Primary and Revision Hip Arthroplasty: 5-year Outcomes and Influence of Age and Comorbidity
ANNE LÜBBEKE, JEFFREY N. KATZ, THOMAS V. PERNEGER, and PIERRE HOFFMEYER
Objective. Revision hip arthroplasty is associated with less favorable short and longterm results than primary total hip arthroplasty (THA). We compared quality of life and satisfaction 5 years after the 2 interventions, to determine the influence of patient characteristics on poorer outcomes after revision, and to analyze if their influence differed for primary and for revision arthroplasty.
Methods. This was a hospital-based prospective cohort study including patients who underwent primary (n = 435) or revision THA (n = 116). Quality of life was measured by Harris Hip Score, Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), and Medical Outcomes Study Short Form-12 questionnaire. Satisfaction was evaluated with a visual analog scale.
Results. Patients undergoing a revision were older, more often obese, and had more medical and orthopedic comorbidities. Five years after surgery, 349 patients with primary THA and 85 with revisions were available for followup. Unadjusted quality of life and satisfaction were significantly lower after revision (Harris Hip Score 76.7 vs 88.1; WOMAC pain 66.4 vs 73.3; WOMAC function 61.6 vs 70.0; satisfaction 7.7 vs 8.9). Adjustment for patient characteristics revealed that this difference was partly explained by the greater morbidity and older age of patients undergoing revision. The influence of age, comorbidities, and preoperative function on 5-year outcomes did not substantially differ for the 2 intervention groups. However, obesity was associated with a stronger negative effect on revision surgery.
Conclusion. Patients and physicians should acknowledge additional risks and consequently lower results associated with revision THA. Better information and medical preparation before surgery may help to improve the success of revision surgery. (First Release Dec 1 2006; J Rheumatol 2007;34:394-400)
Key Indexing Terms:
From the Department of Orthopaedic Surgery, Geneva University Hospital, Geneva, Switzerland; Section of Clinical Sciences, Division of Rheumatology, Immunology and Allergy, and Department of Orthopaedic Surgery, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts, USA; and Quality of Care Service, Geneva University Hospital, Geneva, Switzerland.
Supported by US National Institutes of Health grants NIH P60 AR47782 and NIH K24 AR02123.
A. Lübbeke, MD, MSc, Department of Orthopaedic Surgery, Geneva University Hospitals; J.N. Katz, MD, MS, Section of Clinical Sciences, Division of Rheumatology, Immunology and Allergy, and Department of Orthopaedic Surgery, Brigham and Women's Hospital, Harvard Medical School; T.V. Perneger, MD, PhD, Quality of Care Service, Geneva University Hospital; P. Hoffmeyer, MD, Department of Orthopaedic Surgery, Geneva University Hospital.
Address reprint requests to Dr. A. Lübbeke, Department of Orthopaedic Surgery, Geneva University Hospitals, 24, rue Micheli-du-Crest, CH-1211 Geneva 14, Switzerland. E-mail: email@example.com
Accepted for publication September 22, 2006.
Primary total hip arthroplasty (THA) is a frequently performed orthopedic intervention that significantly improves quality of life and patient satisfaction1-6. Longterm outcomes of THA are favorable, with 85–90% good and excellent results5. Ten percent to 18% of all THA interventions are revision procedures, due mostly to aseptic loosening, but also to recurrent dislocation, technical problems, or septic loosening7,8. In comparison to primary THA, revision THA is associated with more short- and longterm complications, higher mortality rates6,7,9-12, smaller improvements in functional outcome, and lower satisfaction1,6,13-18. Nevertheless, a recent metaanalysis6 points out that the existing literature about quality of life and satisfaction after revision THA is sparse.
Other than technical factors and quality of bone, patient-related factors may also be part of the reasons why revision THA achieves less favorable results than primary THA. Revision patients are older and have more comorbid conditions7. After primary THA, the presence of comorbidities is associated with lower function3,19-22, and poorer functional results decrease satisfaction23-25. While some studies found no association between age and functional outcome after primary THA2,21,26, others reported poorer functional outcomes among older patients5,27,28. In comparison, little is known about how patient age and comorbidity influence longterm clinical outcomes after revision THA1,6,10,25. To our knowledge, no study has specifically investigated the influence of age and medical and musculoskeletal comorbidities on disease-specific quality of life and satisfaction after revision THA in comparison to primary THA. Considering the increasing number of primary THA and consequently of subsequent revisions, as well as the high costs of revision surgery29, it is important to know the effect of age and morbidity on quality of life and patient satisfaction following revision surgery. The issue is critical not only for orthopedic surgeons but also for rheumatologists, because more thorough information about the patient and better preoperative medical preparation can help minimize the risk of complications and improve the success of revision surgery.
