Scientific Articles   |    
Comparison of One and Two-Stage Revision of Total Hip Arthroplasty Complicated by InfectionA Markov Expected-Utility Decision Analysis
Christopher F. Wolf, MD1; Ning Yan Gu, PhD2; Jason N. Doctor, PhD3; Paul A. Manner, MD1; Seth S. Leopold, MD1
1 Department of Orthopaedics and Sports Medicine, University of Washington, Box 356500, 1959 N.E. Pacific Street, Seattle, WA 98195
2 Pharmerit North America, LLC, 4350 East West Highway, Suite 430, Bethesda, MD 20814. E-mail address: ngu@pharmerit.com
3 School of Pharmacy, University of Southern California, 1540 East Alcazar Street, CHP-140, Los Angeles, CA 90089-9004
View Disclosures and Other Information
Disclosure: The authors did not receive any outside funding or grants in support of their research for or preparation of this work. Neither they nor a member of their immediate families received payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity.

Investigation performed at the University of Washington, Seattle, Washington
A commentary by Thomas J. Blumenfeld, MD, is available at www.jbjs.org/commentary and is linked to the online version of this article.

Copyright © 2011 by The Journal of Bone and Joint Surgery, Inc.
J Bone Joint Surg Am, 2011 Apr 06;93(7):631-639. doi: 10.2106/JBJS.I.01256
A commentary by Thomas J. Blumenfeld, MD, is available here
5 Recommendations (Recommend) | 3 Comments | Saved by 3 Users Save Case



Two-stage revisions of total hip arthroplasties complicated by chronic infection result in reinfection rates that are lower than those following single-stage revisions but may also result in increased surgical morbidity. Using a decision analysis, we compared single-stage and two-stage revisions to determine which treatment modality resulted in greater quality-adjusted life years (QALYs).


A review of the literature on the treatment of patients with an infection at the site of a total hip arthroplasty provided probabilities; utility values for common postoperative health states were determined in a previously published study. With these data, we conducted a Markov cohort simulation decision analysis. Sensitivity analysis validated the model, and comparisons were made in terms of QALYs.


The twelve-month model favored direct-exchange revision over the two-stage approach, regardless of whether surgeon or patient-derived utilities were used (0.945 versus 0.896 and 0.897 versus 0.861 QALYs for the patient and surgeon models, respectively). Similar results were observed in a lifetime model with a ten-year life expectancy (7.853 versus 7.771, and 7.438 versus 7.362 QALYs, respectively). The findings were found to be robust in sensitivity analyses in which clinically relevant ranges of input variables were used.


This analysis favored the direct-exchange arthroplasty over the two-stage approach. This study should be considered hypothesis-generating for future randomized controlled trials in which, ideally, health end points will be considered in addition to the eradication of infection.

Level of Evidence: 

Economic and decision analysis Level II. See Instructions to Authors for a complete description of levels of evidence.

Figures in this Article
    Sign In to Your Personal ProfileSign In To Access Full Content
    Not a Subscriber?
    Get online access for 30 days for $35
    New to JBJS?
    Sign up for a full subscription to both the print and online editions
    Register for a FREE limited account to get full access to all CME activities, to comment on public articles, or to sign up for alerts.
    Register for a FREE limited account to get full access to all CME activities
    Have a subscription to the print edition?
    Current subscribers to The Journal of Bone & Joint Surgery in either the print or quarterly DVD formats receive free online access to JBJS.org.
    Forgot your password?
    Enter your username and email address. We'll send you a reminder to the email address on record.

    Forgot your username or need assistance? Please contact customer service at subs@jbjs.org. If your access is provided
    by your institution, please contact you librarian or administrator for username and password information. Institutional
    administrators, to reset your institution's master username or password, please contact subs@jbjs.org


    Accreditation Statement
    These activities have been planned and implemented in accordance with the Essential Areas and policies of the Accreditation Council for Continuing Medical Education (ACCME) through the joint sponsorship of the American Academy of Orthopaedic Surgeons and The Journal of Bone and Joint Surgery, Inc. The American Academy of Orthopaedic Surgeons is accredited by the ACCME to provide continuing medical education for physicians.
    CME Activities Associated with This Article
    Submit a Comment
    Please read the other comments before you post yours. Contributors must reveal any conflict of interest.
    Comments are moderated and will appear on the site at the discretion of JBJS editorial staff.

