To evaluate the effect of obesity on the incidence of adverse events (surgical site infection, dislocation, re-revision, or ≥1 adverse event), functional outcome, residual pain, and patient satisfaction after revision total hip arthroplasty (THA).
We conducted a university hospital-based prospective cohort study including 52 obese and 152 nonobese patients with revision THA performed between 1996 and 2006. We used incidence rates, rate ratios, and hazard ratios (HRs) to compare the incidence of events in obese and nonobese patients and in 4 body mass index (BMI) categories (<25, 25–29.9, 30–34.9, ≥35). Functional outcome and pain were measured 5 years postoperative using the Harris Hip Score.
The incidence rate for ≥1 complication increased with rising BMI (1.8, 3.4, 10.3, and 17.9 cases/100 person-years). The increase was small between normal and overweight patients (adjusted HR 1.5, 95% confidence interval [95% CI] 0.5, 4.7), significantly greater with BMI 30–34.9 (adjusted HR 4.5, 95% CI 1.4, 14.0), and most evident with BMI ≥35 (adjusted HR 10.9, 95% CI 2.9, 41.1). The adjusted HR for surgical site infection (obese versus nonobese) was 4.1 (95% CI 1.1, 15.0) and for dislocation 3.5 (95% CI 1.3, 9.3). Eighty patients had a followup visit at 5 years. Obese patients had moderately lower functional results and higher levels of residual pain, but patient satisfaction was almost similar.
Revision THA is technically challenging, particularly in obese patients, probably due to more difficult anatomic conditions. We found an increased risk of adverse events, notably surgical site infection and dislocation in these patients.
Obesity has been associated with a higher prevalence of symptomatic osteoarthritis of the hip and a subsequent increase in total hip arthroplasty (THA) procedures (1–6). Interestingly, a recent large population-based study (6) did not find an increase in revision procedures among obese patients post primary THA. Prosthetic joint infections, dislocations, and revisions are rare but serious complications after primary THA. There are conflicting results in the current literature concerning the influence of obesity on the occurrence of these complications. A number of studies found an increase in perioperative morbidity (7–9) and complications such as infections or dislocations associated with obesity (8, 10–13), whereas others have reported no differences (14–18). The literature concerning the influence of obesity on revision for aseptic loosening is also inconclusive, which has partially been related to differences in activity level (15, 16, 18–23). However, functional outcome and patient satisfaction after THA have been considered comparable or only slightly lower in obese compared with nonobese patients (11, 13, 17, 24).
In comparison with primary hip arthroplasty, revision hip arthroplasty is a prolonged intervention resulting in more extensive tissue damage and is associated with more short- and long-term complications, as well as a higher mortality rate (25–29). The intervention is considered to be technically challenging, particularly in obese patients probably because of more difficult anatomic conditions. Only a few studies, notably with short-term followup, have reported on the influence of obesity on outcomes after revision THA (12, 30, 31). There is even less literature concerning the effect of obesity on functional outcomes and satisfaction after revision surgery (32, 33).
The aim of this study was first to evaluate the association between obesity and the incidence of main complications (in particular, surgical site infection, dislocation, and re-revision) after revision THA up to 5 years postoperatively. Also, we aimed to determine whether functional outcome, pain, and patient satisfaction at 5 years differed between obese and nonobese patients who underwent revision THA.
PATIENTS AND METHODS
Study design and patient population.
We undertook a prospective etiologic cohort study including all consecutive patients who underwent a revision THA at the university orthopedic department of the only public hospital for the urban and surrounding rural population between March 26, 1996 and July 31, 2006. Overall, 300–350 primary and 25–30 revision THAs were done annually. Revisions were performed by senior orthopedic surgeons. We excluded re-revisions from this study.
Study followup extended through October 31, 2006. For those patients who underwent bilateral revisions during the study period (n = 9), only the first revision THA was included in order to allow for an analysis on the patient level. A total of 205 patients were eligible. One patient was excluded due to missing weight and height information, leaving 204 patients (Table 1). All patients received antibiotic and thrombosis prophylaxis, and the procedures were performed in ultraclean air laminar flow operating rooms using hooded gowns. Gentamicin-loaded bone cement was used in cases with cemented implants.
