While the adverse effect of excessive weight on the risk of osteoarthritis of the hip developing seems established1-4, the impact of obesity on the outcome of total hip arthroplasty remains controversial. As early as 1981, Chandler et al. found that a patient weight of >82 kg was detrimental to the survival of hip prostheses, and they included this variable in their calculation of a risk index for total hip arthroplasty5. In 1983, Surin and Sundholm found that excessive body weight (>80 kg) was associated with aseptic loosening6. More recently, a review of the literature led investigators to conclude that obesity appears to adversely affect the results of total hip and knee arthroplasty7. On the other hand, several studies in which obese and nonobese patients were compared showed no difference in complication rates or the early outcomes of total hip arthroplasty8-12.
In light of these conflicting data, many physicians are reluctant to recommend total hip arthroplasty for patients with a high body mass index13. The increase in stresses on the fixation of the prosthesis resulting from excessive body weight is often the justification for withholding total hip arthroplasty from obese patients even though such patients have been shown to have a lower level of activity14.
Hip resurfacing provides a prosthetic solution that closely matches the natural loading of the proximal part of the femur15. Our previous publications have demonstrated that a greater patient weight was not associated with early loosening of the femoral component of a hip resurfacing prosthesis16,17.
The purpose of this study was to determine whether there is an association between body mass index and the survivorship of a metal-on-metal hip resurfacing prosthetic design. We compared the clinical results of patients with a body mass index of =30 with those of patients with a body mass index of <30.
The study was approved by our institutional review board. Between November 1996 and September 2006, the senior author (H.C.A.) performed 1000 metal-on-metal resurfacing procedures (Conserve Plus, Wright Medical Technology, Arlington, Tennessee) at our institution. We retrospectively reviewed our registry to identify all patients with a body mass index of =30 who had undergone resurfacing arthroplasty during that time period for end-stage arthritis secondary to a variety of diagnoses and who had been followed for a minimum of two years. One hundred and twenty-eight patients (148 hips) met these inclusion criteria. Two patients (three hips) had died twenty-one and twenty-two months after the resurfacing of causes unrelated to the surgery, and one patient (one hip) was lost to follow-up, leaving 125 patients (144 hips) for inclusion in the study group.
We then identified a control group of 533 patients (628 hips) with similar diagnoses who had undergone resurfacing arthroplasty during the same period of time but had a body mass index of <30 at the time of surgery. Two patients (two hips) were lost to follow-up in the control group, leaving 531 patients (626 hips) for inclusion in the study.
The demographic characteristics of the study and control groups are presented in Table I.
The surgical technique and postoperative management have been described previously18-20. The metaphyseal stem of the femoral component was cemented in forty-three hips (30%) in the study group and in 229 hips (37%) in the control group.
Postoperative clinical and radiographic examinations were performed four months, one year, and then annually after the surgery16. The patients were evaluated with use of the UCLA (University of California at Los Angeles) hip scoring system including the UCLA activity score21, calculation of SF-12 (Short Form-12) scores from patient-completed questionnaires22, and Harris hip scores23. Preoperative Harris hip scores were not available for the whole cohort because we began collecting these data three years after the initiation of the series.
Intraoperative photographs were made at the end of femoral bone preparation24. They were reviewed retrospectively, and the size of the femoral head defect was recorded and entered into the calculation of the surface arthroplasty risk index (SARI)17 (see Appendix).
Prosthetic survivorship was determined with use of the Kaplan-Meier method25. The Cox proportional hazards ratio was used to study the effect of body mass index on the survival of the prosthesis, with adjustment for potential covariates. Because the effect of body mass index on survivorship was more linear than dichotomous, we divided the cohort into three subgroups on the basis of whether the patients had a body mass index of <25, 25 to 29, or =30. Student t tests (two-sample, equal variance) were used to compare parametric variables between the study group and the controls. Paired t tests were performed for the same variables to compare preoperative with postoperative scores. The Mann-Whitney U test was used for comparison of nonparametric variables between groups, and the Wilcoxon signed-rank test was used to compare preoperative with postoperative nonparametric scores. A p value of <0.05 was considered significant.
The average follow-up time was 6.2 years (range, 2.0 to 10.0 years) for the study group and 5.8 years (range, 2.0 to 10.2 years) for the control group. The preoperative and postoperative clinical scores are reported in Table II. The UCLA and SF-12 clinical scores were found to be significantly improved, compared with the preoperative values, at the time of follow-up (p < 0.0001) (Table II). (Preoperative Harris hip scores were not available for comparison with final follow-up scores.) The control group had significantly better preoperative UCLA walking (p = 0.0108), function (p = 0.0105), and activity (p = 0.0075) scores and SF-12 physical component scores (p = 0.0003) than the study group. Postoperatively, we found no significant differences between the two groups with regard to the average UCLA pain score (9.5 points in both groups, p = 0.4731) or walking score (9.6 points in the control group and 9.4 points in the study group, p = 0.1153) or the mental component score of the SF-12 survey (53.7 points in the control group and 52.5 points in the study group, p = 0.2033). In contrast, the UCLA function and activity scores were significantly better in the control group (9.6 compared with 9.2 points [p = 0.0013] and 7.6 compared with 7.1 points [p = 0.0021], respectively), as was the physical component score of the SF-12 survey (51.4 compared with 49.3 points [p = 0.0129]). The mean postoperative Harris hip score was 93.8 points (range, 38 to 100 points) in the control group compared with 90.6 points (range, 41 to 100 points) in the study group, which was a significant difference (p = 0.0003).