    * = Required Field
    (if multiple authors, separate names by comma)
    Example: John Doe

    Seth S. Leopold, MD
    Posted on October 21, 2011
    Dr. Romano's comments indeed are very thoughtful!
    University of Washington School of Medicine

    We thank Dr. Romano for his interest and careful review of our work. First, to give credit where credit is due, and to acknowledge an error that Dr. Romano identified in our manuscript: Several papers that were included in our analysis – specifically, those of Haddad, Masri, Takahira, and Tsukayama – appear to have enrolled some patients who may have had infections treated before 3 weeks after surgery. There is some ambiguity about this, as most of these refer to patients “presenting” with infection earlier, but with no mention of when treatment was initiated. Insofar as the spirit of these papers was clearly chronic infection – on the far end, they enrolled patients to 7, 13, 19, and 25 years, respectively – we opted to include them. Moreover, these all were papers in the 2-stage group, and to the extent that success rates for treating early infections are consistently better than for treating chronic infections, including these papers should have, if anything, improved the results in the 2-stage group relative to the direct-exchange group, rather than doing the opposite, as Dr. Romano feared. Dr. Romano is correct that in Koo’s paper some patients were treated before 3 weeks; however, that paper included a table that separated these out, and we included only those patients in Koo’s paper who treated late. The only direct-exchange papers that had any ambiguity about start dates were those of Raut and Younger; the former refers to deep infections and the spirit of the paper is clearly one treating chronic infections, while the latter deals with patients who have major bone loss, again, a phenomenon that is not associated with acute infections. In retrospect, we should have contacted the authors of the papers in question for greater clarity, but for the reasons mentioned, we don’t see this either a fatal flaw or as affecting the findings of the paper – particularly given that the sensitivity analysis would have required the infection rate in the direct exchange rate to increase four to five times what the literature would substantiate in order to offset the difference in disutility generated by the staged approach, and given that most of the papers in question were in the 2-stage group, and the bias, to the extent that any would be introduced by including those papers, would therefore have been in favor of the 2-stage approach.We opted not to include particular approaches to 2-stage or direct revisions (such as using antibiotic-impregnated grafts), so the papers using particular or novel techniques that Dr. Romano identified were not included. Rather, we sought to use the “traditional” approaches to 2-stage and direct-exchange revision. As for not including papers since ’08…at some point we had to stop collecting data and start analyzing it. Peer review also takes time. More importantly on this point, though, the parameter regarding the end date for publications to be reviewed was determined at the outset of the endeavor so as to not introduce bias. It is easy to see that if no end date for evaluating publications is set, then analysts may continue to collect data and stop only when a favored result is obtained. We wanted to avoid this problem. We stopped collecting papers when we started analysis. That was in 2008. We invite Dr. Romano – or any others interested in the subject – to repeat the process whenever the enthusiasm to do so is there. Decision analyses indeed should get updated as the literature evolves. We would read that update with great interest! It is possible that the results of our study have upset some readers. We ourselves were upset by the findings at first blush. Not only did the results cut against the dominant paradigm, but they also are at odds with our own typical approach to the problem of the chronically infected arthroplasty. Before doing the study, the two hip surgeons involved (PAM and SSL), like most in the United States, were avid users of the two-stage approach to the infected THA. We, like most surgeons, remain users of the two-stage approach for most patients; what has changed for us is that now we are more open-minded to the possibility that direct-exchange may be an overlooked option. Dr. Romano raises the possibility of dual publication of results, and he correctly identifies that several authors published more than one paper on the topic of the infected revision that came up in our search. In academic orthopaedics, it is not uncommon for the same clinician-scientist to publish more than one clinical series on the same topic. The alternative explanation, as Dr. Romano implies, is that some of these were either dual publications or inclusion of the same patients in more than one series without pointing that out; this alternative explanation, in either form, would be scientific misconduct. Absent evidence in support of such an allegation, we could not assume it to be the case. As for the findings being “unsubstantiated,” Sensitivity analysis is a particularly powerful tool for substantiating the findings, and testing how robust they are. To make it easier on readers to do this themselves, we published graphs with the manuscript (not in the appendix, but in the paper itself), which allow each reader to change the boundary parameters as (s)he desires. If a reader feels that the literature (or we) underestimated, say, the re-infection rate in the direct-exchange group, that reader can simply consult those graphs and adjust the numbers to see how high the re-infection rate would have to be in that group before it would offset the added morbidity associated with the second stage and the period of recumbency associated with the two-stage approach. As shown in Figure 3, that re-infection rate would have to exceed 60%, which is many times higher than the literature substantiates. While sensitivity analysis is a powerful tool for this purpose, so is clinical intuition. Does Dr. Romano doubt that a second major surgical procedure, a period of substantial immobility averaging over 4 months, and the physical deconditioning that inevitably follows this treatment approach would add morbidity when this approach is applied broadly to a largely-geriatric population? We don’t doubt this, either. Decision analysis allows us to apply utility values and probabilities to the inevitable complications (and, yes, additional deaths) associated with the larger and longer treatment that is the two-stage approach. Patients value health outcomes other than recurrent infection; this study suggests that surgeons should, too. As we stated in the paper, we consider our results to be hypothesis-generating, and not the kind of study that should immediately change the way revision arthroplasty is practiced. The best way to answer the question would be to have a multi-center randomized trial to complement (or refute) our results from this decision analysis. We were very explicit about recommending against sweeping changes to practice in the absence of such a study, and dedicated our entire concluding section to emphasizing this point. Finally, Dr. Romano is concerned that some readers will change practice based on reading an abstract rather than a paper. Mindful of this concern, we were careful to conclude our abstract, similar to the paper itself, with the following recommendation: “This study should be considered hypothesis-generating for future randomized controlled trials in which, ideally, health end points will be considered in addition to the eradication of infection.” But we teach our residents not to quote abstracts, and certainly not to change practice based on reading them – and we trust Dr. Romano does similarly. In fact, his detailed and thoughtful review leaves no doubt on this point. Dr. Romano is indeed a careful reader, and we thank him for taking the time to write, and for sharing his considerable insight with us and with readers of the Journal.Respectfully, Christopher F. Wolf, MD; Ning Yan Gu, PhD; Jason N. Doctor, PhD; Paul A. Manner, MD; Seth S. Leopold, MD

