Institutional review board approval was obtained prior to initiation of the study. The electronic database at our institution was searched to identify all patients who had undergone revision total hip arthroplasty between 1992 and 1999. One hundred and seventy total hip revisions performed in 159 patients with a mean age of 83.8 years (range, 80.0 to 93.8 years) were identified and included in the study. With use of the same database, a control group of patients who were seventy years of age or younger when they underwent revision total hip arthroplasty during the same period was also identified. The control cohort consisted of 162 patients (170 hips) with a mean age of 57.7 years (range, thirty-one to 70.4 years) (Table I). Although strict matching was not performed, the control group was comparable with the study group in terms of many characteristics, including gender, type of revision surgery (revision of the femoral component, the acetabular component, or both components), surgical approach, type of components, method of fixation (all uncemented), and surgeon (Table I).
The medical records and the radiographs of the patients in both groups were reviewed to extract relevant information regarding the indication for the revision, details of the surgery, postoperative hospital course, functional outcome, complications, and mortality.
Coexistent medical morbidities were assessed preoperatively with use of the physical status classification of the American Society of Anesthesiologists (ASA)7.
At our institution, the total joint registry prospectively collects clinical and radiographic data on all patients treated with arthroplasty. Patients are contacted at six weeks, one year, two years, and five years postoperatively and then every five years thereafter. Some patients are evaluated more frequently, as deemed necessary by the treating surgeon or patient.
The functional outcome was measured with use of the Harris hip score, and the general quality of life was assessed with the Short Form-36 (SF-36)8. These data were collected preoperatively and at the time of the last follow-up. The outcome of the arthroplasty and the prevalence of complications within six months after the surgery were determined by a detailed review of the patient's medical and radiographic records.
After the surgery, the patients were either evaluated clinically or interviewed by telephone. When a patient had died, surviving family members were interviewed.
Statistical Analysis
Descriptive statistics and the Fisher exact test or the Student t test were used for analysis of categorical data. Continuous data were analyzed with use of descriptive statistics and the nonpaired t test. Survival analysis was performed with the Kaplan-Meier method with the need for repeat revision surgery or another reoperation as the end point and with perioperative deaths for any reason as censored data. The log-rank test was used to identify significant differences between the survival curves of the study cohorts. A univariate Cox proportional-hazards regression model was used to calculate a hazard ratio estimating relative risk. For the univariate analysis, p < 0.1 was considered significant. For all other tests, p < 0.05 was considered significant.
Medical Optimization
Preoperatively, all patients in both groups underwent a series of medical optimization tests including a physical examination by an internist or cardiologist, a chest radiograph, an electrocardiogram, a complete blood-cell count, and a basic metabolic blood screening panel. When indicated, an echocardiogram or a stress test was done to assess cardiac health.
Pulmonary function was evaluated preoperatively in all patients in the octogenarian group and in ninety-two patients (56.8%) in the control group (p < 0.01). All of the patients in the octogenarian group and eighty-six patients (53.1%) in the control group underwent an echocardiogram preoperatively (p < 0.01). A pharmacologic or physical cardiac stress test was performed on 145 octogenarian patients (91.2%) and forty-one patients (25.3%) in the control group (p < 0.01).
Postoperatively, 32% of the octogenarian patients and 5% of the control group were transferred to a telemetry unit for close postoperative monitoring (p < 0.01).
Surgical Data
The average time from the index procedure to the revision arthroplasty was 8.8 years (range, four months to twenty-three years) in the octogenarian cohort and four years (range, 0.2 to 14.1 years) in the control group (p < 0.05). The mean ASA score was 2.7 (2, 3, or 4) for the octogenarian population and 2.4 (range, 1 to 4) for the control group (p = 0.001) (Table I). Spinal anesthesia was used for the operations on 162 hips (95.3%) in the octogenarian group and 164 hips (96.5%) in the control group. The remainder of the operations were done with general anesthesia. There was no significant difference in the mean operative time, surgical approach, estimated blood loss, components that were revised, or transfusion needs between the two groups (Table I). Uncemented components were used in all hips. The mean length of the hospital stay was 6.9 days (range, three to forty-eight days) in the octogenarian group compared with 4.5 days (range, three to twelve days) in the control group (p = 0.0001).
