To evaluate factors influencing orthopedic surgeons' decision in daily practice to recommend or not recommend total hip arthroplasty (THA) in patients with hip osteoarthritis (OA).
To evaluate factors influencing orthopedic surgeons' decision in daily practice to recommend or not recommend total hip arthroplasty (THA) in patients with hip osteoarthritis (OA).
General practitioners and rheumatologists were asked to prospectively include 1 patient with hip OA for whom a consultation with an orthopedic surgeon was planned to determine whether or not THA was indicated. The following variables were obtained: age, sex, occupational status, body mass index, comorbidities, duration of hip OA, patient's global assessment, Western Ontario and McMaster Universities Osteoarthritis Index pain and functioning subscale scores, New Zealand score, quality of life, and structural parameters on radiographs. The surgeon's decision was obtained by followup questionnaires. Statistical analysis evaluated potential predictors of the surgeon's decision (indication for THA within the next 6 months, yes or no) using univariate and multivariate analysis.
A total of 558 patients were included (249 men, 300 women, mean age 68.4 years, mean disease duration 4.9 years). The surgeon's decision, available for 486 patients, was to prescribe THA in 60.7% of patients. On multivariate analysis, the variables related to the surgeon's decision were the presence or absence of severe cardiovascular disease, Short Form 12 physical subscale score, and amount of joint space narrowing.
While the amount of structural degradation is only slightly or not at all taken into account in numerous criteria and/or recommendations on indications for THA, it is an independent predictor of the surgeon's decision in daily practice. Such a discrepancy should be evaluated and understood in further studies.
Total hip arthroplasty (THA) has been shown to be a highly cost-effective procedure in hip osteoarthritis (OA) that dramatically reduces joint pain and disability in most cases. It is generally recommended after failure of nonsurgical treatment and is usually performed in patients with a severe disease. However, a wide variation in the rate of THA between and within developed countries, which is unlikely to be explained by confounding factors, has been reported (1, 2). In addition, racial, socioeconomic, and sex disparities in the rates of THA have been demonstrated (3–5). Several reasons can be proposed to explain this variation, such as differences in health care systems, population demographics, prevalence of the disease, access to surgery, population perceptions about THA, and/or willingness to consider THA. Finally, another reason for variations is the variability of orthopedic surgeons' opinions regarding the appropriate time to perform surgery, and/or the appropriate patients to undergo the operation. At this time, there are no standardized universal guidelines that provide indications for THA. The accepted orthopedic practice is to delay surgery until the joint pain is severe and functional limitation is advanced (6). However, it has recently been shown that surgery performed later in the natural history of hip OA results in a poorer postoperative functional status (7–9), suggesting that THA should be performed earlier in the course of the disease in some patients. A few studies have evaluated surgeons' opinions on the right time to perform THA. In a survey performed in North America, orthopedic surgeons required severe pain and disability as well as destruction of most of the joint space on radiographs (10). However, there were some variations in surgeons' opinions on factors both related to and not related to OA severity. In another survey recently performed in Europe, orthopedic surgeons and referring physicians considered rest pain, pain with activity, and functional limitations to be the most important criteria, while radiographic changes were of minor importance (11). However, again, there were differences in the approach to indications.
Although surveys are very useful to evaluate which criteria referring physicians and surgeons consider as important in theoretical or virtual indication for THA, they do not make it possible to evaluate and weigh the criteria that really influence these health professionals' decision in clinical practice. Therefore, the objective of the present study was to evaluate the factors influencing the surgeons' decisions in daily practice in a French hip OA population for whom a consultation with an orthopedic surgeon was planned to determine whether THA was indicated.
In 2002, a total of 1,600 French general practitioners and 400 French private practice office-based rheumatologists were randomly selected from a database that includes all French medical practitioners. The participants were asked to prospectively include 1 patient with hip OA (according to the American College of Rheumatology criteria ) for whom a consultation with an orthopedic surgeon was planned within the following 3 months to determine whether THA was indicated. Other criteria for inclusion were age ≥18 years, hip OA as the main reason for the patient's visit, and anteroposterior pelvic radiograph performed within the last year. The criterion for noninclusion was previous THA. The participants had to include the first patient seen between February and July 2003 who satisfied the inclusion and exclusion criteria.