    Carlo L. Romanò
    Posted on October 17, 2011
    Should we trust this paper ?
    Istituto Ortopedico IRCCS Galeazzi - Milano - Italy

    I have read with attention the paper by Christopher F. Wolf, MD, et al.: "Comparison of one and two-stage revision of total hip arthroplasty complicated by infection. A Markov expected-utility decision analysis" and I found it has a number of bias which make debatable the conclusions of this paper. (1) A first limitation concerns the relative death chances of the two surgical approaches, that is quite relevant to the final conclusion proposed by the authors. In this regard, Dr. Wolf and co-workers conclude that, although the two-stage treatment provides a greater chance of eradication of infection, it also yields a greater chance of death. In particular, the authors sum up 8 death out of 321 in the two-stage series versus 3 out of 576 in the one-stage procedure. Given the lack of the reason for the reported deaths, the relatively small number of patients that died after each procedure and considering the complete absence of any description of the pre-operative status of the patients, that may have well explained the different mortality rates, this parameter should not have been considered for further analysis, since the assumption that the choice of one- or two-stage procedure caused the different incidence of death is not substantiated and never previously reported in the literature; instead, the authors only mention this as one of the several limitations of the study in the Discussion section (“Most of the reports in the literature were not sufficiently detailed to permit explicit understanding of the causes of most of the deaths; however, it seems possible that they were related to the second surgical intervention and/or the period of relative recumbency, which lasted, on average, approximately four months. Neither the cohort of articles on the staged revisions nor the cohort on the direct-exchange revisions consistently presented the preoperative health status or American Society of Anesthesiologists grade of the patients in the studies, and we found no randomized comparisons of the two cohorts. These issues place limitations on the inferences that one may draw from these reports.”). (2) A second bias is represented by comparing pooled data reported over different span periods and this has also been recently pointed out by Dr. Blumenfeld in his Commentary to the paper (J Bone Joint Surg Am. 2011;93:e33(1-2). (3) A third bias concerns papers selection, under different regards: (a) The authors state that “The initial literature review yielded 4048 articles. All articles pertaining to direct-exchange or staged revision for the treatment of patients with an infection at the site of a total hip arthroplasty were read carefully. Articles were excluded if the duration of infection was less than three weeks, appropriate postoperative antibiotics based on the findings of cultures and sensitivity testing were not used, detailed follow-up data were not provided, and/or the duration of follow-up was less than two years. Bibliographies of the included articles were then hand-searched to ensure that no eligible articles were overlooked. Non-English-language journal articles were included; unpublished articles or studies presented only as association meeting abstracts were not. Our search yielded eleven articles dealing with a two-stage protocol and eight focusing on a direct-exchange protocol”. 12 papers out of 4048 articles seems a rather strict selection. However, according to the indicated inclusion criteria, some more papers could have been included (cf., among others: Buttaro MA, Pusso R, Piccaluga F. Vancomycin-supplemented impacted bone allografts in infected hip arthroplasty. Two-stage revision results. J Bone Joint Surg Br. 2005 Mar;87(3):314-9. Nusem I, Morgan DA. Structural allografts for bone stock reconstruction in two-stage revision for infected total hip arthroplasty: good outcome in 16 of 18 patients followed for 5-14 years. Acta Orthop. 2006 Feb;77(1):92-7). (b) On the other side, the following papers, included in the data analysis by the authors, did not meet at least one their inclusion criteria (“Articles were excluded if the duration of infection was less than three weeks”): 1. Haddad et al., 2000: “Patients and Methods. …… All had cemented arthroplasties with sepsis presenting between two weeks and seven years after operation (Fig. 1). The original diagnoses are summarised in Table I.