Same-Day vs Different-Day Elective Upper and Lower Endoscopic Procedures by Setting

Key Points

Question  In different outpatient settings, how often are an elective upper endoscopy and an elective lower endoscopy performed on 2 separate days when they could potentially be performed on the same day?

Findings  In this cohort study of all paired upper and lower endoscopic procedures performed within a 90-day period among Medicare beneficiaries from January 1, 2011, to June 30, 2018, the different-day procedure rate was significantly higher in physician offices (47.7%) and freestanding ambulatory surgery centers (22.2%) compared with hospital outpatient departments (13.6%).

Meaning  This disparity in practice pattern by practice setting may represent an opportunity for quality improvement and financial savings for common endoscopic procedures.

Importance  Performing elective upper and lower endoscopic procedures on the same day is a patient-centered and less costly approach than a 2-stage approach performed on different days, when clinically appropriate. Whether this practice pattern varies based on practice setting has not been studied.

Objectives  To estimate the rate of different-day upper and lower endoscopic procedures in 3 types of outpatient settings and investigate the factors associated with the performance of these procedures on different days.

Design, Setting, and Participants  A retrospective analysis was conducted of Medicare claims between January 1, 2011, and June 30, 2018, for Medicare beneficiaries who underwent a pair of upper and lower endoscopic procedures performed within 90 days of each other at hospital outpatient departments (HOPDs), freestanding ambulatory surgery centers (ASCs), and physician offices.

Main Outcomes and Measures  Undergoing an upper and a lower endoscopic procedure on different days, adjusted for patient characteristics (age, sex, race/ethnicity, residence location and region, comorbidity, and procedure indication) and physician characteristics (sex, years in practice, procedure volume, and primary specialty). Adjusted odds ratios (aORs) and 95% CIs were calculated.

Results  A total of 4 028 587 procedure pairs were identified, of which 52.5% were performed in HOPDs, 43.3% in ASCs, and 4.2% in physician offices. The rate of different-day procedures was 13.6% in HOPDs, 22.2% in ASCs, and 47.7% in physician offices. For the 7564 physicians who practiced at both HOPDs and ASCs, their different-day procedure rate changed from 14.1% at HOPDs to 19.4% at ASCs. For the 993 physicians who practiced at both HOPDs and physician offices, their different-day procedure rate changed from 15.8% at HOPDs to 37.4% at physician offices. Patients were more likely to undergo different-day procedures at physician offices and ASCs compared with HOPDs, even after adjusting for patient and physician characteristics (physician office vs HOPD: aOR, 2.02; 95% CI, 1.85-2.20; ASC vs HOPD: aOR, 1.27; 95% CI, 1.23-1.32). Older age (85-94 years vs 65-74 years: aOR, 1.10; 95% CI, 1.08-1.11; 95 years or older vs 65-74 years: aOR, 1.14; 95% CI, 1.03-1.26), black and Hispanic race/ethnicity (black: aOR, 1.15; 95% CI, 1.12-1.17; Hispanic: aOR, 1.12; 95% CI, 1.10-1.14), and residing in the Northeast region (adjusted OR, 1.32; 95% CI, 1.28-1.36) were risk factors for undergoing different-day procedures. Micropolitan location (aOR, 0.94; 95% CI, 0.92-0.96) and rural location (aOR, 0.91; 95% CI, 0.89-0.93), more comorbidities (≥5: aOR, 0.75; 95% CI, 0.74-0.76), physician’s fewer years in practice (aOR, 0.84; 95% CI, 0.81-0.87), physician’s higher procedure volume (aOR, 0.65; 95% CI, 0.62-0.68), and physician’s specialty of general surgery (aOR, 0.86; 95% CI, 0.80-0.91) were protective factors.

Conclusions and Relevance  Physician offices and ASCs had much higher different-day procedure rates compared with HOPDs. This disparity may represent an opportunity for quality improvement and financial savings for common endoscopic procedures.