Given these considerations, we undertook a study of the comparative outcomes of primary and revision THA, with 3 specific objectives: (1) To compare general and disease-specific quality of life and patient satisfaction 5 years postintervention in patients with revision THA and primary THA carried out at the same institution. (2) To assess whether the presumably poorer outcomes after revision THA could be explained by a higher number of medical and musculoskeletal comorbidities and older age. (3) To analyze if the influence of patient characteristics (age, sex, obesity, medical and orthopedic comorbidities) on outcome differed in patients undergoing revision THA and primary THA.
MATERIALS AND METHODS
Study design and setting. We conducted a prospective cohort study among patients who had undergone primary or revision THA at the orthopedic department of a Swiss tertiary hospital, the only public hospital in the city and surrounding region. About 300 THA procedures per year are performed by orthopedic surgeons with different levels of experience.
Study population. The study population is part of a prospective hospital-based cohort of all patients undergoing primary or revision THA at the orthopedic department followed routinely since March 1996. In this study, we consecutively included patients undergoing primary THA between March 1996 and March 1998 and patients undergoing revision THA between March 1996 and December 2000. For those who had a primary or revision procedure of their other hip during the same time period (primary THA n = 36, revision THA n = 8), we included only the first hip in the analysis. Patients whose hip disease was secondary to cancer (n = 6) were excluded, as were those in the primary THA group who had undergone revision surgery before the followup at 5 years and did not yet have a 5-year followup after revision (n = 2). The study population was predominantly of European origin.
Outcome variables and instruments. Primary outcome variables were disease-specific quality of life and satisfaction 5 years after primary or revision THA. Quality of life was measured using 2 instruments: (1) The Harris Hip Score (HHS)30, a hip-specific instrument evaluating pain, function, activity, and motion on a 0–100 scale (where 100 = best). A total score below 70 points is considered a poor result, 70–80 fair, 80–90 good, and 90–100 excellent31. (2) The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), an arthritis-specific quality of life instrument completed by the patient32. We employed the 5-point Likert scale version of the WOMAC 3.0. We used the reduced WOMAC function scale, which includes 7 of the 17 original items for function33. The Cronbach alpha coefficients of the pain and the shortened function scales in our cohort were 0.90 and 0.93, respectively. Results are presented separately for pain and function on a 0–100 scale (100 = best). For the WOMAC instrument an effect size ≥ 0.25 or effects > 6% of the maximal score have been considered the minimal clinically important difference34.
The patient's satisfaction with the result of the intervention was measured using 2 visual analog scales, one for satisfaction with pain relief and the other for satisfaction with function, each scaled between 0 (lowest satisfaction) and 10 (highest satisfaction). We also report the mean of these 2 scales, given that the results were highly correlated (Spearman correlation coefficient 0.8).
Secondary outcome measures. Secondary outcome measures were the following: (1) Merle-d'Aubigné score35,36, another hip-specific physician-assessed measure evaluating pain, function and motion on a scale of 0–6 (best, total score 18). The score correlates closely with the HHS (Spearman correlation coefficient 0.936); (2) Medical Outcomes Study Short Form-12 (SF-12)37, a patient-administered generic health-related quality of life measure consisting of 12 items and comprising 2 summary measures, the physical (7 items) and the mental health component scores (5 items). The summary measures range from 0 to 100 (best).
Covariates. Potential confounders were as follows. (1) Patient characteristics: age, sex, and body mass index (BMI), as both continuous and binary variable (BMI ≥ 30). (2) Comorbidities: American Society of Anesthesiologists (ASA) score investigated as categorical and binary variable (ASA ≥ 3), medical comorbidities, and musculoskeletal disorders. Musculoskeletal disorders were measured using the number of conditions including documented pathology of the lumbar spine (spinal stenosis, recent herniated disc, osteoarthritis of the lumbar spine, ankylosing spondylitis), and osteoarthritis or THA of the contralateral hip and the knees. We dichotomized comorbidities (medical and musculoskeletal) as 0–1 versus 2 or more. (3) Preoperative functional status and pain level: measured by the Merle-d'Aubigné score, preoperative Charnley disability grade38, and previous joint surgery. Charnley disability grades classify patients into 3 categories: (A) one hip affected, (B) both hips affected, and (C) multiple joint disease or other disabilities leading to difficulties in ambulation.
