The techniques of and indications for acromioplasty have evolved considerably since Neer's description of acromioplasty in 19721-7. Acromioplasty is a commonly performed outpatient orthopaedic procedure, and excellent results have been reported in association with both open and arthroscopic approaches in appropriately selected patients8-11. While numerous clinical series have evaluated the outcomes of acromioplasty, to our knowledge there have been no efforts to analyze the frequency of this procedure over time. In the background of widely documented variability in the indications for other commonly performed shoulder procedures12,13, some have suggested that the indications for acromioplasty have become too broad and that this procedure is performed too frequently today. While Neer suggested that extrinsic impingement of the acromion was the most common cause of chronic rotator cuff derangement, advances in magnetic resonance imaging and arthroscopic visualization of pathoanatomy have provided evidence for intrinsic tendon disease14-19, implying that extrinsic mechanical impingement is not the only etiologic factor in rotator cuff tendinitis and that perhaps acromioplasty has a more limited role in the treatment of some impingement lesions. Furthermore, there is mounting evidence that acromioplasty may not offer significant benefit in the treatment of impingement syndrome, when compared with nonoperative treatment, in terms of functional outcomes or pain in either the short term or the long term20-24.
In light of this controversy, we sought to determine the frequency of acromioplasty procedures and how this frequency has changed over time by means of a two-part investigation. In Part A, a state-wide ambulatory surgery database was used to compare the volume and incidence of this procedure with those of other ambulatory orthopaedic procedures in the state of New York over an eleven-year span. In Part B, a national database managed by the American Board of Orthopaedic Surgery (ABOS) was used to compare the volume of arthroscopic acromioplasties with that of other orthopaedic procedures reported by candidates taking Part II of the board certification examination over a ten-year span. The purpose of the present study was to determine how the frequency of acromioplasty procedures has changed on a statewide and national level relative to other orthopaedic procedures over a recent span of time.
The New York Statewide Planning and Research Cooperative System (SPARCS) database was used for Part A of the study. The New York State Department of Health has collected ambulatory surgery data from New York State hospital-based and freestanding licensed ambulatory surgery facilities since 1986 as mandated by the New York State Public Health Law that created SPARCS in 1979. The initial law focused on the collection of inpatient discharges from all state hospitals, and, in 1985, the State Hospital Review and Planning Council adopted additional regulations regarding the reporting of ambulatory surgery data to the New York State Department of Health. The SPARCS ambulatory database contains records for each ambulatory discharge from nonfederal, state-licensed hospitals in New York, submitted according to Universal Data Set specifications by trained medical records personnel in each hospital and verified for accuracy by the Department of Health. The records include information relating to the patient's disposition, age, sex, race, payer, primary and secondary diagnoses, and primary and secondary procedures. SPARCS data have been used for numerous studies investigating procedure volume, incidence rates, epidemiologic trends, and surgical outcomes for various surgical subspecialties, including various orthopaedic procedures such as rotator cuff repair, total shoulder arthroplasty, the treatment of hip fractures, and spine arthrodesis for scoliosis25-31.
The inclusion criteria for Part A included (1) all acromioplasty procedures (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9] code 81.83), including those performed in the context of rotator cuff tear as well as those performed in patients without rotator cuff tears and (2) all other orthopaedic surgical procedures (ICD-9 codes 76.01 through 84.99) that were reported in each year of the SPARCS ambulatory database from 1996 through 2006 (the most recent year for which SPARCS data were available at the time of the present study). Demographic data were collected for each acromioplasty procedure, including the age of the patient on the day of the procedure, sex, payer, ICD-9 primary diagnosis codes (with evaluation of the primary but not secondary diagnosis code associated with any acromioplasty procedure) and ICD-9 procedure codes (with evaluation of all procedure codes represented through the first four possible positions in the primary and secondary procedure code fields), location of the surgery site (hospital-based or freestanding licensed ambulatory surgery center), and procedure year. Only primary ICD-9 diagnosis codes were used in the analysis as they represent the principal orthopaedic diagnosis associated with the procedures performed, in contrast with secondary medical diagnosis codes, which may be unrelated to the procedures performed (e.g., a secondary diagnosis of hypertension). Each case could have only one primary diagnosis code, followed by several secondary diagnosis codes. There was no exclusion of cases on the basis of principal diagnoses, including rotator cuff tears requiring repair. Race and ethnicity data were not available until 2002 and were removed from the analysis.
