Increasing consolidation of total joint arthroplasties at higher-volume centers1,2 and adoption of volume standards3 have been proposed as ways to improve patient outcomes. An association between complication risk following joint arthroplasty and greater hospital and surgeon procedure volume has been demonstrated1,3-8. Some population-based studies have demonstrated a reduction in the risk of mortality and complications, such as acute myocardial infarction, pulmonary embolism, deep infection, dislocation, and pneumonia, with an increase in procedure volume1,3-8. Other studies have not identified similar associations9. Limitations in prior analyses include data restricted to selected states6-9, reliance on a single year of data1,3,5, or a lack of consideration of postdischarge complications3,4. It remains unclear if procedure volume is associated with other clinical outcomes, such as the rate of revision surgery.
The association between the volume of total hip arthroplasties performed and the revision rate up to four years postoperatively8,10 has been previously examined. Although surgeon volume was found to be associated with higher revision rates immediately after the surgery8,10, further follow-up did not demonstrate a significant correlation10. Hospital volume was not found to be associated with short-term revision rates8. In addition, the patient's functional status at three years following surgery did not appear to be associated with the hospital and/or surgeon volume11.
The objective of this study was to determine the association between hospital and surgeon volumes of total hip arthroplasties and the revision rates at 0.5, two, five, and eight years in the Medicare population. Specifically, we examined the hypothesis that the performance of elective total hip arthroplasty at higher-volume institutions and by higher-volume surgeons would not be associated with revision rates that differed significantly from those following arthroplasties performed at lower-volume institutions or by lower-volume surgeons.
Materials and Methods
The 5% systematic Medicare sample (Beneficiary Encrypted File [BEF]) from 1997 to 2004 was used to extract total hip arthroplasty claims data. This sample represented claims from approximately 2.5 million Medicare beneficiaries whose Medicare records were consistently extracted annually. The 5% sample is created by the Centers for Medicare and Medicaid Services (CMS) from the Medicare National Claims History database by selecting beneficiaries whose Health Insurance Claim number contains specific digits12. The 5% Medicare sample has been used previously to examine population-based health outcomes research questions, such as the effect of epidural analgesia on morbidity and mortality following total hip arthroplasty13 and total knee arthroplasty14 as well as the risk of mortality and pulmonary embolism following pelvic, patellar, and long-bone fractures15.
Primary total hip arthroplasties performed from January 1, 1997, to December 31, 2004, were identified according to the claim records submitted by hospitals and surgeons using the appropriate ICD-9-CM (International Classification of Diseases, Ninth Revision, Clinical Modification) and CPT-4 (Current Procedural Terminology, Fourth Revision) codes. The CPT code 27130 (arthroplasty, acetabular and proximal femoral prosthetic replacement) was used for primary total hip arthroplasty. All patients who were younger than sixty-five years old, who were not enrolled in both Part A and Part B of Medicare, who were enrolled in a health maintenance organization (HMO), or who had been diagnosed with bone cancer, metastatic cancer, or joint infection were excluded from the study. We further limited the study population to patients who had undergone elective arthroplasty. Thus, patients with a pelvic or femoral fracture (ICD-9-CM codes 808 and 820 to 821.11) were excluded.
The hospital and surgeon associated with each procedure was identified on the basis of the unique hospital (provider) and surgeon identifier codes in the claims data. Because each hospital's provider identification number was not encrypted in the 5% BEF data, information regarding the institution's characteristics (for example, number of beds, teaching status, and setting [rural or urban]) was obtained from the American Hospital Association's annual survey. The individual hospital and surgeon volumes were considered to be the average number of primary total hip arthroplasties performed annually during the study period.
Total hip arthroplasty revision was defined as revision of any component. Revision rates were examined at 0.5, two, five, and eight years postoperatively as these are short and medium-term time points typical of clinical follow-up. Each patient who underwent a primary total hip arthroplasty was tracked longitudinally to identify subsequent admissions for revision surgery. Revisions were identified by the CPT-4 codes 27134, 27137, and 27138 from surgeon (Part-B) claims records. If no revision was identified, the censored survival period was calculated until the last date of enrollment at or before the 0.5, two, five, or eight-year end point. Patients who died without having undergone a revision were considered censored, and the longevity of their respective implants was calculated up to the date of death. Each beneficiary's enrollment status and date of death were identified in the annual Medicare denominator files.