Follow-up
All patients were followed until failure of the prosthesis requiring repeat revision or death. The average duration of follow-up was 6.8 years (range, three days to 14.7 years) for the octogenarians and six years (range, two days to 11.5 years) for the control group. The Kaplan-Meier survivorship with the need for repeat revision surgery or another reoperation as the end point and with perioperative deaths for any reason as censored data was 95% for the octogenarians and 94% in the control group at one year and 92% for the octogenarians and 90% in the control group at five years (Fig. 1). The log-rank test showed no significant difference between the two groups regarding survivorship (p = 0.3).
Functional Outcome
In the study group, there was a significant improvement in the Harris hip score from a preoperative mean of 47 points (range, 20 to 55 points) to a mean of 85 points (range, 60 to 100 points) at the time of the latest follow-up (p < 0.05). The Harris hip score improved from a preoperative mean of 44.3 points (range, 10 to 60 points) to a mean of 87.9 points (range, 65 to 100 points) at the time of the latest follow-up in the control group (p < 0.05). The difference in the improvement in the Harris hip score between the two groups was not significant (p > 0.05). The mean postoperative Harris hip score for the patients who died before the time of final follow-up was 87.1 points (range, 65 to 100 points).
In the octogenarian group, the scores for the physical and mental health domains of the SF-36 improved, respectively, from a mean of 44.6 points (range, 12 to 67 points) and 62.5 points (range, 18 to 100 points) preoperatively to a mean of 78.1 points (range, 25 to 100 points) and 67.4 points (range, 45.4 to 89.2) points postoperatively (p < 0.05). In the control group, the SF-36 physical and mental health scores improved from a mean of 43.6 points (range, 9 to 67 points) and 57 points (range, 34 to 71 points) preoperatively to a mean of 74.2 points (range, 25 to 100 points) and 68.9 points (range, 33 to 99 points) postoperatively (p < 0.05). The difference in the improvements in the SF-36 physical and mental health scores between the two groups was not significant (p > 0.05). These values compare closely with the SF-36 physical and mental function scores for an age-matched normal population (74.4 and 74.2 points, respectively)9. The mean postoperative scores for the physical and mental health domains of the SF-36 for the patients who died before the final follow-up were 79.3 points (range, 33 to 100 points) and 66 points (range, 45.4 to 84 points), respectively.
Medical Complications
During the six months after the surgery, medical complications developed in thirty octogenarian patients (thirty-eight hips [22.4%]) compared with twenty-five controls (twenty-six hips [15.3%]) (see Appendix) (p > 0.05). Ninety-six percent of the medical complications in the study group and 93% of those in the control group developed during the hospital stay. Medical complications were more likely to develop in patients with preexisting comorbidities. In the octogenarian group, the overall prevalence of medical complications correlated directly with preexisting medical comorbidities as measured with the ASA score (odds ratio = 1.92, 95% confidence interval = 0.025 to 2.86, p = 0.05) (see Appendix). Complications developed after eight (12.9%) of the revisions in octogenarians who were classified as having an ASA score of =2 before the surgery compared with twenty-one revisions (21.6%) in octogenarians with an ASA score of 3 and nine revisions (81.8%) in octogenarians with an ASA score of 4. The age at surgery did not correlate with the prevalence of overall medical complications (odds ratio = 1.026, p = 0.66).
Orthopaedic Complications
Orthopaedic complications occurred in thirty-eight hips (22.4%) in the octogenarian group compared with thirty-seven (21.8%) in the control group (Table II) (p > 0.05). The rate of intraoperative fracture was significantly higher in the octogenarian group (eight hips) than in the control group (no hips) (p = 0.007). Sixteen hips in the control group and four in the octogenarian group dislocated postoperatively (p = 0.01). All four dislocations in the octogenarian group were treated successfully with closed reduction. There was no significant difference between the two groups regarding other orthopaedic complications, including wound-related problems, deep infection, and postoperative periprosthetic fracture (Table II).
