- Email: [email protected]
Who performs endocrine operations in the United States?
Who performs endocrine operations in the United States? Brian D. Saunders, MD, Reid M. Wainess, BS, Justin B. Dimick, MD, Gerard M. Doherty, MD, Gilbert R. Upchurch, MD, and Paul G. Gauger, MD, Ann Arbor, Mich
Background. Endocrine surgery is a discipline that is dedicated to high-quality care of patients with endocrine surgical disease. The relationship between its ‘‘identity’’ as a separate field and clinical practice patterns is not known. Methods. The National Inpatient Sample was searched by the International Classification of Diseases– 9th revision–Clinical Modification codes for parathyroidectomy, thyroidectomy, and adrenalectomy for the years 1988 through 2000. The surgeons who performed these operations were profiled by 2 methods: Method A, by the percentage of the total primary International Classification of Diseases–9th revision– Clinical Modification procedure codes that were the selected endocrine procedures; method B, by absolute number of index endocrine procedures performed per year. Only patients with complete coding data for the surgeons were included. Results. In this sample, surgeons whose practice was comprised of 25% or less of these endocrine procedures performed 11,071 parathyroidectomies (78% of total), 46,210 thyroidectomies (82% of total), and 4209 adrenalectomies (94% of total). In contrast, surgeons whose practice was comprised of more than 75% of these endocrine procedures performed 769 parathyroidectomies (5% of total), 1560 thyroidectomies (3% of total), and 128 adrenalectomies (3% of total). Conclusion. If these data can be extrapolated to indicate generalized practice patterns, the majority of common operations for endocrine disease are performed by surgeons whose practice is not focused on endocrine surgery. However, much of this effect is due to the fact that non–endocrine surgeons far outnumber endocrine surgeons. This understanding of clinical practice patterns will be important to consider during future studies that seek to determine the relationship between surgeon volume and patient outcomes. (Surgery 2003;134:924-31.) From the Department of Surgery, Divisions of Endocrine Surgery and Vascular Surgery, University of Michigan Medical Center, Ann Arbor, Mich
IMPROVEMENTS IN THE UNDERSTANDING of embryology and pathophysiology, as well as the technical aspects of operative therapies, have allowed for the emergence of endocrine surgery as a defined field of concentration within the larger discipline of general surgery. This specialty is dedicated to delivering high-quality and high-volume specialized surgical care to patients with endocrine disease. The indications for operations on endocrine glands have expanded as diagnostic and localization modalities have improved and as intraoperative and postoperative care have become more refined. The spectrum of intraoperative tools Presented at the 24th Annual Meeting of the American Association of Endocrine Surgeons, San Diego, California, May 1l-14, 2003. Reprint requests: Brian D. Saunders, MD, 2920H Taubman Center, 1500 E. Medical Center Dr, Ann Arbor, MI 48109-0331. Ó 2003, Mosby, Inc. All rights reserved. 0039-6060/2003/$30.00 + 0 doi:10.1016/S0039-6060(03)00420-3
924 SURGERY
in the armamentarium of the endocrine surgeon has also grown to include minimally invasive procedures, intraoperative hormone monitoring, intraoperative ultrasound scanning, laparoscopy, and video-assisted surgery. Despite the growing complexity of the field, there remains no formal certification to signify additional expertise in endocrine surgery. The case number and type of endocrine operations that are performed by graduating surgical chief residents varies widely across the country. This fact is known from data that are tracked by the Residency Review Committee.1,2 Currently, however, a thorough understanding of the national patterns of endocrine surgery (as defined by clinicians in practice) is lacking. Knowledge of these clinical practice patterns would allow for a more accurate study of surgical outcomes in the context of surgeon volume and would facilitate the planning of future needs for endocrine surgeon specialists. The current study uses a database that is representative of surgical practice in the United
Surgery Volume 134, Number 6
States to characterize the clinicians who are performing endocrine procedures and to determine ‘‘who’’ is performing endocrine operations. METHODS The National Inpatient Sample (NIS) is a stratified, cross-sectional database that includes approximately 20% of all non-Veterans Affairs hospital discharges in the United States. The NIS was borne out of federal, state, and industry partnership and is managed under the Health Care Cost and Utilization Project of the Agency for Healthcare Research and Quality.3 This study’s data were derived from the version of the NIS database that encompasses calendar years 1988 through 2000. The number of states whose hospital discharges were represented varied over the study period. There were 759 hospitals in 8 states in 1988 and 994 hospitals in 28 states in 2000. An average of 6.5 3 106 hospital discharges per year was included in the NIS database (Table I). The database was searched for the individual years 1988 through 2000, selecting out hospital discharges for which the primary International Classification of Diseases, 9th revision, clinical modification (ICD-9-CM) procedure code was for parathyroidectomy, hemior total thyroidectomy, and adrenalectomy.4 The specific ICD-9-CM procedure codes for parathyroidectomy, thyroidectomy, and adrenalectomy (hereafter referred to collectively as endocrine codes) that were used are indicated in Table II. Endocrine pancreatic procedures (eg, enucleation of islet cell tumor) were excluded because of predictably small numbers and variations in the coding of these pancreatic procedures. The hospital discharges were excluded if there was incomplete or absent procedural coding data for the surgeon (range, 53.5%-66.2%). Because individual surgeons cannot be identified by name or professional affiliation in the NIS, a method of practice profiling was required. We identified 2 separate models that were intended to group surgeons into quartiles. The first (method A) expressed concentration in endocrine surgery as a percentage of any individual surgeon’s caseload during a defined time period (1 year). The second model (method B) grouped surgeons into quartiles on the basis of the absolute number of index endocrine cases that were performed as compared with other surgeons within the same defined time period. Once a hospital discharge was selected, method A required that all of the primary ICD-9-CM procedure codes for the surgeon that were linked
Saunders et al 925
to that discharge were screened to determine the proportion of endocrine procedures that were performed over the study period. The specific algorithm that was used was to divide the total number of primary endocrine procedure codes by the total number of primary procedures codes. The surgeons were then grouped into quartiles (group A, 0%-25%; group B, 26%-50%; group C, 51%-75%; and group D, 76%-100%). Method B used the same dataset of linked surgeons and patients after exclusion for incomplete or absent coding data. The absolute number of index endocrine surgery cases (as defined by the codes given earlier) were determined per surgeon per year. The statistical breakpoints into quartiles were then determined. All descriptive statistical analyses were performed with STATA 6.0 (Stata Corp, College Station, Texas). RESULTS Parathyroid. There were 30,606 patients who were identified whose hospital discharge had a primary procedural ICD-9-CM code for parathyroidectomy. Of these, there was complete procedural coding information for the surgeons for 14,232 patients (46.5%). Associated with these 14,232 patients were 6100 unique surgeons (Table III). Stratification of the surgeons into quartiles by the percent of endocrine procedures (method A) in their practice yielded 5587 surgeons in group A, 343 surgeons in group B, 101 surgeons in group C, and 69 surgeons in group D (Table IV). Group A contained 11,071 patients (or 78%); group B contained 1833 patients (13%); group C contained 559 patients (4%); and, group D contained 769 patients (5%; Fig 1, A). This pattern was validated generally by the use of method B for the same dataset. In specific, the quartile break points indicated that the 2995 surgeons in group A (lowest absolute volume and 49% of all surgeons) performed 1 to 3 parathyroid operations per year. One thousand eight hundred fifty-eight group B surgeons (30% ) performed an average of 4 to 8 operations per year; 1005 group C surgeons (17% ) performed from 9 to 26 operations per year. The quartile with the highest absolute volume was group D, in which 242 surgeons (4%) performed 27 to 182 parathyroid operations per year (Fig 2, A). Thyroid. There were a total of 123,716 patients who were identified whose hospital discharge had a primary procedural ICD-9-CM code for thyroidectomy. Of these, 56,286 patients (45.5%) had complete procedural coding data for the associated
926 Saunders et al
Surgery December 2003
Table I. NIS data sources from 1988 through 2000 Calendar year
Data source
Hospitals (n) Inpatient stays (n)
1988 1989 1993 1995 1997 1999 2000
CA, CO, FL, IL, IA, MA, NJ, WA Added AZ, PA, WI Added CT, KS, MD, NY, OR, SC Added MO, TN Added GA, HI, UT Added ME, VA AZ, CA, CO, CT, FL, GA, HI, IL, IA, KS, KY, MA, MD, ME, MO, NC, NJ, NY, OR PA, SC, TN, TX, UT, VA, WA, WI, WV (added KY, NC, TX, WV)
759 882 913 938 1012 984 994
5,265,756 6,110,064 6,538,976 6,714,935 7,148,420 7,198,929 7,450,992
Table II. Descriptive lexicon that is associated with the ICD-9-CM procedure codes that were used to query the NIS database Parathyroid Thyroid
Adrenal
Procedure code
Lexicon
0681 0689 062 0639 064 0650 0651 0652 072 0721 0722 0729 073
Complete parathyroidectomy Other parathyroidectomy: parathyroidectomy NOS, partial parathyroidectomy Unilateral thyroid lobectomy Other partial thyroidectomy: isthmusectomy, partial thyroidectomy NOS Complete thyroidectomy Substernal thyroidectomy NOS Partial substernal thyroidectomy Complete substernal thyroidectomy Partial adrenalectomy Excision of lesion of adrenal gland Unilateral adrenalectomy Other partial adrenalectomy: partial adrenalectomy NOS Bilateral adrenalectomy: excision of remaining adrenal gland
NOS, Not otherwise specified.