The overall revision rate in each cohort was analyzed with the Kaplan-Meier approach. The covariates for this study included sex, age, race/ethnicity, a diagnosis of arthritis, hospital volume, surgeon volume, hospital teaching status, hospital ownership (public, private, or nonprofit), hospital location (rural or urban), hospital size, and Medicare buy-in status. The Medicare buy-in status is an identifier of patients whose Medicare premiums and deductibles were covered by the state because of their financial status and was used as a proxy for patients with a lower socioeconomic status or in poverty.
Hospital size was characterized on the basis of the number of beds (less than 150, 150 to 299, 300 to 499, and more than 500). To facilitate direct comparison with previous research, we used the same surgeon and institution-volume designations as those used in studies of short-term complication risks of total hip arthroplasty in the Medicare population5. Specifically, the hospital procedure-volume groups were one to ten, eleven to twenty-five, twenty-six to fifty, fifty-one to 100, and more than 100. The surgeon procedure-volume groups were one to five, six to ten, eleven to twenty-five, twenty-six to fifty, and more than fifty. Differences in revision rates were assessed with the log-rank test. The proportional hazards or Cox regression method, with adjustment for the covariates mentioned above as well as with control for comorbidities as measured with the Charlson index16, was used to evaluate the joint effect of multiple covariates at 0.5, two, five, and eight years, with the highest-volume hospital and surgeon categories serving as the reference. The hazards model also included the year as a variable to account for possible differences in revision rates between years. The significance level was defined as a p value of <0.05.
Cohort and Hospital Characteristics
During the eight-year study period (1997 to 2004), 26,036 primary total hip arthroplasties that met our selection criteria were identified from the 5% Medicare sample data; 2746 hip arthroplasties associated with a fracture and 145 associated with bone cancer were excluded from the initial patient sample. On the basis of the 5% sampling design, the selected patient sample extrapolated to a national total of 520,720 total hip arthroplasties in the Medicare population over the study period. The Medicare total hip arthroplasty patient population grew by 39.8% from 1997 to 2004, increasing steadily from an extrapolated national sample of 57,420 to 80,260. By the end of 2004, 3.6% of the patients who had been identified as having undergone a primary total hip arthroplasty in this study had undergone revision surgery. Of the patients who underwent revision surgery, 62% were operated on by the same surgeon who had performed the primary surgery. In the 5% sample, 23,705 of the patients who had had a primary procedure were available for follow-up at six months; 17,222, at two years; 7566, at five years; and 2857, at the end of the study period.
Approximately one-third of the primary total hip arthroplasties were done at hospitals with an average annual volume of more than 100 total hip arthroplasties (Table I). Only 5% of the total hip arthroplasties were carried out at centers with an annual volume of twenty-five procedures or fewer. One-third of all of the primary total hip arthroplasties were performed by surgeons with an annual volume of eleven to twenty-five procedures, whereas about one-sixth were performed by the highest-volume surgeons (more than fifty procedures) (Table I). The distribution of surgeon and hospital volumes was relatively unchanged from 1997 to 2004.
Primary total hip arthroplasties in the Medicare population were performed equally between teaching and non-teaching institutions (see Appendix). Hospital characteristics were relatively unchanged from 1997 to 2004. The Charlson comorbidity index of the patients had a weak but significant correlation with hospital volume (p < 0.0001; Spearman rank correlation, r = −0.031) and surgeon volume (p < 0.0001; Spearman rank correlation, r = −0.034) (see Appendix).
Procedure Volume Effects
Multivariate proportional hazards regression analysis showed that the rate of total hip arthroplasty revision was significantly associated with surgeon volume at six months (p = 0.018) but not at any other time points (Fig. 1). The hospital volume did not have a significant effect on the revision rate at any time point (p > 0.05) (Table II).
