardiograms before major abdominal surgical procedures may be considered “rarely appropriate” according to the Appropriate Use Criteria for Echocardiography. It is important to focus continued efforts on mitigating potential overuse of preoperative echocardiography, as unnecessary testing contributes to increased costs to patients and to the healthcare system.

Dr. Rubin and Dr. Nagele were supported by funding from the National Institutes of Health/National Heart, Lung, and Blood Institute (Bethesda, Maryland; R01 HL126892).

Dr. Rubin is the president of DRDR Mobile Health (Chicago, Illinois), a company that creates mobile applications for health care, including functional capacity assessment applications. He has engaged in consulting for mobile applications as well. He has not taken any salary or money from the company. Dr. Rubin has also consulted as an expert witness. The research in this article had no relationship to any of that work. The other authors declare no competing interests.