Advertisement

Patient complexity by surgical specialty does not correlate with work relative value units

Published:April 24, 2020DOI:https://doi.org/10.1016/j.surg.2020.03.002

      Abstract

      Background

      Understanding the differences in how patient complexity varies across surgical specialties can inform policy decisions about appropriate resource allocation and reimbursement. This study evaluated variation in patient complexity across surgical specialties and the correlation between complexity and work relative value units.

      Study design

      The 2017 American College of Surgeons National Surgical Quality Improvement Program was queried for cases involving otolaryngology and general, neurologic, vascular, cardiac, thoracic, urologic, orthopedic, and plastic surgery. A total of 10 domains of patient complexity were measured: American Society of Anesthesiologists class ≥4, number of major comorbidities, emergency operation, major complications, concurrent procedures, additional procedures, length of stay, non-home discharge, readmission, and mortality. Specialties were ranked by their complexity domains and the domains summed to create an overall complexity score. Patient complexity then was evaluated for correlation with work relative value units.

      Results

      Overall, 936,496 cases were identified. Cardiac surgery had the greatest total complexity score and was most complex across 4 domains: American Society of Anesthesiologists class ≥4 (78.5%), 30-day mortality (3.4%), major complications (56.9%), and mean length of stay (9.8 days). Vascular surgery had the second greatest complexity score and ranked the greatest on the domains of major comorbidities (2.7 comorbidities) and 30-day readmissions (10.1%). The work relative value units did not correlate with overall complexity score (Spearman’s ρ = 0.07; P < .01). Although vascular surgery had the second most complex patients, it ranked fifth greatest in median work relative value units. Similarly, general surgery was the fifth most complex but had the second-least median work relative value units.

      Conclusion

      Substantial differences exist between patient complexity across specialties, which do not correlate with work relative value units. Physician effort is determined largely by patient complexity, which is not captured appropriately by the current work relative value units.

      Introduction

      Efficient health care delivery depends on matching resources with need. Patient complexity is a difficult-to-define conceptual framework, but complexity inherently implies increased resource utilization and effort to provide adequate care. Patient complexity has been defined as a dynamic state in which personal, social, and clinical aspects of the patient’s experience operate as complicating factors.
      • Shippee N.D.
      • Shah N.D.
      • May C.R.
      • Mair F.S.
      • Montori V.M.
      Cumulative complexity: a functional, patient-centered model of patient complexity can improve research and practice.
      Recent assessment of patient complexity across medical specialties suggests wide variation, which has ramifications for health care policy, education, and attributed resources.
      • Tonelli M.
      • Wiebe N.
      • Manns B.J.
      • et al.
      Comparison of the complexity of patients seen by different medical subspecialists in a universal health care system.
      Although major differences between medical specialties and surgical specialties exist, assessment of patient complexity stratified by surgical specialty has yet to be performed. This assessment is particularly important, because surgical care alone accounts for more than half of all Medicare spending and represents 20% of the total health care expenditure of the United States.
      • Kaye D.R.
      • Luckenbaugh A.N.
      • Oerline M.
      • et al.
      Understanding the costs associated with surgical care delivery in the medicare population.
      Therefore, defining relative patient complexity as it applies to surgical specialties is imperative to guide resource allocation of the health care system and to ensure appropriate reimbursement by payers, including the Centers for Medicare and Medicaid Services. Complex patients require greater resources through increased complications, readmissions, physician time, ancillary support, greater length of stay, and need for postdischarge rehabilitation and nursing care.
      Relative value units (RVUs) were created by the Centers for Medicare and Medicaid Services to provide a rank on a common scale of the resources used to provide each service. The RVU is the basis for the Medicare Physician Fee Schedule that is used to determine payments nationally. RVUs include the physician’s work, the expense to the physician’s practice, professional liability, and an adjustment for geographic variation.
      • S.C
      The Basics:.
      The physician’s work is captured by the work RVU (wRVU). The wRVU encompasses the time, technical skill and effort, mental effort and judgment, and stress to provide each service.
      • S.C
      The Basics:.
      Because patient complexity increases each component of the wRVU, it would follow that patient complexity should correlate closely with wRVUs. The RVU determination is assessed via the Relative Value Scale Update Committee (RUC), formulated by a group of physicians convened by the American Medical Association. The RUC often has been criticized, and recent evidence suggests that the RUC estimates on time spent performing various operations are innacurate.
      • Chan D.C.
      • Huynh J.
      • Studdert D.M.
      Accuracy of valuations of surgical procedures in the medicare fee schedule.
      RUC updates to RVUs are a zero-sum game and are based largely on expert opinion rather than on evidence, creating the potential for misappropriation of resources.
      • S.C
      The Basics:.
      Many health care systems have moved toward RVU-based compensation models aimed at incentivizing productivity. Understanding the interplay between patient complexity and RVUs helps determine whether there may be unintended consequences to these commonly used, RVU-based compensation models. Reimbursement fundamentally drives care delivery, and there is a potential to disincentivize care for complex patients who are not recognized adequately with wRVUs.
      • Schroeder S.A.
      • Frist W.
      National Commission on Physician Payment Reform
      Phasing out fee-for-service payment.
      Furthermore, specialties with complex patients may be wholly undervalued by wRVU, which could lead to decreased resources and worse care for the very patients who need it the most. Understanding patient complexity across surgical specialties is thus fundamental to providing equitable and efficient health care. This study aimed to define surgical patient complexity across multiple domains to help benchmark specialties and determine whether wRVU correlates with patient complexity. We hypothesized that patient complexity would vary between specialties and would correlate positively with wRVUs.

      Methods

      This retrospective cohort analysis is presented according to the reporting guidelines of the Strengthening the Reporting of Observational Studies in Epidemiology initiative.
      • von Elm E.
      • Altman D.G.
      • Egger M.
      • et al.
      The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies.
      This study did not contain any protected health information because of the de-identified nature of the data, which is not considered human research and therefore, was considered exempt by the institutional review board of the University of California, San Francisco.