THA = total hip arthroplasty; BMI = body mass index.
Total number of revision THAs performed
Patients excluded because of bilateral revision
Number of eligible revision patients
Patients with missing BMI
Total number of patients included
Patients due for 5-year visit
Patients who had died
Lost to followup
Unable to attend 5-year visit
Patients who completed 5-year followup
Exposure and outcome variables.
The exposure (etiologic factor) of interest was obesity defined as a BMI ≥30 kg/m2, in keeping with previous studies (12, 13, 30, 32, 34). The association between BMI and the primary outcome was also examined using 4 BMI categories (<25, 25–29.9, 30–34.9, and ≥35).
The primary outcome was the incidence of adverse events including the occurrence of ≥1 of the following during the followup period: surgical site infection (prosthetic joint and/or superficial infection), dislocation (first event), or re-revision for any cause. Prosthetic joint infection was diagnosed according to the criteria by Zimmerli et al (35), and internationally recognized criteria (12) were used for the inclusion of superficial infections.
Secondary outcomes were functional status, pain, and patient satisfaction (all continuous variables) 5 years after revision surgery. The Harris Hip Score (HHS; a physician-assessed, hip-specific clinical score evaluating the domains of pain, function, deformity, and motion) (36) was used to measure functional status. Scores range from 0 (worst) to 100 (best). The pain item of the HHS, consisting of a 6-grade response scale (none, slight/occasional, mild, moderate, marked, serious limitation/totally disabling pain) and rated numerically from 0 to 44 (no pain = 44), was used to define the pain outcome. The HHS is the most widely used physician-assessed hip score in orthopedic surgery (37, 38). The total score and the pain subscore have shown high reliability and validity (38–40). Patient satisfaction was measured using a 0–10 visual analog scale, where 10 = best.
In the estimation of the association between obesity and the primary (and secondary) outcomes, the following potential confounders were assessed preoperatively: 1) age; 2) sex; 3) indication for revision defined as a 7-category variable (aseptic loosening of cup, stem, or both; septic loosening; recurrent dislocation; periprosthetic fracture; technical error) as well as dichotomized variable (aseptic loosening versus all other indications for revision); and 4) surgery of the index hip prior to primary THA.
The following clinical scores were also preoperatively assessed: the Merle d'Aubigné score, the Charnley classification grade, and the American Society of Anesthesiologists (ASA) score. The Merle d'Aubigné score (41) is a hip-specific, physician-assessed clinical score evaluating pain, function, and motion on a scale from 0 to 6 (best, total score 18). The score has been used by orthopedic surgeons to evaluate hip arthroplasties for more than 50 years, and it highly correlates with the HHS (Spearman's correlation coefficient 0.81–0.82) (42, 43). The Charnley classification (44) comprises 3 grades: 1 hip affected and the other normal (grade A), both hips affected (grade B), and multiple-joint disease or other disabilities leading to difficulties in ambulation (grade C). The ASA score is obtained from the anesthesia report and evaluated as a binary variable (1–2 versus 3–4). It is a physical status classification system aimed at grading the patient in relation to his physical preoperative status. The ASA 6-point scale ranges from a healthy patient to a patient with an extreme systemic disorder that is an imminent threat to life.
The Merle d'Aubigné score was used preoperatively from March 1996 to June 2003. Complete score information was available for 30 (76.9%) of 39 obese patients and for 88 (76.5%) of 115 nonobese patients operated on during that time period. Four patients had a missing ASA score.
The following information related to the revision surgery was assessed: 1) surgical approach without or with (yes/no) use of trochanteric/proximal-femoral osteotomy indicated in cases where greater femoral and/or acetabular exposure was needed; 2) use of an acetabular ring in case of insufficient acetabular bone (yes/no); 3) femoral head size in millimeters; 4) cementing of stem and/or cup (yes/no); and 5) operation time in minutes.