Data collection. Information about the preoperative period and the intervention was documented routinely by the operating surgeons on a specifically designed data form. Information about medical comorbidities and existing disorders of the contralateral hip, the knees, and the lumbar spine at baseline, as well as data on perioperative complications, were obtained from the medical charts by a trained medical secretary. Four to 5 years after the operation, participants were contacted by telephone and mail to schedule a followup visit including a clinical and radiological examination. At the same time, the WOMAC and SF-12 questionnaires were sent to them. At the followup visit patients were asked to rate their satisfaction with the intervention. Information about orthopedic complications that had occurred since the intervention was obtained either during the visit or by telephone for all those who refused to or could not participate. All followup examinations were done by 2 trained physicians who were not involved in the operations.
Statistical methods. Objective 1. To evaluate the crude association between type of intervention (primary vs revision) and outcome, mean scores and standard deviations were calculated for each outcome score. Mean differences in outcome and 95% confidence intervals were presented. To show the relative magnitude of a difference measured by an instrument we calculated effect sizes (mean unadjusted difference divided by pooled SD of the corresponding mean scores). Effect sizes of 0.2, 0.5, and 0.8 were regarded as small, medium, and large degrees of difference, respectively39. We also presented the proportion of patients with poor results on the HHS, defined by scores < 7031. Further, we examined satisfaction scores ≥ 8, an arbitrary cutoff point.
Objective 2. To evaluate the adjusted association between the type of intervention and the outcome measures, multiple linear regression analysis was performed separately for each score. Analysis included the type of intervention as main predictor, while adjusting for confounding effects of covariates. The covariates age at operation, BMI and preoperative functional status were included as continuous, ASA scores as categorical, and medical and orthopedic comorbidities as binary variables. Charnley disability grades were not included in the final model because of redundancy with the variables medical and orthopedic comorbidities. We compared the unadjusted regression coefficient associated with revision (i.e., the simple difference between primary and reintervention) with the regression coefficient following adjustment for comorbidities and age, and then with further adjustment for complications.
Objective 3. We compared patient characteristics between the 2 intervention groups at baseline, calculating risk ratios and mean differences (with 95% CI). Then, separately for each intervention group, we analyzed the relationship between patient characteristics and outcome using linear regression. We restricted this analysis to the HHS, which discriminated best between the 2 groups. We compared unadjusted regression coefficients (95% CI) for each patient characteristic. Further, we calculated Pearson correlation coefficients for the relationship between BMI and HHS, and produced a scatterplot including a smoothed regression line for each intervention group40.
To determine if age and comorbidities affect outcome differently in primary and revision patients, a separate linear regression model was built for each patient characteristic including the type of intervention, the patient characteristic, and the corresponding interaction term. We then calculated regression coefficients (95% CI) for each interaction term.
Power calculation. We sought to detect a mean difference between the intervention groups of one-third of a standard deviation for any of the continuous outcome measures (such as HHS), i.e., effect size = 0.33. This corresponds to a moderate yet clinically significant difference39. We chose a type 1 error rate of 0.05, 2-sided. For this analysis, the power is 74%. To detect a difference of 0.5 SD, the power would be 98%.
In total, 435 patients with a primary hip arthroplasty and 116 patients with revision arthroplasty were included. In patients with primary THA, 337 (77.5%) had primary osteoarthritis, and 98 (22.5%) had secondary osteoarthritis (avascular necrosis 9.0%, dysplasia 4.1%, inflammatory arthritis 3.4%, post-fracture 3.7%, other 2.3%). In the primary THA group, 85% of the prostheses we used were hybrid (i.e., one cemented and one uncemented component), 12% were cemented, and 3% uncemented. In the revision group, 64% of the total revisions were cemented, 31% hybrid, and 4% uncemented. In 65 hips (56%) an acetabular ring was used because of insufficient bone stock. Ninety-six (82.8%) of the 116 prostheses were revised for aseptic loosening, 13 (11.2%) for septic loosening, 4 (3.4%) for a periprosthetic fracture, 2 (1.7%) for instability of the cup, and one (0.9%) for recurrent dislocation. Total revision was performed in 82 hips (70.7%), stem revision alone in 21 hips (18.1%), and cup revision alone in 13 hips (11.2%). On average, revisions were performed 121 months (SD 75.3, interquartile range 54–179) after the primary THA. In both intervention groups, 75% of the operations were performed by senior surgeons.