      Data source and cohort

      We analyzed retrospectively the Participant User File of the 2017 American College of Surgeons National Surgical Quality Improvement Program (NSQIP). The NSQIP is a clinical database that collects prospective data on adult patients (≥18 years of age) from participating institutions. The database has been used heavily, and details on the sampling strategy, data collection, variables, and outcomes have been reported elsewhere.
      • Montroy J.
      • Breau R.H.
      • Cnossen S.
      • et al.
      Change in adverse events after enrollment in the national surgical quality improvement program: a systematic review and meta-analysis.
      Our analysis was restricted to patients undergoing a primary procedure with the following surgical specialties as defined within the database: otolaryngology (ENT) and general, neurologic, vascular, cardiac, thoracic, urologic, orthopedic, and plastic surgery. Because the NSQIP only collects data on adult patients, pediatric and congenital surgical patients were not included. The NSQIP does not collect data on transplant or trauma surgery patients. Therefore, these patients also are not included in this study or represented. Major comorbidities were grouped by system using previously described methods.
      • Iannuzzi J.C.
      • Deeb A.P.
      • Rickles A.S.
      • Sharma A.
      • Fleming F.J.
      • Monson J.R.
      Recognizing risk: bowel resection in the chronic renal failure population.
      ,
      • Iannuzzi J.C.
      • Chandra A.
      • Rickles A.S.
      • et al.
      Resident involvement is associated with worse outcomes after major lower extremity amputation.
      Major complications included postoperative venous thromboembolism, renal injury and failure, sepsis, deep tissue infection, failure to wean off of ventilator, reintubation, pneumonia, stroke, cardiac arrest, myocardial infarction, bleeding, and return to the operating room.
      • Boitano L.T.
      • Iannuzzi J.C.
      • Tanious A.
      • et al.
      Predicting postoperative destination through preoperative evaluation in elective open aortic aneurysm repair.
      All variables utilized had <10% of data missing. Missing data were grouped with the referent group to create a conservative estimate for all independent variables.

      Domains of complexity

      Currently, there are no accepted measures of surgical patient complexity.
      • Grant R.W.
      • Ashburner J.M.
      • Hong C.S.
      • Chang Y.
      • Barry M.J.
      • Atlas S.J.
      Defining patient complexity from the primary care physician's perspective: a cohort study.
      Surgical patient complexity in this study included assessments across the domains of surgical care, including preoperative factors, intraoperative measures, and postoperative outcomes. A total of 10 domains were evaluated (Fig 1). These domains were chosen to include metrics from all phases of surgical care, including patient characteristics, inpatient utilization, and postoperative outcomes. Preoperative measures included the American Society of Anesthesiologists (ASA) class ≥4, number of major comorbidities by system, and emergency operations. ASA class and the presence of major comorbidities are global indicators of a patient’s surgical and anesthetic risks and have been demonstrated to correlate positively with morbidity and mortality.
      • Hackett N.J.
      • De Oliveira G.S.
      • Jain U.K.
      • Kim J.Y.
      ASA class is a reliable independent predictor of medical complications and mortality following surgery.
      Emergency operations can be more physically and emotionally demanding, more technically complex, and generally have compromised outcomes.
      • Scott J.W.
      • Olufajo O.A.
      • Brat G.A.
      • et al.
      Use of national burden to define operative emergency general surgery.
      Operative measures included undergoing a concurrent procedure with a different specialty, requiring an additional procedure, and length of hospital stay. Requiring a concurrent or additional procedure indicates the requirement of multiple procedures and coordination among other teams or services, both of which can take substantial effort and work from the primary surgeon. Postoperative measures included non-home discharge to a rehabilitation or skilled nursing facility and 30-day rate of readmissions, major complications, and mortality. Non-home discharge demonstrates that the patient requires a high level of care or rehabilitation and also can complicate discharge planning.
      • Hakkarainen T.W.
      • Ayoung-Chee P.
      • Alfonso R.
      • Arbabi S.
      • Flum D.R.
      Structure, process, and outcomes in skilled nursing facilities: understanding what happens to surgical patients when they cannot go home. A systematic review.
      Readmissions and complications are linked directly to the primary surgeon and hospital and require additional effort, energy, and work. Furthermore, these metrics affect reimbursement and may play more substantial roles in future models of insurance reimbursement.
      Figure thumbnail gr1
      Fig 1Domains of complexity. Currently there are no accepted measures of patient complexity, and therefore, a broad spectrum of surgical care including preoperative, intraoperative, and postoperative measures were examined.

      Ranking and complexity score

      To examine how patient complexity differed by surgical specialty, specialties were first ranked from 1 (most complex) to 9 (least complex) by each individual domain of complexity. This was accomplished by first determining the proportion or mean for each domain (depending on whether a categorical or continuous measure) according to specialty. We then ranked the specialties from the greatest proportion or mean to the least. Points were assigned to each rank where the greatest ranked specialty received 9 points and the least ranked specialty received 1 point. This was performed for each domain of complexity. Points were then summed to create an overall complexity score, translating to an overall complexity rank. To examine how overall patient complexity for each specialty compared with general surgery, a ratio was created using the overall complexity score of general surgery as the referent.

      Work RVUs

      The primary wRVU was defined as collected in NSQIP, encompassing only the index operation. Surgical specialties then were ranked by median wRVUs. The correlation between median wRVU and patient complexity was assessed using Spearman’s rank correlation test. To examine how overall patient complexity correlated with wRVU, an observed-to-expected ratio for wRVU was created using the overall complexity score for the most complex surgical specialty. A ratio was created comparing each individual specialty with the most complex specialty. This was then multiplied by the median wRVUs for the most complex specialty to create an adjusted wRVU. The adjusted wRVU was then used to create the observed-to-expected ratio, using the median wRVU for each specialty.

      Statistical analysis

      All statistical analyses were performed using STATA v 15.0 (StataCorp LLC, College Station, TX, USA). Patients first were stratified by their primary operating specialty. Summary statistics were reported using median and interquartile range (IQR) for continuous variables and frequency and percentage for categorical variables. Differences between specialties were calculated using a X2 test for categorical variables and a Kruskal-Wallis test for continuous variables. Statistical significance was set at 2-sided P < .05. A 95% confidence interval (CI) was calculated for each estimated proportion. For each categorical domain of complexity, an unadjusted odds ratio was calculated using general surgery as the referent. A Spearman’s rank correlation was used to examine the correlation between primary wRVU and overall complexity score.