Patient weight and height were obtained at the preoperative entrance examination, just before surgery. Information about the potential confounding variables and the surgical intervention was systematically documented by the operating surgeon on specifically designed data collection forms. The data were checked by a trained medical secretary and 1 of the investigators (AL). We retrieved information regarding primary outcome events from each patient. We contacted all participants who were 5 years postoperative for a followup visit that included a clinical and radiologic examination. Information about the occurrence of the primary outcome events since the intervention, which were not treated at our institution, was obtained either during the followup visit or, for all those who were unable to join the followup visit, by phone. Followup examinations were completed by 2 trained physicians who had not performed the operations.
The distribution of preoperative baseline characteristics (potential confounders) and technique- and implant-related variables was examined for obese and nonobese patients. To assess the association between obesity and the primary outcome, we calculated person-times from the date of operation until the first occurrence of any of the above defined events, death, end of the study (October 31, 2006), or loss to followup. Then we estimated the crude incidence rate and crude incidence rate ratio (IRR) with their 95% confidence intervals (95% CIs). We repeated the analyses after stratification for age, sex, ASA score, Charnley classification grade, and indication for revision (dichotomized) to identify possible confounding or effect modification using 2 BMI groups, with the nonobese group as the reference group. Moreover, we employed Cox proportional hazards analysis and presented adjusted hazard ratios (HRs). The proportional hazards assumption was checked by examining graphs of the log minus log survival functions. The analyses above were performed for any first primary outcome event, and then repeated for the separate events if data allowed for it. Due to the small numbers of primary outcome events, we adjusted only for the most important confounders, which were ASA score for surgical site infection, and age and ASA score for dislocation. No adjustment was made for the outcome re-revision as only 8 patients had this event. Additionally, incidence rates and IRRs for the first primary outcome event were assessed across the 4 BMI categories (<25, 25–29.9, 30–34.9, and ≥35) with the normal weight category (<25) as the reference group. We used Cox proportional hazards analysis to adjust for age, sex, and ASA score. The Kaplan-Meier approach was employed to estimate the one minus survival function (cumulative incidence function).
To assess the association between obesity and the secondary outcomes, we calculated mean scores (with SDs) and the (crude) mean differences with 95% CIs for obese and nonobese patients using the unpaired Student's t-test. We repeated the analyses after stratification for age, sex, Charnley classification grade, ASA score, and indication for revision. Multivariable linear modeling was used to adjust the effect of BMI on all 3 secondary outcomes for age, sex, preoperative function and pain, ASA score, and Charnley classification grade.
Among the 204 patients, 114 were women and 90 were men, with a mean age of 71.6 years (range 32–94 years) and a mean BMI of 26.7 (range 15–44). Twenty-five percent (n = 52) of the revisions were performed in patients with a BMI ≥30. Table 2 presents the distribution of the baseline characteristics (potential confounders) across obese and nonobese patients. The 2 groups mainly differed with respect to age, ASA score, Charnley classification grade, and indication for revision. Obese participants were younger, more often classified as Charnley grade C, and had higher ASA scores. Moreover, they were revised for aseptic loosening less often overall (less loosening of the cup or of both components, more stem loosening), but more often for recurrent dislocation or technical error.
Table 2. Distribution of baseline characteristics among obese and nonobese patients*
Values are the number (percentage) unless indicated otherwise. BMI = body mass index; RR = relative risk; 95% CI = 95% confidence interval; ASA = American Society of Anesthesiologists; THA = total hip arthroplasty.
F or continuous variables.
Used preoperatively from March 1996 to June 2003. Complete score information was available for 30 (77%) of 39 obese patients and 88 (77%) of 115 nonobese patients operated on during that time period.
Charnley classification grade C compared with grade A and B combined.
In obese patients the revision was performed, on average, 92 months after the primary hip THA, compared with 125 months in nonobese patients (mean difference 33 months, 95% CI 9, 58). The 2 groups did not significantly differ with respect to the use of osteotomy. A trochanteric or proximal femoral osteotomy was performed in 31 (60%) obese patients and 79 (52%) nonobese patients. In the obese group, an acetabular ring was used in 17 (43%) of the 40 revisions involving a cup replacement as compared with 89 (66%) of 134 cup replacements in the nonobese group. In the large majority of patients, a 28-mm head was chosen, except for 3 cases from each group with a 22-mm or a 32-mm head. A cemented cup was inserted in 67% of obese patients versus 80% of nonobese patients, and a cemented stem in 83% versus 74%. The groups did not differ with respect to operation time (209 versus 210 minutes).