Complications since primary THA included 14 (3.2%) dislocations and 3 (0.7%) prosthetic joint infections; and after revision arthroplasty, 11 dislocations (9.5%) and 2 infections (1.7%). Twelve revision patients (10.3%) had a loss of fixation of the trochanteric osteotomy.
At 5 years, 349 (80.2%) patients in the primary THA group had either the followup visit (n = 314) or only a telephone interview and questionnaire (n = 35). In the revision group, 85 (73.3%) patients had either a followup visit (n = 77) or only a telephone interview and questionnaire (n = 8). The mean followup was 5 years (range 4–6 yrs). Fifty-one patients (11.7%) of the primary group and 17 (14.7%) of the revision group had died since the intervention. The remaining 35 cases (8.0%) in the primary group and 14 (12.1%) in the revision group (i.e., nonparticipants) had either left the region or were unable or unwilling to attend.
Participants versus nonparticipants. In the primary THA group, nonparticipants did not differ from participants with regard to age, BMI, sex distribution, or comorbidities. In the revision group, nonparticipants were more often female (71.4% vs 52.9% in participants) and were less likely to have had 2 or more medical comorbidities (21.4% vs 45.9% in participants). They did not differ substantially for age, BMI, or presence of other musculoskeletal disorders. In nonparticipants of the primary THA group (35 patients), 2 dislocations and no infection occurred, as compared to 12 dislocations and 3 infections in participants (349 patients). In nonparticipants of the revision THA group (14 patients), 2 dislocations and one infection occurred, as compared to 7 dislocations and one infection in participants (85 patients). Thus complications were more frequent in nonparticipants in both the primary and revision THA groups.
Baseline characteristics of participants. Compared with patients undergoing primary THA, patients undergoing revision THA were older and more often obese, and had higher ASA scores, more medical diseases, and a higher number of musculoskeletal disorders of the lumbar spine and lower limb. Women undergoing a primary THA had significantly lower preoperative Merle d'Aubigné scores, but not so before revision THA (Table 1).
Outcomes of surgery. Five years after the intervention the mean Merle-d'Aubigné score had improved substantially in both groups: in the primary THA by 6.1 points (95% CI 5.8, 6.3) and in revision THA by 3.9 points (95% CI 3.3, 4.7). However, after revision, 31% (n = 24) of the patients reported poor disease-specific quality of life (HHS < 70) compared to 9% (n = 29) after primary THA. Further, satisfaction was high (score ≥ 8 points) in only 67% (n = 51) of revision patients compared to 84% (n = 260) of primary THA patients. Higher levels of satisfaction were significantly associated with the absence of complications in both the revision (73% satisfied vs 39% of those with complications) and the primary THA group (86% satisfied vs 56% of those with complications). Patients revised for aseptic loosening had a mean HHS of 78.2 compared to 67.5 after septic loosening (mean difference 10.7 points, 95% CI –0.8, 22.3). The mean satisfaction score was 7.8 compared to 7.1 after septic loosening (mean difference 0.7, 95% CI –0.9, 2.3).
All unadjusted mean scores were lower in the revision group, except for the mental component score of the SF-12 (Table 2). The differences in outcome were most important on the HHS and the Merle-d'Aubigné score, both hip-specific physician-assessed instruments, and on the satisfaction scores (effect sizes 0.6–0.7). The differences measured with the patient-assessed WOMAC (pain and function) and the SF-12 physical component score were small to moderate, but still were of clinical importance (effect sizes 0.3–0.5). After adjustment for preoperative functional status, ASA score, medical and musculoskeletal comorbidities, BMI, sex, and age, the differences were attenuated but remained clinically relevant for HHS, Merle-d'Aubigné score, and satisfaction. WOMAC pain and function and the SF-12 physical component score showed a difference of less than one-third of a standard deviation after adjustment.
To determine whether less favorable functional outcomes after revision were due to the higher number of complications (including dislocation, infection, or trochanteric fixation problems) after revision, we adjusted for their presence or absence. Again, this lessened the differences, but they still remained important for HHS, Merle-d'Aubigné score, and the satisfaction evaluation (Table 2).
Functional outcome according to patient characteristics
Primary THA. In univariate analysis, older age, female sex, increasing BMI, lower preoperative function, a higher ASA score, and 2 or more medical or orthopedic comorbidities were significantly associated with lower HHS scores (Table 3).
