Characteristics of Practice Among Rural and Urban General Surgeons in North Carolina

Transcription

1 ORIGINAL ARTICLES Characteristics of Practice Among Rural and Urban General Surgeons in North Carolina Jennifer King, BA,* Erin P. Fraher, PhD,* Thomas C. Ricketts, PhD,* Anthony Charles, MD, George F. Sheldon, MD, and Anthony A. Meyer, MD, PhD Objective: To examine variation in the practice patterns of individual general surgeons and how they differ between rural and urban areas of North Carolina. Summary of Background Data: Traditional physician supply analyses often rely on head counts and do not take into account how physicians practice patterns differ. Practice characteristics including the volume and the breadth of services that a physician provides may be especially important in understanding the supply and distribution of specialists, such as general surgeons. Methods: Cross-sectional study using physician licensure data linked with administrative records on all inpatient hospital discharges and all surgeries performed at freestanding ambulatory surgery centers in North Carolina in Results: Total procedure volumes varied widely (interquartile range: ). The average general surgeon in a rural county performed 54 different procedures at least once during the year, compared to 59 in counties with small urban areas and 62 in metropolitan counties. The 10 procedures that a general surgeon performed most frequently accounted for 72% of that surgeon s total annual procedures in rural counties, 67% in counties with small urban areas, and 66% in metropolitan counties. These rural metropolitan differences were smaller after controlling for secondary specialty and other surgeon characteristics. Conclusions: There was significant variation in the volume and scope of procedures that North Carolina general surgeons performed in the year. Many general surgeons in metropolitan areas performed an array of procedures that was broader than those in rural areas. (Ann Surg 2009;249: ) Medicine is becoming increasingly specialized, 1 3 a trend exemplified by the field of general surgery. Over 70% of general surgeons pursue additional fellowship training after they receive primary certification; this compares to 55% in The American Board of Surgery is exploring abbreviated training for subspecialists, including joint training programs that lead to combined certification in general surgery and vascular surgery in 1 year less than the time it takes to complete the separate programs. In addition, in 2006 the Board instituted a primary certificate in vascular surgery with no prerequisite of certification in general surgery. 5 Additional From the *Cecil G. Sheps Center for Health Services Research, The University of North Carolina at Chapel Hill, North Carolina; and Department of Surgery, The University of North Carolina at Chapel Hill School of Medicine, Chapel Hill, North Carolina. Supported by North Carolina Area Health Education Program and American College of Surgeons Health Policy Institute. Reprints: Jennifer King, Cecil G. Sheps Center for Health Services Research, The University of North Carolina at Chapel Hill, 725 Martin Luther King Jr Blvd, CB 7590, Chapel Hill, NC jking@schsr.unc.edu. Copyright 2009 by Lippincott Williams & Wilkins ISSN: /09/ DOI: /SLA.0b013e3181a6cd surgical subspecialties that have required general surgery board certification are also exploring opting out of this structure. This phenomenon has implications for policies related to the size of the surgical workforce. Traditional physician supply analyses often rely on head counts, with little information on how physicians practice patterns may differ. Three important practice characteristics are the content, volume, and breadth of services that a general surgeon performs given increasing specialization, counting the number of physicians with a general surgery specialty may not adequately describe the availability of the full range of general surgical care services. 4,6 Adding to this challenge is the perception that the breadth of a general surgeon s practice varies substantially depending on whether the surgeon practices in an urban or rural location. Some research has found that rural surgeons tend to perform a broader range of procedures and feel less-well prepared by their residency experience than urban surgeons. 7,8 However, there is a lack of published empirical data on the extent to which breadth of practice varies from individual surgeon to individual surgeon and how this is affected by geography. 