surgeon and were thus included in our study. There were 17,475 unique surgeons who were associated with these 56,286 patients (Table III). The use of method A to stratify these surgeons into quartiles yielded 16,257 surgeons in group A, 748 surgeons in group B, 260 surgeons in group C, and 210 surgeons in group D (Table IV). The number of patients per quartile was 46,210 in group A (82%); 6425 in group B (11%), 2091 in group C (4%), and 1560 in group D (3%; Fig 1, B). As with the parathyroid data, this pattern is validated with the use of method B for the same dataset. In specific, the quartile break points indicated that the 10,620 surgeons in group A (lowest absolute volume; 61% of all surgeons) performed 1 to 2 thyroid operations per year. Four thousand one hundred six surgeons in group B (23%) performed an average of 3 to 5 operations per year; 2187 group C surgeons (13%) performed from 6 to 15 operations per year. The quartile with the highest absolute volume was group D, in which 562 surgeons (3%) performed 16 to 182 thyroid operations per year (Fig 2, B).
Adrenal. There were a total of 13,229 patients who were identified whose hospital discharge had a primary procedural ICD-9-CM code for adrenalectomy. Of these, 4468 patients had complete procedural coding data for their surgeons to allow inclusion. This represents 33.8% of the total number of adrenalectomies that were identified in the database. There were 3050 unique surgeons who were identified with these 4468 patients (Table III). Stratification of these surgeons into quartiles with method A yielded 2946 surgeons in group A, 58 surgeons in group B, 19 surgeons in group C, and 27 surgeons in group D (Table IV). The number of patients per quartile was 4209 in group A (94%), 103 in group B (2%), 28 in group C (less than 1%), and 128 in group D (3%; Fig 1, C ). Again, this distribution of case volume is essentially validated by using Method B for the same dataset. The quartile break points indicated that the 1,375 surgeons in group A (lowest absolute volume and 45% of all surgeons) performed 1 adrenal operation per year. Four hundred seventyone group B surgeons (15%) performed an average
Saunders et al 927
Surgery Volume 134, Number 6
Table III. Patients (total and ultimately included) and unique surgeons who were identified in the NIS database (1988-2000) with a primary endocrine procedure code Gland
Total patients (n)
Patients with surgeon identifier (n)
Patients included (%)
Unique surgeons (n)
Parathyroid Thyroid Adrenal
30,606 123,716 13,229
14,232 56,286 4,468
46.5 45.5 33.8
6,100 17,475 3,050
Table IV. Distribution of surgeons (accrued over the entire study period) to quartiles an the basis of surgeon percentage of endocrine procedures (method A) Gland
Total surgeons profiled (n)
Quartile A: 0-25% (n)
Quartile B: 26%-50% (n)
Quartile C: 51%-75% (n)
Quartile D: 76%-100% (n)
Parathyroid Thyroid Adrenal
6,100 17,475 3,050
5,587 (91.6%) 16,257 (93.0%) 2,946 (96.6%)
343 (5.6%) 748 (4.3%) 58 (1.9%)
101 (1.7%) 260 (1.5%) 19 (0.6%)
69 (1.1%) 210 (1.2%) 27 (0.9%)
of 2 operations per year; 679 group C surgeons (23%) performed from 3 to 6 operations per year. The quartile with the highest absolute volume was group D, in which 525 surgeons (17%) performed 7 to 167 adrenal operations per year (Fig 2, C ). DISCUSSION This study attempted to characterize the practice patterns of endocrine surgery within the context of the overall practice of surgery in the United States. The great majority of operations for endocrine disease are performed by surgeons whose clinical practice is not focused heavily in endocrine surgery. Additionally, the number of surgeons whose practice is focused on endocrine surgery comprises a very small percentage of the total number of surgeons who are performing common operations for endocrine disease. There have been limited efforts to determine the volume of operations for endocrine disease that are performed throughout the United States. Many published reports in endocrine surgery come from single centers. There is, however, a growing medical and lay interest in the relationship between surgeon volume and patient outcome. To date, much of this work has focused on mortality rates and the inhospital length of stay for selected complex general surgery and vascular surgery operations.5,6 Since the foundation of both national and international endocrine surgical associations, there has been an emphasis on the improvement of the care that is given to patients with endocrine disease, the dissemination of new diagnostic and therapeutic advancements, and the teaching of future practitioners of endocrine surgery. Endocrine surgery as a separate field remains self-identified and
without formal certification. There are those who argue against specific fellowship training of general surgery graduates, claiming that procedures such as parathyroidectomy, thyroidectomy, and adrenalectomy should be in the armamentarium of most well-trained general surgeons. Endocrine surgical procedures continue to be performed by endocrine surgeons, general surgeons, transplantation surgeons, urologists, and otorhinolaryngologists. Previously, there was no large study that sought to identify and to quantify the profile of the clinicians who actually were performing common operations for endocrine disease. The NIS was designed to provide a research database for conducting national and regional studies of inpatient care throughout the United States. It has a large size, reflects all payers, and is longitudinal in nature. These qualities allow for research on health care delivery, cost, use, access, and quality at the patient, physician, market, and state levels. We were able to take advantage of the design of the NIS to characterize clinicians who perform operations for endocrine disease. Importantly, the database contains an accurate crosssection of the hospital discharges in the United States and is not limited to academic health centers, as is the case for many other series that have reported endocrine surgery. Another desirable feature of the NIS is the inability for individual surgeons to be identified by name or by declared surgical specialty. The validity of the database continues to be documented through many other studies that have made use of it.7-12 There are some limitations to the NIS database. Of the patients who were identified in this study as having undergone parathyroid, thyroid, or adrenal
928 Saunders et al
Fig 1. Contribution of individual surgeon quartiles to the total number of patients who underwent operation, as determined by the percentage of practice calculation and accrued over the entire study period (method A). A, Parathyroid; B, thyroid; C, adrenal.