Repeat Revisions and Other Reoperations
There were thirteen repeat revisions and four other reoperations in the octogenarian group compared with twenty-three repeat revisions and six other reoperations in the control population (Table II) (p = 0.08). In the octogenarian group, repeat revision arthroplasty was performed for treatment of a postoperative periprosthetic fracture in four hips, a periprosthetic infection in two hips, and aseptic loosening of one or both components in seven hips. Both the femoral and the acetabular component were loose in four hips, only the femoral component was loose in one hip, and only the acetabular component was loose in two hips. Four irrigation and débridement procedures were performed to treat prolonged wound drainage, which was due to a superficial surgical wound infection in three hips and to a deep wound infection in one hip. Three hips with a deep infection required a reoperation, which included débridement and retention of the components in one hip and two-stage reimplantation in two hips.
In contrast, there were twenty-three repeat revisions in the control group. Eight were done because of aseptic loosening of one or both components; three, because of a periprosthetic infection; two (of the femoral implant), because of periprosthetic fracture; and ten, because of recurrent dislocation. There were six other reoperations: five were done for treatment of prolonged wound drainage and one, for open reduction and internal fixation of a postoperative fracture. Sixteen hips (9.4%) dislocated (p = 0.01). Six hips were treated successfully by closed reduction alone, whereas ten required a subsequent repeat revision.
Mortality Data
There were no intraoperative deaths in either group. One patient in the octogenarian group died of a myocardial infarction in the hospital three days after the surgery, and another patient in that group died of staphylococcal pneumonia twenty-eight days after the surgery. In the control group, one patient died of a myocardial infarction in the hospital two days after the surgery. There was no significant difference between the two groups regarding the in-hospital mortality (p > 0.05) or mortality within thirty days (p > 0.05).
The prevalence of mortality at the time of final follow-up was significantly higher in the octogenarian group than in the control group (p < 0.0001). At the time of final follow-up, 100 octogenarian patients (58.8% of the hips) had died, at a mean of 5.3 years (range, three days to 11.3 years) following the revision total hip arthroplasty. Ninety-four percent of the 100 patients had a well-functioning hip at the time of death. The mean age at the time of death was 89.6 years (range, 81.4 to 101.8 years), whereas the mean age of the surviving patients was 92.2 years (range, 86.4 to 106.5 years) at the time of final follow-up. In comparison, there were only twelve deaths in the control group; they occurred at a mean of 3.1 years (range, two days to 6.2 years) following the index surgery.
The prevalence of mortality was higher for the patients with preexisting comorbidities. The mortality rate for the patients with an ASA score of 3 or 4 was higher than that for patients with lower scores (p > 0.05) (see Appendix). The mortality rate was not significantly different between men and women (p < 0.627). The latter finding was still true after adjustment for age and ASA scores.
Analysis of Risk Factors
Univariate analysis provided preliminary data suggesting that female gender, preexisting comorbidities (as determined on the basis of the ASA score), higher estimated blood loss, increased operative time, and a lower preoperative score for the mental health measure of the SF-36 may be important factors predicting the development of complications and/or mortality. When these factors were subjected to multiple logistic regression analysis, only preexisting comorbidities (odds ratio = 2.3, 95% confidence interval = 1.9 to 3.5), female gender (odds ratio = 1.12, 95% confidence interval = 0.8 to 2.2), and increased operative time (odds ratio = 0.69, 95% confidence interval = 0.45 to 1.6) were found to be significant predictors of major complications.
Primary total hip arthroplasty is generally successful in octogenarians, but these patients have been reported to be at increased risk for the development of postoperative complications, with prevalences ranging from 48% to 93%10-15. Revision arthroplasty can also provide excellent pain relief and improvement in function for octogenarians3-6, but, because of its inherent surgical complexities and longer operative times, revision arthroplasty would be expected to be associated with an even higher complication rate in these patients16,17. The prevalence of orthopaedic complications following revision arthroplasty in the very elderly has been reported to range from 30% to 61%3-6,16,17. In our study, the prevalence of orthopaedic complications—except for intraoperative periprosthetic fractures, which were more common in the octogenarian group, and dislocation, which was more frequent in the patients who were seventy years old or younger—was not significantly different between the octogenarian and control groups.