      Results

      Baseline characteristics

      Overall, 936,496 patients were identified. The cohort included 52.7% females, with 79.2% of the white race, and a median age of 59 years (IQR: 46–70) (Table I). After stratification by surgical specialty, vascular surgery patients had the greatest median age (68 years; IQR = 59–76); plastic surgery patients were most frequently female (82.6%), and cardiac surgery patients were most frequently white (83.1%) (Table I). Operative time differed across surgical specialties, with cardiac surgery having the greatest median operative time (225 minutes; IQR = 165–292), and general surgery having the least median operative time (75 minutes; IQR = 44–132).
      Table IPatient demographics and comorbidities by surgical specialty
      CardiacVascularThoracicNeurologicGeneralOrthopedicUrologyPlasticsENTP value
      Calculated using a χ2 test for categorical variables and a Kruskal-Wallis test for continuous variables.
      Demographics
       Count4,01256,98112,29854,856444,203243,99161,14031,12427,891<.001
       Age
      Reported as median (interquartile range).
      (y)
      66 (58–74)68 (59–76)65 (46–72)59 (48–69)56 (42–67)63 (52–72)66 (57–74)49 (38–59)49 (31–63)<.001
       White
      Reported as n (percent).
      3,334 (83.1%)45,040 (79.0%)10,180 (82.8%)45,553 (83.0%)339,172 (76.4%)202,531 (83.0%)50,273 (82.2%)23,732 (76.3%)21,495 (77.1%)<.001
       Female
      Reported as n (percent).
      1,186 (29.6%)22,260 (39.1%)5,990 (48.7%)26,075 (47.5%)252,229 (56.8%)131,742 (54.0%)11,940 (19.5%)25,710 (82.6%)16,492 (59.1%)<.001
       Operative time
      Reported as median (interquartile range).
      (min)
      225 (165–292)105 (66–167)125 (73–197)127 (82–196)75 (44–132)79 (54–109)90 (39–184)115 (66–189)81 (36–146)<.001
      Comorbidities
      Reported as n (percent).
       Pulmonary1,451 (36.2%)11,114 (19.5%)3,978 (32.4%)4,621 (8.4%)35,287 (7.9%)17,905 (7.3%)5,304 (8.7%)1,008 (3.2%)1,696 (6.1%)<.001
       Cardiac3,008 (75.0%)42,944 (75.4%)6,405 (52.1%)27,061 (49.3%)180,953 (40.7%)123,962 (50.8%)33,722 (55.2%)7,912 (25.4%)8,796 (31.5%)<.001
       Renal820 (20.4%)14,586 (25.6%)1,715 (14.0%)5,462 (10.0%)45,089 (10.2%)30,157 (12.4%)10,767 (17.6%)1,421 (4.6%)1,609 (5.8%)<.001
       Dialysis112 (2.8%)4,282 (7.5%)120 (1.0%)239 (0.4%)6,094 (1.4%)1,317 (0.5%)677 (1.1%)157 (0.5%)106 (0.4%)<.001
       Morbid obesity265 (6.6%)4,203 (7.4%)670 (5.5%)5,349 (9.8%)66,034 (14.9%)27,689 (11.4%)3,862 (6.3%)2,255 (7.3%)2,807 (10.1%)<.001
       Malnutrition60 (1.5%)2,086 (3.7%)759 (6.2%)923 (1.7%)14,277 (3.2%)4,103 (1.7%)1,042 (1.7%)520 (1.7%)670 (2.4%)<.001
       Diabetes mellitus1,284 (32.0%)19,381 (34.0%)2,058 (16.7%)9,548 (17.4%)65,628 (14.8%)37,223 (15.3%)11,502 (18.8%)2,674 (8.6%)3,077 (11.0%)<.001
       Ascites5 (0.1%)111 (0.2%)23 (0.2%)21 (0.04%)1,976 (0.4%)132 (0.1%)44 (0.1%)14 (0.04%)6 (0.02%)<.001
       Smoking798 (19.9%)18,569 (32.6%)3,369 (27.4%)11,567 (21.1%)74,237 (16.7%)34,358 (14.1%)9,791 (16.0%)3,955 (12.7%)4,967 (17.8%)<.001
       Steroids123 (3.1%)2,686 (4.7%)671 (5.5%)3,232 (5.9%)18,040 (4.1%)8,362 (3.4%)1,648 (2.7%)697 (2.2%)646 (2.3%)<.001
       Bleeding disorder405 (10.1%)12,665 (22.2%)442 (3.6%)1,231 (2.2%)14,509 (3.3%)8,331 (3.4%)1,557 (2.6%)454 (1.5%)348 (1.3%)<.001
       Anemia997 (24.9%)18,397 (32.3%)2,668 (21.7%)6,463 (11.8%)81,771 (18.4%)34,977 (14.3%)8,393 (13.7%)3,776 (12.1%)2,117 (7.6%)<.001
       Disseminated cancer30 (0.8%)382 (0.7%)1,367 (11.1%)2,420 (4.4%)13,177 (3.0%)1,748 (0.7%)1,685 (2.8%)244 (0.8%)475 (1.7%)<.001
       Functionally Dependent76 (1.9%)4,087 (7.2%)240 (1.9%)1,615 (2.9%)9,269 (2.1%)9,083 (3.7%)1,149 (1.9%)533 (1.7%)248 (0.9%)<.001
      Calculated using a χ2 test for categorical variables and a Kruskal-Wallis test for continuous variables.
      Reported as median (interquartile range).
      Reported as n (percent).