surgery, 53.5% to 66.2% were necessarily excluded because of insufficient coding data for the surgeons who were associated with the hospital discharge abstract. (A hospital discharge abstract is the electronic database record of a patient’s complete hospital stay.) Although this will affect the total
Surgery December 2003
numbers, this would not affect the conclusion of this analysis because of the large sample size that is afforded by the NIS database and because of the random nature of the exclusions. In effect, these exclusion criteria limit our sample to approximately 40% of patients in this 20% stratified, crosssectional sample of non-Veterans Affairs hospitals across the United States. This is essentially a ‘‘biopsy’’ of the dataset, and therefore, it can serve only to indicate patterns and trends that are not absolute or universal. There is some danger in the extrapolation of the numbers in this study to the experience of an individual surgeon because of this factor. There are a greater number of ICD-9-CM codes for thyroid surgery than for parathyroid or adrenal surgery (6 vs 5 and 2, respectively). This could allow more specific coding on discharge abstracts and thus a potentially higher capture rate for thyroid surgery in our study. There was no separate ICD-9CM procedure code for laparoscopic adrenalectomy during the study period. Therefore, some of these procedures may have been missed in our search algorithm because of coding variations. The NIS contains hospital discharge abstract data from inpatients only. Therefore, operations for endocrine disease, particularly thyroid and parathyroid operations, that were performed on an outpatient basis will not be included in this study. The NIS includes only non-federal hospitals. This necessarily excludes procedures that are performed at Veterans Affairs hospitals. Most Veterans Affairs hospitals are staffed by surgeons who are affiliated with academic health centers. However, it is not common for a surgeon to practice only endocrine surgery at a Veterans Affairs hospital. Because it is a reasonable assumption that many of the surgeons whose practices focus on endocrine surgery are located at an academic health center, this study may fail to capture some of the cases that have been performed by these surgeons at a Veterans Affairs hospital. Predictably, our data demonstrate that those surgeons whose practice is comprised of more than 76% endocrine surgical procedures do in fact perform more parathyroidectomies, thyroidectomies, and adrenalectomies per surgeon than those surgeons whose practice is less than 25% endocrine surgical procedures in terms of individual volume. However, despite this, most patients who undergo operations for endocrine disease in the United States have the procedure performed by surgeons whose practice is comprised of less than 25% endocrine surgery. This is a result of the fact that surgeons who perform endocrine surgery but do
Surgery Volume 134, Number 6
not specialize in this field far outnumber those surgeons whose practice is limited essentially to endocrine surgery. The surgeons in this study whose practice is comprised of greater than 76% endocrine surgical procedures make up only 1.1% of the surgeons who are performing parathyroidectomies, 1.2% of the surgeons who are performing thyroidectomies, and 0.9% of the surgeons who are performing adrenalectomies. When profiled by absolute number of operations, the surgeons in the highest quartile are only 4% of the surgeons who are performing parathyroidectomies, 3% of the surgeons who are performing thyroidectomies, and 17% of the surgeons who are performing adrenalectomies. Put another way, 50% of patients who undergo parathyroid operations have the procedure performed by a surgeon that does 8 or fewer parathyroid operations per year, and 50% of patients who undergo thyroid operations have them performed by surgeons who do an average of 5 or fewer of these operations per year. For adrenalectomy, 50% of patients have the procedures performed by a surgeon who averages 2 or fewer procedures per year. There are a number of reasons that can be postulated to explain the fact that most patients who undergo common endocrine surgical procedures in the United States are being treated by surgeons whose practice is comprised of less than 25% endocrine surgery. As noted earlier, there are simply many fewer surgeons who specialize in endocrine surgery. There are approximately 300 members who are currently in the American Association of Endocrine Surgeons. This number stands in contrast to the more than 50,000 current members in the American College of Surgeons. This may, in part, be due to limited exposure to endocrine surgery during formal residency training. Endocrine disease is relatively uncommon, and most graduating chief residents do not have adequate exposure to common operations for endocrine disease during their residency.1,2 Further, most graduating chief residents do not have direct training under a surgeon who specializes in endocrine surgery.13 This study does not examine the outcomes that were associated the patients who were identified. Much of the published literature on outcomes and complication rates of endocrine surgeries has emanated from the highest volume endocrine surgeons at academic health centers that correlate to quartile D in the method used in this study. The present study shows that this group of surgeons is treating a minority of the patients who undergo operations for endocrine disease. We currently do not know the actual outcomes of operations for
Saunders et al 929
Fig 2. Distribution of the surgeons to quartiles, as determined by the absolute number of index endocrine ICD-9-CM procedure codes per year. A, Parathyroid; B, thyroid; C, adrenal.
endocrine disease that were performed by surgeons in the other 3 quartiles. This information is especially important because reoperative endocrine surgery is associated with increased morbidity and cost.14 This trend of improved outcomes that relate to increased volume dates back to the founding of surgery on endocrine glands. Kocher was the first high-volume thyroid surgeon. As he accrued more patients throughout his career, there was a dramatic decrease in mortality rate from thyroidectomy.15 Two recent studies have specifi-
930 Saunders et al
cally looked at high-volume parathyroid and thyroid surgeons. Chen et al16 and Sosa et al17 have demonstrated improved cure rates and decreased complication rates and length of stay for a single, high-volume endocrine surgeon when compared with larger numbers of general surgeons who intermittently perform operations for endocrine disease. There are few ICD-9-CM diagnostic codes for outcomes of endocrine surgeries. The NIS database only looks at hospital discharge data. Many of the complications from operations for endocrine disease (ie, recurrent laryngeal nerve dysfunction, hypoparathyroidism, and adrenal hormone insufficiencies) may not be evident or documented fully until follow-up after hospital discharge and would not be captured by analysis of this database. The NIS therefore may not be suitable for outcome-based analysis of most endocrine surgical procedures. If an association of surgical volume to outcome exists, this would then support a referral pattern of patients who require endocrine surgery to highvolume surgeons. Not only would this allow patients to be seen and operated on by endocrine surgeons, but also it would allow for them to be evaluated by and receive therapy from other specialists who are necessary for their care. The development of interdisciplinary teams that include endocrine surgeons, medical endocrinologists, nuclear medicine physicians, radiologists, and pathologists has been a goal of many practitioners in the field of endocrine surgery.18 However, one cannot use this study to determine the relationship between volume and patient outcome. Although it is likely that a volume outcome effect exists in endocrine surgery, it has not been documented rigorously since the reports of Sosa et al17 and Chen et al16 and will require further careful study. Additionally, there is currently a paucity of published, benchmarking outcome data that are based on series from lower volume surgeons.19 It remains complete conjecture as to the threshold whereby a low-volume endocrine surgeon becomes a high-volume endocrine surgeon. If a true volume/outcome effect is proved eventually for endocrine surgical procedures, there may then be an increased need for specialisttrained endocrine surgeons to provide the care that is required by these patients. Currently, there is no separate board certification process for endocrine surgery. Many general surgeons who perform operations for endocrine disease get no more training than that provided during residency. Harness et al1,2 published results of a Residency
Surgery December 2003
