In the United States the treatment of an infection at the site of a total hip arthroplasty is usually performed in two surgical stages, whereas in Europe single-stage surgery, or direct-exchange arthroplasty, is often employed. To date, the literature indicates that the two-stage approach is favored, as the ability to eradicate the infection is typically improved. Depending on the organism identified, the patient's own host factors (medical comorbidities), and the ability to deliver appropriate antibiotics, the quoted success rate for a two-stage approach is 85% to 90%1.
While, from the surgeon's perspective, treatment success depends on the ability to eradicate the infection, a more comprehensive analysis would include an examination of other possible treatment outcomes and assessment of their impact on the patient's quality of life. Alternative outcomes that impact the patient's quality of life include failure of the initial surgical treatment, necessitating repeat two-stage surgery or Girdlestone arthroplasty, or complications of the preferred treatment (e.g., intraoperative fracture, dislocation, a period of inactivity during treatment, and death either at or after the time of surgery). This expanded analysis allows an examination of the utility of the procedure. Utilities are patient-preference-based measures that are used to examine how patients value the status of health generated by the outcome of the intervention.
In their article, Wolf et al. examine the available literature on two-stage and direct-exchange surgery for the treatment of infection. To be included in the analysis, each paper had to specify the outcome of treatment, including the failures requiring a subsequent reoperation, the complications and their treatment, and the number of deaths. The authors found eleven articles on two-stage treatment, comprising a patient group of 321 and published between 1996 and 2007, and eight articles on the direct-exchange treatment, with a total patient group of 576 and published between 1986 and 2003. From the pooled sets of literature, the authors identified the probabilities for success and the rates of death, repeat treatment due to failure to eradicate the infection, and mechanical complications treated operatively or nonoperatively. Patient and surgeon-derived utility values for each outcome were identified. The probabilities and utilities were then analyzed with use of a Markov model2. In a Markov analysis, starting from an initial defined health problem (in this case, an infection at the site of a total hip arthroplasty), one can examine each possible subsequent outcome for the utility of the outcome (infection eradication would have a high utility, and death would have the lowest utility). As each possible outcome represents a transition from one health state to another (termed "a Markov state"), the new health state can be examined with regard to the subsequent utility by reentry into the Markov analysis. This type of analysis has, to date, been used primarily to examine the outcomes of treatment of medical conditions. A PubMed search with use of the term "Markov decision analysis and medical" identified 521 articles, whereas a search with use of "Markov decision analysis and orthopaedics" identified three articles.
The results of this paper are striking and bear careful evaluation. In comparing the two surgical approaches, the authors concluded that, although the two-stage treatment yields a greater chance of eradication of infection, it also yields a greater chance of death. When the utility of the procedures was examined, the direct-exchange treatment was otherwise favored for all outcomes assessed. This utility may be understood as patients undergoing a direct exchange not having to undergo a second surgical procedure (unless the direct exchange fails), avoiding the decreased function inherent in the time interval between the procedures of a two-stage approach, avoiding the longer recovery time necessitated by two operations, and having a lower probability of death.
One must remember the limitations of the analysis, as it is based on pooled data obtained from literature reported over a span of either eleven years (for the two-stage procedures) or seventeen years (for the direct-exchange procedures). Recumbency periods, organism profiles, and surgical techniques all have changed over time. Also, the numbers examined in each data set were small.
The authors are to be commended for carefully couching their findings in terms of the relevance to clinical decision-making and for suggesting that the findings be considered "hypothesis-generating for future randomized controlled trials" rather than definitive and suggestive of the need for a change in current treatment standards. This paper not only makes an important contribution to the understanding of the risks, benefits, and utility of direct-exchange versus two-stage treatment of infection at the site of a total hip arthroplasty, but also highlights the use of an important analytic tool that may aid in decision-making in orthopaedic patient care.