      Domains of complexity by specialty

      When domains of patient complexity were considered individually, there were statistically significant differences between surgical specialties (Tables II and III). Vascular and cardiac surgery patients were consistently the most complex across a majority of the domains examined. Cardiac surgery was most complex across 4 individual domains: ASA class ≥4 (78.5% versus least ranked specialty in this category [least] 1.6%), mean length of hospital stay (9.8 days versus [least] 1.5 days), major complications (56.9% versus [least] 5.1%), and mortality (3.4% versus [least] 0.1%). When compared with general surgery patients, cardiac surgery patients had more than 65 times the odds of having an ASA class ≥4, 11 times the odds of having a major complication, and 3 times the odds of 30-day mortality. Vascular surgery patients were ranked greatest for number of major comorbidities (2.73; 95% CI = 2.72–2.74 versus [least] 0.82; 95% CI = 0.81–0.84) and readmissions (10.1%; 95% CI = 9.8–10.3% versus [least] 3.0%; 95% CI = 2.8–3.2%).
      Table IIPreoperative and intraoperative domains of complexity by surgical specialty
      ASA class ≥ 4Number of major comorbiditiesEmergent surgeryConcurrent procedureAdditional ProcedureMean Length of Hospital Stay (days)
      Proportion (95% CI)Unadjusted OR (95% CI)
      Reported with general surgery as referent.
      P valueMean (95% CI)P valueProportion (95% CI)Unadjusted OR (95% CI)
      Reported with general surgery as referent.
      P valueProportion (95% CI)Unadjusted OR (95% CI)
      Reported with general surgery as referent.
      P valueProportion (95% CI)Unadjusted OR (95% CI)
      Reported with general surgery as referent.
      P valueMean (95% CI)P Value
      Cardiac78.5% (77.2–79.8)65.1 (60.3–70.3)<.0012.35 (2.31–2.40)<.0019.2% (8.3–10.1)0.65 (0.58–0.72)<.0011.9% (1.5–2.4)0.37 (0.29–0.46)<.00160.7% (59.2–62.3)3.20 (3.00–3.41)<.0019.8 (9.5–10.1)<.001
      Vascular27.0% (26.7–27.4)6.62 (6.47–6.77)<.0012.73 (2.72–2.74)<.0018.2% (8.0–8.4)0.57 (0.55–0.59)<.0011.5% (1.4–1.6)0.28 (0.26–0.30)<.00137.6% (37.2–38.0)1.25 (1.23–1.27)<.0015.0 (4.9–5.1)<.001
      Thoracic11.7% (11.2–12.3)2.37 (2.24–2.51)<.0011.99 (1.97–2.02)<.0012.4% (2.2–2.7)0.16 (0.14–0.18)<.0012.7% (2.5–3.0)0.533 (0.48–0.60)<.00163.1% (62.2–64.0)3.54 (3.41–3.67)<.0016.1 (5.9–6.2)<.001
      Neurologic6.8% (6.6–7.0)1.30 (1.25–1.35)<.0011.45 (1.44–1.46)<.0014.7% (4.5–4.9)0.32 (0.30–0.33)<.0012.6% (2.4–2.7)0.50 (0.47–0.53)<.00161.8% (61.4–62.3)3.36 (3.29–3.42)<.0014.0 (3.9–4.1)<.001
      General5.3% (5.2–5.4)Referent1.410 (1.406–1.414)<.00113.5% (13.4–13.6)Referent5.0% (4.9–5.1)Referent32.6% (32.4–32.7)Referent3.29 (3.27–3.31)<.001
      Orthopedic4.0% (3.9–4.1)0.75 (0.73–0.76)<.0011.391 (1.386–1.396)<.0014.8% (4.7–4.9)0.323 (0.316–0.329)<.0010.83% (0.79–0.86)0.16 (0.15–0.17)<.00122.7% (22.6–22.9)0.61 (0.60–0.62)<.0012.44 (2.42–2.45)<.001
      Urology4.4% (4.2–4.5)0.81 (0.78–0.85)<.0011.49 (1.48–1.50)<.0011.8% (1.6–1.9)0.114 (0.107–0.121)<.0012.6% (2.5–2.8)0.51 (0.49–0.54)<.00130.7% (30.4–31.1)0.92 (0.90–0.94)<.0012.21 (2.18–2.24)<.001
      Plastics1.6% (1.4–1.7)0.29 (0.26–0.31)<.0010.82 (0.81–0.84)<.0011.4% (1.2–1.5)0.09 (0.08–0.10)<.00114.3% (13.9–14.7)3.17 (3.06–3.28)<.00163.6% (63.0–64.1)3.61 (3.53–3.70)<.0011.54 (1.48–1.59)<.001
      ENT2.4% (2.2–2.6)0.44 (0.40–0.47)<.0010.99 (0.97–1.00)<.0011.4% (1.3–1.6)0.09 (0.08–0.10)<.0013.4% (3.1–3.6)0.66 (0.62–0.70)<.00130.6% (30.1–31.2)0.91 (0.89–0.94)<.0011.50 (1.46–1.55)<.001
      ASA, American Society of Anesthesiology; CI, confidence interval; OR, odds ratio.
      Reported with general surgery as referent.
      Table IIIPostoperative domains of complexity by surgical specialty
      Non-home discharge30-day major complication30-day readmission30-day mortality
      Proportion (95% CI)Unadjusted OR (95% CI)
      Reported with general surgery as referent.
      P valueProportion (95% CI)Unadjusted OR (95% CI)
      Reported with general surgery as referent.
      P valueProportion (95% CI)Unadjusted OR (95% CI)
      Reported with general surgery as referent.
      P valueProportion (95% CI)Unadjusted OR (95% CI)
      Reported with general surgery as referent.
      P value
      Cardiac14.8% (13.7–15.9)3.86 (3.53–4.22)<.00156.9% (55.3–58.4)11.2 (10.5–12.0)<.0018.7% (7.8–9.6)1.61 (1.44–1.80)<.0013.4% (2.9–4.0)3.10 (2.61–3.69)<.001
      Vascular15.7% (15.4–16.0)4.15 (4.04–4.26)<.00121.8% (21.5–22.2)2.38 (2.33–2.43)<.00110.1% (9.8–10.3)1.89 (1.83–1.95)<.0012.6% (2.5–2.8)2.38 (2.25–2.53)<.001
      Thoracic6.6% (6.1–7.0)1.56 (1.45–1.68)<.00117.4% (16.7–18.0)1.79 (1.71–1.88)<.0017.7% (7.2–8.1%)1.40 (1.31–1.50)<.0012.0% (1.7–2.2)1.76 (1.54–2.00)<.001
      Neurologic15.2% (14.9–15.5)3.99 (3.88–4.10)<.00110.9% (10.6–11.2)1.04 (1.01–1.07).0055.8% (5.6–6.0)1.04 (1.00–1.08).0471.2% (1.1–1.3)1.05 (0.96–1.14)0.268
      General4.3% (4.2–4.4)Referent10.5% (10.4–10.6)Referent5.6% (5.5–5.7)Referent1.12% (1.09–1.15)Referent
      Orthopedic17.3% (17.2–17.5)4.68 (4.59–4.76)<.0018.3% (8.2–8.4)0.77 (0.75–0.78)<.0013.47% (3.40–3.54)0.61 (0.59–0.62)<.0010.71% (0.67–0.74)0.63 (0.59–0.66)<.001
      Urology2.4% (2.3–2.5)0.54 (0.51–0.57)<.0019.2% (9.0–9.5)0.86 (0.84–0.89)<.0016.1% (5.9–6.3)1.10 (1.06–1.14)<.0010.47% (0.42–0.53)0.42 (0.37–0.47)<.001
      Plastics2.1% (2.0–2.3)0.49 (0.45–0.53)<.0016.0% (5.8–6.3%)0.55 (0.52–0.57)<.0013.2% (3.0–3.4)0.56 (0.52–0.60)<.0010.11% (0.08–0.016)0.10 (0.07–0.14)<.001
      ENT1.5% (1.3–1.6)0.33 (0.30–0.37)<.0015.1% (4.9–5.4)0.46 (0.44–0.49)<.0013.0% (2.8–3.2)0.52 (0.49–0.56)<.0010.20% (0.15–0.26)0.18 (0.14–0.23)<.001
      CI, confidence interval; OR, odds ratio.
      Reported with general surgery as referent.
      Orthopedic surgery patients required the most postoperative rehabilitation services, with the greatest proportion of patients requiring non-home discharge (17.3%; 95% CI = 17.2–17.5% versus [least] 1.5%; 95% CI = 1.3–1.6%). Plastic surgery patients had the greatest proportion of patients requiring a concurrent procedure (14.3%; 95% CI = 13.9–14.7% versus [least] 0.83%; 95% CI = 0.79–0.86%) and an additional procedure (63.6%; 95% CI = 63.0–64.1% versus [least] 22.7%; 95% CI = 22.6–22.9%). General surgery patients had the greatest proportion of patients undergoing an emergency operation (13.5%; 95% CI = 13.4–13.6% versus [least] 1.4%; 95% CI = 1.2–1.5%).