The following characteristics were collected at entry: age, sex, work status, body mass index (BMI), severe general comorbidities that might influence the surgeon's decision (severe cardiovascular disease [yes/no], respiratory insufficiency [yes/no], evolutive cancer [yes/no]), duration of index hip OA, bilateral hip OA (yes/no), associated symptomatic knee OA (yes/no), patient's global assessment of OA condition (0–10 visual numeric scale), Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain and function subscale scores, New Zealand hip OA score, quality of life (Short Form 12 [SF-12] score), past and present hip OA treatment (analgesics, nonsteroidal antiinflammatory drugs [NSAIDs], slow-acting drugs, rehabilitation, walking sticks), and duration of treatment.
In all patients, an anteroposterior pelvic radiograph performed within the last year was analyzed. The structural variables were obtained in 2 ways: 1) the participating physician's global assessment of index hip OA joint space narrowing (absent, mild, moderate, severe) and osteophytes (absent, small, medium, extensive), and 2) a digitized photograph of the radiographs obtained by the participating physicians using an ordinary digital camera; these photographs were centralized and assessed by 3 independent trained readers. The Kellgren/Lawrence and the Osteoarthritis Research Society International (OARSI) classification for osteophytes and joint space narrowing were obtained, as was the localization of joint narrowing (superolateral, superior, medial, global) (13, 14).
Patients had to complete and mail a followup questionnaire after visiting the orthopedic surgeon or after cancelling the appointment. The questionnaire included the following questions: “Have you seen the orthopedic surgeon (yes or no)? If yes, what was the surgeon's decision (indication for THA within the next 6 months, indication for THA delayed to after the next 6 months, no indication for THA, new appointment planned for decision)?”; “If the surgeon decided to plan THA during the next 6 months, do you agree to be operated on? If yes, what are your expectations regarding surgery (low, moderate, significant) for the following 3 items: improved hip mobility, decreased pain, improved functional capacity?”
The analysis was conducted in several steps. The first step consisted of a description of the characteristics of the population, as well as those of hip OA at the time of inclusion, and a description of the followup variables, in particular the surgeons' decisions regarding the indication for THA.
The second step aimed at evaluating potential predictors of the surgeon's decision using univariate analysis. The decision was dichotomized into indication for THA within the next 6 months, yes or no. The chi-square and Student's t-tests were used to evaluate a relationship between the decision and variables that might, according to the literature and/or the authors' opinion, be related to the decision, i.e., demographics, comorbidities, pain, functional limitation, quality of life, structure, and concomitant therapy. These variables were as follows: age (evaluated as a continuous parameter, then as a categorical parameter, i.e., <62 years, 62–81 years, and >81 years), sex, general comorbidities, work status, duration of hip OA, bilateral hip OA, association with symptomatic knee OA, patient's global assessment, WOMAC pain and function subscale scores, New Zealand hip OA score, mental and physical SF-12 scores, structural parameters (osteophyte and joint space narrowing scores, localization of joint narrowing), and current treatment.
The third step used a stepwise multiple logistic regression analysis in which the variable to be explained was the surgeon's indication for THA within the next 6 months (yes or no). The stepwise selection method combined backward and forward methods (15). A significance level of 0.2 on univariate analysis was required to allow a variable into the model, and a significance level of 0.05 was required for a variable to stay in the model. Before being entered in this last analysis, in order to facilitate interpretation of odds ratios, the continuous variables were changed into categorical variables using the 25th, 50th, and 75th percentiles of scores for the population. Finally, to assess the stability of our model, we used the method recommended by Austin and Tu (16). Therefore, we first created 1,000 bootstrap samples using random deletion and replacement of cases from the initial data set. Then we applied, to each bootstrap sample, the stepwise selection procedure including the variables with P values less than 0.2, which were selected using a P value threshold of 0.05.