We also compared a subset of twenty-seven patients who had a body mass index of =35 with the other patients from the study group (body mass index of =30 but <35) and found no significant differences in their postoperative clinical scores, except for the mental component score of the SF-12 (48.9 compared with 53.4 points, respectively, p = 0.040).
The five-year survivorship of the prosthesis in the study group was 98.6% (95% confidence interval, 94.5% to 99.7%), with revision for any reason used as the end point. In comparison, the five-year survivorship in the control group was 93.6% (95% confidence interval, 90.4% to 95.8%). This difference was significant (log-rank test, p = 0.0401) as illustrated in Figure 1.
After we divided the entire study cohort into three subgroups according to whether the body mass index was <25, 25 to 29, or =30 and adjusted for the size of the femoral defect, the risk of revision decreased twofold as the mean body mass index increased from one subdivision to the next (Cox proportional hazards ratio, 2.01 [95% confidence interval, 1.17 to 3.41]; p = 0.013).
The five-year survivorship was 90.6% (95% confidence interval, 84.0% to 94.5%) in the subgroup with a body mass index of <25, 95.3% (95% confidence interval, 91.2% to 97.5%) in the subgroup with a body mass index of 25 to 29, and 98.6% (95% confidence interval, 94.5% to 99.7%) in the subgroup with a body mass index of =30 (our study group). The difference in survivorship between the subgroup with a body mass index of <25 and the study group was significant (p = 0.013).
The mean SARI score for the study group (see Appendix) was 1.5 points (range, 0 to 5 points); this was significantly lower (p < 0.0001) than the score for the control group (mean, 2.6 points; range, 0 to 6 points), as a result of a lower mean patient activity score and a greater mean patient weight. However, there was no significant difference in the average femoral cyst size (p = 0.860) between the study group (10.9 mm; range, 0 to 29 mm) and the controls (10.7 mm; range, 0 to 35 mm).
The average diameter of the femoral component head was greater in the study group (48.3 mm; range, 36 to 54 mm) than it was in the control group (46.8 mm; range, 36 to 54 mm) (p = 0.0001). A comparative analysis of those variables is presented in Table III.
Radiographic analysis revealed that three (2.1%) of 142 unrevised hips in the study group had radiolucencies. These radiolucencies appeared on the anteroposterior radiographs at an average of 27.7 months (range, twenty-two to thirty-seven months) after the surgery. All three radiolucent lines appeared about an uncemented femoral stem. The patients (two women with body mass indices of 30.0 and 30.8 and one man with a body mass index of 30.8) were asymptomatic at the time of the latest follow-up. None of these radiolucencies had progressed by an average of 5.2 years (range, 4.5 to 6.8 years) since their appearance.
Ten (1.7%) of the 595 unrevised hips in the control group had radiolucencies, which were also all about uncemented femoral stems. These radiolucencies appeared at an average of 22.6 months (range, nine to forty-two months). The prevalence of radiolucencies about unrevised uncemented stems did not differ between the study group (three of 105; 2.9%) and the control group (ten of 399; 2.5%) (p = 0.8421).
Two hips underwent revision in the study group. One was converted to a conventional total hip replacement five months after the surgery, following a fracture of the femoral neck. The other reoperation involved a reconstruction of the acetabular wall and revision of the acetabular component after it protruded through the medial acetabular wall three days after the surgery, in a patient with poor bone quality. In contrast, in the control group of 531 patients (626 hips), there were thirty-one revisions: twenty due to loosening of the femoral component, seven due to femoral neck fracture, two due to hematogenous infection, and one due to recurrent subluxations. The reason for one revision was undetermined, as it was performed at an outside institution and insufficient data were collected to classify the mode of failure.
There were six complications in the study group, all of which resolved without sequelae, yielding an overall complication rate of 4.2%. One patient sustained a dislocation when he fell one month after the surgery. He was treated with closed reduction, and the hip was stable at four years since this event. One hip became infected two months postoperatively. The infection was presumed to be hematogenous since primary wound-healing was uneventful. The hip was treated with débridement, antibiotics, and retention of the prosthesis. It was not necessary to revise the components, and at the time of writing there had been no episodes of infection for 4.5 years. Two patients had a femoral nerve palsy presenting as an inability to extend the knee; both palsies resolved completely, after one year in one patient and nine months in the other. One patient had thrombophlebitis three months after the surgery; it was treated with Coumadin (warfarin) without sequelae. Finally, one patient had a hematoma, which resolved without any specific treatment.
There were thirty-eight complications (a rate of 6.1%) in the control group; these included eight dislocations, five infections (including those in the two hips that were revised), fourteen femoral nerve palsies, and nine blood-related events (four thromboembolic events and five episodes of bleeding or hematoma). In addition, two patients underwent removal of heterotopic bone. All patients recovered. The difference in the complication rates between the study and control groups was not significant (p = 0.389).