(…)” 2. Takahira et al., 2003: “Patients and Methods. ….Infection presented 4 days to 19.4 years after the last THA operation or hemiarthroplasty (…)” 3. Koo et al., 2001: “Patients and Methods. …the interval between the previous index operation and the revision range from 10 days to 12 years (...)” 4. Masri et al., 2007: “Results. The diagnosis of infection was made 2 weeks to 13 years after the original operation. (…)”. 5. Tsukayama et al., 1996: “Results. The mean interval between the index operation and the diagnosis of infection was 2.6 years (range, seven days to twenty-five years). (…)”.while in the following articles, the indicated inclusion criteria were not mentioned and hence should also have been excluded: 1. Raut et al., 1994: no reference to the time of onset of infection (only refers to deep infection); 2. Younger et al., 1998: no reference to the time of onset of infection, it only refers to patients with major bone loss. (c) The authors do not explain why in their analysis, published in April 2011, only papers available until May 2008 were considered; a search through all year 2008, 2009 and eventually 2010 would have been possible and could have at least provided an overall similar number of patients, treated according to the different procedures; (d) It is not mentioned nor discussed the finding that among the 8 papers reporting on one-stage procedures 4 come from the same group (Raut, Wrobleski, et al.), covering the same time period, making it highly possible that the patients considered in the 4 papers are also the same. As a matter of fact, 354 out of 576 patients (61.5%), included in the one-stage evaluation, come from these 4 papers. Even in the hypothesis that those authors were able to report one year on 57 patients (Raut et al., 1994), the year after on 183 different patients (Raut et al, 1995) and then again on 15 more patients (Raut 1996), how can we generalize data that, for more that 61% come from one or two surgeons? How could the authors, if “All articles pertaining to direct-exchange or staged revision (…) were read carefully”, not appreciate that 4 out of 8 papers were from the same hospital and 3 of them were published from the same first author? (e) Similarly, as to concern the selection of the 12 cited papers on two-stage, 4 comes from the same 2 groups (Hsieh et al. and Younger, Masri et al.). In particular, the 2 papers by Hsieh et al. cover the same time period and it is again therefore highly possible that in the two papers the same patients have been included. In total, 122 out of 321 patients (38.0%) included in the two-stage evaluation come from these 4 papers. (f) Unfortunately, not only this paper selection bias is not discussed by the authors, but the Journal put this relevant material only in Appendix 1, that is available only to subscribers, on the internet and is not reported within the article. As far as I know, most of the readers stop on the abstract, some go through the paper and very few take the time to further consult the Appendix of the papers they read. In this way the message that would remain for most of the orthopaedic community is what we find in the abstract: “This analysis favored the direct-exchange arthroplasty over the two-stage approach” and “...the findings were found to be robust...”.Given the relevancy of the topic and the possible consequences of such analysis on the decision making process of the surgeons and of the health care providers in the world, I believe it necessary to make the orthopaedic community aware that the conclusions drawn in this paper (“This analysis favored the direct-exchange arthroplasty over the two-stage approach”) may not be correct and the observation made by the authors that “Unfortunately, to our knowledge, no meta-analyses or randomized controlled trials related to this issue have been reported” should not be seen as a justification for an analysis based on insufficient or questionable data but should, at the opposite, prompt us to be extremely careful before spreading unsubstantiated hypotheses as conclusions.

    Related Content
    The Journal of Bone & Joint Surgery
    JBJS Case Connector
    Topic Collections
    Related Audio and Videos
    PubMed Articles
    Clinical Trials
    Readers of This Also Read...
    JBJS Jobs
    LA - Ochsner Health System
    OH - University Hospitals Case Medical Center
    DC - Children's National Medical Center
    WV - Charleston Area Medical Center