9 Such information is important for policy decisions about the surgery workforce and the breadth of surgical education. To understand the structure of general surgeons practices, this study investigated the volume and scope of procedures performed by general surgeons in North Carolina and how they differed between rural and urban areas. We evaluated 2 main hypotheses: that there were differences between individual surgeons in the volume and scope of procedures performed, and that there were differences between rural and urban surgeons in the volume and scope of procedures performed. METHODS Data Sources Data on general surgeons are from the North Carolina Health Professions Data System (HPDS) at the Cecil G. Sheps Center for Health Services Research at the University of North Carolina. This data set is derived from the North Carolina Medical Board s licensure files and contains records on all licensed physicians in the state on October 31 of each year. Because state law requires physicians to hold a license to practice in North Carolina, the HPDS contains records for all practicing physicians at that point in time. All data elements, including information on medical education, specialty, and demographic characteristics, are self-reported by physicians when they apply for a new or renewed license. We used HPDS licensure data for 2004 in this analysis. The population of interest is active North Carolina physicians who are not employed by the federal government and who identify their primary specialty as general surgery there were 648 such physicians in To obtain data on the procedures performed by general surgeons, we linked the HPDS to records from the North Carolina Hospital Discharge Database, a data set maintained by the Sheps Center under contract with the North Carolina Division of Health Annals of Surgery Volume 249, Number 6, June 2009

2 Annals of Surgery Volume 249, Number 6, June 2009 Rural and Urban Practice Patterns Services Regulation. As set forth by the Medical Care Data Act of 1995, the database contains records on all discharges from nonpsychiatric, nonfederal hospitals in the state and all surgery procedures for free-standing ambulatory surgery facilities. 10 Each record in the discharge and ambulatory surgery databases contains the attending physician s Medicare Unique Physician Identifier Number (UPIN) and 6 ICD-9 procedure codes. To link the licensure data and discharge data, a common physician identifier was necessary. Because the UPIN is not included in the HPDS licensure data, we merged the 2004 UPIN Directory from the Centers for Medicare and Medicaid Services to the HPDS data by first name, last name, and middle initial. Not all general surgeons in the HPDS data matched to a record in the UPIN file; we searched the online UPIN Registry for any unmatched surgeons. (The UPIN Registry was maintained by NHIC, Corp., under contract with the Centers for Medicare and Medicaid Services, and was available at until May 23, Because the National Provider Identifier has replaced the UPIN, the UPIN Registry is no longer available). We were unable to find a UPIN for 47 general surgeons (7%); these surgeons were excluded from the analysis. Compared with general surgeons in our final analysis file, those missing the UPIN on average were slightly older, reported less hours per week in clinical care, and were more likely to practice in a metropolitan county (Table 1). After merging the files, we found that 57 surgeons in the licensure data had zero procedures in the discharge files. Compared with general surgeons in our final analysis file, those with no procedures on average were slightly older, reported fewer hours per week in clinical care, were more likely to be female, and were more likely to practice in a metropolitan county (Table 1). The final analysis file contained 544 general surgeons. Practice Content Variables To characterize the procedures that general surgeons performed, we used the Agency for Healthcare Research and Quality s Clinical Classification Software (CCS) to group ICD-9 procedure codes into 231 mutually exclusive clinical categories. 12 Most CCS codes represent single types of procedures, but some represent groups of procedures that occur infrequently (eg, Other non-or lower GI therapeutic procedures ). We also used the Agency for Healthcare Research and Quality s Procedure Classes, which group ICD-9 codes into 4 categories: minor diagnostic, minor therapeutic, major diagnostic, and major therapeutic. 13 Procedures are defined as major if they are operating room procedures; minor procedures are nonoperating room procedures. We also grouped the 231 procedure categories into 6 broad content areas: alimentary tract, abdomen, breast skin and soft tissue, head and neck surgery, vascular surgery, and surgical critical care. These categories correspond to the American Board of Surgery essential content areas (Four ABS American Board of Surgery content areas are not separately commented on here. To report the pediatric, trauma/burn, and surgical oncology content areas separately would require contextual detail not included in the CCS codes. As a result, for example, in this analysis trauma exploratory laparotomy with bowel resection would be categorized within abdomen (exploratory laparotomy) and alimentary (bowel resection). Because there are very few endocrine procedures, we categorized these procedures in the larger content area within which they fall. For example, we categorized procedures like total thyroidectomy as head and neck and adrenalectomy as abdomen). Of the 231 CCS categories, 114 did not fall into any of these content areas these procedures were classified as not part of traditional general surgery training. We aggregated the discharge records by attending physician and created summary variables at the individual surgeon level to measure the number of procedures that each surgeon performed in each of the 231 procedure categories, in each of the 4 procedure classes, and in each