For Part B, the ABOS database was used. This database comprises cases reported by candidates taking Part II of their orthopaedic surgery board certification examination. Board certification in orthopaedic surgery consists of passing a written examination (Part I) and a practice-based oral examination (Part II). To qualify for Part II, candidates must pass Part I, followed by a period of practice of twenty-two months. Candidates must report all surgical procedures performed at all hospitals and ambulatory surgery centers during a defined six-month period. Since 1999, the ABOS has maintained records of these cases on a secure online database that includes procedure dates, primary and secondary diagnosis codes (ICD-9), and procedure codes (Current Procedural Terminology [CPT] 2009 Professional Edition), among other variables. Unlike the SPARCS database, which uses ICD-9 procedure codes, the ABOS database uses CPT codes, allowing for the specific identification of all arthroscopic acromioplasties. The ABOS database has been previously used to investigate practice trends in orthopaedic surgery, including the treatment of distal radial fractures and intertrochanteric hip fractures32,33.
The inclusion criteria for Part B included (1) any arthroscopic acromioplasty procedures (CPT code 29826), including those performed in the context of a rotator cuff tear as well as those performed in patients without a rotator cuff tear, and (2) any other orthopaedic surgical procedures that were reported in each year of the ABOS ambulatory database from 1999 through 2008 (the most recent year for which ABOS data were available at the time of the present study). In addition, the ABOS database records procedure volume per individual surgeon; therefore, the volume of cases was reported as the mean number of procedures per candidate. These cases were separated into two categories—those performed by all candidates and those reported by candidates performing at least one arthroscopic acromioplasty. The latter subgroup was created to better identify trends among surgeons who perform this procedure and to exclude surgeons who have never performed an arthroscopic acromioplasty. Although this second subgroup analysis may have increased the apparent number of acromioplasties per number of orthopaedic surgeons taking Part II of the board examination, it may have been a more accurate representation of the average number of acromioplasties performed by surgeons who actually perform this procedure as part of their case mix. This subgroup analysis was not performed in Part A as the SPARCS database did not include information regarding individual surgeon volume, making it impossible to analyze procedures per surgeon. In order to maintain homogeneity between cases compared over time, the analysis was limited to cases reported only by candidates taking Part II of the board certification examination, and the exclusion criteria included cases reported by surgeons taking (1) recertification, (2) sports medicine subspecialty, or (3) maintenance of certification examinations as these data among subspecialty groups were not available for the first several years of the database. As was the case for Part A, in Part B there was no exclusion of cases on the basis of principal diagnoses, including rotator cuff tears requiring repair.
This research protocol was reviewed by the institutional review board of Columbia-Presbyterian Medical Center and was judged to be exempt from requiring consent. For Part A, the SPARCS database is de-identified with regard to patient identity. For Part B, candidates applying for Part II of the examination are informed that the data they report may be used for research, and they grant permission to use case data on signing their application.
Statistical Methods
For both Parts A and B, descriptive statistics were reported on demographic variables, procedure volumes, and incidence rates. For Part A, a multivariate logistic regression model was constructed with procedure type as the response variable, controlling for age, sex, payer, facility type, and year of procedure. With the exception of age, which was a continuous variable, all tested effects were categorical. Variables were not entered into the model in a stepwise fashion, and instead these variables were chosen a priori, derived from the clinical experience that these factors were expected to have a clinical bearing on the likelihood of surgery. As the ABOS database did not contain variables recorded in the SPARCS database that would allow for the creation of a detailed multivariate logistic regression model, for Part B, a simple univariate logistic regression model was created with procedure type as the response variable as a function of year of procedure. Ninety-five percent confidence intervals were reported for all odds ratios in the logistic regression analysis. The level of significance for all tests was p < 0.05.