Low-value care is recognized as a driver of increasing health care costs and is associated with avoidable patient harm in the United States. A study that examined 26 low-value care measures applicable to insurance claims data estimated that 42% of all Medicare beneficiaries received at least 1 low-value service in 2009, resulting in $8.5 billion of Medicare spending.¹ Besides these 26 measures, there is a widespread prevalence of other low-value services across different areas of care.^2-4 In gastroenterology, one form of low-value care is the performance of elective upper and lower endoscopic procedures on 2 different days when they can be safely performed as a combined procedure on the same day.^5,6 In these cases, the different-day approach is associated with unnecessary repeated venipuncture and sedation, use of an endoscopy suite, added physician professional fees, and patient inconvenience.

A profit motive associated with physician ownership is one possible driver of overuse in outpatient settings, evidenced by studies that identified greater use of low-value imaging studies, physical therapy, and common outpatient procedures in physician-owned facilities.^7-10 We hypothesized that the pattern of different-day upper and lower endoscopic procedures might also differ based on the practice setting. We studied the performance of different-day elective upper and lower endoscopic procedures in hospital outpatient departments (HOPDs), freestanding ambulatory surgery centers (ASCs), and physician offices and investigated the factors associated with the use of different-day endoscopic procedures.

We evaluated the 100% Medicare fee-for-service carrier claims between January 1, 2011, and June 30, 2018. Using Current Procedural Terminology (CPT) and Healthcare Common Procedure Coding System (HCPCS) codes, we identified upper and lower endoscopic procedures (eTable 1 and eTable 2 in the Supplement). We included patients who underwent both elective upper and lower endoscopic procedures within a 90-day period in HOPDs, ASCs, and physician offices (eFigure in the Supplement). Based on the assumption that a new condition was unlikely to develop during a short period, we focused on elective upper and lower endoscopic procedures performed within 90 days of each other. This study was approved by the Johns Hopkins University School of Medicine Institutional Review Board. Patient consent was not required because only administrative claims data were used.

For patients who had a same-day procedure preceded or followed by another upper or lower endoscopic procedure on a different date, we excluded the preceding or following single procedure so that these patients only contributed 1 same-day paired procedure to the analysis. We also excluded patients who had procedures on at least 3 different days within a 90-day timeframe because they might have more complex clinical conditions. We excluded patients whose first procedure occurred in the last 90 days of the study period so as to fully capture the subsequent paired procedures. Finally, we excluded patients who had not been enrolled for at least 12 months in Medicare Parts A and B before the procedure date so as to have all claims available for identifying comorbidities, patients with missing demographic information, and patients younger than 18 years.

Patients’ demographic information, including age, sex, race/ethnicity, and zip code of residence, were obtained from the Master Beneficiary Summary Files of Medicare data. Patients’ zip codes were mapped to Federal Information Processing Standard codes to determine state and region of residence. Using a “crosswalk” (a data file that maps the elements in one system to the corresponding elements in another system) provided by the National Bureau of Economic Research, Federal Information Processing Standard codes were further mapped to Core Based Statistical Area codes to determine the type of residence (metropolitan, micropolitan, and rural).¹¹ We reviewed inpatient, outpatient, and carrier claims within 1 year before the procedure date to determine a patient’s comorbidity status.¹² We used 31 comorbidity categories previously validated for predicting patient outcomes.^13-15 To define a comorbidity, we required at least 1 diagnosis from inpatient claims or at least 2 diagnoses recorded more than 30 days apart from outpatient and carrier claims.¹² The Elixhauser comorbidity score was calculated as the total number of comorbidities present.¹⁶

The place of service code and performing physician’s National Provider Identifier were available in carrier claims and were used to determine the setting and the endoscopist performing each procedure. For different-day procedures, we calculated the time between the first and second procedures. We used the principal diagnosis code (the International Classification of Diseases, Ninth Revision, Clinical Modification code and the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Clinical Modification code) on the procedural visit to determine the procedure indication (codes for screening or surveillance, 530.85/K22.7, 555/K50, 556/K51, and V01-V89/Z00-Z99; codes for gastrointestinal symptoms or functional disorders, 787, 789/R10-R19, 530.0/K22.0, 530.5/K22.4, 536/K30-K31, 564/K56, and K58-K59; codes for peptic ulcer or gastritis, 531-535/K25-K29; codes for anemia, 280-285/D50-D64; codes for gastrointestinal bleeding, 578/K92.0-K92.2; codes for gastrointestinal neoplasm, 239.0/D49.0, 150-159/C15-C26, and 210-211/D10-D13; codes for esophageal disease other than Barrett esophagus, 530.0-530.7, 530.81-530.84, 530.86-530.89, 530.9/K20-K21, K22.0-K22.6, K22.8-K22.9, and K23; and codes for intestinal diverticular disease, 562/K57). For a lower endoscopy, for which the CPT and HCPCS codes represented screening or surveillance procedures (HCPCS codes G0105 and G0121 or other colonoscopy codes billed with a modifier PT or 33), the indication was reclassified as screening or surveillance regardless of the principal diagnosis. The indication of upper endoscopy was determined exclusively by the principal diagnosis.