of the 6 content areas in To describe total procedure volume, we calculated the number of times each general TABLE 1. Characteristics of NC General Surgeons, 2004 Excluded From Analysis Final Analysis File Total Missing UPIN No Procedures Total Metropolitan Micropolitan Rural N Primary practice location Metropolitan county 73.6% 85.1% 87.7% 71.1% Micropolitan county 19.6% 12.8% 10.5% 21.1% Rural county 6.8% 2.1% 1.8% 7.7% Age * H/wk in clinical care Female 8.2% 8.5% 10.5% 7.9% 8.5% 7.0% 4.8% Secondary specialty No secondary specialty 57.7% 53.2% 56.1% 58.3% 51.7% 75.7% 71.4% Vascular surgery 16.2% 12.8% 12.3% 16.9% 17.8% 15.7% 11.9% Critical care surgery 6.3% 2.1% 5.3% 6.8% 9.3% 0.9% 0.0% Oncology surgery 4.3% 2.1% 1.8% 4.8% 5.7% 2.6% 2.4% Thoracic surgery 3.9% 4.3% 3.5% 3.9% 4.4% 2.6% 2.4% Abdominal surgery 2.9% 2.1% 3.5% 2.9% 3.4% 0.9% 4.8% Primary care 1.5% 2.1% 5.3%* 1.1% 1.0% 0.9% 2.4% Colon and rectal surgery 1.2% 0.0% 1.8% 1.3% 1.6% 0.0% 2.4% Other 5.9% 21.3% 10.5%* 4.0% 5.2% 0.9% 2.4% Results for Total are based on physicians who indicated they are active, nonfederal physicians with a primary specialty of general surgery. Results for Final Analysis File include physicians for whom UPIN was available and who performed at least 1 procedure in * P 0.05 (0.01), 2-tailed t test of means relative to final analysis file. P 0.05 (0.01), 2-tailed t test of means relative to metropolitan Lippincott Williams & Wilkins

3 King et al Annals of Surgery Volume 249, Number 6, June 2009 TABLE 2. Outcome Measures Calculation Example, Procedures Performed by Individual General Surgeon Number CCS Procedure Category Performed in 2004 Cholecystectomy and common duct exploration 107 Intraoperative cholangiogram 82 Inguinal and femoral hernia repair 42 Appendectomy 31 Other hernia repair 29 Hemorrhoid procedures 25 Other OR lower GI therapeutic procedures 25 Colonoscopy and biopsy 22 Colorectal resection 22 Upper gastrointestinal endoscopy; biopsy 21 Excision of skin lesion 17 Other non-or lower GI therapeutic procedures 16 Lumpectomy; quadrantectomy of breast 16 Other non-or therapeutic procedures on skin and 16 breast Total 471 Subtotal: top 10 procedures 406 Outcome measures Total procedure volume 471 Number of different types of procedures 14 Share of procedures concentrated in top 10 procedures 86% surgeon performed any procedure in the year. To measure the diversity of individual surgeons practices, we calculated the number of different procedures the surgeon performed at least once during the year. Finally, to measure the extent to which surgeons focused on a small number of procedures, we calculated the share of each surgeon s total procedure volume accounted for by the 10 procedures that the surgeon performed most frequently. An example of these calculations is provided in Table 2. Geographic Area and Surgeon Characteristics In the licensure files, physicians report the county of their primary practice location. We used the Office of Management and Budget s Core Based Statistical Area (CBSA) designations to classify the surgeon s practice location as metropolitan, micropolitan, or rural. 14 In brief, metropolitan counties are those with at least 1 urbanized area with a population of 50,000 or more and micropolitan counties have an urbanized area with a population of at least 10,000 but less than 50,000. Contiguous counties are included in the CBSA if they meet a specified level of commuting to or from the central counties. We also used other self-reported information from the licensure data, including gender, medical school graduation year, hours per week spent in clinical care, and secondary specialty. Analysis We performed bivariate and multivariate analyses. To gain a broad understanding of general surgeons activities, we compared the distributions of the outcome variables in metropolitan, micropolitan, and rural counties. To isolate the relationship between county type and breadth of practice, we also used ordinary least squares regression to estimate the relationship between county type and each of the 2 breadth-of-practice outcome variables described above, controlling for individual surgeon characteristics secondary specialty, hours per week spent in clinical care, gender, and years since medical school graduation. Because surgeons with very small procedure volumes could distort attempts to characterize the breadth of practices in North Carolina, we excluded surgeons who performed fewer than 50 procedures in the year (27 surgeons in 2004) in analyses of the average number of procedures performed in specific procedure categories and measures of breadth of practice. RESULTS General Surgeons in North Carolina In 2004, there were 648 active general surgeons in North Carolina, or 0.76 general surgeons per 10,000 residents. This supply ratio was 0.81 in metropolitan counties, 0.65 in micropolitan counties, and 0.61 in rural counties. About 70% of the 544 general surgeons in the final analysis file had a primary practice location in a metropolitan county (Table 1). The average general surgeon was 47.9 years old. Surgeons in rural counties were older than those in metropolitan and micropolitan areas and less likely to be female. The average hours per week spent in clinical care was similar across