Source of Funding
There were no external sources of funding for the present study.
Descriptive Statistics for Part A
A table in the Appendix summarizes cases recorded in the New York State SPARCS ambulatory surgery database from 1996 through 2006. Of the 1.92 million orthopaedic surgery ambulatory procedures, 138,585 were acromioplasties. In 1996, there were 5571 acromioplasties among 129,973 orthopaedic surgery ambulatory procedures, representing 4.29% of all orthopaedic surgery ambulatory procedures and a population incidence of 30.0 per 100,000 (see Appendix), based on annual New York State census data34. In 2006, there were 19,743 acromioplasties among 231,760 orthopaedic surgery ambulatory procedures, representing 8.5% of all orthopaedic surgery ambulatory procedures and a population incidence of 101.9 per 100,000. Over eleven years, the volume of acromioplasties increased by 254.4% (Fig. 1), whereas the volume of all orthopaedic surgery ambulatory procedures increased by only 78.3% over this entire span. This increase was similar in hospital-based centers, which had a 262.9% increase, and freestanding ambulatory surgery centers, which had a 242.5% increase (see Appendix). In addition, the percentage of acromioplasties as a proportion of all orthopaedic surgery ambulatory procedures nearly doubled, and the population incidence more than tripled.
A table in the Appendix presents the most common primary ICD-9 diagnoses associated with acromioplasty procedures in the SPARCS database. The most common primary diagnosis codes associated with acromioplasties in New York State during this time were impingement syndrome (42.8% of all acromioplasties), followed by sprains and strains of the rotator cuff (13.6% of all acromioplasties), bursitis of the shoulder (13.3% of all acromioplasties), complete rupture of the rotator cuff (5.9% of all acromioplasties), and superior glenoid labrum lesions (3.6% of all acromioplasties). Over time, there were shifts in the relative percentage of each of these common primary diagnoses. In 1996, impingement syndrome was the primary diagnosis for 57.0% of all acromioplasties, whereas in 2006, it was the primary diagnosis for 35.7% of all acromioplasties. In 1996, sprains and strains of the rotator cuff were the primary diagnosis for 16.4% of all acromioplasties, whereas in 2006, these injuries were the primary diagnosis for 10.9% of all acromioplasties. In 1996, bursitis was the primary diagnosis for 7.5% of all acromioplasties, whereas in 2006, this condition was the primary diagnosis for 16.4% of all acromioplasties. In 1996, complete rupture of the rotator cuff was the primary diagnosis for 2.3% of all acromioplasties, whereas in 2006, this injury was the primary diagnosis for 6.8% of all acromioplasties. Last, in 1996 there were no primary diagnoses of superior glenoid labrum lesions, whereas in 2006, these lesions were the primary diagnosis for 7.6% of all acromioplasties.
Multivariate Logistic Regression Analysis for Part A
In the logistic regression model, acromioplasty was the response variable, controlling for age, sex, payer, surgical facility type, and procedure year. Regression model assessment revealed that the acromioplasty logistic had a percent concordance of 66.6 and c value of 0.673, indicating good predictive properties35. There were significant effects of age, sex, payer, surgical facility type, and year on the likelihood of having an acromioplasty as compared with all other ambulatory orthopaedic procedures. For every additional ten years of patient age, a patient was 1.24 times (95% confidence interval, 1.23 to 1.24 times) more likely to have an acromioplasty procedure (p < 0.0001). Males were 1.7 times (95% confidence interval, 1.70 to 1.74 times) more likely to have an acromioplasty procedure than females were (p < 0.0001). Patients receiving Workers’ Compensation were 2.4 times (95% confidence interval, 2.38 to 2.47 times) more likely to have an acromioplasty procedure than were patients with Blue Cross insurance (p < 0.0001). Patients with commercial insurance were 1.05 times (95% confidence interval, 1.03 to 1.07 times) more likely to have an acromioplasty procedure than were patients with Blue Cross insurance (p < 0.0001). Patients with Medicare were 0.65 times (95% confidence interval, 0.63 to 0.66 times) as likely to have an acromioplasty procedure than were those with Blue Cross insurance (p < 0.0001). Patients with Medicaid were 0.42 times (95% confidence interval, 0.40 to 0.45 times) as likely to have an acromioplasty procedure as patients with Blue Cross insurance (p < 0.0001). Hospital-based centers were 2.2 times (95% confidence interval, 2.12 to 2.38 times) more likely than freestanding ambulatory centers to be the location for acromioplasty procedures (p < 0.0001). In 2006 compared with 1996, patients had a 2.4 times (95% confidence interval, 2.30 to 2.43 times) increased likelihood of having an acromioplasty procedure as opposed to all other orthopaedic surgery ambulatory procedures (p < 0.0001).
