Heiko Pohl, MD; Douglas Robertson, MD, MPH; H. Gilbert Welch, MD, MPH
Disclaimer: The findings, statements, and views expressed are those of the authors and do not necessarily represent those of the Medicare Payment Advisory Commission, the U.S. Department of Veterans Affairs, or the U.S. government. All authors had full access to all of the data in the study and take full responsibility for the integrity of the data and the accuracy of the data analysis.
Acknowledgment: The authors thank Mark Miller, Kevin Hayes, and Carol Frost of the Medicare Payment Advisory Commission for obtaining the data and assisting with the analysis. They also thank Dr. Corey A. Siegel and Dr. L. Campbell Levy from the Section of Gastroenterology at Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire, for their dedicated assistance with additional analysis in the revision process.
Financial Support: The research underlying this article was completed with support from the Medicare Payment Advisory Commission.
Potential Conflicts of Interest: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M13-0046.
Reproducible Research Statement: Study protocol: Study algorithm available from Dr. Pohl (e-mail, firstname.lastname@example.org). Statistical code: Not available. Data set: Available from Dr. Pohl (e-mail, email@example.com).
Requests for Single Reprints: Heiko Pohl, MD, Department of Gastroenterology, Veterans Affairs Medical Center, 215 North Main Street, White River Junction, VT 05009; e-mail, firstname.lastname@example.org.
Current Author Addresses: Drs. Pohl, Robertson, and Welch: Department of Gastroenterology, Veterans Affairs Medical Center, 215 North Main Street, White River Junction, VT 05009.
Author Contributions: Conception and design: H. Pohl, D. Robertson, H.G. Welch.
Analysis and interpretation of the data: H. Pohl, D. Robertson, H.G. Welch.
Drafting of the article: H. Pohl, H.G. Welch.
Critical revision of the article for important intellectual content: H. Pohl, D. Robertson, H.G. Welch.
Final approval of the article: H. Pohl, D. Robertson, H.G. Welch.
Statistical expertise: H.G. Welch.
Administrative, technical, or logistic support: H.G. Welch.
Collection and assembly of data: H.G. Welch.
Pohl H., Robertson D., Welch H.; Repeated Upper Endoscopy in the Medicare Population: A Retrospective Analysis. Ann Intern Med. 2014;160:154-160. doi: 10.7326/M13-0046
Download citation file:
Published: Ann Intern Med. 2014;160(3):154-160.
Esophagogastroduodenoscopy (EGD) is done often for various indications. Little is known about the frequency of repeated EGD and the diagnoses that drive it.
To describe the frequency of repeated EGD in the Medicare population and determine diagnoses most often associated with it.
Among a 5% random sample of Medicare beneficiaries, Current Procedural Terminology (CPT) codes were used to identify patients who had an index EGD between 2004 and 2006. Diagnoses from the International Classification of Diseases, Ninth Revision, Clinical Modification, reported for the index endoscopy were divided into 3 diagnostic groups on the basis of whether the index diagnosis suggested that repeated EGD was expected, uncertain, or not expected.
Proportion of patients with repeated EGD within 3 years of an index EGD.
Approximately 12% of Medicare beneficiaries had an EGD between 2004 and 2006 (n = 108 785). Of these, 33% (n = 36 331) had at least 1 repeated EGD within 3 years. Of all patients with initial EGDs, 10% (n = 11 370) had an associated diagnosis suggesting a need for follow-up examination, whereas 61% (n = 66 307) did not. Of all patients with repeated examinations, 54% (n = 19 687) came from the group in which repeated EGD was not expected. When new clinical events were excluded, 43% of all patients with repeated EGDs (n = 15 706) did not have a diagnosis at index or repeated EGD that justified a repeated examination.
Whether individual procedures were clinically indicated cannot be verified.
One in 3 Medicare beneficiaries who received an EGD had a repeated EGD within 3 years. Nearly one half of repeated examinations were done in patients with diagnoses at index or repeated EGD that did not suggest the need for a repeated examination.
Esophagogastroduodenoscopy (EGD) is commonly done, and concerns have been raised about potential overuse. However, little is known about how frequently EGD is repeated.
This study found that approximately 12% of Medicare beneficiaries had EGD over 3 years; of these, about one third had repeated EGD within this period. About one half of all repeated EGDs followed an initial procedure coded in a manner not suggestive of a need to repeat the examination.
The accuracy of diagnostic codes for identifying the need for follow-up EGD is unknown.
A substantial proportion of Medicare recipients have repeated EGDs, many of which might not be required. Further evaluation of possible overuse is warranted.