Review Committee database survey several years ago that showed the inadequate level of exposure of graduating general surgery chief residents to common and uncommon endocrine surgical procedures. The latest follow-up to this data was again discussed by Harness18 in his 2000 American Association of Endocrine Surgeons Presidential lecture. Formal fellowship training in endocrine surgery is not easily accessible and may need to become more widespread if endocrine surgical procedures become limited to those surgeons who specialize in endocrine surgery. Most operations for endocrine disease in the United States are performed by surgeons whose practice is less than 25% endocrine. This is reflective of the fact that the number of surgeons who specialize in endocrine surgery is significantly smaller than the number of surgeons whose practice does not focus on endocrine surgery. This pattern will play a significant role in the analyses of surgical outcome and the planning of future care models for patients with endocrine disease. REFERENCES 1. Harness JK, Organ CH Jr, Thompson NW. Operative experience of US general surgery residents in thyroid and parathyroid disease. Surgery 1995;118:1063-70. 2. Harness JK, Organ CH Jr, Thompson NW. Operative experience of US general surgery residents with diseases of the adrenal gland, endocrine pancreas, and other less common endocrine organs. World J Surg 1996;20:885-91. 3. Steiner C, Elixhauser A, Schnaier J. The healthcare cost and utilization project: an overview. Effective Clin Pract 2002;5:143-51. 4. Public Health Service, Health Care Financing Administration. International Classification of Diseases, 9th revision, Clinical Modification. Washington (DC): US Department of Health and Human Resources;1991. 5. Gordon TA, Burleyson GP, Tielsch JM, Cameron JL. The effects of regionalization on cost and outcome for one general high risk surgical procedure. Ann Surg 1995;221: 43-9. 6. Dimick JB, Stanley JC, Axelrod DA, Kazmers A, Henke PK, Jacobs LA, et al. Variation in death rate after abdominal aortic aneurysmectomy in the United States: impact of hospital volume, gender, and age. Ann Surg 2002;235: 579-85. 7. Shen JJ. Severity of illness, treatment environments, and outcomes of treating acute myocardial infarction for Hispanic Americans. Ethn Dis 2002;12:488-98. 8. Cowan JA, Dimick JB, Thompson BG, Stanley JC, Upchurch GR Jr. Surgeon volume as an indicator of outcomes after carotid endarterectomy: an effect independent of specialty practice and hospital volume. J Am Coll Surg 2002;195: 814-21. 9. Chang RK, Chen AY, Klitzner TS. Clinical management of infants with hypoplastic left heart syndrome in the United States, 1988-1997. Pediatrics 2002;110:292-8. 10. Xiao H, Campbell ES, Song KS. A trend analysis of organ transplantation among ethnic groups. J Natl Med Assoc 2002;94:15-20.
Surgery Volume 134, Number 6
11. Nietart PJ, Silverstein MD, Silver RM. Hospital admissions, length of stay, charges, and in-hospital death among patients with systemic sclerosis. J Rheumatol 2001;28:2031-7. 12. Howard TS, Hoffman LH, Stang PE, Simeos EA. Respiratory syncytial virus pneumonia in the hospital setting: length of stay, charges, and mortality. J Pediatr 2000;137:227-32. 13. Prinz RA. Presidential address: endocrine surgical training: some ABC measures. Surgery 1996;120:905-12. 14. Doherty GM, Weber B, Norton JA. Cost of unsuccessful surgery for primary hyperparathyroidism. Surgery 1994; 116:954-8. 15. Becker WF. Pioneers in thyroid surgery. Ann Surg 1977; 185:493-504. 16. Chen H, Zeiger MA, Gordon TA, Udelsman R. Parathyroidectomy in Maryland: effects of an endocrine center. Surgery 1996;120:948-52. 17. Sosa JA, Bowman HM, Tielsch JM, Powe NR, Gordon TA, Udelsman R. The importance of surgeons experience for clinical and economic outcomes from thyroidectomy. Ann Surg 1998;228:320-30. 18. Harness JK. Presidential address: interdisciplinary care: the future of endocrine surgery. Surgery 2000;128:873-80. 19. Reeve TS, Curtin A, Fingleton L, Kennedy P, Machie W, Porter T, et al. Can total thyroidectomy be performed as safely by general surgeons in provincial centers as by surgeons in specialty endocrine surgical units? Arch Surg 1994;129:834-6.
DISCUSSION Dr Herbert Chen (Madison, Wis). If you were a surgeon who treated 500 cases a year, for instance, and one half of them were endocrine cases (you would have 250 endocrine), then you would be put in the B category; whereas if you treated only 100 cases and 80 of them were endocrine, you would be in the high group but would have performed only one third of the number of endocrine cases as the first example. Therefore, have you looked at your data with regard to absolute number of endocrine cases and seen that the data might shake out a little differently than you have presented here? Dr Saunders. I agree that a potential limitation of this study is the specific method that was used to define the quartiles. A limitation of these data is that there are very few surgeons in our highest quartile. We have not yet performed that analysis, but I think that recategorizing the surgeons according to the absolute number of the number of operations performed probably will support the results of the model as presented. Dr Ashok R. Shaha (New York, NY). Not to open Pandora’s box, but I am sure that, when you started the study and started analyzing the data, you and your coauthors must have gone through a lot of soul-searching about who is an endocrine surgeon, and then you built up these quartiles about the practice. What was your discussion? How did you define an endocrine surgeon? Ultimately you came up with a high-volume/low-volume endocrine surgeon. What should be the ideal number? Dr Saunders. At this point in our analysis, we do not have a discreet number that would define a threshold
Saunders et al 931
between a low-volume and a high-volume endocrine surgeon. What this study looked at was the percentage of the practice of the surgeons who were identified that was comprised of endocrine surgery. The creation of quartiles was fairly simplistic, but we did not mean to imply where the high-volume and low-volume cut-offs would be. I think that there is a lot more study that would need to be done to determine where the true threshold lies. Dr Richard A. Hodin (Boston, Mass). Did you have a chance to look at different regions throughout the United States. I am aware of the data in our area in Massachusetts, and I can tell you it would not look anything like the data you presented. Boston may be different from a lot of other areas, but for sure, the vast majority of patients who have endocrine surgery have it performed by what would be the higher quartile of surgeons. Dr Saunders. The NIS database can be stratified according to region of the country. We did not do that at this point, but the data do allow that to be done. I agree with you. I have a feeling that, depending on city or area of the country, the numbers will look a little bit different. Dr Karen R. Borman ( Jackson, Miss). Did you, by any chance, look for a self-designated specialty designator for these physicians? For example, in some of the other Department of Human Health and Services databases, they will be able to tell you whether these were otolaryngologists or general surgeons who performed these operations. Dr Saunders. One of the characteristics of the NIS is that one cannot identify specific surgeons. Therefore, unfortunately, there was no way of identifying which surgical subspecialty each of these surgeons belonged to. Dr Clive S. Grant (Rochester, Minn). Could I just clarify my understanding of the data. The study period was approximately 12 or 13 years, and the highest quartiles I saw on that graph showed that, depending on the type of operation, the surgeons might have performed 6 to 14 total cases. This was during the entire study period? If this is true, is the data valid? It would imply that the busiest surgeons only averaged about one case per year. Dr Saunders. I agree with you, that is one of the limitations of this data and this database. The database is somewhat incomplete. However, I think it is made up for by its large numbers. The take-home message of this is more the trends and the practice patterns rather than the specific numbers. In fact, probably it is the numbers in quartile A that are the most accurate because it is the largest number of surgeons. It is a limitation of this database that the number of surgeons in the highest quartile is the lowest. Dr Lawrence A. Danto (Truckee, Calif ). Does the database allow you to look at and did you look at the payment patterns or the payment mechanisms for patients as they affect referral patterns? Dr Saunders. This database reflects all payers, other than those patients at a federal hospital, but it does not allow you to look at reimbursement or payment. Therefore, we did not analyze this factor as it may relate to referral patterns.
