Studnicki J, Fisher JW, Tsulukidze MM, Taylor YJ, Salandy S, Laditka JN. Career Phase of Board-Certified General SurgeonsWorkload Composition and Outcomes. Arch Surg. 2011;146(11):1307-1313. doi:10.1001/archsurg.2011.265
Author Affiliations: Department of Public Health Sciences (Drs Studnicki, Fisher, and Laditka) and Doctoral Program in Health Services Research (Dr Tsulukidze and Mss Taylor and Salandy), College of Health and Human Services, University of North Carolina, Charlotte.
Objective To examine surgeon career phase and its association with surgical workload composition and outcomes of surgery.
Design Cross-sectional study.
Setting The study used data from calendar years 2004 through 2006 from 4 Florida general surgeon (GS) cohorts determined by years since board certification.
Participants American Board of Surgery–certified GSs regardless of subspecialty (n = 1187) performing 460 881 operations on adults 18 years or older.
Main Outcome Measures Workload composition based on the Clinical Classification System, complications identified by patient safety indicators, and in-hospital mortality. Poisson regression with robust error variance estimated adjusted rate ratios (RRs) for complications and mortality.
Results Compared with late-career surgeons, the rate of complications from cardiovascular procedures was higher for surgeons in the early-career phase (RR, 1.23; 95% CI, 1.06-1.44) and the late middle–career phase (1.18; 1.02-1.37). The mortality rate for cardiovascular procedures also was higher for early-career surgeons (RR, 1.23; 95% CI, 1.04-1.46). For digestive procedures, early-career surgeons had lower complication rates than late-career surgeons (RR, 0.86; 95% CI, 0.75-0.99).
Conclusion Late-career GSs perform both better and worse compared with early-career GSs, relative to their workload composition and proportional surgical volume. Factors such as training and case complexity may contribute to these career-phase differences.
For more than 46 million surgical procedures performed in the United States each year, reducing adverse events is important.1 An extensive body of literature has examined the role of hospital volume and found that higher volume is associated with better outcomes for many surgical procedures.2,3 Findings from a 2003 study4 of Medicare claims data suggest an inverse relationship between surgeon volume and mortality for 8 types of procedures after adjusting for patient and hospital characteristics. A systematic review of the volume-outcome literature by Halm et al3 covering 1988 to 2000 found a statistically significant association between patient outcomes and physician volumes in 69% of studies reviewed. More recently, the role of surgeon experience has been examined. A study examining inpatient complications, mortality, and hospital length of stay after colon resection found that increasing years after board certification were associated with reduced morbidity and mortality after adjusting for patient and hospital characteristics.5
For general surgeons (GSs) with training in a broad range of surgical areas, volume may not be concentrated within a specific set of procedures. Ritchie Jr et al6 examined the relationship between the workloads and practice patterns of 2434 nonspecialist GSs in the United States during a 3-year period and found that, on average, surgeons performed 398 operations per year. Operations involving the abdomen and the alimentary tract comprised 41% of the total, followed by breast and endoscopic operations with 13% each and vascular operations with 10%. Operations involving trauma and the head and neck comprised less than 2% of those performed. Workload and practice patterns were found to vary by practice type, geographic location, and rural or urban status. In a comparison of 10-year and 20-year practice cohorts, the 10-year cohort performed a slightly higher number of operations, but outcomes were not examined.
An important related question is how specialization by GSs early or late in their careers affects patient outcomes. In a review of the effect of surgeon experience and specialization on cancer surgery outcomes, Bilimoria et al7 found that specialized surgeons had better outcomes and that time since American Board of Surgery (ABS) certification was a significant predictor of patient outcomes. This finding is supported by research suggesting that accumulation of knowledge and skills during years in practice results in superior quality of care and better outcomes.8 However, limited research has examined the relationship between surgeon case load volume and career phase and the association between these factors and patient outcomes. The hypothesis under consideration in this research is that the career phase of GSs, as indicated by years since board certification, is associated with variation in case load composition, cumulative and cross-sectional volume, surgical complications, and in-hospital mortality.
A cross-sectional study design was used to examine the composition, workload, and outcomes of GS practice. Surgeons were categorized into 4 cohorts based on their career phase as expressed in terms of years since board certification: less than 10 years (early), 10 to 19 years (early middle), 20 to 29 years (late middle), and 30 years or more (late).