We obtained physician information from Medicare Data on Provider Practice and Specialty and from the Physician Compare National Downloadable File.^17,18 Physician characteristics of interest included sex, year of graduation from medical school, primary specialty, practice setting and region, and procedure volume during the study period.

The rate of different-day procedures was calculated by dividing the volume of different-day procedures by the sum of the volumes of different-day and same-day procedures. We plotted the annual rate of different-day procedures by practice setting (HOPDs, ASCs, and physician offices) between 2011 and 2018. We also plotted the distribution of the number of days between 2 procedures for different-day paired procedures by practice setting.

For physicians who performed more than 10 procedures in a given setting, we calculated their physician-level different-day procedure rate, which was a crude rate calculated by dividing a physician’s volume of different-day procedures by his or her sum of the volumes of same-day and different-day procedures. We defined an extreme outlier as a physician who had a rate of elective different-day procedures greater than 50%. This metric was intended to delineate inliers and outliers for quality improvement purposes and is not intended to grade physicians by their specific rate of different-day procedures. For this reason, risk adjustment was deliberately not performed, and, instead, differences in risk were built into a consensus threshold of what defined an outlier physician. We plotted the distribution of this physician-level metric by practice setting. For physicians who practiced at multiple settings, we examined whether their practice patterns varied based on the practice setting.

We estimated the annual incremental cost of avoidable different-day procedures based on the allowed charge amounts in Medicare 2017 claims. The procedure cost consisted of the endoscopist fee, anesthesia fee, pathology fee, and facility fee. We calculated the median spending per same-day paired procedures vs per different-day paired procedures by state and practice setting to use as standardized prices.¹ We then calculated the number of avoidable different-day procedures in each setting (ASCs and physician offices) by state, using the HOPD’s rate of different-day procedures as the reference. The incremental cost was calculated by multiplying the number of avoidable different-day procedures by the difference in standardized prices between different-day paired procedures and same-day paired procedures, summed across settings and states. Using this state-specific and setting-specific standardized price, we accounted for the price variation across regions and settings.

We compared patient and physician characteristics between HOPDs, ASCs, and physician offices using Pearson χ² tests for categorical variables and 1-way analysis of variance for continuous variables. We used generalized estimating equations to fit a multivariable logistic regression model that examined patient and physician characteristics associated with the use of a different-day procedure. To account for correlation of errors within a facility, we calculated cluster-robust SEs at the facility level, which was robust to the nested correlation structure within cluster units.^19,20 The model covariates included patient characteristics, such as age, sex, race/ethnicity, residence location (metropolitan, micropolitan, and rural), region, Elixhauser comorbidity score, place of service, and procedure indication, as well as physician characteristics, such as sex, years since medical school graduation, procedure volume, and primary specialty. We included procedure year as a continuous covariate in the model to estimate the change in the use of different-day procedures over time.

The outcome in our main model was undergoing upper and lower endoscopic procedures 1 to 90 days apart. We also examined a separate model with the outcome of undergoing upper and lower endoscopic procedures 1 to 30 days apart with the covariates unchanged. We repeated the 90-day and 30-day models for paired procedures in which either procedure was performed for screening or surveillance because these were the most common clinical scenarios for which same-day procedures were highly indicated. All statistical analyses were performed using SAS Enterprise Guide, version 7.1 (SAS Institute Inc). All P values were from 2-sided tests and results were deemed statistically significant at P < .05.