geographic areas. Over half of those who reported a primary specialty of general surgery did not report a secondary specialty (58.3%). General surgeons in metropolitan counties were much more likely to report a second specialty (48.3%) compared with those in micropolitan (24.3%) and rural counties (28.6%). The most common secondary specialty was vascular surgery, reported by 16.9% of general surgeons. Although 6.8% of general surgeons overall reported a secondary specialty in critical care surgery, almost all of these surgeons have a primary practice location in a metropolitan county (less than 1% of surgeons in micropolitan areas and no surgeons in rural counties reported this secondary specialty). Procedure Volume and Type As a group, North Carolina general surgeons performed over 300,000 procedures in 2004 (Table 3). In metropolitan counties, over half of all procedures performed by general surgeons (54%) were major therapeutic procedures, compared with 46% in micropolitan counties and 35% in rural counties. In nonmetropolitan areas, minor diagnostic procedures made up a larger portion of the procedures performed by general surgeons (34% in rural; 23% in micropolitan; 17% in metropolitan). Alimentary tract procedures accounted for 43% of all procedures performed by general surgeons in 2004; these procedures made up a larger share of all procedures in micropolitan and rural counties compared with metropolitan counties. Procedures in the breast/skin/soft tissue, abdomen, and vascular content areas accounted for larger shares of total procedures in metropolitan areas compared with micropolitan and rural counties. the individual surgeon level, there was a wide range of annual procedure volumes (Fig. 1). The average general surgeon performed 554 procedures. One-quarter of general surgeons performed less than 356 procedures, and one-quarter performed more than 700 procedures. A difference of similar magnitude was present for major therapeutic procedures one-quarter of the general surgeons performed less than 169 and one-quarter performed more than 381 (results not shown). The average number of total procedures performed annually was similar across the state, ranging from 551 in metropolitan counties to 568 in rural counties. However, the distribution of procedure volumes in rural counties was wider: the interquartile range (25th percentile to 75th percentile) was 275 to 783. Some surgeons performed very small numbers of procedures as mentioned above, 27 surgeons performed less than 50 procedures in Almost all of these surgeons, 24, reported a metropolitan practice location, and 3 reported practicing in a micropolitan county. In terms of gender and age, these low volume Lippincott Williams & Wilkins

4 Annals of Surgery Volume 249, Number 6, June 2009 Rural and Urban Practice Patterns TABLE 3. Total Procedures Performed by NC General Surgeons by Procedure Type and Content Area, 2004 Total Metropolitan Micropolitan Rural N % N % N % N % Total 301, ,176 64,263 23,855 Procedure type Major therapeutic 153, , , Minor therapeutic 72, , , Major diagnostic 16, , Minor diagnostic 58, , , Content area Alimentary tract 128, , , , Breast/skin/soft tissue 60, , , Abdomen 39, , Vascular 32, , Head/neck Critical care Not general surgery 28, , Data are based on active nonfederal physicians with a primary specialty of general surgery who performed at least 1 procedure in 2004 (N 544). Procedure types are based on Clinical Classification Software from the Agency for Healthcare Research and Quality. Procedures are classified by content area based on authors clinical experience. FIGURE 1. Number of procedures performed by NC general surgeons, Data are based on active nonfederal physicians with a primary specialty of general surgery who performed at least one procedure in 2004 (N 544). surgeons were similar to the rest of the analysis file. On average, this group reported spending 47.0 hours in clinical care per week. As described above, we calculated the number of times each general surgeon performed each of the 231 CCS procedures, and then calculated the average across all general surgeons for each procedure. Table 4 shows the 50 procedures with the highest averages. There was significant variation in the number of these procedures performed by individual general surgeons. For example, in 2004 the average general surgeon performed 49 procedures that were classified in the cholecystectomy and common duct exploration category, and 90% of general surgeons performed at least 5 of these procedures over the year. However, 5% (24 surgeons) did not perform any procedures in the cholecystectomy and common duct exploration category during the year. For some procedures, like GI laparoscopy, this variation was wider: 62% of general surgeons performed at least 1 GI laparoscopy in 2004, but only 24% performed more than 5. There was geographic variation in the average number of specific procedures performed, as seen in the 10 most common procedure categories (Table 5). Compared with surgeons in metropolitan areas, the average surgeon in nonmetropolitan areas performed more colonoscopy and biopsy; upper gastrointestinal endoscopy/biopsy; and other non-or lower GI therapeutic procedures. While there were some significant differences between metropolitan and micropolitan counties, surgeons in rural counties were the most distinct. All surgeons in rural counties performed 7 of the 10 procedures at least once this was not true of any procedure in micropolitan or metropolitan counties. Breadth of Surgeons Practices The average general surgeon performed 61 different types of procedures at least once during 2004 (Fig. 2). On average, general surgeons in rural counties performed 8 fewer different types of procedures than those in metropolitan counties over the course of the 2009 Lippincott Williams & Wilkins