Background. Endocrine surgery is a discipline that is dedicated to high-quality care of patients with endocrine surgical disease. The relationship between its ‘‘identity’’ as a separate field and clinical practice patterns is not known. Methods. The National Inpatient Sample was searched by the International Classification of Diseases– 9th revision–Clinical Modification codes for parathyroidectomy, thyroidectomy, and adrenalectomy for the years 1988 through 2000. The surgeons who performed these operations were profiled by 2 methods: Method A, by the percentage of the total primary International Classification of Diseases–9th revision– Clinical Modification procedure codes that were the selected endocrine procedures; method B, by absolute number of index endocrine procedures performed per year. Only patients with complete coding data for the surgeons were included. Results. In this sample, surgeons whose practice was comprised of 25% or less of these endocrine procedures performed 11,071 parathyroidectomies (78% of total), 46,210 thyroidectomies (82% of total), and 4209 adrenalectomies (94% of total). In contrast, surgeons whose practice was comprised of more than 75% of these endocrine procedures performed 769 parathyroidectomies (5% of total), 1560 thyroidectomies (3% of total), and 128 adrenalectomies (3% of total). Conclusion. If these data can be extrapolated to indicate generalized practice patterns, the majority of common operations for endocrine disease are performed by surgeons whose practice is not focused on endocrine surgery. However, much of this effect is due to the fact that non–endocrine surgeons far outnumber endocrine surgeons. This understanding of clinical practice patterns will be important to consider during future studies that seek to determine the relationship between surgeon volume and patient outcomes. (Surgery 2003;134:924-31.) From the Department of Surgery, Divisions of Endocrine Surgery and Vascular Surgery, University of Michigan Medical Center, Ann Arbor, Mich
IMPROVEMENTS IN THE UNDERSTANDING of embryology and pathophysiology, as well as the technical aspects of operative therapies, have allowed for the emergence of endocrine surgery as a defined field of concentration within the larger discipline of general surgery. This specialty is dedicated to delivering high-quality and high-volume specialized surgical care to patients with endocrine disease. The indications for operations on endocrine glands have expanded as diagnostic and localization modalities have improved and as intraoperative and postoperative care have become more refined. The spectrum of intraoperative tools Presented at the 24th Annual Meeting of the American Association of Endocrine Surgeons, San Diego, California, May 1l-14, 2003. Reprint requests: Brian D. Saunders, MD, 2920H Taubman Center, 1500 E. Medical Center Dr, Ann Arbor, MI 48109-0331. Ó 2003, Mosby, Inc. All rights reserved. 0039-6060/2003/$30.00 + 0 doi:10.1016/S0039-6060(03)00420-3
924 SURGERY
in the armamentarium of the endocrine surgeon has also grown to include minimally invasive procedures, intraoperative hormone monitoring, intraoperative ultrasound scanning, laparoscopy, and video-assisted surgery. Despite the growing complexity of the field, there remains no formal certification to signify additional expertise in endocrine surgery. The case number and type of endocrine operations that are performed by graduating surgical chief residents varies widely across the country. This fact is known from data that are tracked by the Residency Review Committee.1,2 Currently, however, a thorough understanding of the national patterns of endocrine surgery (as defined by clinicians in practice) is lacking. Knowledge of these clinical practice patterns would allow for a more accurate study of surgical outcomes in the context of surgeon volume and would facilitate the planning of future needs for endocrine surgeon specialists. The current study uses a database that is representative of surgical practice in the United
Surgery Volume 134, Number 6
States to characterize the clinicians who are performing endocrine procedures and to determine ‘‘who’’ is performing endocrine operations. METHODS The National Inpatient Sample (NIS) is a stratified, cross-sectional database that includes approximately 20% of all non-Veterans Affairs hospital discharges in the United States. The NIS was borne out of federal, state, and industry partnership and is managed under the Health Care Cost and Utilization Project of the Agency for Healthcare Research and Quality.3 This study’s data were derived from the version of the NIS database that encompasses calendar years 1988 through 2000. The number of states whose hospital discharges were represented varied over the study period. There were 759 hospitals in 8 states in 1988 and 994 hospitals in 28 states in 2000. An average of 6.5 3 106 hospital discharges per year was included in the NIS database (Table I). The database was searched for the individual years 1988 through 2000, selecting out hospital discharges for which the primary International Classification of Diseases, 9th revision, clinical modification (ICD-9-CM) procedure code was for parathyroidectomy, hemior total thyroidectomy, and adrenalectomy.4 The specific ICD-9-CM procedure codes for parathyroidectomy, thyroidectomy, and adrenalectomy (hereafter referred to collectively as endocrine codes) that were used are indicated in Table II. Endocrine pancreatic procedures (eg, enucleation of islet cell tumor) were excluded because of predictably small numbers and variations in the coding of these pancreatic procedures. The hospital discharges were excluded if there was incomplete or absent procedural coding data for the surgeon (range, 53.5%-66.2%). Because individual surgeons cannot be identified by name or professional affiliation in the NIS, a method of practice profiling was required. We identified 2 separate models that were intended to group surgeons into quartiles. The first (method A) expressed concentration in endocrine surgery as a percentage of any individual surgeon’s caseload during a defined time period (1 year). The second model (method B) grouped surgeons into quartiles on the basis of the absolute number of index endocrine cases that were performed as compared with other surgeons within the same defined time period. Once a hospital discharge was selected, method A required that all of the primary ICD-9-CM procedure codes for the surgeon that were linked
Saunders et al 925
to that discharge were screened to determine the proportion of endocrine procedures that were performed over the study period. The specific algorithm that was used was to divide the total number of primary endocrine procedure codes by the total number of primary procedures codes. The surgeons were then grouped into quartiles (group A, 0%-25%; group B, 26%-50%; group C, 51%-75%; and group D, 76%-100%). Method B used the same dataset of linked surgeons and patients after exclusion for incomplete or absent coding data. The absolute number of index endocrine surgery cases (as defined by the codes given earlier) were determined per surgeon per year. The statistical breakpoints into quartiles were then determined. All descriptive statistical analyses were performed with STATA 6.0 (Stata Corp, College Station, Texas). RESULTS Parathyroid. There were 30,606 patients who were identified whose hospital discharge had a primary procedural ICD-9-CM code for parathyroidectomy. Of these, there was complete procedural coding information for the surgeons for 14,232 patients (46.5%). Associated with these 14,232 patients were 6100 unique surgeons (Table III). Stratification of the surgeons into quartiles by the percent of endocrine procedures (method A) in their practice yielded 5587 surgeons in group A, 343 surgeons in group B, 101 surgeons in group C, and 69 surgeons in group D (Table IV). Group A contained 11,071 patients (or 78%); group B contained 1833 patients (13%); group C contained 559 patients (4%); and, group D contained 769 patients (5%; Fig 1, A). This pattern was validated generally by the use of method B for the same dataset. In specific, the quartile break points indicated that the 2995 surgeons in group A (lowest absolute volume and 49% of all surgeons) performed 1 to 3 parathyroid operations per year. One thousand eight hundred fifty-eight group B surgeons (30% ) performed an average of 4 to 8 operations per year; 1005 group C surgeons (17% ) performed from 9 to 26 operations per year. The quartile with the highest absolute volume was group D, in which 242 surgeons (4%) performed 27 to 182 parathyroid operations per year (Fig 2, A). Thyroid. There were a total of 123,716 patients who were identified whose hospital discharge had a primary procedural ICD-9-CM code for thyroidectomy. Of these, 56,286 patients (45.5%) had complete procedural coding data for the associated
926 Saunders et al
Surgery December 2003
Table I. NIS data sources from 1988 through 2000 Calendar year
Data source
Hospitals (n) Inpatient stays (n)
1988 1989 1993 1995 1997 1999 2000
CA, CO, FL, IL, IA, MA, NJ, WA Added AZ, PA, WI Added CT, KS, MD, NY, OR, SC Added MO, TN Added GA, HI, UT Added ME, VA AZ, CA, CO, CT, FL, GA, HI, IL, IA, KS, KY, MA, MD, ME, MO, NC, NJ, NY, OR PA, SC, TN, TX, UT, VA, WA, WI, WV (added KY, NC, TX, WV)
759 882 913 938 1012 984 994
5,265,756 6,110,064 6,538,976 6,714,935 7,148,420 7,198,929 7,450,992
Table II. Descriptive lexicon that is associated with the ICD-9-CM procedure codes that were used to query the NIS database Parathyroid Thyroid
Adrenal
Procedure code
Lexicon
0681 0689 062 0639 064 0650 0651 0652 072 0721 0722 0729 073
Complete parathyroidectomy Other parathyroidectomy: parathyroidectomy NOS, partial parathyroidectomy Unilateral thyroid lobectomy Other partial thyroidectomy: isthmusectomy, partial thyroidectomy NOS Complete thyroidectomy Substernal thyroidectomy NOS Partial substernal thyroidectomy Complete substernal thyroidectomy Partial adrenalectomy Excision of lesion of adrenal gland Unilateral adrenalectomy Other partial adrenalectomy: partial adrenalectomy NOS Bilateral adrenalectomy: excision of remaining adrenal gland
NOS, Not otherwise specified.