The study used 2004-2006 Florida Patient Discharge Data, available from the Florida Agency for Health Care Administration. Data included demographic characteristics such as sex, race, ethnicity, age, patients' residence zip code and county, the diagnosis-related group, admission source, length of stay, year and quarter of admission, discharge disposition, payer category, and a unique hospital and surgeon identifier. Each record also provided the principal and secondary diagnosis and procedure codes. Using unique surgeon identifiers, patient discharge data were linked with the practitioner profile available from the Florida Department of Health. The practitioner profile includes self-reported information regarding practicing physicians, such as program type (internship, residency, fellowship, or other) for graduate medical education, the year the licensee began practicing, the date the licensee began attending medical school, the title of the degree awarded to the licensee, the date of graduation, the educational institution name, and the specialty certification name. The date of board certification was obtained from the ABS online database.
Only surgeons with ABS certification in general surgery were included in the study. For each surgeon identified as the operating physician for an operation occurring during the study period, data from the Florida Practitioner Profile Database were used to identify certification in general surgery regardless of subspecialty. Then, surgeons were matched with ABS certification data to verify board certification status. A total of 1300 GSs performing 528 985 operations during the study period were identified. Of these, 100 surgeons performing 39 401 operations were excluded because certification status and date could not be verified through the ABS database. Also, 28 703 operations performed on children were excluded, along with 13 surgeons who had performed operations only on children during the period. Thus, 1187 surgeons performing 460 881 operations on adults 18 years and older remained in this analysis.
Patient safety indicators (PSIs) were used to measure complications associated with each discharge record. The PSIs are a set of indicators developed by the Agency for Healthcare Research and Quality to provide information regarding potentially preventable in-hospital complications and adverse events. For the purposes of this study, we selected 11 PSIs based on clinical relevance for general surgery: death in low-mortality diagnosis-related groups (PSI 2), foreign body left in patient during surgical procedure (PSI 5), iatrogenic pneumothorax (PSI 6), selected infections due to medical care (PSI 7), postoperative hemorrhage or hematoma (PSI 9), postoperative physiologic and metabolic derangements (PSI 10), postoperative respiratory failure (PSI 11), postoperative pulmonary embolism or deep vein thrombosis (PSI 12), postoperative sepsis (PSI 13), postoperative wound dehiscence in abdominopelvic surgical patients (PSI 14), and accidental puncture and laceration (PSI 15). Patients were identified as having a complication if any of the 11 PSIs were present on the discharge record.
The composition of GS workload was characterized using Clinical Classification Software (CCS), a tool that uses the International Classification of Diseases, Ninth Revision, Clinical Modification to classify diagnosis and procedure codes into clinically relevant categories.9 The CCS was designed for use with hospital discharge data. Previous studies have applied the CCS to inpatient data to determine reasons for hospitalization10,11 and to characterize the workload composition of GSs.12
The single-level CCS includes 231 surgical procedure categories, which are grouped by major organ systems into 16 broader categories. The following categories based on site of the procedure were used: nervous system; endocrine system; eye; ear; nose, mouth, and pharynx; respiratory system; cardiovascular system; hemic and lymphatic system; digestive system; urinary system; male genital organs; female genital organs; obstetric; musculoskeletal system; integumentary system; and miscellaneous diagnostic and therapeutic. For each of the categories, total volume of procedures was calculated for all surgeons combined and then for each cohort of surgeons.
To accommodate differential patient risks, we used version 3.5 of the Comorbidity Software distributed by the Agency for Healthcare Research and Quality.13 The comorbidity measure used by the software was developed by Elixhauser and colleagues14 to predict length of hospital stay, total charges, and in-hospital mortality.
Descriptive statistics, including means and percentages, were calculated for all surgeons combined and for each cohort separately. For unadjusted analysis, we used the χ2 test to examine associations between career phase and workload composition and patient characteristics.
For adjusted analysis, Poisson regression with robust error variance15 was used to estimate rate ratios (RRs) for complications and in-hospital mortality. This method uses generalized estimating equations16 to adjust for clustering of patients within physicians and for physicians within hospitals. Clusters were defined using a combination of surgeon identification and hospital identification to account for surgeons practicing in more than one hospital. Models were adjusted for patient characteristics (age, sex, race/ethnicity, health insurance coverage type, and comorbid conditions in the Elixhauser Comorbidity Index), hospital characteristics (ownership type and mean volume of surgical procedures during the study period), and mean surgeon volume. One condition included in the Elixhauser Comorbidity Index, peptic ulcer disease, was excluded from the model because of small frequencies (<50 observations) within surgeon cohorts. Analyses were conducted using SAS statistical software, version 9.2 (SAS Institute, Inc, Cary, North Carolina). All tests were 2-sided, with statistical significance determined by P < .05.