The followup period for the primary outcome ranged from 3 to 74 months. During followup, 3 (5.8%) of 52 obese patients and 22 (14.5%) of 152 nonobese patients had died, and 4 (7.7%) obese and 6 (3.9%) nonobese patients were lost to followup.
Obese patients contributed a mean ± SD of 33 ± 25 person-months, and nonobese patients contributed a mean ± SD of 41 ± 21 person-months. Overall, 20 complications occurred in 17 (33%) of 52 obese patients compared with 18 events in 13 (9%) of 152 nonobese patients. Surgical site infections were reported in 6 (11.5%) obese patients (3 superficial, 3 prosthetic joint) and 4 (2.6%) nonobese patients (1 superficial, 3 prosthetic joint). The crude incidence rate was 4.6 times higher in obese patients (Table 3). After adjustment for ASA score, the adjusted HR was 4.1 (95% CI 1.1, 15.0). After stratification for sex, we observed a higher incidence rate of infection in obese versus nonobese women (IRR 15.5, 95% CI 1.7, 138.8) but no significant increase in obese versus nonobese men (IRR 2.0, 95% CI 0.3, 12.2). Dislocation was observed in 10 (19.2%) obese patients and 10 (6.6%) nonobese patients. The crude incidence rate was 3.1 times higher in obese individuals. After adjustment for age and ASA score, the adjusted HR was 3.5 (95% CI 1.3, 9.3). Eight re-revisions were undertaken, with 4 re-revisions in each group (7.7% versus 2.6%). In each group, 2 re-revisions were performed for septic loosening; 1 for recurrent dislocation and 1 for early aseptic loosening of the cup. The crude incidence rate of re-revision for any cause was 2.8 times higher in obese patients (95% CI 0.7, 11.1). Loss of fixation of the trochanteric osteotomy was observed in 5 (17.2%) obese patients and 5 (9.3%) nonobese patients.
Table 3. Incidence rates for each of the adverse events across obese and nonobese patients*
BMI <30 (n = 152)
BMI ≥30 (n = 52)
Unadjusted IRR (95% CI)
Adjusted HR (95% CI)
BMI = body mass index; IRR = incidence rate ratio; 95% CI = 95% confidence interval; HR = hazard ratio.
Complication (1 event of any primary outcome) adjusted for age, sex, and American Society of Anesthesiologists (ASA) score.
Surgical site infection (including superficial and deep infection) adjusted for ASA score.
Dislocation adjusted for age and ASA score using Cox proportional hazards models.
The incidence rate for the occurrence of ≥1 adverse event increased with rising BMI (Table 4 and Figure 1). This increase was small between normal and overweight patients (adjusted HR 1.5), but it became significantly greater in the group with a BMI 30–34.9 (adjusted HR 4.5) and was most evident in the group with a BMI ≥35 (adjusted HR 10.9), although the width of the 95% CI increased due to the low patient number in this latter group. Adjustment was performed for age, sex, and ASA score. Further adjustment for indication for revision and Charnley classification grade did not substantially change the results.
Table 4. Occurrence of first primary outcome event (surgical site infection, dislocation, or re-revision) according to 4 BMI categories*
<25 (n = 83)
25–29.9 (n = 69)
30–34.9 (n = 41)
≥35 (n = 11)
BMI = body mass index; IRR = incidence rate ratio; HR = hazard ratio; 95% CI = 95% confidence interval.
Adjusted for age, sex, and American Society of Anesthesiologists score using Cox proportional hazards models.