Descriptive Statistics for Part B
A table in the Appendix summarizes cases that were submitted to the ABOS by candidates taking Part II of the orthopaedic surgery board examination. Of the 866,655 orthopaedic surgery procedures reported, 33,121 were arthroscopic acromioplasties. In 1999, the 648 candidates who took Part II of the orthopaedic surgery board examination reported an average (and standard deviation) of 2.6 ± 5.1 arthroscopic acromioplasties per candidate and 117.3 ± 58.3 orthopaedic surgery procedures per candidate. In 2008, 664 candidates averaged 6.3 ± 11.2 arthroscopic acromioplasties per candidate and 132.6 ± 67.3 orthopaedic surgery procedures per candidate. Over a span of ten years, the mean volume of acromioplasties per candidate increased by 142.3% (Fig. 2) whereas the mean volume of all orthopaedic surgery procedures increased by only 13.0%. Unlike Part A, in which a New York State population incidence for acromioplasty procedures was calculated, this incidence was not calculated for Part B as the ABOS database of cases submitted by board certification candidates represents a small percentage of the total cases performed by all orthopaedic surgeons across the nation.
When cases that had been submitted only by candidates who had reported at least one arthroscopic acromioplasty were analyzed, the results were similar. In 1999, the 302 candidates performing at least one arthroscopic acromioplasty performed a mean of 5.6 ± 6.2 arthroscopic acromioplasties per candidate and 121.6 ± 54.5 total orthopaedic surgery cases per candidate. In 2008, the 342 candidates performing at least one arthroscopic acromioplasty performed a mean of 12.2 ± 13.1 arthroscopic acromioplasties per candidate and a mean of 136.0 ± 61.2 total orthopaedic surgery cases per candidate. Over the span of ten years, the mean volume of acromioplasties per candidate increased by 117.9% (Fig. 2) whereas the mean volume of all orthopaedic surgery procedures increased by only 11.8%.
A table in the Appendix presents the most common primary ICD-9 diagnoses associated with arthroscopic acromioplasty procedures in the ABOS database. The most common primary diagnosis codes associated with arthroscopic acromioplasties during this time were impingement syndrome (31.4% of all acromioplasties), followed by bursitis of the shoulder (21.2%), complete rupture of the rotator cuff (19.8%), sprains and strains of the rotator cuff (10.6%) and other specified disorders of the rotator cuff (4.1%). Unlike the SPARCS database, in which superior glenoid labrum lesions were ranked as the fifth most common primary diagnosis code, in the ABOS database superior glenoid labrum lesions were ranked as the seventh most frequent primary diagnosis over the time period analyzed (representing 2.4% of all acromioplasties). Similar to Part A, there were shifts in the relative percentage of each of these common primary diagnoses over time. In 1999, impingement syndrome was the primary diagnosis for 37.3% of all arthroscopic acromioplasties, whereas in 2008, this disorder was the primary diagnosis for 27.1% of all arthroscopic acromioplasties. In 1999, bursitis was the primary diagnosis for 25.1% of all arthroscopic acromioplasties, whereas in 2008, it was the primary diagnosis for 17.3% of all arthroscopic acromioplasties. In 1999, complete rupture of the rotator cuff was the primary diagnosis for 9.4% of all arthroscopic acromioplasties, whereas in 2008, it was the primary diagnosis for 25.5% of all arthroscopic acromioplasties. In 1999, sprains and strains of the rotator cuff were the primary diagnosis for 11.5% of all arthroscopic acromioplasties, whereas in 2008 these injuries were the primary diagnosis for 10.9% of all arthroscopic acromioplasties. Last, in 1999, other specified disorders of the rotator cuff comprised 4.6% of all arthroscopic acromioplasties, whereas in 2008, such disorders comprised 3.2% of all arthroscopic acromioplasties.