5 King et al Annals of Surgery Volume 249, Number 6, June 2009 TABLE Average Number of Procedures Performed by NC General Surgeons, 50 Most Common Procedures, Procedure Category Average Maximum Shared Performing at Least 1 Shared Performing at Least 5 1 Cholecystectomy and common duct exploration % 90% 2 Colonoscopy and biopsy % 55% 3 Upper gastrointestinal endoscopy; biopsy % 61% 4 Lumpectomy; quadrantectomy of breast % 74% 5 Inguinal and femoral hernia repair % 82% 6 Other hernia repair % 86% 7 Other non-or lower GI therapeutic procedures % 44% 8 Excision of skin lesion % 78% 9 Appendectomy % 83% 10 Other OR lower GI therapeutic procedures % 81% 11 Other therapeutic procedures; hemic and lymphatic system % 65% 12 Intraoperative cholangiogram % 49% 13 Other non-or therapeutic procedures on skin and breast % 76% 14 Breast biopsy and other diagnostic procedures on breast % 47% 15 Debridement of wound; infection or burn % 33% 16 Colorectal resection % 72% 17 Other vascular catheterization; not heart % 72% 18 Other OR upper GI therapeutic procedures % 32% 19 Blood transfusion % 55% 20 Excision; lysis peritoneal adhesions % 64% 21 Other OR procedures on vessels other than head and neck % 32% 22 Other OR gastrointestinal therapeutic procedures % 59% 23 Respiratory intubation and mechanical ventilation % 43% 24 Other therapeutic procedures % 28% 25 Hemorrhoid procedures % 40% 26 Other therapeutic procedures on muscles and tendons % 42% 27 Creation; revision and removal of arteriovenous fistula % 15% 28 Mastectomy % 33% 29 Other OR therapeutic procedures on skin and breast % 38% 30 Laparoscopy (GI only) % 24% 31 Suture of skin and subcutaneous tissue % 22% 32 Other non-or gastrointestinal therapeutic procedures % 39% 33 Incision of pleura; thoracentesis; chest drainage % 26% 34 Proctoscopy and anorectal biopsy % 24% 35 Incision and drainage; skin and subcutaneous tissue % 26% 36 Enteral and parenteral nutrition % 21% 37 Contrast arteriogram of femoral and lower extremity arteries % 10% 38 Thyroidectomy; partial or complete % 15% 39 Arterio- or venogram (not heart and head) % 12% 40 Amputation of lower extremity % 23% 41 Contrast aortogram % 8% 42 Skin graft % 13% 43 Small bowel resection % 18% 44 Biopsy of liver % 18% 45 Gastrostomy; temporary and permanent % 14% 46 Other diagnostic radiology and related techniques % 14% 47 Other therapeutic endocrine procedures % 9% 48 Hemodialysis % 13% 49 Other OR therapeutic procedures; male genital % 14% 50 Mammography % 11% Data are based on active nonfederal physicians with a primary specialty of general surgery who performed at least 50 procedures in 2004 (N 517). Procedure categories are based on Clinical Classification Software from the Agency for Healthcare Research and Quality Lippincott Williams & Wilkins

6 Annals of Surgery Volume 249, Number 6, June 2009 Rural and Urban Practice Patterns TABLE 5. Average Number of Procedures Performed by NC General Surgeons, 10 Most Common Procedures, 2004 Metropolitan Micropolitan Rural Least 1 Least 5 Mean Max Least 1 Least 5 Mean Max Least 1 Least 5 Procedure Category Mean Max Cholecystectomy and common duct % 88% % 95% 40.8* % 95% exploration Colonoscopy and biopsy % 47% % 68% % 86% Upper gastrointestinal endoscopy; biopsy % 57% % 65% % 88% Lumpectomy; quadrantectomy of breast % 70% % 86% % 76% Inguinal and femoral hernia repair % 78% % 94% 16.5 * % 83% Other hernia repair % 84% % 92% % 88% Other non-or lower GI therapeutic procedures % 36% % 54% % 86% Excision of skin lesion % 76% % 81% % 86% Appendectomy % 80% % 92% 13.6* % 86% Other OR lower GI therapeutic procedures % 81% % 83% % 79% Data are based on active nonfederal physicians with a primary specialty of general surgery who performed at least 50 procedures in 2004 (N 517). Procedure categories are based on Clinical Classification Software from the Agency for Healthcare Research and Quality. P 0.05 (0.01), 2-tailed t test of mean relative to metropolitan. * P 0.05 (0.01), 2-tailed t test of mean relative to micropolitan. FIGURE 2. Number of different types of procedures performed by NC general surgeons, Data are based on active nonfederal physicians with a primary specialty of general surgery who performed at least 50 procedures in 2004 (N 517). The line in the middle of each box represents the median, and the boundaries of the boxes represent the 25th and 75th percentiles. The whiskers extend to 1.5 times the interquartile range. Dots outside the whiskers represent outliers. *P 0.01, 2-tailed t test of null hypothesis that mean is equal to metropolitan. year. In metropolitan and micropolitan counties, there were several outliers that performed procedures from a very small number of