      Overall complexity score and rank

      The individual rankings for each domain of complexity were then summed to create an overall complexity score, which was utilized to create an overall complexity rank (Fig 2). When specialty type was ranked across all 10 complexity domains, the order from most to least complex was cardiac surgery, vascular surgery, thoracic surgery, neurologic surgery, general surgery, urologic surgery, orthopedic surgery, ENT, and plastic surgery (Table IV).
      Figure thumbnail gr2
      Fig 2Rankings by each domain of complexity and overall ranking. The individual rankings for each domain of complexity were summed to create an overall complexity score, which was utilized to create an overall rank.
      Table IVSurgical specialties ranked by overall complexity score
      Overall complexity score
      Specialties were ranked by individual complexity domains and the markers summed to create the overall complexity score.
      Ratio of complexity score
      Created using the overall complexity score of general surgery as the referent.
      1. Cardiac751.42
      2. Vascular711.34
      3. Thoracic661.25
      4. Neurologic571.08
      5. General531.00 (Referent)
      6. Orthopedic400.70
      7. Urology370.75
      8. Plastics300.57
      9. ENT210.40
      Specialties were ranked by individual complexity domains and the markers summed to create the overall complexity score.
      Created using the overall complexity score of general surgery as the referent.
      Several sensitivity analyses examining other potential domains of patient complexity were completed. A sensitivity analysis also including operative time as a domain of complexity did not alter the rank order. A sensitivity analysis ranking specialty by the 5-item modified frailty index yielded a ranking that was very similar to the “Major Comorbidities” ranking (Supplemental Table I).
      • Subramaniam S.
      • Aalberg J.J.
      • Soriano R.P.
      • Divino C.M.
      New 5-factor modified frailty index using American College of Surgeons NSQIP data.
      A sensitivity analysis including patient age as a domain of complexity resulted in a single change to the overall ranking, with urology and orthopedic surgery switching ranks so that urology was ranked sixth and orthopedic surgery ranked seventh (Supplemental Table II). The lack of any substantial change in the overall ranking suggests that patient age in and of itself may not be an adequate marker of complexity and that comorbidities and ASA class are better markers of patient physiology and complexity. It was, therefore, not included as a final domain of complexity. Because of variables that are unavailable in the NSQIP, sensitivity analyses including rankings using the Charlson or Elixhauser comorbidity indices were unable to be completed.

      Work relative value units

      On Spearman’s rank test, there was no correlation between overall complexity score and median wRVU (Spearman’s ρ=0.07; P < .01). Specialties then were ranked by median wRVUs (Table V). Consistent with the overall complexity ranking, cardiac surgery had the greatest median wRVUs (33; IQR: 33.1–43.3), and ENT had the least (11.2; IQR: 4.4–15.6). Although top and bottom median wRVUs were conserved, vascular surgery had the second greatest overall complexity but only ranked fifth in median wRVUs. The discrepancy between patient complexity and wRVUs also was apparent for general surgery, which ranked fifth in overall complexity but was ranked eighth in median wRVUs.
      Table VSurgical specialties ranked by work relative value units (wRVU)
      Median wRVU (IQR)Observed/expected ratio of wRVU
      Calculated utilizing the overall complexity score and median wRVU of cardiac surgery.
      1. Cardiac33.8 (33.1–43.3)1.00 (Referent)
      2. Thoracic23.5 (14.5–25.8)0.79
      2. Neurological23.5 (15.4–27.5)0.91
      4. Orthopedic20.7 (12.4–20.7)1.15
      5. Vascular20.5 (10.2–22.5)0.64
      6. Plastics15.9 (10.4–17.1)1.18
      7. Urology15.3 (8.0–26.8)0.92
      8. General11.9 (9.5–20.8)0.50
      9. ENT11.2 (4.4–15.6)1.18
      IQR, interquartile range; wRVU, work relative value units.
      Calculated utilizing the overall complexity score and median wRVU of cardiac surgery.
      An analysis of the observed-to-expected ratio of wRVUs demonstrated that vascular and general surgery were outliers with observed-to-expected ratios of 0.64 and 0.50, respectively (Table V). Other specialties, including orthopedic surgery, plastic surgery, and ENT, were high outliers with observed-to-expected ratios of 1.15, 1.18, and 1.18, respectively.