Since Rudolph Schindler introduced the Wolf–Schindler semiflexible gastroscope in 1932 (1), upper endoscopy has steadily improved. Over the past few decades, technologic enhancements have transformed the procedure from a simple visual inspection to an advanced diagnostic and therapeutic tool. Endoscopic imaging now allows high-definition examination of the gastrointestinal epithelium by using magnification, optical filters, and even real-time microscopy (2–4). Endoscopic devices can now ligate and inject varices and resect neoplasia and early cases of cancer (5, 6).
Esophagogastroduodenoscopy (EGD) has become routine not only in the specialized setting of varices and Barrett neoplasia (7) but also in the evaluation of the more common symptoms of abdominal pain and dysphagia. Consequently, use of EGD has exploded in the United States—more than doubling between 1989 and 2006 (8–10)—with approximately 7 million procedures performed annually (10–12).
Although EGD has a clear role in the evaluation of the gastrointestinal tract, it can be overused. Unnecessary procedures not only increase patient risk but also burden health care resources. Overuse is considered to be the most important contributor to the high cost of U.S. health care (13). The American Board of Internal Medicine's Choosing Wisely campaign now addresses concerns about overuse for many disciplines (14). The American College of Physicians has issued best practice guidelines on the use of EGD in patients with gastroesophageal reflux disease to minimize inappropriate use (15), and the American Gastroenterological Association has targeted overuse of EGD in Barrett esophagus on its list of 5 things that patients and physicians should question (16). This concern was supported by an investigation of patients with Barrett esophagus without dysplasia that found that over 3 years, the typical patient had more than 2 excess EGDs than expected by guidelines (17).
Although overuse of EGD is increasingly recognized in the setting of Barrett esophagus, few data examine the problem across the many indications for which EGD is performed in daily clinical practice. Overuse can be considered in terms of 2 broad categories. First, the procedure may be performed on too many patients (for example, patients with indications of unproven benefit). Second, too many procedures may be performed on the same patient (for example, examinations are repeated within a short time). In this study, we focus on the second category by examining the frequency of repeated EGD in a nationally representative sample of the Medicare population.
Medicare administrative data were used to determine the rate of repeated upper endoscopy within 3 years of an initial examination. Diagnostic codes were used to provide insight into which studies were most likely to be repeated. This study was determined to be exempt from review by the Dartmouth Committee for the Protection of Human Subjects, Hanover, New Hampshire.
We analyzed a 5% random sample of patients who received Medicare between 2004 and 2009. Beneficiaries enrolled in Medicare Part A only or a risk-contract HMO were excluded because of incomplete claims data. As in prior work on repeated testing (18), we restricted the analysis to beneficiaries who were “at risk” for repeated EGD—that is, those who were alive for the entire 6-year period. Of the 1 579 553 fee-for-service beneficiaries enrolled in 2004, a total of 936 370 lived through 2009.
For this population, which was continuously enrolled in fee-for-service Medicare from 2004 through 2009, we pulled all claims with a Current Procedural Terminology (CPT) code for EGD (Appendix Table 1) during the same 6 years. To ensure a 3-year follow-up window, we excluded all beneficiaries whose earliest EGD occurred after 31 December 2006. This produced the cohort in which we studied repeated EGD: 108 785 beneficiaries who had an EGD between 2004 and 2006. Each patient's first EGD occurring during these 3 years is subsequently called the “index EGD.”
Appendix Table 1.Upper Gastrointestinal Endoscopy CPT Codes Applied for the Analysis of Upper Endoscopy Use in the Medicare Population
We began by determining the most common International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnoses appearing on the claim index EGD (Appendix Figure). Although 698 distinct ICD-9-CM diagnoses appeared in our cohort, we chose to focus on common diagnoses (those that appeared on 100 or more EGDs). This restriction resulted in 70 ICD-9-CM diagnoses that appeared in 103 711 patients with an index EGD, or 95% of all index EGDs. The diagnoses of the remaining 5074 patients were not classified.
Study flow diagram.
EGD = esophagogastroduodenoscopy.
* The proportions in the Repeated EGD cells do not sum to 100% because 5% of repeated examinations were in the unclassified category.
Two investigators then divided these 70 diagnoses into 14 meaningful diagnostic categories (for example, Barrett esophagus or gastritis or duodenitis). Using a modified Delphi process, we further divided the 14 diagnostic categories into 3 broad diagnostic groups: those in which repeated EGD was expected, uncertain, or not expected given the diagnosis at index endoscopy. A finding of Barrett esophagus or varices, for example, would be expected to result in repeated EGD, whereas a finding of gastritis or duodenitis or nonspecific findings would not.