We speculate that octogenarian patients had a significantly higher prevalence of intraoperative periprosthetic fracture because of lower bone mass, which can predispose them to fractures11,15. The most commonly reported orthopaedic complication in octogenarians after total hip replacement is dislocation, with a prevalence ranging from 3.7% to 20%3-6. Yet, the prevalence of dislocation in our study was higher in the younger group. We believe that the main reason for this finding is that more constrained liners were utilized in the octogenarians than in the younger patients, perhaps because the surgeons thought it necessary to use a more constrained device in the very elderly.
We suggest that the low prevalence of medical complications following revision arthroplasty in the octogenarians may be explained by a number of factors. First, because of the anticipated complexity of revision surgery in the elderly, the medical optimization for the octogenarian patients in this cohort was more stringent and extensive than it was for their younger counterparts. Compared with the younger patients, the octogenarian patients had more medical tests such as stress echocardiography, pulmonary function tests, and other extensive investigations. Second, the medical surveillance of the octogenarian patients during the postoperative period was closer and more stringent than that for the younger control group, as evidenced by the sixfold greater rate of postoperative monitoring in the telemetry unit for the octogenarian group. Hence, many medical complications may have been prevented or detected early and treated appropriately.
Despite the presence of preexisting comorbidities in a large number of octogenarians and the complexity of the procedure in some cases, there were no intraoperative deaths and only two in-hospital deaths in that group. The latter finding attests to the improvements in medical and orthopaedic care that have occurred over the last two decades7. For example, the majority of the patients in this cohort had the surgery under regional hypotensive anesthesia, which minimized intraoperative blood loss and the possibility of subsequent complications7. Although the early mortality rate was low in the octogenarian group, a large number of these patients died soon after the revision surgery. Thus, revision arthroplasty provided excellent benefit to these patients but for only a limited period of time. Nonetheless, patients who were still alive at the time of final follow-up continued to enjoy the functional improvement that was provided by the revision arthroplasty.
This study had several strengths. First, it consisted of a large cohort of patients who had undergone cementless revision total hip arthroplasty in a single institution. Second, it was a case-control study intending to remove the influence of important confounding variables and place the findings in a proper perspective. Third, we succeeded in identifying distinct risk factors that may result in complications and/or mortality following revision arthroplasty in the very elderly. Preexisting comorbidities as determined by the ASA score are an important factor predictive of outcome. This finding is in agreement with those in previous studies that have demonstrated that healthier patients (those with an ASA score of 1 or 2) are more likely to reach their statistical life expectancy6. In our series, the prevalence of complications in female patients was also found to be higher than that in male patients18. This was true even after we adjusted for age and ASA scores. The univariate analysis also showed increased operative time and higher blood loss to be important factors predisposing patients to complications and/or mortality. Interestingly, age per se did not seem to increase the risk of complications in this series. We agree with others that the influence of age on the prevalence of complications is dependent on patient comorbidities at the time of surgery5,6. Although it is reasonable to assume that elderly patients are more likely to have preexisting comorbidities, this may not always be the case. Hence, older yet healthier patients may be just as likely to undergo uneventful surgery as their younger counterparts.
This study had certain limitations. It was retrospective, with all of the inherent limitations of such a study design with regard to uniformity of data collection. It is possible that some of the complications or other important events may have escaped documentation or later detection. Second, the study did not involve a detailed radiographic review. It is possible that some complications, such as avulsion of the greater trochanter or nondisplaced periprosthetic fracture, or other relevant radiographic findings may have gone undetected.
Despite the above-mentioned limitations, the findings of this study are encouraging in that revision arthroplasty was demonstrated to provide excellent improvement in function with an acceptably low rate of complications and mortality in octogenarian patients.