Revision THA. In univariate analysis, higher BMI and lower preoperative function were significantly associated with lower HHS scores. Men, older patients, and those with more comorbidities also had lower HHS scores (Table 3).
Comparison of primary versus revision THA. Higher BMI was associated with lower HHS scores in both groups (Figure 1), but this effect was stronger in the revision group. In contrast to the results after primary THA, women did not achieve worse results than men after revision surgery. We detected no substantial differences between the 2 groups regarding the influence of age, preoperative functional status, ASA scores, or medical and orthopedic comorbidities (Table 3).
The majority of patients with either primary or revision THA achieved good to excellent clinical results 5 years after the operation. Nevertheless, patient satisfaction and functional outcome were lower after revision THA. This finding confirms previous reports. In addition, we found that the less favorable results associated with revision THA were only partly explained by the greater morbidity and older age of this patient group. Moreover, we found that obesity had a stronger negative effect on outcome in revision patients, and that female sex was associated with worse functional results only in the primary THA group, both before and 5 years post-surgery. The influence of the other risk factors was similar in both types of intervention.
Our results are consistent with the findings of a recent metaanalysis6 about functional outcome after revision THA and of 2 large population-based studies1,25 comparing primary and revision THA patients. In a cohort study of Medicare patients, Katz, et al25 reported 21% of poor results (HHS < 70) after primary (vs 9% in our study) and 43% after revision THA (31% in our study). In a large case-control study nested within the Norwegian arthroplasty register cohort1, 84% of the patients were satisfied with primary THA (vs 84% in our study) compared to 61% with revision THA (67% in our study).
In addition to the influence of patient characteristics, poorer results of revision surgery have been explained by suboptimal results of the primary THA, the longer and technically more demanding procedure, poorer bone quality, and a higher rate of complications such as dislocations, infections and trochanteric fixation problems7,11. The incidence of mechanical failure is also higher after revision11,12. Moreover, Robinson, et al16 suggested that even in the absence of complications, functional status may decline more rapidly after revision surgery than after primary THA.
Patient characteristics. We found that obesity was more strongly associated with unfavorable outcomes after revision THA than after primary THA. Few studies1,25 have reported on patient characteristics in relation to functional outcome comparing primary and revision THA. Katz, et al25 also reported a relationship between poor functional results and obesity in both intervention groups, but in contrast to our study, did not find a stronger association related to revision.
We found that men had better outcomes than women after primary THA, but not after revision THA. Katz, et al25 reported similar findings, but Espehaug, et al1 reported worse functional outcomes in women with both types of intervention.
Previous studies have reported that lower preoperative functional status3, older age1,5,27,28, medical comorbidities3,19,20, and musculoskeletal disorders21,27 are related to lower functional results after primary THA. We found similar associations in the revision group as well. This suggests that risk factors for poor outcome are globally similar in both groups of patients.
Study strengths and limitations. Our study was community-based and included a standardized clinical followup. Outcome instruments were validated and widely used, and we analyzed both patient and physician-assessed outcome measures. Assessment was done by 2 independent trained examiners in order to prevent observer bias.
As for limitations, the sample size of the revision group was small, hence statistical power to detect subgroup differences was limited. Second, no information on intraoperative bone quality was available. Third, we lacked information about socioeconomic status. As socioeconomic status is related to postoperative quality of life and satisfaction3,21, this may have caused confounding, but only if socioeconomic status was associated with the probability of revision. Fourth, exclusion of hips revised for septic loosening would have yielded slightly better results in the revision group, but the number of hips included was small (11 hips available for followup). Fifth, nonparticipants unable to participate because of poor health or death were probably more likely to experience worse outcomes. Because 5-year mortality was higher after revision THA than after primary THA (14.7% vs 11.7%), inclusion of those results could have led to a slight increase of the observed differences. Finally, preoperation information for the WOMAC and SF-12 scores was not available.
In conclusion, functional outcome and satisfaction were lower after revision THA than after primary THA. This difference was partly explained by the older age and greater morbidity of patients undergoing revision surgery. Functional outcome after primary as well as after revision THA was more likely to be poor in patients with obesity, lower preoperative function, 2 or more other affected joints, older age, and more comorbidities. However, obesity appeared to have a stronger negative effect on the outcomes of revision surgery than on primary THA. In this study, female gender was only associated with worse functional results after primary THA.