surgeon and were thus included in our study. There were 17,475 unique surgeons who were associated with these 56,286 patients (Table III). The use of method A to stratify these surgeons into quartiles yielded 16,257 surgeons in group A, 748 surgeons in group B, 260 surgeons in group C, and 210 surgeons in group D (Table IV). The number of patients per quartile was 46,210 in group A (82%); 6425 in group B (11%), 2091 in group C (4%), and 1560 in group D (3%; Fig 1, B). As with the parathyroid data, this pattern is validated with the use of method B for the same dataset. In specific, the quartile break points indicated that the 10,620 surgeons in group A (lowest absolute volume; 61% of all surgeons) performed 1 to 2 thyroid operations per year. Four thousand one hundred six surgeons in group B (23%) performed an average of 3 to 5 operations per year; 2187 group C surgeons (13%) performed from 6 to 15 operations per year. The quartile with the highest absolute volume was group D, in which 562 surgeons (3%) performed 16 to 182 thyroid operations per year (Fig 2, B).
Adrenal. There were a total of 13,229 patients who were identified whose hospital discharge had a primary procedural ICD-9-CM code for adrenalectomy. Of these, 4468 patients had complete procedural coding data for their surgeons to allow inclusion. This represents 33.8% of the total number of adrenalectomies that were identified in the database. There were 3050 unique surgeons who were identified with these 4468 patients (Table III). Stratification of these surgeons into quartiles with method A yielded 2946 surgeons in group A, 58 surgeons in group B, 19 surgeons in group C, and 27 surgeons in group D (Table IV). The number of patients per quartile was 4209 in group A (94%), 103 in group B (2%), 28 in group C (less than 1%), and 128 in group D (3%; Fig 1, C ). Again, this distribution of case volume is essentially validated by using Method B for the same dataset. The quartile break points indicated that the 1,375 surgeons in group A (lowest absolute volume and 45% of all surgeons) performed 1 adrenal operation per year. Four hundred seventyone group B surgeons (15%) performed an average
Saunders et al 927
Surgery Volume 134, Number 6
Table III. Patients (total and ultimately included) and unique surgeons who were identified in the NIS database (1988-2000) with a primary endocrine procedure code Gland
Total patients (n)
Patients with surgeon identifier (n)
Patients included (%)
Unique surgeons (n)
Parathyroid Thyroid Adrenal
30,606 123,716 13,229
14,232 56,286 4,468
46.5 45.5 33.8
6,100 17,475 3,050
Table IV. Distribution of surgeons (accrued over the entire study period) to quartiles an the basis of surgeon percentage of endocrine procedures (method A) Gland
Total surgeons profiled (n)
Quartile A: 0-25% (n)
Quartile B: 26%-50% (n)
Quartile C: 51%-75% (n)
Quartile D: 76%-100% (n)
Parathyroid Thyroid Adrenal
6,100 17,475 3,050
5,587 (91.6%) 16,257 (93.0%) 2,946 (96.6%)
343 (5.6%) 748 (4.3%) 58 (1.9%)
101 (1.7%) 260 (1.5%) 19 (0.6%)
69 (1.1%) 210 (1.2%) 27 (0.9%)
of 2 operations per year; 679 group C surgeons (23%) performed from 3 to 6 operations per year. The quartile with the highest absolute volume was group D, in which 525 surgeons (17%) performed 7 to 167 adrenal operations per year (Fig 2, C ). DISCUSSION This study attempted to characterize the practice patterns of endocrine surgery within the context of the overall practice of surgery in the United States. The great majority of operations for endocrine disease are performed by surgeons whose clinical practice is not focused heavily in endocrine surgery. Additionally, the number of surgeons whose practice is focused on endocrine surgery comprises a very small percentage of the total number of surgeons who are performing common operations for endocrine disease. There have been limited efforts to determine the volume of operations for endocrine disease that are performed throughout the United States. Many published reports in endocrine surgery come from single centers. There is, however, a growing medical and lay interest in the relationship between surgeon volume and patient outcome. To date, much of this work has focused on mortality rates and the inhospital length of stay for selected complex general surgery and vascular surgery operations.5,6 Since the foundation of both national and international endocrine surgical associations, there has been an emphasis on the improvement of the care that is given to patients with endocrine disease, the dissemination of new diagnostic and therapeutic advancements, and the teaching of future practitioners of endocrine surgery. Endocrine surgery as a separate field remains self-identified and
without formal certification. There are those who argue against specific fellowship training of general surgery graduates, claiming that procedures such as parathyroidectomy, thyroidectomy, and adrenalectomy should be in the armamentarium of most well-trained general surgeons. Endocrine surgical procedures continue to be performed by endocrine surgeons, general surgeons, transplantation surgeons, urologists, and otorhinolaryngologists. Previously, there was no large study that sought to identify and to quantify the profile of the clinicians who actually were performing common operations for endocrine disease. The NIS was designed to provide a research database for conducting national and regional studies of inpatient care throughout the United States. It has a large size, reflects all payers, and is longitudinal in nature. These qualities allow for research on health care delivery, cost, use, access, and quality at the patient, physician, market, and state levels. We were able to take advantage of the design of the NIS to characterize clinicians who perform operations for endocrine disease. Importantly, the database contains an accurate crosssection of the hospital discharges in the United States and is not limited to academic health centers, as is the case for many other series that have reported endocrine surgery. Another desirable feature of the NIS is the inability for individual surgeons to be identified by name or by declared surgical specialty. The validity of the database continues to be documented through many other studies that have made use of it.7-12 There are some limitations to the NIS database. Of the patients who were identified in this study as having undergone parathyroid, thyroid, or adrenal
928 Saunders et al
Fig 1. Contribution of individual surgeon quartiles to the total number of patients who underwent operation, as determined by the percentage of practice calculation and accrued over the entire study period (method A). A, Parathyroid; B, thyroid; C, adrenal.