We identified 4 028 587 paired upper and lower endoscopic procedures performed within 90 days of each other for Medicare beneficiaries between January 1, 2011, and June 30, 2018, of which 52.5% were performed in HOPDs, 43.3% in ASCs, and 4.2% in physician offices. The overall rate of different-day procedures was 18.7%. Compared with HOPDs, physician offices had a 3.5-fold higher rate of different-day procedures, and ASCs had a 1.6-fold higher rate of different-day procedures (47.7% in physician office vs 22.2% in ASCs vs 13.6% in HOPDs). The rate of different-day procedures decreased from 2011 to 2018, decreasing from 50.9% to 45.1% at physician offices, from 26.0% to 19.2% at ASCs, and from 16.8% to 11.5% at HOPDs (Figure 1).

Patients at ASCs and physician offices were slightly older compared with those at HOPDs (median age, 71.3 years at ASCs vs 71.1 years a physician offices vs 71.0 years at HOPDs) (Table 1). Patients at physician offices were more likely to be Asian (6.9% at physician offices vs 2.5% at ASCs vs 1.6% at HOPDs), reside in metropolitan areas (89.1% at physician offices vs 86.7% at ASCs vs 71.8% at HOPDs) and in the Northeast region (50.3% at physician offices vs 15.1% at ASCs vs 16.0% at HOPDs), and have fewer comorbidities (median Elixhauser comorbidity score, 2 at physician offices vs 2 at ASCs vs 3 at HOPDs) compared with patients at ASCs and HOPDs. For different-day procedures, the median time between the first and the second procedure was the shortest at physician offices and the longest at HOPDs (16 days at physician offices vs 21 days at ASCs vs 28 days at HOPDs). At physician offices, 68.1% of the different-day procedures were performed 1 to 30 days apart, compared with 61.4% at ASCs and 52.2% at HOPDs. In all 3 settings, we found the number of days apart to be multiples of 7 (ie, 1 week) for about 37% of the different-day procedures (37.0% for ASCs, 36.3% for HOPDs, and 36.8% for physician offices) (Figure 2). The elective upper endoscopic procedure preceded the elective lower endoscopic procedure for most different-day paired procedures. For both same-day and different-day paired procedures, the lower endoscopic procedures were most commonly performed for colorectal cancer screening or disease surveillance, and the upper endoscopic procedures were most commonly performed for esophageal diseases, peptic ulcer or gastritis, or other symptomatic or functional gastrointestinal disorders.

Physician-level rates of different-day procedures were the highest at physician offices and lowest at HOPDs (Figure 3). There were 1587 physicians who performed more than 10 paired procedures at physician offices, of whom 44.0% were extreme outliers (ie, a rate of different-day procedures >50%), 27.3% had a rate of different-day procedures greater than 90%, and 11.6% had a rate of different-day procedures of 100%. For the 9764 physicians at ASCs, 15.9% were extreme outliers, and 4.7% had a rate of different-day procedures greater than 90%. For the 17 792 physicians at HOPDs, 6.3% were extreme outliers, and 1.3% had a rate of different-day procedures greater than 90%.

Even the same physician’s practice pattern varied based on the practice setting. For the 7564 physicians who practiced at both ASCs and HOPDs, their practice at ASCs demonstrated a 19.4% rate of different-day procedures compared with 14.1% at HOPDs. For the 993 physicians who practiced at both physician offices and HOPDs, their practice at physician offices demonstrated a 37.4% rate of different-day procedures compared with 15.8% at HOPDs.