categories. Three outliers in rural counties performed a very large number of different procedures; however, 2 of these surgeons would not be considered outliers in the metropolitan distribution. After controlling for individual and county characteristics, the difference in the number of different types of procedures performed between surgeons in rural counties and those in metropolitan counties was smaller ( 3.9, 95% confidence interval: 9.6, 1.9) (Table 6). Controlling for other factors in the model, female general surgeons had less diverse practices than male surgeons, as did surgeons who had been out of medical school longer. Relative to general surgeons who reported no other specialty, those with secondary specialties of vascular, thoracic, critical care, and colon and rectal surgery performed a larger number of different types of procedures annually. On average, the 10 procedures that a general surgeon performed most frequently in 2004 accounted for 68% of that surgeon s total procedures. This figure was higher for the average surgeon in rural counties (72%) than those in micropolitan and metropolitan areas (67% and 66%, respectively) (Fig. 3). There was more variation in this measure among surgeons in metropolitan areas compared with those in nonmetropolitan areas. After controlling for covariates, the difference between rural and metropolitan surgeons in the share of practice concentrated in 10 procedures was 3.8 percentage points (95% confidence interval: 0.9, 6.8). Holding other factors in the model constant, female surgeons and those who had been out of medical school for longer periods of time had more focused practices. Secondary specialties of vascular and critical care surgery were associated with less-focused practices relative to no secondary specialty. DISCUSSION The substantial variation in the volume and breadth of procedures performed by individual North Carolina general surgeons has implications for our understanding of the supply and distribution of surgical care. First, of the 601 general surgeons who reported 2009 Lippincott Williams & Wilkins

7 King et al Annals of Surgery Volume 249, Number 6, June 2009 TABLE 6. Ordinary Least Squares Estimates of Breadth of Practice, NC General Surgeons 2004 No. Different Types of Procedures Performed Percent of Total Procedure Volume Concentrated in Top 10 Procedures Coefficient 95% CI Coefficient 95% CI Primary practice location Micropolitan county , , 1.9 Rural county , , 6.2 Age , , 0.29 Age squared* , Female , , 8.6 Secondary specialty Vascular surgery , , 4.0 Abdominal surgery , , 6.0 Thoracic surgery , , 2.4 Oncology surgery , , 3.4 Critical care surgery 18 12, , 11 Primary care , , 8.3 Other , , 9.9 Colon and rectal surgery , , 3.5 Constant , , 64 Data are based on active nonfederal physicians with a primary specialty of general surgery who performed at least 50 procedures in 2004 (N 517). Omitted categories are metropolitan county, male, and no secondary specialty. 95% confidence intervals are based on robust standard errors. *Specification tests showed a nonlinear relationship between age and the number of different types of procedures performed. F-test for joint significance of age and age-squared 21.79, P FIGURE 3. Percent of total procedure volume concentrated in top 10 procedure types, NC general surgeons, Data are based on active nonfederal physicians with a primary specialty of general surgery who performed at least 50 procedures in 2004 (N 517). The line in the middle of each box represents the median, and the boundaries of the boxes represent the 25th and 75th percentiles. The whiskers extend to 1.5 times the interquartile range. Dots outside the whiskers represent outliers. *P 0.01, 2-tailed t test of null hypothesis that mean is equal to metropolitan. **P 0.05, 2-tailed t test of null hypothesis that mean is equal to micropolitan. being in active practice in the licensure files and who were matched to the discharge data, 57 had no record of performing inpatient or ambulatory surgery procedures during the year. Even if we assume that the 47 general surgeons who were not matched to the discharge files (because they were missing UPIN) performed procedures in the year, traditional head counts that assume active general surgeons are providing general surgical care would overestimate the workforce by 57 persons, or 9%. An additional 4% (27) of general surgeons performed 50 or fewer procedures during the year. These 84 surgeons were disproportionately located in metropolitan counties only 9 were in micropolitan counties and 1 was in a rural county which suggests that this over-count is greatest in metropolitan areas. It is puzzling that these general surgeons who performed no or very few procedures during the year reported spending an average of more than 40 hours per week in clinical care on their licensure renewal applications. It is possible that these surgeons were involved in patient care activities other than inpatient or ambulatory surgery center procedures, or that the self-reported work hours are inaccurate. Further, among all surgeons with at least 1 procedure in the year, the most productive one-quarter of North Carolina surgeons performed at least twice as many procedures as the lowest volume quarter. Differences in the intensity of these procedures may be driving some of this variation surgeons who perform more complex, time-intensive procedures may perform a smaller Lippincott Williams & Wilkins