Medicare patients who were operated on by the highest-volume surgeons (more than fifty procedures per year) had a lower revision rate at six months (0.7%) than those operated on by surgeons with an annual procedure volume of six to ten (1.3%) or eleven to twenty-five (1.3%). The relative risk of revision for patients in each surgeon-volume group decreased over time relative to the revision rate for the patients treated by the highest-volume surgeons. No consistent associations between the total hip arthroplasty revision rate and the hospital's teaching status (except at five and eight years [p ≤ 0.015]), ownership (except at five and eight years [p ≤ 0.025]), location, or number of beds were found (Table III). Non-teaching hospitals tended to be associated with a higher revision risk, while nonprofit hospitals tended to be associated with a lower revision risk. With regard to patient demographics, age and a diagnosis of arthritis were found to have significant (p < 0.05) associations with the total hip arthroplasty revision rate (Table III). Younger patients (sixty-five to sixty-nine years old) and patients with a diagnosis other than arthritis generally had greater risks of revision. The numbers of revision procedures performed by each surgeon and hospital-volume group are presented in a table in the Appendix.
Our data showed that, during the period from 1997 to 2004, Medicare patients who were treated with a total hip arthroplasty by low-volume surgeons (six to ten and eleven to twenty-five procedures per year) had a significantly greater risk of having a revision at six months postoperatively. With longer-term follow-up, no significant association between surgeon volume and the total hip arthroplasty revision rate was seen. Furthermore, we did not find any significant association between the total hip arthroplasty revision rate and the hospital volume at any time point.
Although the effect of volume on outcome has been examined in a number of previous studies, most focused on mortality and complications, particularly in the short term1,3-5,7,9. There have been some limited investigations of procedure-volume effects on revision rates extending beyond the early follow-up period as well as investigations of the temporal changes in volume effects8,10,17; our findings corroborated the observations in those studies.
Using 1995 and 1996 national Medicare data, Losina et al. showed that patients in whom a total hip arthroplasty was carried out by a surgeon who had performed fewer than twelve procedures per year were more likely to undergo revision than those operated on by higher-volume surgeons10. However, they reported a significant association (p < 0.0001) between revision and surgeon volume only during the first eighteen postoperative months, and not between nineteen and forty-eight months. Although we were unable to directly compare our data with those of Losina et al., our study also demonstrated that lower surgeon volume resulted in significantly higher rates of total hip arthroplasty revision within six months postoperatively. Similarly, Kreder et al. found that patients treated with total hip arthroplasty by lower-volume surgeons in Washington state had a higher risk of revision at three months and one year than did those treated by higher-volume surgeons8. Furthermore, their data showed that the relative risk of revision between the lower and higher-volume surgeons was not as great at one year as it was at three months. Whether the increased risk of revision for patients treated by lower-volume surgeons would have continued into the future is unknown.
Our finding of no association between hospital volume and the total hip arthroplasty revision rate at the time of short-term follow-up is consistent with the results of Kreder et al.8. We also determined that hospital volume effects were not significant at up to eight years after the surgery. Katz et al. provided insight into the volume effect on the Harris hip scores and satisfaction of Medicare patients at three years following total hip arthroplasty in Ohio, Pennsylvania, and Colorado11. They did not find any significant difference in the adjusted odds ratio for the Harris hip score between low and high-volume hospitals, although patients treated at low-volume hospitals were less satisfied, possibly because they had worse prognostic features and less preoperative function.
Our findings are also supported by a previous study by Judge et al. of hip and knee arthroplasties performed in the National Health Service system of the United Kingdom from 1997 to 200217. Although those authors identified an inverse relationship between thirty-day in-hospital mortality and hospital volume for both total hip arthroplasty and total knee arthroplasty, they found no association between hospital volume and readmission rates and five-year revision rates. In addition, they found some evidence of higher risks of readmission and revision for patients treated at high-volume centers compared with those treated at intermediate-volume centers. It was hypothesized that this finding was due to the greater number of complex cases treated at higher-volume centers. In our study, we attempted to account for confounding effects by adjusting for comorbidities and excluding patients who had undergone a nonelective procedure.