Univariate Logistic Regression Analysis for Part B
While the ABOS database did not contain variables recorded in the SPARCS database that would allow for the creation of a detailed multivariate logistic regression model controlling for age, sex, payer, and surgical facility type, a simple univariate logistic regression model was created with acromioplasty as the response variable as a function of procedure year. Among all candidates reporting cases to the ABOS, candidates were 2.2 times (95% confidence interval, 2.07 to 2.32 times) more likely to have performed an arthroscopic acromioplasty in 2008 as compared with 1999 (p < 0.0001). Among candidates performing at least one acromioplasty, candidates were 2.0 times (95% confidence interval, 1.93 to 2.17 times) more likely to have performed an acromioplasty in 2008 as compared with 1999 (p < 0.001).
Neer remarked that acromioplasty should be reserved for "carefully selected patients with mechanical impingement," and proposed that this procedure should only be performed for patients with reasonable life expectancy and persistent disability despite at least one year of nonoperative treatment1,36. Since Neer's landmark work, acromioplasty has become one of the most frequently performed procedures in orthopaedics, as corroborated by the data in the present report. The present data also document a dramatic increase in the volume and population-based incidence of cases recorded by the New York Department of Health on a state level as well as the mean number of cases per surgeon as reported by candidates taking Part II of their orthopaedic surgery board certification examination on a national level. These increases in acromioplasty were well above those observed in all orthopaedic procedures overall in both databases. In the state of New York in 2006 compared with 1996, patients were 2.4 times more likely to have an acromioplasty as opposed to all other ambulatory orthopaedic procedures, and in the United States in 2008 compared with 1999 candidates reporting cases to the ABOS were 2.2 times more likely to report performing an arthroscopic acromioplasty as opposed to all other orthopaedic procedures, both significant findings.
These findings are concordant with previous data documenting increasing utilization of surgical procedures overall in recent years, both within the specialty of orthopaedic surgery and, specifically, within the subspecialty of shoulder surgery12,37,38. For instance, in the arena of total knee replacement, Katz et al. found that from 1985 to 1990 the number of Medicare-funded total knee replacements increased each year at an annual rate of 18.5%, doubling overall during this short period37. These authors cited a number of reasons for this increase, including an increase in the number of patients with knee arthritis, improved surgical technology and surgeon expertise, better access to orthopaedic care, expanding surgical indications as the technology of total knee arthroplasty became more disseminated into the community, and variations in physician behavior. In the arena of shoulder surgery, Sherman et al., utilizing the SPARCS database, found a dramatic increase in the total number of rotator cuff repairs, from 6656 in 1997 to 10,128 in 200238. In their analysis, the proportion of ambulatory cases increased from 57% to 82% during this time period. The investigation did not compare the volume of rotator cuff repair to that of any other procedure, making it difficult to conclude whether these increases were specific to rotator cuff repair or whether this trend represented a global increase in health care utilization across the board.
There are several possible explanations for the increasing frequency of acromioplasty procedures as documented in the current investigation. One reason may be related to improvements in imaging modalities. Innovations in magnetic resonance imaging technology—including fast spin-echo imaging and fat saturation—and improved resolution of ultrasonography have increased the sensitivity of diagnosing rotator cuff tears39,40. It is possible that greater use of magnetic resonance imaging and ultrasonography has led to an increased rate of diagnosis of rotator cuff tears that are subsequently treated surgically with acromioplasty. Furthermore, magnetic resonance imaging findings such as tendinosis and partial-thickness tears that are detected with greater frequency may be more aggressively treated than in the past. Another possible factor to account for the increasing frequency of acromioplasty may be the increasing dissemination of arthroscopy in the practice of orthopaedic surgery today. Arthroscopy can show the exact location and size of partial and full-thickness tears and labral tears as well as the presence of ligamentous laxity. While diagnostic arthroscopy on its own has limited indications, the increasing availability of arthroscopy in orthopaedic practice provides the opportunity to diagnose and potentially to treat shoulder pathology in one setting.