There were no revisions in the subgroup of twenty-seven patients (22% of the study group) who had a body mass index of =35 at the time of surgery. However, three of the six complications in the study group occurred in this subgroup.
The calculation of the body mass index (patient weight [kilograms] divided by height [in meters] squared) and the body mass index cut points suggested by the American Obesity Association are commonly accepted and have provided us with an operational definition of obesity for this study. The validity of using the body mass index to infer physical fitness can be questioned26,27 because the measurement of weight does not discriminate between lean body mass and fat content.
The most important findings of this study are the absence of component loosening and the low prevalence of femoral neck fractures in the study group, and the positive effect of a greater body mass index on survivorship of the prosthesis at the time of the latest follow-up. The rate of femoral neck fracture was comparable between the study and control groups (0.7% and 1.1%, respectively), but the rate of loosening of an uncemented femoral component was greater in patients with a lower body mass index. There was no femoral loosening or radiolucencies in any of the hips in which the stem was cemented, regardless of body mass index. These findings are consistent with the positive association between weight and medium-term survival of resurfacing prostheses that we had suggested in previous publications16,17. There are several possible explanations for this result, among which is the increased average diameter of the prosthetic femoral head (1.5 mm greater) needed to perform the resurfacing in the patients from the study group, a difference attributable to the greater proportion of men in that group (Table III). As previously reported, the area of bone-cement contact increases linearly with the size of the femoral component28.
Also, because a higher body mass index is correlated with greater bone mineral density of the proximal part of the femur29, the study group may have benefited from having better bone quality. Anecdotally, we have found it more difficult to ream the femoral head in patients with a high body mass index. However, preoperative bone mineral density scans were not performed on our patients, and such observations remain hypothetical.
The study group had a lower postoperative activity score (even though 62% still regularly engaged in some sort of sports activity), and this probably contributed to the reduced rate of aseptic loosening in comparison with that in the control group. The three study patients with a femoral stem radiolucency had low UCLA activity scores (5 points for one and 6 points for two), and this might explain the absence of progression of the radiolucencies over time.
Our results agree with those reported by Wendelboe et al.30, who indicated that obesity does not increase the risk of revision surgery following total hip or knee replacements. This observation was confirmed in a recent publication by McLaughlin and Lee31, who found no difference between obese and nonobese patients with regard to revision rates following cementless total hip arthroplasty.
In a recent study, Mont et al. suggested that a body mass index of >35 be considered a contraindication for hip resurfacing32. However, our study showed that a body mass index of >35 does not in itself compromise prosthetic survival, at least in the intermediate term. We do believe that excessive weight, when combined with poor positioning of the femoral component (a component left proud with uncovered reamed bone) and other risk factors33, can increase the risk of femoral neck fracture.
Recent changes in our operative technique have resulted in dramatic reductions in the rates of radiolucencies and femoral component loosening28. These changes include better removal of cystic material, the use of dome and lesser trochanteric suction during application of the cement, an increased number of drilled holes for a larger bone-cement interface, and cementation of the metaphyseal stem in hips with risk factors (a component size of <48 mm and a femoral head defect of >1 cm).
The main limitation of this study is the absence of a control group of obese patients who received a conventional stemmed total hip prosthesis. Therefore, the scope of our conclusions is restricted and the relative merits of resurfacing compared with those of total hip arthroplasty in obese patients cannot be determined with the current data. Also, most of the patients in the study group were male, and these results cannot be inferred to a population of obese postmenopausal women; there were only eleven such patients in our series, although none of them sustained a femoral neck fracture.
The present study demonstrated slightly inferior results in the study group, when compared with the control group, in terms of function, activity, and the physical component of the SF-12. However, these differences were also present preoperatively, and we assume that a patient's body mass index is unlikely to decrease after total joint arthroplasty34. Thus, these differences should probably be attributed to the effect of obesity on functional scores rather than on the procedure itself.
The overall complication rate in the study group was low (4.2%) and comparable with that in the nonobese patient group (6.1%), and there was no significant difference between the groups with regard to any category of complications.
Technically, the procedure is more challenging in patients with increased girth due to either fat or muscle, or both, because a more extensive exposure is needed, not only to deliver the head for preparation, but also to translate the head out of the way to completely visualize and access the acetabulum during reaming and implantation of the socket. We strongly recommend debulking the head (reducing the size of the head with a cylindrical reamer that is larger than the templated size19), especially when the femoral head is large (>48 mm), to facilitate access to the acetabulum. Also, an ample incision length and full gluteal release from the linea aspera are very helpful in creating enough exposure to optimize the orientation of the socket.
Bearing in mind the increasing prevalence of obesity in the general American population35 and its effect on the development of osteoarthritis1-4, hip resurfacing appears to be a promising prosthetic solution. At this time, we believe that a body mass index of =30 is not a relative contraindication to resurfacing. However, weight reduction should still be encouraged to facilitate the procedure and enhance the clinical results.