surgery, 53.5% to 66.2% were necessarily excluded because of insufficient coding data for the surgeons who were associated with the hospital discharge abstract. (A hospital discharge abstract is the electronic database record of a patient’s complete hospital stay.) Although this will affect the total
Surgery December 2003
numbers, this would not affect the conclusion of this analysis because of the large sample size that is afforded by the NIS database and because of the random nature of the exclusions. In effect, these exclusion criteria limit our sample to approximately 40% of patients in this 20% stratified, crosssectional sample of non-Veterans Affairs hospitals across the United States. This is essentially a ‘‘biopsy’’ of the dataset, and therefore, it can serve only to indicate patterns and trends that are not absolute or universal. There is some danger in the extrapolation of the numbers in this study to the experience of an individual surgeon because of this factor. There are a greater number of ICD-9-CM codes for thyroid surgery than for parathyroid or adrenal surgery (6 vs 5 and 2, respectively). This could allow more specific coding on discharge abstracts and thus a potentially higher capture rate for thyroid surgery in our study. There was no separate ICD-9CM procedure code for laparoscopic adrenalectomy during the study period. Therefore, some of these procedures may have been missed in our search algorithm because of coding variations. The NIS contains hospital discharge abstract data from inpatients only. Therefore, operations for endocrine disease, particularly thyroid and parathyroid operations, that were performed on an outpatient basis will not be included in this study. The NIS includes only non-federal hospitals. This necessarily excludes procedures that are performed at Veterans Affairs hospitals. Most Veterans Affairs hospitals are staffed by surgeons who are affiliated with academic health centers. However, it is not common for a surgeon to practice only endocrine surgery at a Veterans Affairs hospital. Because it is a reasonable assumption that many of the surgeons whose practices focus on endocrine surgery are located at an academic health center, this study may fail to capture some of the cases that have been performed by these surgeons at a Veterans Affairs hospital. Predictably, our data demonstrate that those surgeons whose practice is comprised of more than 76% endocrine surgical procedures do in fact perform more parathyroidectomies, thyroidectomies, and adrenalectomies per surgeon than those surgeons whose practice is less than 25% endocrine surgical procedures in terms of individual volume. However, despite this, most patients who undergo operations for endocrine disease in the United States have the procedure performed by surgeons whose practice is comprised of less than 25% endocrine surgery. This is a result of the fact that surgeons who perform endocrine surgery but do
Surgery Volume 134, Number 6
not specialize in this field far outnumber those surgeons whose practice is limited essentially to endocrine surgery. The surgeons in this study whose practice is comprised of greater than 76% endocrine surgical procedures make up only 1.1% of the surgeons who are performing parathyroidectomies, 1.2% of the surgeons who are performing thyroidectomies, and 0.9% of the surgeons who are performing adrenalectomies. When profiled by absolute number of operations, the surgeons in the highest quartile are only 4% of the surgeons who are performing parathyroidectomies, 3% of the surgeons who are performing thyroidectomies, and 17% of the surgeons who are performing adrenalectomies. Put another way, 50% of patients who undergo parathyroid operations have the procedure performed by a surgeon that does 8 or fewer parathyroid operations per year, and 50% of patients who undergo thyroid operations have them performed by surgeons who do an average of 5 or fewer of these operations per year. For adrenalectomy, 50% of patients have the procedures performed by a surgeon who averages 2 or fewer procedures per year. There are a number of reasons that can be postulated to explain the fact that most patients who undergo common endocrine surgical procedures in the United States are being treated by surgeons whose practice is comprised of less than 25% endocrine surgery. As noted earlier, there are simply many fewer surgeons who specialize in endocrine surgery. There are approximately 300 members who are currently in the American Association of Endocrine Surgeons. This number stands in contrast to the more than 50,000 current members in the American College of Surgeons. This may, in part, be due to limited exposure to endocrine surgery during formal residency training. Endocrine disease is relatively uncommon, and most graduating chief residents do not have adequate exposure to common operations for endocrine disease during their residency.1,2 Further, most graduating chief residents do not have direct training under a surgeon who specializes in endocrine surgery.13 This study does not examine the outcomes that were associated the patients who were identified. Much of the published literature on outcomes and complication rates of endocrine surgeries has emanated from the highest volume endocrine surgeons at academic health centers that correlate to quartile D in the method used in this study. The present study shows that this group of surgeons is treating a minority of the patients who undergo operations for endocrine disease. We currently do not know the actual outcomes of operations for
Saunders et al 929
Fig 2. Distribution of the surgeons to quartiles, as determined by the absolute number of index endocrine ICD-9-CM procedure codes per year. A, Parathyroid; B, thyroid; C, adrenal.
endocrine disease that were performed by surgeons in the other 3 quartiles. This information is especially important because reoperative endocrine surgery is associated with increased morbidity and cost.14 This trend of improved outcomes that relate to increased volume dates back to the founding of surgery on endocrine glands. Kocher was the first high-volume thyroid surgeon. As he accrued more patients throughout his career, there was a dramatic decrease in mortality rate from thyroidectomy.15 Two recent studies have specifi-
930 Saunders et al
cally looked at high-volume parathyroid and thyroid surgeons. Chen et al16 and Sosa et al17 have demonstrated improved cure rates and decreased complication rates and length of stay for a single, high-volume endocrine surgeon when compared with larger numbers of general surgeons who intermittently perform operations for endocrine disease. There are few ICD-9-CM diagnostic codes for outcomes of endocrine surgeries. The NIS database only looks at hospital discharge data. Many of the complications from operations for endocrine disease (ie, recurrent laryngeal nerve dysfunction, hypoparathyroidism, and adrenal hormone insufficiencies) may not be evident or documented fully until follow-up after hospital discharge and would not be captured by analysis of this database. The NIS therefore may not be suitable for outcome-based analysis of most endocrine surgical procedures. If an association of surgical volume to outcome exists, this would then support a referral pattern of patients who require endocrine surgery to highvolume surgeons. Not only would this allow patients to be seen and operated on by endocrine surgeons, but also it would allow for them to be evaluated by and receive therapy from other specialists who are necessary for their care. The development of interdisciplinary teams that include endocrine surgeons, medical endocrinologists, nuclear medicine physicians, radiologists, and pathologists has been a goal of many practitioners in the field of endocrine surgery.18 However, one cannot use this study to determine the relationship between volume and patient outcome. Although it is likely that a volume outcome effect exists in endocrine surgery, it has not been documented rigorously since the reports of Sosa et al17 and Chen et al16 and will require further careful study. Additionally, there is currently a paucity of published, benchmarking outcome data that are based on series from lower volume surgeons.19 It remains complete conjecture as to the threshold whereby a low-volume endocrine surgeon becomes a high-volume endocrine surgeon. If a true volume/outcome effect is proved eventually for endocrine surgical procedures, there may then be an increased need for specialisttrained endocrine surgeons to provide the care that is required by these patients. Currently, there is no separate board certification process for endocrine surgery. Many general surgeons who perform operations for endocrine disease get no more training than that provided during residency. Harness et al1,2 published results of a Residency
Surgery December 2003