Among 1187 GSs in the analysis, the largest percentage had been board certified for 10 to 19 years (38.4%), followed by 20 to 29 years (26.8%), less than 10 years (19.2%), and 30 years or more (15.6%) (Table 1). Most surgeons (90.3%) practiced in a metropolitan area and had graduated from US medical schools. Yet, one-third of surgeons 20 to 29 years after certification were foreign medical graduates, reflecting the larger share of foreign medical graduates among newly licensed physicians 3 decades ago. Important combinations of certification and training represented within our study population of board-certified GSs include the following major categories of dually certified surgeons: 172 ABS/American Board of Thoracic Surgery certified, 107 ABS/vascular surgery subspecialty certified, 102 ABS/American Board of Plastic Surgery certified, and 49 ABS/American Board of Colon and Rectal Surgery certified (data not shown).
Surgical procedural volume examined by career phase showed several trends (Table 2). The mean surgeon volume peaked in the early middle–career phase, plateaued in the late middle–career phase, and decreased for surgeons in the late-career phase for all CCS categories. However, for 2 CCS categories, the late-career physicians had significantly higher (cardiovascular) and lower (digestive) percentages relative to total workload than the 3 less-experienced cohorts. For all GSs, digestive and cardiovascular procedures accounted for nearly 75% of total workload. Although significant differences were observed in other CCS categories, the numbers were too small to represent a meaningful difference in practice composition.
Examination of patient characteristics by career phase also revealed notable trends (Table 3). Late-career GSs treated disproportionally larger numbers of Hispanics than all other cohorts. Also, patients within the cohorts generally reflected the age of their surgeon. That is, with the exception of the patients aged 60 to 69 years, late-career surgeons had proportionately fewer young patients and proportionately older patients than surgeons in earlier phases of their careers. Consistent with the patient age distribution, late-career GSs also saw more patients covered by Medicare and fewer covered by commercial health insurance.
The rate of complications from cardiovascular procedures was higher for surgeons in their early-career phase (RR, 1.23; 95% CI, 1.06-1.44) and late middle–career phase (1.18; 1.02-1.37) when compared with surgeons in their late-career phase (Table 4). The mortality rate for cardiovascular procedures also was higher for early-career surgeons (RR, 1.23; 95% CI, 1.04-1.46). Conversely, for digestive system procedures, complication rates were lower for early-career surgeons (RR, 0.86; 95% CI, 0.75-0.99) when compared with late-career surgeons.
Few studies have examined the relationship between GS career phase, workload composition, and patient outcomes. Our findings indicate that increasing career phase was associated with an increased emphasis on cardiovascular and decreased emphasis on digestive system procedures. Likewise, patient outcomes were better for late-career surgeons when cardiovascular system procedures were examined but worse for digestive system procedures. These findings support the relevance of specialization as measured by proportional volume and provide a new context for examining the volume-outcome relationship.
Specialization has been found to enhance surgical proficiency and, in turn, to improve patient outcomes. A systematic review by Chowdhury et al17 included 163 studies examining 42 different types of surgical procedures and covering 13 surgical specialties. Twenty of 22 studies examining the performance of specialized as opposed to generalist GSs reported significantly better outcomes for the specialist surgeons. The review also noted a varying magnitude of the benefit of high surgeon volume and specialization across the specialties. The study concluded that surgeon volume and surgeon specialization are independently positively associated with patient outcomes. Of interest, our study finds a relationship between reduced total volume but increased proportional volume and improved results for cardiovascular system procedures. We found the opposite, although we had expected a pattern of diminished results associated with lower proportional volume for digestive system procedures.
A comprehensive review of the literature by Halm et al3 also noted that surgeon volume was a more important determinant of outcomes than hospital volume in many studies. This occurred with regard to coronary artery bypass grafting, carotid endarterectomy, surgical prodecures for ruptured abdominal aortic aneurysm, and procedures for colorectal cancer, all of which are included in the scope of general surgery.
Our finding that cumulative (rather than cross-sectional) surgical volume may be a valid indicator of improved outcomes also has been supported by previous research. A study17 that controlled for volume discrepancies between specialists and GSs to examine independent contributions of volume and specialization found that the benefit from specialization remained significant. This finding suggests that improved outcomes can be explained partly by cumulative volume.
Similar to our finding suggesting that surgeons with less than 10 years' experience have worse outcomes for cardiovascular system procedures, Wurtz et al18 found that new surgeons in 2 surgical subspecialties had higher class I surgical site infection rates compared with their experienced colleagues. Also, their operation time was longer, and they did not operate on sicker patients. However, with the accumulated experience (as measured by cumulative cases, which can be considered advancement in career phase), their operating room times and surgical site infection rates decreased toward the level attained by their experienced colleagues. However, the study was based on data from 1 hospital.