Unlike previous studies of total hip arthroplasty volume that focused on mortality4,5 and complications4,5,9 or shorter-term revision rates8,10, we evaluated total hip arthroplasty revisions at up to eight years postoperatively. Our data sample was also from a more recent patient cohort, which provides a better understanding of current trends. In addition, we accounted for potential confounding factors such as age, sex, race/ethnicity, comorbidities, Medicare buy-in status, and a diagnosis of arthritis. We found a significant difference in the patient case-mix, in terms of comorbidities, between low and high-volume hospitals and between low and high-volume surgeons. However, this appears to be due to the power to detect small differences in a large-patient-population study and may not correspond to substantial clinical differences. The relative proportion of patients in each comorbidity subgroup appeared comparable among the hospital-volume groups and among the surgeon-volume groups.
Our study was limited by the lack of clinical information on outcome measures such as pain and physical functional status11 in the Medicare data. In the absence of a national implant registry, the administration of a study that includes clinical outcomes on a national scale would be extremely resource-intensive and may only capture a modest proportion of eligible patients11; thus, it could suffer from self-selection or other forms of bias. Furthermore, it was not possible to evaluate the specific reason for the revisions as a result of the limitations of the ICD-9 codes. However, detailed complication codes (for example, for osteolysis, loosening, and so on) were introduced in October 2005, and that should allow more detailed investigations in the future. Also, the coding in the Medicare data for identifying which side had been operated on was incomplete, with approximately 65% and 40% of the claims lacking that information in 1997 and 2004, respectively. However, in a previous study, we found no significant difference in total hip arthroplasty revision rates between Medicare beneficiaries for whom the operatively treated side was known and those for whom it was not known (p = 0.11)18. Although approximately two-thirds of all total joint arthroplasty recipients are Medicare beneficiaries19, the findings from this study do not extend to younger patients, who typically are not insured by Medicare. Research using other relevant data sources is needed to provide additional analysis about volume effects in this younger population.
High-volume centers may have better standardization of services such as structured protocols, greater resources, and more integrated delivery of care17,20, which may account for better outcomes. As such, high volume has been considered a proxy for higher-quality processes of care6. Our findings, combined with those of other researchers4,5,8,11, indicate that the likely higher quality of care in larger-volume hospitals primarily affects mortality and complications but not longer-term revision rates.
Tables showing the distributions of primary total hip arthroplasties according to hospital characteristics and patient morbidity and the cumulative numbers of total hip arthroplasty revisions are available with the electronic versions of this article, on our web site at jbjs.org (go to the article citation and click on “Supplementary Material”) and on our quarterly CD/DVD (call our subscription department, at 781-449-9780, to order the CD or DVD).
NOTE: The authors thank Kate Sutton, MA, ELS, for assistance with editing of the manuscript.
Disclosure: In support of their research for or preparation of this work, one or more of the authors received, in any one year, outside funding or grants in excess of $10,000 from Stryker Orthopaedics. In addition, one or more of the authors or a member of his or her immediate family received, in any one year, payments or other benefits in excess of $10,000 or a commitment or agreement to provide such benefits from a commercial entity (Stryker Orthopaedics). No commercial entity paid or directed, or agreed to pay or direct, any benefits to any research fund, foundation, division, center, clinical practice, or other charitable or nonprofit organization with which the authors, or a member of their immediate families, are affiliated or associated.
Investigation performed at Exponent, Inc., Philadelphia, Pennsylvania
1. Katz JN, Barrett J, Mahomed NN, Baron JA, Wright RJ, Losina E. Association between hospital and surgeon procedure volume and the outcomes of total knee replacement. J Bone Joint Surg Am. 2004;86:1909-16.
2. Losina E, Kessler CL, Wright EA, Creel AH, Barrett JA, Fossel AH, Katz JN. Geographic diversity of low-volume hospitals in total knee replacement: implications for regionalization policies. Med Care. 2006;44:637-45.
