Katz et al., utilizing the Medicare data base, attempt to answer two questions: (1) On a numerical basis, does a revision of a primary total hip arthroplasty occur in the same hospital as the primary procedure, or in a different hospital? and (2) Do higher-volume hospitals perform a disproportionate number of revision arthroplasties (compared with the number of primary arthroplasties performed at the hospital and the expected number of revisions that would be generated)? The results have potentially important health policy implications.
Previous studies involving the use of two large databases, the Medicare claims database and the Nationwide Inpatient Sample database, have given insight into the relationship between surgeon volume and the early outcomes of total hip arthroplasty, early complication rates after primary and revision total hip arthroplasty, and the types of revisions performed. In 2008, Manley et al. reported on the effect of surgical volume on early revision rates less than six months after the primary total hip arthroplasty and showed that patients who were managed by a surgeon who performed more than fifty total hip arthroplasties annually were less likely to have undergone a revision total hip arthroplasty than those who were managed by a surgeon who performed fewer than twenty-five total hip arthroplasties annually1. Two separate publications by similar groups of authors have shown the dislocation rates at ninety days after total hip arthroplasty to be between 3% and 4% (3.1% and 3.9%)2,3. The reported results for dislocation rates after revision total hip arthroplasty increased to 8.4%2 and 14.4%3. Bozic et al. recently reported on the types of arthroplasty revisions, and the causes for the revision surgery, after examining 51,345 revision total hip arthroplasties performed between October 1, 2005, and December 31, 20064. Over 40% of the time, the revision involved all components, and the most common reasons for revision were instability/dislocation (22.5%), mechanical loosening (19.7%), and infection (14.8%).
In the current study, Medicare claims data on patients who underwent a primary total hip arthroplasty between July 1995 and June 1996, followed by revision surgery performed through December 31, 2006, were examined. Hospital volume was stratified into four groups on the basis of the number of primary total hip arthroplasties performed in the year study period: twenty-five or fewer, twenty-six to fifty, fifty-one to 100, and >100. The number of revisions performed over the ensuing ten-year period for these same volume strata were evaluated. From an initial cohort of 58,521 primary total hip arthroplasties, a total of 4460 revisions were performed. After excluding twelve patients, 4448 patients were examined. The overall trend was that lower-volume hospitals generated more revisions but performed fewer revision arthroplasties, whereas higher-volume hospitals generated fewer revisions yet performed more revision arthroplasties. Hospitals that performed twenty-five or fewer total hip arthroplasties in the study year performed 21,068 total hip arthroplasties, generated 1710 failed primaries, and performed 1476 revisions. Hospitals that performed >100 total hip arthroplasties in the study year performed 5889 total hip arthroplasties, generated 438 failed primaries, and performed 531 revisions. In the Discussion, the authors state, "The number of revisions performed in the cohort and the number of primary total hip replacements that eventuated in revision were not distributed evenly across volume strata. We found that the highest-volume hospitals performed 21% more revision total hip replacements than they generated. Similarly, in the lowest-volume hospitals, the number of revisions performed was 14% lower than the number of revisions generated. This issue is pertinent because of the documented associations between a lower volume of revision total hip replacement and an increased risk of adverse outcomes."
The study by Katz et al. suggests two further areas of potential inquiry. Specifically, the type of revision performed (e.g., head and liner exchange, one component, two component, etc.) relative to the hospital volume strata, and whether the surgeon who performed the initial total hip arthroplasty performed the revision, would be valuable additions to our understanding of the environments in which these arthroplasties are performed.
The policy implications of this study and the previously cited studies are of importance. In the United States, until a national joint registry becomes a reality, the outcomes of surgery can currently be only retrospectively inferred by utilizing large databases. To date, we can summarize the literature as showing that higher volume leads to a lower early complication rate, that instability (despite the presumed use of larger-diameter femoral heads for total hip arthroplasty) remains an important cause of revision surgery (as do mechanical loosening and infection), and, finally, that the treatment of these complications with all-component revision remains a costly intervention, with average total charges for revision total hip arthroplasty reported to be $54,5534. Katz et al. add to this understanding the information that lower-volume hospitals generate more revisions than would be expected on the basis of their primary total hip arthroplasty volume, whereas higher-volume hospitals perform more revisions than would be expected. Given the current health-care climate comprising rising costs, rising patient expectations and demands, and the increasing demands of the payers for the limitation or elimination of complications (the so-called "never" events) the reader who performs total hip arthroplasty would be well advised to examine the data presented while carefully critiquing his or her own results.
Dr. Katz and his colleagues are to be commended for an important addition to our indirect understanding of the outcomes of a common orthopaedic procedure.