In their systematic review of the relationship between clinical experience and quality of health care, Choudhry et al19 found an overall inverse relationship between the number of years in practice (measured as years since graduation from medical school, years since certification, or physician age) and the quality of care provided by the physician. A study by Rhee20 included in the review showed that most appropriate care was provided by physicians who had been practicing 6 to 15 years; physicians with more or fewer years of experience provided less appropriate care. That study, however, acknowledged several limitations, including insufficient control for patient comorbidity, other physician factors, and the clustering of patients within physicians. Also, trends toward greater specialization across the career trajectory were not addressed.
An increasing trend toward specialization among general surgery residents has led to what some describe as a “crisis” in the GS workforce.21 The total number of trained GSs who practice as generalists has decreased, and in 2007, less than 25% of chief residents pursuing certification by the ABS indicated that they would not pursue a subspecialty.21 Now, GSs tend to specialize sooner after completion of their residencies,22,23 suggesting the need to go beyond examinations of the number of available surgeons to determine the types of services they provide.22
A principal limitation of our ability to interpret the findings associated with career phase is related to the many possible combinations of training represented by GSs and the composition of their surgical workload in terms of complexity and technical difficulty at the procedure level. Combinations of dual board certification and other training characteristics are likely to influence the composition of surgeons' practices. For example, we would expect that surgeons with ABS/American Board of Thoracic Surgery certification would be more oriented toward cardiovascular system procedures and that their case mix of those procedures is likely to be of greater complexity than that of other GSs. These important differences in training and complexity composition may influence the cohort-level comparisons in ways that are not discernable through the present study.
As with all studies using administrative data, we were unable to examine and control for clinical detail, which may affect the outcomes notably. However, many studies3,24,25 using clinical data also have shown a significant effect of volume on outcome. We also accounted for only a limited number of structural and organizational factors in our analysis. Other factors that often are likely to contribute to improved surgical outcomes (eg, multidisciplinary team, staffing ratios, and availability of technology and specialized equipment, including specialized intensive care units and operating rooms) were unavailable.
Our assessment of complications using PSIs lacked the “present on admission” indicator, which was legislatively mandated to be added to administrative claims data in 2007.26 However, 8 PSIs used in our study (PSI 5, PSI 6, PSI 7, PSI 9, PSI 10, PSI 11, PSI 13, and PSI 15) have been found to remain potential patient safety problems after eliminating conditions reported as present on admission.26,27 Two PSIs used in our analysis (PSI 2 and PSI 14) have numerators based on procedure codes or discharge status26 that cannot be present on admission. After eliminating conditions reported as present on admission, the only PSI with less likelihood still to be considered a potential patient safety problem was PSI 12, postoperative pulmonary embolism or deep vein thrombosis.26 Thus, PSIs selected for the analysis in our study are valid measures of in-hospital complications even in the absence of present-on-admission prompts in the data set.
Each CCS category encompasses multiple types of surgical procedures with widely varying degrees of complexity, the examination of which was beyond the scope of this article. However, we recognize that changes in volume may be accompanied by variation in the technical difficulty or complexity of the procedures.
Finally, the nature of this cross-sectional study did not permit us to follow the cohorts for an extended period. However, our longitudinalized approach to study design relies on the assumption that the patterns of practice represented by the 3-year period are an accurate representation of GS career experience.
In conclusion, investigators of the volume-outcome relationship have called for further scrutiny of underlying mechanisms of the volume effect.3,28 This study has shown that practice patterns and proportional surgical volume vary across career categories of GSs in Florida. The underlying contributory factors to these cohort differences, such as training and practice complexity, require further analysis. The improved understanding of the relationship among physician workload, volume, and clinical outcomes may provide important policy and practice implications.
Correspondence: M. Maka Tsulukidze, MD, Doctoral Program in Health Services Research, College of Health and Human Services, University of North Carolina, 9201 University City Blvd, Charlotte, NC 28223-0001 (firstname.lastname@example.org).
Accepted for Publication: June 3, 2011.
Author Contributions:Study concept and design: Studnicki and Fisher. Acquisition of data: Studnicki and Fisher. Analysis and interpretation of data: Studnicki, Fisher, Tsulukidze, Taylor, Salandy, and Laditka. Drafting of the manuscript: Studnicki, Tsulukidze, Taylor, and Salandy. Critical review of the manuscript for important intellectual content: Studnicki, Fisher, Tsulukidze, Taylor, Salandy, and Laditka. Study supervision: Studnicki.
Financial Disclosure: None reported.