3. Hervey SL, Purves HR, Guller U, Toth AP, Vail TP, Pietrobon R. Provider volume of total knee arthroplasties and patient outcomes in the HCUP-Nationwide Inpatient Sample. J Bone Joint Surg Am. 2003;85:1775-83.
4. Doro C, Dimick J, Wainess R, Upchurch G, Urquhart A. Hospital volume and inpatient mortality outcomes of total hip arthroplasty in the United States. J Arthroplasty. 2006;21(6 Suppl 2):10-6.
5. Katz JN, Losina E, Barrett J, Phillips CB, Mahomed NN, Lew RA, Guadagnoli E, Harris WH, Poss R, Baron JA. Association between hospital and surgeon procedure volume and outcomes of total hip replacement in the United States Medicare population. J Bone Joint Surg Am. 2001;83:1622-9.
6. SooHoo NF, Lieberman JR, Ko CY, Zingmond DS. Factors predicting complication rates following total knee replacement. J Bone Joint Surg Am. 2006;88:480-5.
7. Soohoo NF, Zingmond DS, Lieberman JR, Ko CY. Primary total knee arthroplasty in California 1991 to 2001: does hospital volume affect outcomes? J Arthroplasty. 2006;21:199-205.
8. Kreder HJ, Deyo RA, Koepsell T, Swiontkowski MF, Kreuter W. Relationship between the volume of total hip replacements performed by providers and the rates of postoperative complications in the state of Washington. J Bone Joint Surg Am. 1997;79:485-94.
9. Khatod M, Barber T, Paxton E, Namba R, Fithian D. An analysis of the risk of hip dislocation with a contemporary total joint registry. Clin Orthop Relat Res. 2006;447:19-23.
10. Losina E, Barrett J, Mahomed NN, Baron JA, Katz JN. Early failures of total hip replacement: effect of surgeon volume. Arthritis Rheum. 2004;50:1338-43.
11. Katz JN, Phillips CB, Baron JA, Fossel AH, Mahomed NN, Barrett J, Lingard EA, Harris WH, Poss R, Lew RA, Guadagnoli E, Wright EA, Losina E. Association of hospital and surgeon volume of total hip replacement with functional status and satisfaction three years following surgery. Arthritis Rheum. 2003;48:560-8.
12. Research Data Assistance Center. Medicare frequently asked questions. http://http://www.resdac.umn.edu
/Medicare/MedicareFAQ.asp. Accessed 2008 Jan 3.
13. Wu CL, Anderson GF, Herbert R, Lietman SA, Fleisher LA. Effect of postoperative epidural analgesia on morbidity and mortality after total hip replacement surgery in Medicare patients. Reg Anesth Pain Med. 2003;28:271-8.
14. Wu CL, Hurley RW, Anderson GF, Herbert R, Rowlingson AJ, Fleisher LA. Effect of postoperative epidural analgesia on morbidity and mortality following surgery in Medicare patients. Reg Anesth Pain Med. 2004;29:525-33, 515-9.
15. Barrett JA, Baron JA, Beach ML. Mortality and pulmonary embolism after fracture in the elderly. Osteoporos Int. 2003;14:889-94.
16. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613-9.
17. Judge A, Chard J, Learmonth I, Dieppe P. The effects of surgical volumes and training centre status on outcomes following total joint replacement: analysis of the Hospital Episode Statistics for England. J Public Health (Oxf). 2006;28:116-24.
18. Kurtz SM, Ong KL, Schmier J, Mowat F, Saleh K, Dybvik E, Kärrholm J, Garellick G, Havelin LI, Furnes O, Malchau H, Lau E. Future clinical and economic impact of revision total hip and knee arthroplasty. J Bone Joint Surg Am. 2007;89 Suppl 3:144-51.
19. Agency for Healthcare Research and Quality. http://hcupnet.ahrq.gov/. Accessed 2007 Mar 15.
Copyright 2008 by The Journal of Bone and Joint Surgery, Incorporated
20. Sharkey PF, Shastri S, Teloken MA, Parvizi J, Hozack WJ, Rothman RH. Relationship between surgical volume and early outcomes of total hip arthroplasty: do results continue to get better? J Arthroplasty. 2004;19:694-9.