Five years after revision, 31 obese and 79 nonobese patients were due for followup. Among the obese group, 2 (6.5%) patients had died, 3 (9.7%) were lost to followup, and 2 (6.5%) were unable to attend. Of the nonobese patients, 13 (16.5%) had died, 5 (6.3%) were lost to followup, and 5 (6.3%) were unable to attend. Of those still alive, 24 (82.8%) obese patients and 56 (84.8%) nonobese patients were seen at the 5-year visit. The mean ± SD time to followup was 57 ± 8 months.
Five years postoperatively, obesity was associated with moderately lower functional status (HHS) and a lower pain subscore, which remained after adjustment for age, sex, preoperative function and pain (Merle d'Aubigné score), ASA score, and Charnley classification grade (Table 5). The association with patient satisfaction was less evident, after adjustment the 95% CI included 0.
Table 5. Harris hip score (HHS), pain, and patient satisfaction at 5 years for obese and nonobese patients*
Values are the mean ± SD unless otherwise indicated. BMI = body mass index; 95% CI = 95% confidence interval.
Adjusted for age, sex, and Charnley classification grade (all 80 patients included in analysis).
Adjusted for age at operation, sex, preoperative Merle d'Aubigné score, American Society of Anesthesiologists (ASA) score, and Charnley classification grade. Due to missing values on preoperative Merle d'Aubigné score and ASA score, only 66 patients could be included for adjustment.
82.8 ± 14.7
71.4 ± 17.0
11.4 (3.9, 18.9)
9.2 (2.0, 16.7)
8.9 (1.9, 15.9)
39.2 ± 7.2
33.9 ± 9.6
5.3 (0.8, 9.7)
5.0 (0.9, 9.1)
5.4 (1.1, 9.8)
8.2 ± 1.8
7.2 ± 2.7
1.0 (−0.2, 2.3)
0.8 (−0.3, 1.9)
1.1 (−0.1, 2.3)
The primary aim of this study was to determine whether obesity was associated with a higher incidence of major events in patients undergoing revision THA. We found significantly more events, notably surgical site infections and dislocations, in obese patients. Secondarily, we found moderately lower results on the HHS but almost similar satisfaction in obese patients 5 years postoperatively.
In comparison with primary THA, revision surgery is associated with greater soft tissue damage and prolonged operation time, which may explain the increased incidence of superficial and deep infections after revision. Larger soft tissue dissection and subsequent increased muscle weakness in revision THA might be one of the responsible factors for higher dislocation rates (45). In turn, obesity has been associated with higher rates of wound healing complications (46) and higher morbidity. Furthermore, in obese patients the intervention can be more technically challenging with respect to exposure, implant positioning, and soft tissue closure (47).
Literature about the influence of obesity on the occurrence of adverse events after revision THA is sparse. Most studies have reported on small patient groups and short followup periods. Perka et al (31), in a retrospective study including 229 patients of whom 31 had a BMI ≥30, found no increase in perioperative morbidity and mortality within 90 days. Another study (12) evaluated the risk of surgical site infection after revision according to 3 BMI categories. The authors did not observe an increase in infection in obese patients during the study followup, which was limited to the in-hospital postoperative period.
In contrast, a recent matched cohort study (30) reported a much higher risk of dislocation (relative risk 6.3) in morbidly obese patients as compared with a group of normal-weight and overweight patients 12 to 28 months after revision THA. The mean age of their patients was 57 years, compared with 72 years in our study. All patients were operated upon by 1 experienced surgeon. The even higher relative risk for dislocation in obese patients in their study might be due to the low dislocation rate (3%) in their young, nonobese patients.
In a previous study evaluating the effect of obesity on complications after primary THA, we observed a 4-times higher rate of infection similar to our current findings. The only difference is that in the present study, we included both superficial and prosthetic infections (11). However, the increase in dislocation due to obesity was less important after primary THA in that study than it was in this study (2 times versus 3 times higher).
Our findings are also in accordance with several other studies reporting on increased postoperative complications in obese patients after primary THA (7, 8, 12, 13). Two of these studies found higher rates of postoperative infections (8, 12). Stickles et al (13) evaluated 1-year orthopedic complication rates according to 5 BMI categories (<25, 25–29, 30–34, 35–39, and ≥40) using univariate analyses, and they found a significant increase with rising BMI, similar to what we observed in our data. A few studies reported similar short-term (orthopedic) complication rates (14–18), but their numbers of patients, and as a consequence their numbers of adverse events, were small.