To understand one of the most common reasons that diagnoses were categorized as those in which repeated EGD was not expected, it is important to realize that ICD-9-CM code diagnoses fall into 2 broad categories: findings (for example, peptic ulcer) or indications or symptoms (for example, abdominal pain). Per Medicare coding rules, diagnoses should be coded by finding if an abnormality is identified that relates to the indication or by indication if no abnormality is detected (19). For example, a patient with abdominal pain whose EGD shows a peptic ulcer would be diagnosed with this condition; however, if no abnormality is found to explain the pain, it should be coded as abdominal pain. Therefore, a diagnostic code of a simple symptom, if coded properly, should reflect the indication for the test and the lack of a meaningful clinical finding on that examination.
In the main analysis, we examined the proportion of patients who had at least 1 repeated EGD within 3 years after the index endoscopy. We then obtained the proportion of patients with repeated EGD for each diagnostic category and in each diagnostic group (those in which repeated EGD was expected, uncertain, or not expected). Finally, we calculated the contribution of each diagnostic category and group to all repeated procedures, including those repeated in the unclassified category (Table 1).
Table 1. Diagnoses at Index EGD and Frequency of Repeated Examinations
Our additional analyses were restricted to patients whose diagnosis at index EGD did not suggest the need for repeated EGD (Appendix Figure) but who nevertheless had a repeated examination. We recognize that a portion of these patients may have developed a new clinical problem that required repeated EGD; therefore, we examined the diagnoses from the repeated (that is, second) EGD in these patients. We sought to understand 2 scenarios. First, we evaluated how often patients in this subgroup had the same diagnosis at index and repeated EGD. We asserted that patients with the same diagnosis at both EGDs (in the broad category in which a repeated examination was not expected) were unlikely to have a condition that justified a repeated procedure. For example, a patient with a diagnosis of dyspepsia, abdominal pain, or nonspecific symptoms at both EGDs was unlikely to need repeated EGD (Table 2). Second, we aimed to understand how often patients developed a new clinical problem that required repeated EGD independent of the initial diagnosis. For example, repeated EGD would be justified for a patient with dyspepsia as the initial diagnosis who then developed diarrhea or weight loss or presented with gastrointestinal bleeding. Using a modified Delphi process, we established combinations of diagnostic categories at index and repeated EGDs that would strongly support the need for a second endoscopy (for example, a combination of dyspepsia and diarrhea would justify repeated EGD, but a combination of gastritis and dyspepsia would not) (Table 3). Appendix Table 2 provides a complete list of pairings of index and repeated EGD diagnoses for which repeated EGD was justified. For both subanalyses, we calculated the proportion of EGDs within the diagnostic categories and their contribution to all repeated procedures.
Table 2. Patients With Repeated EGD Not Expected and Diagnosis Unchanged*
Table 3. Patients With Repeated EGD Not Expected and New Clinical Event*
Appendix Table 2.Pairing of Index and Repeated EGD Diagnoses That Did or Did Not Justify a Repeated Procedure Among All Patients for Whom Repeated EGD Was Not Expected*
To facilitate text presentation, all proportions were rounded to the nearest percentage point; the tables provide more precision. Abstraction of claims data and categorization analysis were performed using SAS, version 9.2 (SAS Institute, Cary, North Carolina). Subsequent analyses were performed in Microsoft Excel, version 14.3.8 (Microsoft, Redmond, Washington).
This study had no external funding.
Approximately 12% of Medicare beneficiaries had at least 1 EGD between 2004 and 2006 (n = 108 785). Index diagnoses suggested that repeated EGD was expected in 10% of patients (n = 11 370) and not expected in 61% (n = 66 307). The need for a repeated procedure was uncertain in 24% (n = 26 034) and was not classified in the remaining patients (n = 5074).
Among all patients with 1 upper endoscopy, the procedure was repeated in 33% within 3 years (n = 36 331). Table 1 details the proportions of repeated EGDs by diagnostic group. Fifty percent of patients in whom repeated EGD was expected, 30% in whom it was not expected, and 35% in whom it was uncertain had a repeated examination. Table 1 also shows our findings for each of the 14 diagnostic categories. For example, 60% of patients with an index diagnosis of varices and 59% with an index diagnosis of Barrett esophagus (both in the group in which repeated EGD was expected) had a repeated examination within 3 years. In contrast, 29% of patients with gastritis or duodenitis and 30% with abdominal pain or dyspepsia (both in the group in which repeated EGD was not expected) had repeated EGD.