      Discussion

      This study evaluated surgical patient complexity across 9 surgical specialties using 10 domains for patient complexity. The overall complexity score demonstrated that cardiac surgery had the greatest complexity across patient domains captured in NSQIP. The complexity score was conserved largely across domains, with little variation in specialty rank. Across specialties, there was wide variation in patient complexity as assessed using the current scoring system. This variation becomes concerning, because wRVUs did not correlate with overall complexity, grossly undervaluing general and vascular surgery. General surgery, on average, was undervalued by 50%, and vascular surgery was undervalued by 36%. This analysis suggests these specialties might require further adjustment of their assigned wRVUs and increased recognition regarding the incurred health care burden for these specialties.
      The conceptual framework regarding patient complexity is one fraught with difficulty. Although the specific components and weight of these individual domains are debatable, this study found that various domains of complexity were largely conserved. Compensation models using RVUs have been rolled out across the country despite major criticisms about how RVUs incentivize volume over value, all of which contributes to physician burnout.
      • Luong P.
      • Bojansky A.M.
      • Kalra A.
      Academic physician compensation in the United States: should providers’ work at academic medical centres be judged by just one metric, the relative value unit (RVU)?.
      There have been calls for major changes to RVUs as well, because they do not incorporate time spent with challenging patients, education, and any effort that is not specifically captured in the RVU.
      • Luong P.
      • Bojansky A.M.
      • Kalra A.
      Academic physician compensation in the United States: should providers’ work at academic medical centres be judged by just one metric, the relative value unit (RVU)?.
      ,
      • Moazzez A.
      • Ozao-Choy J.
      • de Virgilio C.
      Need for more objective data to determine relative value units for surgical effort [e-pu ahead of print].
      Incentivization through RVUs can change practice, as has been observed with the placement of inferior vena cava filters with a drastic decrease in placement after bundled payments decreased the inferior vena cava filter RVU.
      • Glocker R.J.
      • TerBush M.J.
      • Hill E.L.
      • et al.
      Bundling of reimbursement for inferior vena cava filter placement resulted in significantly decreased utilization between 2012 and 2014.
      Calls to change practice based on RVU/minute also have been made in hip arthroplasty, where redo arthroplasty had considerably decreased RVUs/minute compared with the primary procedure, and the authors
      • Sodhi N.
      • Dalton S.E.
      • Garbarino L.J.
      • et al.
      Not all primary total hip arthroplasties are equal—So is there a difference in reimbursement?.
      suggest that because this finding could encourage orthopedic surgeons to avoid complex redo procedures. This type of incentivization can be bad for patients if not monitored closely and evaluated.
      Discrepancies between RUC estimated operative times and measured operative times across surgical specialties has identified that current RVU allocation may not be appropriate. Chan et al
      • Chan D.C.
      • Huynh J.
      • Studdert D.M.
      Accuracy of valuations of surgical procedures in the medicare fee schedule.
      assessed operative time estimates in NSQIP compared with RUC estimates and found that vascular surgery was undervalued in the amount of US$30 million in missed revenue. There was also an underestimate of time required in cardiac surgery, equating to US$130 million of missed revenue. In contrast to the current study, there was no apparent discrepancy for general surgery, which might in part be attributable to the large variation in types of procedures being performed under the umbrella of general surgery and associated variation in operative times. Given the breadth of general surgery, future studies should further examine how patient complexity varies within general surgery or across operations commonly performed by general surgeons and how these correlate with wRVUs. Regardless, it is still concerning that such a broad and varied specialty is so undervalued.
      We expected cardiac and vascular surgery most likely to be in the top 2 ranking positions for the majority of complexity domains, including ASA class. ASA classification has been reported to be an important predictor of morbidity and mortality and is a measure of a patient’s overall fitness for surgery.
      • Hackett N.J.
      • De Oliveira G.S.
      • Jain U.K.
      • Kim J.Y.
      ASA class is a reliable independent predictor of medical complications and mortality following surgery.
      In the current study, cardiac and vascular surgery had the greatest proportion of patients with ASA class 4 or greater, which includes patients with severe systemic disease that is a constant threat to life and moribund patients who are not expected to survive without the operation. A total of 79% of cardiac surgery patients had an ASA class 4 or greater, which is 3 times greater than the next greatest group (vascular surgery). This is not surprising, because ASA 4 or greater includes patients with coronary artery disease/stents, ongoing cardiac ischemia, severe valve dysfunction, and a severe decrease in ejection fraction, which are all common indications for cardiac surgery. Vascular surgery had the greatest mean number of comorbidities as grouped by major organ systems, which may be explained by the fact that atherosclerosis is a fundamentally systemic process and is also a common indication for vascular surgery. The number of comorbidities has been associated with poor surgical outcomes, and the top 3 complex specialties by number of comorbidities were the top 3 ranked groups for major complications, readmissions, and mortality.
      • Quail J.M.
      • Lix L.M.
      • Osman B.A.
      • Teare G.F.
      Comparing comorbidity measures for predicting mortality and hospitalization in three population-based cohorts.
      Although cardiac surgery was the most complex, it appears that cardiac surgeons may be recognized for these efforts, as evidenced by being ranked greatest for median wRVUs. In contrast, general and vascular surgery were largely undervalued by adjusted wRVU and received 50% and 36% fewer RVUs, respectively, than would be expected based on the patient complexity score. This finding is troubling, given that vascular surgery has been found to be undervalued based on operative time (1.5%, equating to US$30 million less in payments) and also represented the oldest patient population in the current study.
      • Chan D.C.
      • Huynh J.
      • Studdert D.M.
      Accuracy of valuations of surgical procedures in the medicare fee schedule.
      Vascular surgeons have been reported to spend the most time working on patient-related activities compared with any other medical or surgical specialty.
      • Leigh J.P.
      • Tancredi D.
      • Jerant A.
      • Kravitz R.L.
      Annual work hours across physician specialties.
      This observation has important implications for the future workforce, because there is already a reported shortage of general and vascular surgeons.
      • Satiani B.
      • Williams T.E.
      • Go M.R.
      Predicted shortage of vascular surgeons in the United States: population and workload analysis.