Literature about pain and function after revision THA is lacking. Lower functional results (on the HHS score) in obese patients undergoing revision THA have been reported in 2 studies at 3 and 5 years postoperative (32, 33). In contrast to the present study, those analyses were unadjusted. Davis et al (48) analyzed predictors of pain and functional outcomes 2 years after revision THA using the Western Ontario and McMaster Universities Osteoarthritis Index, and they emphasized the importance of preoperative pain and function for the 2-year results. However, they did not include BMI among their predictors.
Previous studies have reported that obesity was associated with greater pain and disability in patients with hip or knee osteoarthritis (49, 50). However, we found similar preoperative Merle d'Aubigné scores (10.5 in nonobese versus 10.1 in obese patients) together with significant differences in Charnley classification grades between the 2 groups. Part of the explanation for this could be that the function domain, which we expect to mainly be related to disability, counts only for one-third of the total Merle d'Aubigné score. An additional subscore analysis of our data revealed that the difference of 0.4 points on the preoperative Merle d'Aubigné score between obese and nonobese patients was mostly due to lower function subscores in the obese group. With respect to the pain issue, we are unable to provide a clear reason why obese and nonobese patients did not differ at baseline in our study.
This prospective, hospital-based study compared the occurrence of several main adverse events in addition to pain, functional outcome, and patient satisfaction 5 years postoperatively in obese and nonobese patients undergoing first revision THA. To our knowledge, the followup period was longer than in any previous study. Analyses were performed for 2 and 4 BMI categories. We used incidence rates and Cox proportional hazards analysis to account for patient differences in length of followup and censoring, and to adjust for baseline differences (confounding) across BMI categories. Data collection and clinical followup were standardized, and the clinical assessment was performed by 2 independent surgeons in order to avoid observer bias.
The study has several weaknesses. First, it was limited by the relatively small number of adverse events resulting in large confidence intervals. For the same reason, we were unable to adjust for all potentially confounding factors. However, adjustment for several confounders was made for the combined complications, and for the complications infection and dislocation alone we adjusted for the most important confounders. Second, obese and nonobese patients differed by several baseline characteristics, which could be due to differences in the adverse events following primary THA and/or differential selection at the time of indication for primary or revision arthroplasty. Because 2 different clinical scores evaluating pain, function, and mobility were used at baseline and at followup, we were unable to analyze score differences. No information on intraoperative bone quality was available, but reports have shown that obese subjects tend to have greater bone mineral density and a lower risk of hip fractures (51–53). In addition, the need for an acetabular ring indicating insufficient acetabular bone stock was higher in the nonobese group, and severity of revision based on bone stock loss was not found to be significantly related to pain and function in the study by Davis et al. Finally, our study was conducted at 1 large academic center; however, baseline characteristics and indications for revisions did not substantially differ from descriptions of other hospital- or community-based revision cohorts in the literature (30, 54–56). Further studies including a larger number of patients are needed to assess external validity and to evaluate the extent to which different techniques and implants can influence the results.
Revision THA is a technically-challenging intervention, particularly in obese patients, probably because of more difficult anatomical conditions. Our results revealed that obesity was associated with higher event rates; notably, surgical site infection and dislocation. Furthermore, we found moderately lower HHS with higher levels of pain 5 years postoperative. Surgeons, patients, and referring physicians should be aware of an increased risk of adverse events in this patient group. Further studies are necessary to evaluate whether changes in medical preparation, surgical technique, and implant choice can help reduce the adverse event rate in obese patients undergoing revision THA.
Dr. Lübbeke had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Lübbeke, Moons, Hoffmeyer.
Acquisition of data. Lübbeke, Garavaglia.
Analysis and interpretation of data. Lübbeke, Moons.
Manuscript preparation. Lübbeke, Moons.
Statistical analysis. Lübbeke, Moons.
The authors would like to thank Richard Stern, MD for his helpful comments.