Although one half of patients in the group in which a repeated EGD was expected had repeated EGD, this group accounted for only 16% of all repeated upper endoscopies (Figure). In contrast, the group in which repeated examination was not expected accounted for 54% of all repeated upper endoscopies. The final column of Table 1 shows detailed data on the sources of repeated examination. For example, only 6% of repeated procedures were among patients with Barrett esophagus (those in whom repeated EGD was expected) and 20% were among patients with gastritis or duodenitis (those in whom repeated EGD was not expected).
Repeated EGD in the Medicare population.
The left portion of the figure shows the number of beneficiaries having upper endoscopy and the proportion of procedures repeated in 3 diagnostic groups. The right portion shows the distribution of diagnostic groups among all repeated examinations. EGD = esophagogastroduodenoscopy.
Table 2 expands on the 19 687 patients whose index diagnosis did not suggest the need for a repeated examination (that is, the group in which repeated EGD was not expected) but who nonetheless had one. Almost one third of these patients (32%) had the same diagnosis at index and repeated endoscopies, and two thirds received a diagnosis in the same diagnostic group (those in which repeated EGD was not expected). The most common identical diagnostic categories were gastritis or duodenitis (40%) and dysphagia (34%). Overall, patients with an unchanged diagnosis for which repeated EGD was not expected accounted for 18% of all repeated procedures.
Table 3 focuses on the same 19 687 patients with the acknowledgment that some may have had a new clinical event that would justify repeated EGD (for example, bleeding, new diagnosis of dysphagia, new-onset diarrhea, Barrett esophagus, or varices). In 20% of these patients, the repeated EGD found a new diagnosis that may have justified the repeated procedure. This proportion was similar across diagnostic categories. Conversely, the remaining 80% of patients did not have a new diagnosis to justify repeated EGD. Overall, 43% of repeated EGDs were done in patients with diagnoses at index or repeated EGD that did not suggest the need for a repeated examination.
Our study examined how often upper endoscopies in the Medicare population are repeated within a short interval. Our analysis reveals 3 important findings. First, 12% of Medicare beneficiaries had an EGD within a 3-year period. Second, of those who had an upper endoscopy, 33% had a repeated examination within 3 years. Third, 43% of all patients with repeated EGDs did not seem to have a diagnosis at initial or repeated endoscopy that justified a repeated procedure. Taken together, these results suggest that EGD may be substantially overused.
A MEDLINE search identified few studies examining possible overuse of upper endoscopy. One study determined the frequency of repeated EGD in patients with peptic ulcer disease and found that younger patients and those with ulcers that seemed benign had repeated EGDs at a higher-than-expected rate (20). Similarly, a study by Crockett and colleagues (17) on patients with nondysplastic Barrett esophagus reported that 65% of patients experienced “oversurveillance.” Studies on open-access endoscopy suggest that 10% to 40% of upper endoscopies may not be indicated when criteria from the American Society for Gastrointestinal Endoscopy are applied (21–26).
Overuse is a challenging construct for researchers. Although clinicians and policymakers acknowledge the general problem (13, 15, 27), evidence for overuse within a specific clinical setting is open to alternative explanations. To identify where EGD overuse is occurring, we considered 2 broad categories. First, the procedure may be performed on too many patients. Although this was not the focus of our investigation, we were surprised to find that 1 in 8 Medicare beneficiaries had received the procedure between 2004 and 2006; this number seems high. We invite clinicians to judge for themselves but cannot say much more about this finding.
Our study was designed to examine the second category of overuse: whether the procedure is repeated too often in the same patient. We consider a third of patients having repeated EGD within 3 years as high. One possible explanation is that the procedure is repeated in patients who truly need it (for example, those with Barrett esophagus or esophageal varices). However, our additional analysis suggests that most procedures are done in patients with a diagnosis that would not suggest the need for repeated EGD. In the group in which repeated EGD was not expected, 30% of patients had a repeated procedure.
To fully grasp these findings, it is important to understand the 7 diagnostic categories that constitute this group. Four of the categories relate to symptoms—diagnoses that, under Medicare coding rules, are allowed only if no abnormality is identified. In other words, the results of these examinations are “normal.” One category includes diagnoses occurring outside of the upper gastrointestinal tract (for example, in the colon), again suggesting normal EGD results. The 2 remaining categories (that is, gastritis or duodenitis and nonspecific findings) would be recognized by most gastroenterologists as generally trivial findings with minimal clinical relevance. Given this information, our finding could be restated as 30% of EGDs with “normal” or “near-normal” results are repeated within 3 years and constitute more than one half of all repeated EGDs.