The authors thank Guido Garavaglia, MD, for his assistance with patient followup, and Christophe Barea, PhD, for his assistance with computer programming.
2. Ethgen O, Bruyere O, Richy F, Dardennes C, Reginster JY. Health-related quality of life in total hip and total knee arthroplasty. A qualitative and systematic review of the literature. J Bone Joint Surg Am 2004;86:963-74. [MEDLINE]
3. Fortin PR, Clarke AE, Joseph L, et al. Outcomes of total hip and knee replacement: preoperative functional status predicts outcomes at six months after surgery. Arthritis Rheum 1999;42:1722-8. [MEDLINE]
8. Malchau H, Herberts P, Eisler T, Garellick G, Soderman P. The Swedish Total Hip Replacement Register. J Bone Joint Surg Am 2002;84 Suppl 2:2-20.
9. de Thomasson E, Guingand O, Terracher R, Mazel C. Perioperative complications after total hip revision surgery and their predictive factors. A series of 181 consecutive procedures [French]. Rev Chir Orthop Reparatrice Appar Mot 2001;87:477-88. [MEDLINE]
12. Lie SA, Havelin LI, Furnes ON, Engesaeter LB, Vollset SE. Failure rates for 4762 revision total hip arthroplasties in the Norwegian Arthroplasty Register. J Bone Joint Surg Br 2004;86:504-9. [MEDLINE]
14. Dawson J, Fitzpatrick R, Frost S, Gundle R, McLardy-Smith P, Murray D. Evidence for the validity of a patient-based instrument for assessment of outcome after revision hip replacement. J Bone Joint Surg Br 2001;83:1125-9. [MEDLINE]
17. Talbot N, Bannister G. Revision hip arthroplasty in the octagenarian. Is it worth it? Hip Int 2001;11:152-7.
19. Greenfield S, Apolone G, McNeil BJ, Cleary PD. The importance of co-existent disease in the occurrence of postoperative complications and one-year recovery in patients undergoing total hip replacement. Comorbidity and outcomes after hip replacement. Med Care 1993;31:141-54. [MEDLINE]
24. Liang MH, Katz JN, Phillips C, Sledge C, Cats-Baril W. The total hip arthroplasty outcome evaluation form of the American Academy of Orthopaedic Surgeons. Results of a nominal group process. The American Academy of Orthopaedic Surgeons Task Force on Outcome Studies. J Bone Joint Surg Am 1991;73:639-46. [MEDLINE]
25. Katz JN, Phillips CB, Baron JA, et al. Association of hospital and surgeon volume of total hip replacement with functional status and satisfaction three years following surgery. Arthritis Rheum 2003;48:560-8. [MEDLINE]
27. Nilsdotter AK, Petersson IF, Roos EM, Lohmander LS. Predictors of patient relevant outcome after total hip replacement for osteoarthritis: a prospective study. Ann Rheum Dis 2003;62:923-30. [MEDLINE]
30. Harris WH. Traumatic arthritis of the hip after dislocation and acetabular fractures: treatment by mold arthroplasty. An end-result study using a new method of result evaluation. J Bone Joint Surg Am 1969;51:737-55.
32. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol 1988;15:1833-40. [MEDLINE]
34. Angst F, Aeschlimann A, Stucki G. Smallest detectable and minimal clinically important differences of rehabilitation intervention with their implications for required sample sizes using WOMAC and SF-36 quality of life measurement instruments in patients with osteoarthritis of the lower extremities. Arthritis Rheum 2001;45:384-91. [MEDLINE]
35. Merle D'Aubigne R. Functional results of arthroplasty of the hip. Acta Orthop Belg 1953;19:81-103.
36. Ovre S, Sandvik L, Madsen JE, Roise O. Comparison of distribution, agreement and correlation between the original and modified Merle d'Aubigne-Postel Score and the Harris Hip Score after acetabular fracture treatment: moderate agreement, high ceiling effect and excellent correlation in 450 patients. Acta Orthop 2005;76:796-802. [MEDLINE]
38. Charnley J. Numerical grading of clinical results. In: Charnley J, editor. Low friction arthroplasty of the hip: theory and practice. Berlin: Springer-Verlag; 1979:20-4.
39. Cohen J. Statistical power analysis for the behavioral sciences. Mahwah, NJ: Lawrence Erlbaum Associates; 1988.
40. Cleveland WS. Robust locally weighted regression and smoothing scatterplots. J Am Stat Assoc 1979;74:829-36.