Another possible reason for the increasing frequency of acromioplasty may be related to the increasing life expectancy of patients. This factor likely accounts for little of the overall increased utilization of this procedure, as in Part A the mean age of patients increased only slightly, for a mean difference of 2.1 years over the eleven years analyzed. Another possible reason may be related to an increasing emphasis on training for shoulder surgery and arthroscopy in academic centers, as previous research has demonstrated the value of training resources for residents in shoulder arthroscopy, including hands-on instructional courses, arthroscopy in cadaver laboratories, and computer simulation models41. Perhaps in recent years young surgeons have been more prepared to perform shoulder surgery, both arthroscopic and open, resulting in an increased performance of acromioplasty. Furthermore, there remains the possibility that patients are more aware of surgical options to treat all causes of shoulder pain and have an increasing preference for surgery over nonoperative management and are now seeking operative treatment of shoulder pain with greater frequency than in previous years. There are data to support that there has been an increase in patient interest specifically in arthroscopic shoulder surgery42. In addition, it may be that we are now seeing the satisfaction of a previously unmet need of patients with sources of shoulder pain that can be treated effectively with acromioplasty, including subacromial impingement, partial and complete rotator cuff tears, and painful bursitis. Last, it is possible that the indications for acromioplasty have become increasingly more liberal, particularly in the context of an increasing number of shoulder-related specialists such as sports medicine specialists, shoulder and elbow specialists, and arthroscopists, as a body of literature supports that increasing specialization in medical and surgical fields has historically led to increasing surgical utilization43-45. There is, in fact, research documenting a high variation among orthopaedic surgeons in terms of the indications for commonly performed shoulder procedures, including, most notably, rotator cuff repair12,13.
An analysis of the diagnoses for which acromioplasties were performed for Parts A and B of the study revealed similar trends on the statewide and national levels. While impingement syndrome represented the most common primary diagnosis overall, there were certain primary diagnoses for acromioplasties that were surprising, namely, superior glenoid labral lesions, which ranked as the fifth most common principal diagnosis in the SPARCS database and the seventh most common principal diagnosis in the ABOS database. While it is likely that there was combined labral pathology and acromial/rotator cuff pathology that required acromioplasty in some patients, one would not expect an isolated labral tear to be the primary diagnosis for which acromioplasty is performed in any patient. The high frequency with which acromioplasties were found to be performed in the setting of labral lesions may be due to coding errors on the part of surgeons or administrators in the collection of data or may be due to poorly indicated acromioplasties in patients found to have isolated labral tears. It is interesting to note the differential relative proportion of complete rotator cuff tears that represented the primary diagnosis for each database. In the SPARCS database, a primary diagnosis of complete rotator cuff tear was present in 5.9% of all acromioplasties (2.3% of all acromioplasties in 1996 and 6.8% of all acromioplasties in 2006). In contrast, in the ABOS database, a primary diagnosis of rotator cuff disease was present in 19.8% of all acromioplasties (9.4% of all acromioplasties reported by ABOS candidates in 1999 and 25.5% of all acromioplasties in 2008). While in both databases the relative proportion of complete rotator cuff tear as a primary diagnosis increased by approximately threefold over the time period studied, there was a much higher relative proportion of rotator cuff tears as a primary diagnosis for cases in the ABOS database. One reason to account for this discrepancy may be that the ABOS database specifically identified arthroscopic acromioplasties, whereas the SPARCS database identified open and arthroscopic procedures; it may be that patients with rotator cuff tears simply comprised a greater proportion of arthroscopic acromioplasties captured in the ABOS database as compared with open or arthroscopic acromioplasties captured in the SPARCS database. Furthermore, as the ABOS database may be presumed to include younger surgeons who may have had more recent training in arthroscopic techniques, it may be that these surgeons attempted more arthroscopic acromioplasties combined with rotator cuff repair in comparison with their counterparts in the New York database. Last, the differences in the relative proportion of rotator cuff tears treated with acromioplasty in each database may be related to differences in the coding practices between the two groups of surgeons, data entry personnel, and database administrators. It is interesting to note that while the ABOS database is reflective of candidates taking Part II of the orthopaedic board examination, who are arguably under greater scrutiny regarding the appropriate indications for procedures such as acromioplasty as compared with cases performed by experienced surgeons captured in the SPARCS database, there were similar trends in both the overall increase of acromioplasties and the relative increase in the number of procedures for the treatment of rotator cuff tears. As both groups of surgeons with different incentives displayed similar trends, it is likely that the increase in frequency of acromioplasties is not explained by inappropriately indicated cases.