      ,
      • Ellison E.C.
      • Pawlik T.M.
      • Way D.P.
      • Satiani B.
      • Williams T.E.
      Ten-year reassessment of the shortage of general surgeons: Increases in graduation numbers of general surgery residents are insufficient to meet the future demand for general surgeons.
      The relatively low reimbursements despite high work load, complex patients, and time required to care for these patients may impede future recruitment efforts and further exacerbate shortages in general and vascular surgery. Addressing this discrepancy will be important to ensure appropriate vascular surgery coverage going forward and may be an opportunity to improve health care efficiency by devoting more resources to improving care for vascular surgery patients.
      General surgery had the greatest proportion of cases that were emergency and a younger patient population, likely reflecting the inclusion of acute care surgery services within general surgery. Although there is a high degree of variation within general surgery, it remained the most undervalued specialty, because adjusted wRVUs appeared to be half of what they would otherwise be expected based on patient complexity. Study of emergency biliary, hernia, and colorectal operations (likely reflective of acute care surgery) has demonstrated that wRVUs for emergency procedures, although fraught with greater operative times, lengths of stay, and postoperative complications, were not substantially different than their elective case counterparts.
      • Schwartz D.A.
      • Hui X.
      • Velopulos C.G.
      • et al.
      Does relative value unit-based compensation shortchange the acute care surgeon?.
      This finding suggests that acute care surgeons may not be adequately recognized for their work in challenging clinical scenarios. Poor reimbursement of acute care surgery may disincentivize trainees from becoming acute care surgeons or taking emergency general surgery call, which potentially may contribute to the shortage of acute care surgeons and may worsen access to emergency general surgery in low-education, underserved, and rural communities.
      • Khubchandani J.A.
      • Shen C.
      • Ayturk D.
      • Kiefe C.I.
      • Santry H.P.
      Disparities in access to emergency general surgery care in the United States.
      This lack of recognition around emergency cases likely explains some of the discrepancies observed in the current study between wRVU and overall complexity score. The discrepancy in wRVU and complexity, however, may be partially offset through shorter operative times. An analysis of 11 common general surgery procedures, ranging from a simple mastectomy to a total thyroidectomy and the Whipple procedure, demonstrated a poor correlation between wRVU and adverse events but a moderate correlation with median operative time (R2 = 0.76).
      • Shah D.R.
      • Bold R.J.
      • Yang A.D.
      • Khatri V.P.
      • Martinez S.R.
      • Canter R.J.
      Relative value units poorly correlate with measures of surgical effort and complexity.
      Earlier data, however, still suggest that Medicare reimbursement for common general surgery procedures has decreased steadily since the year 2000.
      • Haglin J.M.
      • Eltorai A.E.M.
      • Richter K.R.
      • Jogerst K.
      • Daniels A.H.
      Medicare reimbursement for general surgery procedures: 2000 to 2018.
      Although likely multifactorial, these results suggest that the discount in wRVU observed in the current study may be partially counterbalanced by decreased overall operative time compared with other specialties, because general surgery had the least operative time of any specialty.
      Orthopedic, plastic, and ENT surgery all were overvalued by wRVU when compared with overall complexity score. Plastic surgery may even be erroneously overestimated in terms of patient complexity by inclusion of concurrent and additional procedures in the metrics for patient complexity. If concurrent and additional procedures were removed from the score, plastic surgery would have the least overall complexity score. When RUC estimates were compared with NSQIP benchmarks, plastic surgery was undervalued by 1.8%, and orthopedic surgery had the greatest discrepancy, receiving an additional US$160 million.
      • Chan D.C.
      • Huynh J.
      • Studdert D.M.
      Accuracy of valuations of surgical procedures in the medicare fee schedule.
      This overpayment alone represents 1.7% of total Medicare Part B payments, which is consistent with the current study, which overvalued orthopedic surgery wRVUs by 15%. Technical skill is included as a component of wRVU, and some have argued that these specialties are highly technical, which may make up for differences in patient complexity not captured in this analysis. But capturing relative technical skill is difficult and subjective, especially because time spent training is essentially conserved across surgical subspecialties. These results collectively provide further evidence that redistribution of health care resources may be necessary to ensure that certain specialties are not overly favored, which may result in disparities within the workforce followed by inadequate or absent care.
      This study has some limitations. Although comorbidity indices have quantitated a patient’s risk of morbidity and mortality, there are no validated measures of surgical patient complexity that better reflect the work, energy, and cost that goes into taking care of patients. Clinical experience was utilized to select preoperative, intraoperative, and postoperative variables that would correlate with the level of surgical care required and overall perioperative resource utilization. Our proposed scoring system admittedly has limitations, including the absence of socioeconomic status, frailty, and complex care coordination, none of which were measured in this study but which can increase health care burden. In addition, the NSQIP does not collect data indicating whether an operation is an index or re-do procedure, which may represent additional complexity. Although attempts were made to examine a broad spectrum of surgical care, it is possible that patient complexity was not captured adequately. In addition, the current measures of patient complexity are not independent of each other. Furthermore, all of the measures of complexity are weighted equally, which may not emphasize particularly impactful domains, such as mortality or major complications. Because surgical specialties were examined as defined in NSQIP, general surgery represents a broad field with a high degree of variation and cases. Future studies should further examine how patient complexity varies across general surgery subspecialties and operations within general surgery. Finally, it is important to note that these analyses do not reflect the technical complexity, or the training required to complete the operations.
      In conclusion, there were substantial differences between patient complexity across surgical specialties which did not correlate with wRVUs. These results provide evidence that general and vascular surgery may be undervalued based on relative complexity and median wRVU. Efforts should be made to ensure that wRVU determinations are evidence based to align health care resources better with measures of productivity. Future study should continue to evaluate domains of patient complexity and whether hospital reimbursement follows similar trends of physician reimbursement as assessed by wRVU.