Even after adjustment for patient and physician characteristics, the odds of undergoing different-day procedures was still twice as high for patients treated at physician offices than those treated at HOPDs (adjusted odds ratio [aOR], 2.02; 95% CI, 1.85-2.20) and 1.3 times as high for patients treated at ASCs than those treated at HOPDs (aOR, 1.27; 95% CI, 1.23-1.32). The results remained similar when we restricted the model to paired procedures in which either procedure was performed for screening or surveillance (Table 2). Other prominent patient-level independent risk factors for undergoing different-day procedures included being 85 years of age or older vs 65 to 74 years (85-94 years: aOR, 1.10; 95% CI, 1.08-1.11; ≥95 years: aOR, 1.14; 95% CI, 1.03-1.26), black or Hispanic race/ethnicity vs white (black: aOR, 1.15; 95% CI, 1.12-1.17; Hispanic: aOR, 1.12; 95% CI, 1.10-1.14), residing in the Northeast or other regions vs the South (Northeast: aOR, 1.32; 95% CI, 1.28-1.36; other regions: aOR, 1.42; 95% CI, 1.31-1.54), and upper endoscopy indications of gastrointestinal bleeding (aOR, 1.25; 95% CI, 1.20-1.30) or gastrointestinal symptoms or functional disorder (aOR, 1.17; 95% CI, 1.15-1.19). Physician characteristics associated with decreased odds of different-day procedures included female sex (aOR, 0.95; 95% CI, 0.92-0.99), higher procedure volume (51-200 vs ≤50: aOR, 0.75; 95% CI, 0.72-0.78; ≥201 vs ≤50: aOR, 0.65; 95% CI, 0.62-0.68), and primary specialty of general (including colorectal) surgery vs gastroenterology (aOR, 0.86; 95% CI, 0.80-0.91).

In 2017, 47 292 different-day paired procedures were performed in ASCs, and 9722 different-day paired procedures were performed in physician offices. Had ASCs and physician offices performed their procedure pairs at HOPDs’ rate of different-day procedures (11.2% in 2017), ASCs would have avoided 19 776 different-day paired procedures, and physician offices would have avoided 7346 different-day paired procedures. Using the state-specific and setting-specific standardized spending of a same-day procedure pair as the reference price, we found that the incremental cost associated with avoidable different-day paired procedures was $9 246 341 at ASCs and $3 345 870 at physician offices. The estimated cost included the Medicare-allowed amount for the endoscopist fee, anesthesia fee, pathology fee, and facility fee but did not include other costs, such as the patients’ and designated drivers’ lost time from work.

Reducing low-value care in favor of patient-centered, high-value care is a growing priority in addressing increasing health care costs.^2,4,21,22 Our study highlights a potential area of health care waste around a very common set of procedures that is highly measurable and addressable. Using a claims-based outlier metric of low-value care, outlier practice patterns could be identified for a more exhaustive review using clinical data. We do not suggest that all elective upper and lower endoscopic procedures performed within 90 days of each other be performed on the same day, nor do we suggest that physicians should be evaluated by their percentage of same-day and different-day endoscopic procedures; instead, we suggest that there are extreme outlier physician practice patterns that should be evaluated more closely and that these practice patterns should possibly be the focus of an educational or peer-to-peer notification.

Our study observed a sustained higher rate of different-day endoscopic procedures at physician offices and ASCs compared with HOPDs. Although rates of different-day procedures have consistently decreased during the past 8 years across all 3 settings, probably owing to increased awareness to deliver patient-centered care or increased anesthesia safety,^23,24 as of 2018, physician offices were still performing different-day procedures at 3.9 times the rate as that of HOPDs, and ASCs were still performing different-day procedures at 1.7 times the rate as that of HOPDs. Physicians changed their practice pattern based on practice settings, exhibiting a 2.4-fold higher different-day procedure rate in their physician office practice (37.4%) than in their HOPD practice (15.8%) and a 1.4-fold higher rate in their ASC practice (19.4%) than in their HOPD practice (14.1%). A patient’s odds of undergoing a different-day procedure increased 2-fold at physician offices and 1.3-fold at ASCs compared with HOPDs, even after adjusting for patient and physician characteristics.

These findings suggest that the disparity is measurable in a large claims database and thus amenable to quality improvement collaboratives and other efforts, including payment reform, which could be applied to extreme outlier practice patterns. Current Medicare reimbursement applies a multiple procedure payment reduction for upper and lower endoscopy performed on the same day, resulting in a lower reimbursement rate for same-day procedures. This reimbursement mechanism may be a financial incentive driving the disparity we observed in practice pattern by practice setting. Although the greater profit motive at physician offices and ASCs might be the most likely driver behind their higher rate of different-day endoscopies, there may be other reasons, such as limited availability of devices and scheduling of endoscopists who focus on 1 type of endoscopy.