8 Annals of Surgery Volume 249, Number 6, June 2009 Rural and Urban Practice Patterns number of procedures during the year. Nevertheless, the volume of surgical procedures provided by North Carolina general surgeons varies widely. Differences in breadth of practice also are relevant to workforce supply issues. Considering the number of different procedures performed at least once during the year, the one-fourth of surgeons with the most diverse practices performed procedures from at least 38% more categories than surgeons in the least diverse one-fourth. And the degree to which surgeons focused their practices on their 10 most common procedures ranged from 37% to 100%, with 13% of general surgeons focusing at least four-fifths of their practices on 10 procedures. General surgeons with narrower scopes of practice may not be comfortable performing the range of surgical procedures that general surgeons are initially trained to provide. There is concern about the impact of this phenomenon, especially for the availability of emergency surgical care. 15 These results also have implications for surgical education an issue directly linked to the size and geographic distribution of the workforce. Amid the move toward fast-track subspecialization in surgery, there have been calls for special rural training tracks, based on the unique challenges faced by surgeons in more isolated areas and the conception that rural surgeons are in greater need of broad-based training than surgeons who plan to practice in urban areas. 9,16 18 We did find some differences between rural and urban surgeons. Compared with their metropolitan counterparts, rural surgeons performed more minor diagnostic procedures and fewer major therapeutic procedures, and alimentary tract procedures played a larger role in their practices. This likely is driven in part by the fact that the average rural general surgeon performed many more colonoscopy and biopsy and upper gastrointestinal endoscopy and biopsy procedures than the average metropolitan general surgeon. Other studies have had similar findings with regard to the role of endoscopic procedures in rural areas. 6,7,19 However, our results show that North Carolina general surgeons in metropolitan areas often had practices that were as broad as or broader than their rural counterparts. While some general surgeons in metropolitan counties were more specialized than those in rural counties, we did not find evidence that the scope of practice for rural general surgeons was significantly more diverse. We also found that the average rural general surgeon dedicated a larger share of his total procedure volume to his top 10 procedures than the average metropolitan surgeon; according to this measure, rural surgeons had slightly more focused practices. These results contrast with other studies that concluded that rural general surgeons have broader practices than those in urban areas. 720 Methodological differences may account for some of this difference. Some studies have relied on surveys or case logs to characterize scope of practice, while our study used administrative data on all inpatient and ambulatory procedures performed in a year. Further, we measured diversity and breadth of practice at the individual surgeon level, rather than comparing the range of procedures performed by all surgeons in a geographic area. Our study was limited to one state, North Carolina. This state may be well-suited for a comparison of rural and urban practice patterns; about one-quarter of North Carolina general surgeons have primary practice locations in nonmetropolitan counties compared with 7.8% of general surgeons nationally, according to tabulations of the American Medical Association Masterfile. 21 And 31% of North Carolina s population lives in nonmetropolitan counties, compared with 16% of all US residents. 22 However, the practice patterns we observed among general surgeons in North Carolina could be different than patterns elsewhere. A range of factors influence physicians practice patterns, including differences in patient populations, physician characteristics, local delivery systems, and local practice norms. Based on crude indicators such as age, gender, and insurance status, the North Carolina population is very similar to the overall US population. 23 And, while North Carolina general surgeons were slightly older in 2004 (average age was 48.4) and less likely to be female (8.2%) than general surgeons nationally (average age was 45.6 years and 11.3% were women), the differences are small. 