      Funding/Support

      This project was supported by the National Center for Advancing Translational Sciences, National Institutes of Health , through UCSF-CTSI Grant Number TL1 TR001871 with additional student research support from the Society for Vascular Surgery Student Research Fellowship Award and the American Heart Association Student Scholarship (J.L.R.). The authors acknowledge additional support from the National Institutes of Health , National Institute on Aging  by awards R03AG060097 (C.D.S.) and R01AG044425 , R21AG054208 , and R01AG058616 (E.F.). The authors further acknowledge support from the National Institutes of Health, National Cancer Institute award P0524688 (E.F.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center for Research Resources or the National Institutes of Health. The funding organizations were not involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation, review or approval of the manuscript.

      Conflict of interest/Disclosure

      Dr. Julie Ann Sosa is a member of the Data Monitoring Committee of the Medullary Thyroid Cancer Consortium Registry supported by GlaxoSmithKline, Novo Nordisk, Astra Zeneca, and Eli Lilly. Dr. Emily Finlayson is a founding shareholder of Ooney, Inc. The other authors have no conflicts of interest to report.

      Supplementary materials

      References

        • Shippee N.D.
        • Shah N.D.
        • May C.R.
        • Mair F.S.
        • Montori V.M.
        Cumulative complexity: a functional, patient-centered model of patient complexity can improve research and practice.
        J Clin Epidemiol. 2012; 65: 1041-1051
        • Tonelli M.
        • Wiebe N.
        • Manns B.J.
        • et al.
        Comparison of the complexity of patients seen by different medical subspecialists in a universal health care system.
        JAMA Netw Open. 2018; 1e184852
        • Kaye D.R.
        • Luckenbaugh A.N.
        • Oerline M.
        • et al.
        Understanding the costs associated with surgical care delivery in the medicare population.
        Ann Surg. 2020; 271: 23-25
        • S.C
        The Basics:.
        Relative Value Units (RVUs). 2015; 5 (Available from: https://www.nhpf.org/library/the-basics/Basics_RVUs_01-12-15.pdf. Accessed December 2, 2019.)
        • Chan D.C.
        • Huynh J.
        • Studdert D.M.
        Accuracy of valuations of surgical procedures in the medicare fee schedule.
        N Engl J Med. 2019; 380: 1546-1554
        • Schroeder S.A.
        • Frist W.
        • National Commission on Physician Payment Reform
        Phasing out fee-for-service payment.
        N Engl J Med. 2013; 368: 2029-2032
        • von Elm E.
        • Altman D.G.
        • Egger M.
        • et al.
        The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies.
        Lancet. 2007; 370: 1453-1457
        • Montroy J.
        • Breau R.H.
        • Cnossen S.
        • et al.
        Change in adverse events after enrollment in the national surgical quality improvement program: a systematic review and meta-analysis.
        PLoS One. 2016; 11e0146254
        • Iannuzzi J.C.
        • Deeb A.P.
        • Rickles A.S.
        • Sharma A.
        • Fleming F.J.
        • Monson J.R.
        Recognizing risk: bowel resection in the chronic renal failure population.
        J Gastrointest Surg. 2013; 17: 188-194
        • Iannuzzi J.C.
        • Chandra A.
        • Rickles A.S.
        • et al.
        Resident involvement is associated with worse outcomes after major lower extremity amputation.
        J Vasc Surg. 2013; 58: 827-831 e1
        • Boitano L.T.
        • Iannuzzi J.C.
        • Tanious A.
        • et al.
        Predicting postoperative destination through preoperative evaluation in elective open aortic aneurysm repair.
        J Surg Res. 2019; 235: 543-550
        • Grant R.W.
        • Ashburner J.M.
        • Hong C.S.
        • Chang Y.
        • Barry M.J.
        • Atlas S.J.
        Defining patient complexity from the primary care physician's perspective: a cohort study.
        Ann Intern Med. 2011; 155: 797-804
        • Hackett N.J.
        • De Oliveira G.S.
        • Jain U.K.
        • Kim J.Y.
        ASA class is a reliable independent predictor of medical complications and mortality following surgery.
        Int J Surg. 2015; 18: 184-190
        • Scott J.W.
        • Olufajo O.A.
        • Brat G.A.
        • et al.
        Use of national burden to define operative emergency general surgery.
        JAMA Surg. 2016; 151e160480
        • Hakkarainen T.W.
        • Ayoung-Chee P.
        • Alfonso R.
        • Arbabi S.
        • Flum D.R.
        Structure, process, and outcomes in skilled nursing facilities: understanding what happens to surgical patients when they cannot go home. A systematic review.
        J Surg Res. 2015; 193: 772-780
        • Subramaniam S.
        • Aalberg J.J.
        • Soriano R.P.
        • Divino C.M.
        New 5-factor modified frailty index using American College of Surgeons NSQIP data.
        J Am Coll Surg. 2018; 226: 173-181 e8
        • Luong P.
        • Bojansky A.M.
        • Kalra A.
        Academic physician compensation in the United States: should providers’ work at academic medical centres be judged by just one metric, the relative value unit (RVU)?.
        Eur Heart J. 2018; 39: 3633-3634
        • Moazzez A.
        • Ozao-Choy J.
        • de Virgilio C.
        Need for more objective data to determine relative value units for surgical effort [e-pu ahead of print].
        JAMA Surg. 2019; https://doi.org/10.1001/jamasurg.2019.2319
        • Glocker R.J.
        • TerBush M.J.
        • Hill E.L.
        • et al.
        Bundling of reimbursement for inferior vena cava filter placement resulted in significantly decreased utilization between 2012 and 2014.
        Ann Vasc Surg. 2017; 38: 172-176
        • Sodhi N.
        • Dalton S.E.
        • Garbarino L.J.
        • et al.
        Not all primary total hip arthroplasties are equal—So is there a difference in reimbursement?.
        Ann Transl Med. 2019; 7: 74
        • Quail J.M.
        • Lix L.M.
        • Osman B.A.
        • Teare G.F.
        Comparing comorbidity measures for predicting mortality and hospitalization in three population-based cohorts.
        BMC Health Serv Res. 2011; 11: 146
        • Leigh J.P.
        • Tancredi D.
        • Jerant A.
        • Kravitz R.L.
        Annual work hours across physician specialties.
        Arch Intern Med. 2011; 171: 1211-1213
        • Satiani B.
        • Williams T.E.
        • Go M.R.
        Predicted shortage of vascular surgeons in the United States: population and workload analysis.
        J Vasc Surg. 2009; 50: 946-952
        • Ellison E.C.
        • Pawlik T.M.
        • Way D.P.
        • Satiani B.
        • Williams T.E.
        Ten-year reassessment of the shortage of general surgeons: Increases in graduation numbers of general surgery residents are insufficient to meet the future demand for general surgeons.
        Surgery. 2018; 164: 726-732
        • Schwartz D.A.
        • Hui X.
        • Velopulos C.G.
        • et al.
        Does relative value unit-based compensation shortchange the acute care surgeon?.
        J Trauma Acute Care Surg. 2014; 76 (discussion 92-94): 84-92
        • Khubchandani J.A.
        • Shen C.
        • Ayturk D.
        • Kiefe C.I.
        • Santry H.P.
        Disparities in access to emergency general surgery care in the United States.
        Surgery. 2018; 163: 243-250
        • Shah D.R.
        • Bold R.J.
        • Yang A.D.
        • Khatri V.P.
        • Martinez S.R.
        • Canter R.J.
        Relative value units poorly correlate with measures of surgical effort and complexity.
        J Surg Res. 2014; 190: 465-470
        • Haglin J.M.
        • Eltorai A.E.M.
        • Richter K.R.
        • Jogerst K.
        • Daniels A.H.
        Medicare reimbursement for general surgery procedures: 2000 to 2018.
        Ann Surg. 2020; 271: 17-22