Review Committee database survey several years ago that showed the inadequate level of exposure of graduating general surgery chief residents to common and uncommon endocrine surgical procedures. The latest follow-up to this data was again discussed by Harness18 in his 2000 American Association of Endocrine Surgeons Presidential lecture. Formal fellowship training in endocrine surgery is not easily accessible and may need to become more widespread if endocrine surgical procedures become limited to those surgeons who specialize in endocrine surgery. Most operations for endocrine disease in the United States are performed by surgeons whose practice is less than 25% endocrine. This is reflective of the fact that the number of surgeons who specialize in endocrine surgery is significantly smaller than the number of surgeons whose practice does not focus on endocrine surgery. This pattern will play a significant role in the analyses of surgical outcome and the planning of future care models for patients with endocrine disease. REFERENCES 1. Harness JK, Organ CH Jr, Thompson NW. Operative experience of US general surgery residents in thyroid and parathyroid disease. Surgery 1995;118:1063-70. 2. Harness JK, Organ CH Jr, Thompson NW. Operative experience of US general surgery residents with diseases of the adrenal gland, endocrine pancreas, and other less common endocrine organs. World J Surg 1996;20:885-91. 3. Steiner C, Elixhauser A, Schnaier J. The healthcare cost and utilization project: an overview. Effective Clin Pract 2002;5:143-51. 4. Public Health Service, Health Care Financing Administration. International Classification of Diseases, 9th revision, Clinical Modification. Washington (DC): US Department of Health and Human Resources;1991. 5. Gordon TA, Burleyson GP, Tielsch JM, Cameron JL. The effects of regionalization on cost and outcome for one general high risk surgical procedure. Ann Surg 1995;221: 43-9. 6. Dimick JB, Stanley JC, Axelrod DA, Kazmers A, Henke PK, Jacobs LA, et al. Variation in death rate after abdominal aortic aneurysmectomy in the United States: impact of hospital volume, gender, and age. Ann Surg 2002;235: 579-85. 7. Shen JJ. Severity of illness, treatment environments, and outcomes of treating acute myocardial infarction for Hispanic Americans. Ethn Dis 2002;12:488-98. 8. Cowan JA, Dimick JB, Thompson BG, Stanley JC, Upchurch GR Jr. Surgeon volume as an indicator of outcomes after carotid endarterectomy: an effect independent of specialty practice and hospital volume. J Am Coll Surg 2002;195: 814-21. 9. Chang RK, Chen AY, Klitzner TS. Clinical management of infants with hypoplastic left heart syndrome in the United States, 1988-1997. Pediatrics 2002;110:292-8. 10. Xiao H, Campbell ES, Song KS. A trend analysis of organ transplantation among ethnic groups. J Natl Med Assoc 2002;94:15-20.
Surgery Volume 134, Number 6
11. Nietart PJ, Silverstein MD, Silver RM. Hospital admissions, length of stay, charges, and in-hospital death among patients with systemic sclerosis. J Rheumatol 2001;28:2031-7. 12. Howard TS, Hoffman LH, Stang PE, Simeos EA. Respiratory syncytial virus pneumonia in the hospital setting: length of stay, charges, and mortality. J Pediatr 2000;137:227-32. 13. Prinz RA. Presidential address: endocrine surgical training: some ABC measures. Surgery 1996;120:905-12. 14. Doherty GM, Weber B, Norton JA. Cost of unsuccessful surgery for primary hyperparathyroidism. Surgery 1994; 116:954-8. 15. Becker WF. Pioneers in thyroid surgery. Ann Surg 1977; 185:493-504. 16. Chen H, Zeiger MA, Gordon TA, Udelsman R. Parathyroidectomy in Maryland: effects of an endocrine center. Surgery 1996;120:948-52. 17. Sosa JA, Bowman HM, Tielsch JM, Powe NR, Gordon TA, Udelsman R. The importance of surgeons experience for clinical and economic outcomes from thyroidectomy. Ann Surg 1998;228:320-30. 18. Harness JK. Presidential address: interdisciplinary care: the future of endocrine surgery. Surgery 2000;128:873-80. 19. Reeve TS, Curtin A, Fingleton L, Kennedy P, Machie W, Porter T, et al. Can total thyroidectomy be performed as safely by general surgeons in provincial centers as by surgeons in specialty endocrine surgical units? Arch Surg 1994;129:834-6.
DISCUSSION Dr Herbert Chen (Madison, Wis). If you were a surgeon who treated 500 cases a year, for instance, and one half of them were endocrine cases (you would have 250 endocrine), then you would be put in the B category; whereas if you treated only 100 cases and 80 of them were endocrine, you would be in the high group but would have performed only one third of the number of endocrine cases as the first example. Therefore, have you looked at your data with regard to absolute number of endocrine cases and seen that the data might shake out a little differently than you have presented here? Dr Saunders. I agree that a potential limitation of this study is the specific method that was used to define the quartiles. A limitation of these data is that there are very few surgeons in our highest quartile. We have not yet performed that analysis, but I think that recategorizing the surgeons according to the absolute number of the number of operations performed probably will support the results of the model as presented. Dr Ashok R. Shaha (New York, NY). Not to open Pandora’s box, but I am sure that, when you started the study and started analyzing the data, you and your coauthors must have gone through a lot of soul-searching about who is an endocrine surgeon, and then you built up these quartiles about the practice. What was your discussion? How did you define an endocrine surgeon? Ultimately you came up with a high-volume/low-volume endocrine surgeon. What should be the ideal number? Dr Saunders. At this point in our analysis, we do not have a discreet number that would define a threshold
Saunders et al 931
between a low-volume and a high-volume endocrine surgeon. What this study looked at was the percentage of the practice of the surgeons who were identified that was comprised of endocrine surgery. The creation of quartiles was fairly simplistic, but we did not mean to imply where the high-volume and low-volume cut-offs would be. I think that there is a lot more study that would need to be done to determine where the true threshold lies. Dr Richard A. Hodin (Boston, Mass). Did you have a chance to look at different regions throughout the United States. I am aware of the data in our area in Massachusetts, and I can tell you it would not look anything like the data you presented. Boston may be different from a lot of other areas, but for sure, the vast majority of patients who have endocrine surgery have it performed by what would be the higher quartile of surgeons. Dr Saunders. The NIS database can be stratified according to region of the country. We did not do that at this point, but the data do allow that to be done. I agree with you. I have a feeling that, depending on city or area of the country, the numbers will look a little bit different. Dr Karen R. Borman ( Jackson, Miss). Did you, by any chance, look for a self-designated specialty designator for these physicians? For example, in some of the other Department of Human Health and Services databases, they will be able to tell you whether these were otolaryngologists or general surgeons who performed these operations. Dr Saunders. One of the characteristics of the NIS is that one cannot identify specific surgeons. Therefore, unfortunately, there was no way of identifying which surgical subspecialty each of these surgeons belonged to. Dr Clive S. Grant (Rochester, Minn). Could I just clarify my understanding of the data. The study period was approximately 12 or 13 years, and the highest quartiles I saw on that graph showed that, depending on the type of operation, the surgeons might have performed 6 to 14 total cases. This was during the entire study period? If this is true, is the data valid? It would imply that the busiest surgeons only averaged about one case per year. Dr Saunders. I agree with you, that is one of the limitations of this data and this database. The database is somewhat incomplete. However, I think it is made up for by its large numbers. The take-home message of this is more the trends and the practice patterns rather than the specific numbers. In fact, probably it is the numbers in quartile A that are the most accurate because it is the largest number of surgeons. It is a limitation of this database that the number of surgeons in the highest quartile is the lowest. Dr Lawrence A. Danto (Truckee, Calif ). Does the database allow you to look at and did you look at the payment patterns or the payment mechanisms for patients as they affect referral patterns? Dr Saunders. This database reflects all payers, other than those patients at a federal hospital, but it does not allow you to look at reimbursement or payment. Therefore, we did not analyze this factor as it may relate to referral patterns.