21 In regards to local delivery systems, one relevant factor in North Carolina is the presence of 6 general surgery residency programs. All 6 programs are located in metropolitan counties and 5 are at teaching hospitals (The 6 general surgery residency programs in North Carolina are located at University of North Carolina Hospitals Chapel Hill ; Carolinas Medical Center Charlotte ; Duke University Medical Center Durham ; East Carolina University Affiliated Hospitals Greenville ; New Hanover Regional Medical Center Wilmington ; and Wake Forest University School of Medicine Winston-Salem ). These training programs may influence the surgical practice patterns in urban and rural areas of the state. Depending on the referral patterns between rural surgeons and their counterparts in metropolitan tertiary care settings, certain procedures may be more or less likely to be performed in rural communities. Also, it may be that rural hospitals lack the infrastructure (eg, technology and intensive care units) to support certain surgical procedures and thus these procedures are referred out of rural communities. We do not know whether these dynamics are different in North Carolina than in other states. In short, similar analyses of national or other state-level data will be necessary to understand whether our findings extend to other geographic areas. One limitation of this analysis is that physician specialty on the licensure file is self-reported. However, studies of the American Medical Association Physician Masterfile, which also relies on self-reported specialty, have found that these self-reported data generally are consistent with medical training and board certification. 24 Also, the number of general surgeons in our analysis is similar to the number of general surgeons in the ABS Diplomates listing: as of 2008, the ABS listing contains 537 surgeons who have a current ABS certificate and no other certificate and who have a North Carolina mailing address (T.W. Biester, communication, October 2008). Although the use of administrative data on inpatient and ambulatory procedures is a strength of this analysis, it presents the potential for error in the measurement of procedures performed by individual surgeons. For example, it is possible that the surgeon reported as attending in the discharge database did not actually perform all procedures on the record. Despite these limitations, this study provides empirical information about the extent to which general surgeons practice patterns vary from individual to individual and the role that urban or rural practice location plays in this variation. Such factors will be important to consider when making future decisions related to medical education and the size and distribution of the general surgery workforce. REFERENCES 1. Donini-Lenhoff FG, Hedrick HL. Growth of specialization in graduate medical education. JAMA. 2000;284: Barondess JA. Specialization and the physician workforce: drivers and determinants. JAMA. 2000;284: Brotherton SE, Rockey PH, Etzel SI. US graduate medical education, JAMA. 2005;294: Stitzenberg KB, Sheldon GF. Progressive specialization within general surgery: adding to the complexity of workforce planning. J Am Coll Surg. 2005;201: Vascular Surgery Board of the American Board of Surgery. Primary certificate passes final hurdle. VSB-ABS Newsletter; Available at: Lippincott Williams & Wilkins

9 King et al Annals of Surgery Volume 249, Number 6, June 2009 gery.org/default.jsp?certvsqe_primarycert&ref index_pd. Accessed May 14, Ritchie WP, Rhodes RS, Biester TW. Work loads and practice patterns of general surgeons in the United States, : a report from the American Board of Surgery. Ann Surg. 1999;230: Heneghan SJ, Bordley J IV, Dietz PA, et al. Comparison of urban and rural general surgeons: motivations for practice location, practice patterns, education requirements. J Am Coll Surg. 2005;201: Breon TA, Scott-Conner CE, Tracy RD. Spectrum of general surgery in rural Iowa. Curr Surg. 2003;60: Finlayson SR. Surgery in rural America. Surg Innov. 2005;12: North Carolina General Statutes, chap 131E, article 11a. 11. Deleted in proof. 12. HCUP CCS Fact Sheet. Healthcare Cost and Utilization Project (HCUP). November Agency for Healthcare Research and Quality, Rockville MD. Available at: jsp. Last modified November 28, HCUP Procedure Classes. Healthcare Cost and Utilization Project (HCUP). September Agency for Healthcare Research and Quality, Rockville, MD. Available at: procedure.jsp. Last modified September 27, About Metropolitan and Micropolitan Statistical Areas. US Census Bureau, Population Division, Population Distribution Branch. Available at: Last modified June 7, Russell TR. The surgical workforce: averting a patient access crisis. Surg Clin NAm. 2007;87: Cogbill TH. Training surgeons for rural America. Am Surg. 2007;73: Hunter JG, Deveney KE. Training the rural surgeon: a proposal. Bull Am Coll Surg. 2003;88: Shively EH, Shively SA. Threats to rural surgery. Am J Surg. 2005;190: Landercasper J, Bintz M, Cogbill TH, et al. Spectrum of general surgery in rural America. Arch Surg. 1997;132: Gates RL, Walker JT, Denning DA. Workforce patterns of rural surgeons in West Virginia. Am Surg. 2003;69: Authors tabulations of 2006 AMA Masterfile. 22. CO-EST2007-alldata. US Census Bureau, Population Division. Available at: Accessed September 10, The Kaiser Family Foundation, statehealthfacts.org. Data Source: Urban Institute and Kaiser Commission on Medicaid and the Uninsured estimates based on the Census Bureau s March 2006 and 2007 Current Population Survey (CPS: Annual Social and Economic Supplements). Accessed September 10, American Medical Association. Physician education, licensure, and certification. Available at: Accessed May 13, Lippincott Williams & Wilkins