Another interesting finding in Part A of the current investigation was that the payer had an effect on the likelihood of having an acromioplasty in New York State. For example, patients with Blue Cross hospitalization insurance were less likely to have an acromioplasty procedure than those with Workers’ Compensation or commercial insurance but were more likely to have an acromioplasty procedure than those with Medicare or Medicaid. However, when one examines this pattern over the span of the eleven years analyzed, this distribution was relatively static. For instance, the relative proportion of cases involving patients receiving Workers’ Compensation remained relatively constant, ranging from a low of 20.6% of all cases in 2000 to a high of 23.6% of all cases in 1996. However, there was a slight decrease in the percentage of cases involving patients with commercial insurance, from 21.5% of all cases in 1996 to 15.2% of cases in 2006. This adds further evidence that a patient's insurance status can affect the likelihood of access to care for both medical and surgical interventions46-49.
The current investigation had several limitations. One limitation was that Part A used ICD-9 procedure codes rather than CPT codes. While CPT codes can distinguish between arthroscopic and open acromioplasty, there exists only one ICD-9 procedure code for acromioplasty, which makes this selection of cases less specific. Indeed, it would have been useful to distinguish open from arthroscopic acromioplasties in Part A of the investigation, which was an advantage of Part B. Another limitation common to any investigation that uses administrative diagnosis and procedure codes to investigate clinical or epidemiologic trends is the possibility of coding errors in the reporting and recording of such codes. While the accuracy of the SPARCS database has been previously examined and validated50, it remains possible that there were small coding inaccuracies, as inaccuracies in ICD-9 coding in general practice have been previously documented51. Another limitation of the current investigation is the lack of an ideal surgical control group with which to compare the frequency of the acromioplasty procedure. All orthopaedic procedures were chosen as the most ideal control as this group of procedures represents a large, broad, constant group of procedures, although it is one that is obviously extremely heterogeneous. A final limitation of the current investigation is that one must use caution in extrapolating the results from either the SPARCS database or the ABOS database to the actual practice patterns of all orthopaedic surgeons on a national level. The SPARCS database is a single state record, and the case mix of surgeons in New York State may differ from that of surgeons practicing in other states. Similarly, the ABOS database reflects the practice of the relatively small group of recently trained orthopaedic surgeons in the United States, which may not represent the case mix of more experienced orthopaedic surgeons in the nation. Specifically, the case mix of candidates and their indications for procedures performed are potentially highly scrutinized, and there is therefore a strong incentive for surgeons to be appropriate in their indications for procedures performed during their board collection period for concern of scrutiny by the ABOS; Part B therefore may underestimate the overall volume of arthroscopic acromioplasties. Nonetheless, we have shown in two independent databases nearly identical findings regarding the increased frequency of acromioplasty during a recent span of years.
In summary, the present data document a substantial increase in the overall volume and population-based incidence of acromioplasty on both a limited state level and the national level for both open and arthroscopic procedures. The reasons for this increase have yet to be determined and are likely multifactorial, with patient-based, surgeon-based, and systems-based factors all playing a role.