That patients with normal results at index endoscopy may have a new clinical event that requires a repeated endoscopy (for example, gastrointestinal bleeding, weight loss, or dysphagia) is plausible. However, only a few repeated examinations in the group in which repeated examination was not expected (20%) had a diagnosis that seemed to justify repeated EGD (Table 3). Furthermore, we were surprised that the diagnostic categories for most patients in the “normal” or “near-normal” group remained the same (Table 2). These data strengthen our analysis and support the argument that many repeated EGDs are unnecessary.
Of note, our findings raise questions about not only overuse but also underuse. Only 58% of patients with Barrett esophagus and 60% of patients with varices had a repeated examination. Even when considering that some of these patients might not require repeated EGD (histopathology did not confirm a diagnosis of Barrett esophagus, variceal bleeding prophylaxis was started, or older age or comorbid conditions eliminated the need), we would expect most of these patients to have a repeated examination within 3 years.
Our results that show overuse and underuse may also be related to limitations in our methods. For example, patients with Barrett esophagus may have been misclassified with a diagnosis of reflux disease, which, in our analysis, would not suggest the need for a repeated examination. However, our findings suggesting overuse and underuse are similar to those in the colonoscopy literature. For example, 1 study of patients with adenoma requiring surveillance colonoscopy simultaneously documented substantial overuse among patients with normal results (26%) and underuse among those with high-risk polyps (58%) (28). The combination of concomitant overuse and underuse may reflect a fragmented health care system in which tests are repeated because the results of previous examinations are unknown or not repeated because of inadequate call-back systems.
Our analysis has limitations. First, our diagnostic data are only as good as physicians' efforts in providing them. In our clinical experience, physicians are more reliable in coding acute, obvious, and consequential disease (in this case, gastrointestinal bleeding or varices) than normal (or near-normal) findings.
Second, to our knowledge, no information is available on the quality of abstraction of Medicare data on upper endoscopies. A recent study on abstraction of colonoscopy data in the Medicare population suggests highly accurate coding for a polyp diagnosis, with sensitivity and specificity greater than 90% (29). Similar results have been reported for abstraction of data on mammography, prostate resection, and hip fracture (30, 31).
To gain some understanding of the quality of the coding data, we performed a subanalysis in the group in which repeated examination was not expected and examined CPT codes to see whether they suggested some important intervention (that is, supported a need for repeated examination) that was not captured by the diagnostic codes on which we exclusively relied. We found that 92% of patients in this group had a simple diagnostic EGD with or without biopsy, with the remaining patients having an intervention, including dilation in 4% and gastrostomy placement in 2% (Appendix Table 3). Among patients in the “dysphagia” or “other symptoms” category, approximately one third had an intervention (typically esophageal dilation or gastrostomy placement), which raises concerns for some misclassification. However, the overall low proportion of complex procedures in this group supports the adjudication based on diagnostic codes.
Appendix Table 3. Proportions of Patients With an Intervention at Index EGD Among All Patients for Whom Repeated EGD Was Not Expected*
Third, we recognize that a diagnostic or procedural code may not provide the real reason for the examination. We recognize that decisions to perform procedures are dictated not simply by clinical indications but also by the expectations of referring physicians. We did not evaluate referral patterns and acknowledge that we cannot separate the roles of the referring physician from those of the endoscopist in repeated examinations. Other factors, such as financial incentives, medical legal concerns, and patient expectations, undoubtedly play a role in the overuse of repeated testing. Our findings are derived from the Medicare population and may not be generalizable to upper endoscopy practice compensated outside of that program. For example, if financial incentives are a major driver of overuse, repeated EGD outside of Medicare (which has lower payment rates than private insurers in some markets) may be even more pronounced.
In conclusion, upper endoscopy is commonly performed and commonly repeated in the Medicare population. One third of Medicare beneficiaries who have an upper endoscopy are likely to have a repeated examination within 3 years. Because almost one half of these procedures seem not to be justified by the diagnosis at index or repeated endoscopy, our findings suggest substantial overuse of repeated EGD. Minimizing overuse might reduce unnecessary and potentially risky examinations and would also lead to substantial cost savings. Although we applaud the engagement of the American Gastroenterological Association in the Choosing Wisely campaign, our findings suggest that it faces a tremendous challenge. Better data systems to inform physicians about previous examinations may partially address some overuse, but endoscopists—and the physicians who refer patients to them—must also carefully consider whether another upper endoscopy is truly warranted.
The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.
Gastroenterology/Hepatology, Healthcare Delivery and Policy, Peptic Disease, Gastroesophageal Reflux Disease.
Results provided by:
Copyright © 2016 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use
This PDF is available to Subscribers Only