Two previous reports described the use of same-day vs different-day upper and lower endoscopic procedures using population-based data in the United States. One study was based on a 5% random sample of Medicare claims between 2007 and 2009 and found that, for paired upper and lower endoscopic procedures performed within 180 days, 35.3% of the procedures were performed on different days, and the rate of different-day procedures varied by geographical region.⁵ This prior study found that the Northeast region had the highest rate of different-day procedures and the West had the lowest, which was consistent with our findings. The other study was based on the 2000-2004 Clinical Outcomes Research Initiative national endoscopic database and found that 11.2% of all patients undergoing an upper or lower endoscopic procedure had same-day procedures, but it did not report the frequency of different-day procedures.²⁵ Using the most recent data, we demonstrated that the practice patterns of same-day vs different-day endoscopic procedures varied by practice setting.

In addition to being associated with clinical waste, different-day upper and lower endoscopic procedures are also associated with avoidable patient inconvenience and risks of complications. Different-day procedures expose patients to the discomfort of a second venipuncture and increased risk of anesthesia-associated complications. Different-day procedures may also result in an increased rate of undetected adverse events, particularly if patients reside at a distance from the index hospital where their endoscopy procedure(s) were performed.²⁶ A previous study has reported that patients undergoing another endoscopic procedure within 30 days prior to an upper endoscopic procedure are predisposed to a 2.6-fold higher risk of infection and that those undergoing another endoscopic procedure within 30 days prior to a lower endoscopy are predisposed to a 1.5-fold higher risk of infection.²⁷ These underappreciated risks are an important reason why performing different-day procedures should be avoided when it is safe and clinically indicated to perform them on the same day.

This study has the limitations of claims data research (ie, it is a population-based rather than clinical-level study). Specifically, we are unable to capture patient factors that may warrant different-day procedures. However, we would expect patients with more complicated clinical conditions to have their procedures performed as inpatients or to present to HOPDs (higher acuity) rather than to physician offices or ASCs. Despite this limitation, we observed that HOPDs had the lowest rates of different-day endoscopic procedures, even with a presumed highest complexity of cases. Thus, the data that we report likely underestimated the true difference in rates if these clinical factors were available in this national data set. Second, this study was limited to Medicare beneficiaries and may not be generalizable to privately insured patient populations; however, it is quite likely that similar trends would be observed for these patients as well. Third, although our model results were adjusted for patient and physician characteristics that were potential confounders, we could not eliminate the possibility of unmeasured confounding.

Based on the findings of this study, we propose that outlier performance above a given consensus threshold rate of elective different-day endoscopic procedures within 90 days may be a meaningful measure of low-value care. Sharing individual performance reports with physicians about this metric is one potential intervention that has promise. Recent studies have shown that such peer comparison using data transparency can affect physician behavior and reduce clinical waste.^28-32 This approach can identify an extreme outlier pattern and can be used as a screening tool for a more exhaustive clinical data review.³³

Accepted for Publication: December 19, 2019.

Corresponding Author: Martin A. Makary, MD, MPH, Department of Surgery, Johns Hopkins University, Blalock 665, 600 N Wolfe St, Baltimore, MD 21287 (mmakary1@jhmi.edu).

Published Online: May 13, 2019. doi:10.1001/jamainternmed.2018.8766

Author Contribution: Dr Makary 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.

Concept and design: Wang, Shin, Hartman, Disney, Kalloo, Makary.

Acquisition, analysis, or interpretation of data: Wang, Hutfless, Fain, Bull-Henry, Daniels, Abdi, Singh.

Drafting of the manuscript: Wang, Shin, Hartman.

Critical revision of the manuscript for important intellectual content: Wang, Hutfless, Shin, Disney, Fain, Bull-Henry, Daniels, Abdi, Singh, Kalloo, Makary.

Statistical analysis: Wang, Daniels.

Obtained funding: Makary.

Administrative, technical, or material support: Shin, Hartman, Disney, Fain, Singh, Makary.

Supervision: Makary.

Conflict of Interest Disclosures: Dr Hutfless reported receiving nonfinancial support from the American Gastroenterology Association; and grants from Luitpold Pharmaceuticals, Medivators, and Consumer’s Checkbook outside the submitted work. Dr Singh reported receiving personal fees from AbbVie, Akcea Therapeutics, and Ariel Precision Medicine outside the submitted work. No other disclosures were reported.