Of the 2,000 doctors contacted, 558 included 1 patient. A total of 425 patients were included by a general practitioner (GP; 26.56% of the GPs participating in the study) and 132 by a rheumatologist (33% of the rheumatologists participating in the study; 1 missing data). The GPs and rheumatologists differed in terms of sex (97% men versus 74% men; P < 0.0001). The GPs worked more frequently in a rural area (30.3% versus 0.76%; P < 0.0001). Of the 558 patients, 249 were men (45.4%) and 300 were women (54.6%), with a mean ± SD age of 68.4 ± 10.3 years. Most of the patients were retired. The population and hip OA characteristics are provided in Tables 1, 2, and 3. Most of the patients presented with severe hip OA. Bilateral hip OA and associated symptomatic knee OA were frequently observed. General comorbidities were rare, except for severe cardiovascular disease. The participating physicians' assessments of joint narrowing and osteophytes were usually obtained. In contrast, in more than one-third of the patients, centralized scoring could not be performed due to the poor quality of the digitized radiograph photograph. Most of the patients were being treated with analgesics. NSAIDs and slow-acting drugs were frequently prescribed. However, only a small proportion of patients were treated by rehabilitation.
|Age, mean ± SD years||68.4 ± 10.3|
|Age, no. of patients|
|Body mass index, mean ± SD kg/m2||26.8 ± 4.1|
|Currently working||96 (17.4)|
|Not working due to invalidity||11 (2)|
|Severe cardiovascular disease||69 (12.4)|
|Respiratory insufficiency||13 (2.3)|
|Evolutive cancer||5 (0.9)|
|Bilateral hip OA||197 (36)|
|Symptomatic knee OA||188 (33.7)|
|Disease duration, years||4.9 ± 4.6|
|WOMAC pain subscale (0–100, best to worst)||52.6 ± 15.2|
|WOMAC function subscale (0–100, best to worst)||54.9 ± 16.3|
|New Zealand score (0–100, best to worst)||49.5 ± 16.4|
|Patient's global assessment (0–10, best to worst)||6.3 ± 1.6|
|SF-12 physical subscale (0–100, worst to best)||30 ± 6.3|
|SF-12 mental subscale (0–100, worst to best)||44.2 ± 10.4|
|Current treatment, %|
|Treating physician's global assessment of joint narrowing†|
|Treating physician's global assessment of osteophytes‡|
|OARSI joint space narrowing grade§|
|OARSI osteophyte grade§|
|Localization of minimal joint narrowing¶|
The surgeon's decision regarding THA was available in 87.1% of the patients (486 patients, 377 included by a GP and 109 by a rheumatologist). For 60.7% of these 486 patients, the decision was made to perform THA during the 6 following months.
On univariate analysis and for patients with and without the indication for THA, respectively, the variables related to the decision to perform surgery were duration of hip OA, WOMAC pain subscale, WOMAC function subscale, New Zealand score, patient's global assessment, SF-12 physical score, physicians' assessment of joint space narrowing and osteophytes, OARSI joint space narrowing, and Kellgren/Lawrence radiographic stages (Table 4). The patients treated with slow-acting drugs were less likely to have THA indicated, whereas those using walking sticks were more likely to have surgery indicated.
|Surgery not indicated in the next 6 months||Surgery indicated in the next 6 months||P|
|No.||Mean ± SD or no. (%)||No.||Mean ± SD or no. (%)|
|Age||190||67.5 ± 11||292||69.4 ± 9.2||0.06|
|Male sex||185||85 (45.9)||293||133 (45.4)||0.9|
|Body mass index (kg/m2)||188||26.7 ± 4||288||26.8 ± 4.2||0.8|
|Severe cardiovascular disease||191||29 (15.2)||294||29 (9.9)||0.08|
|Respiratory insufficiency||190||6 (3.2)||295||7 (2.4)||0.6|
|Evolving cancer||191||3 (1.6)||295||2 (0.7)||0.4|
|Retired||148 (77.9)||244 (84.1)|
|Currently working||36 (18.9)||43 (14.8)|
|Not working due to invalidity||6 (3.2)||3 (1.03)|
|Hip OA duration||186||5.7 ± 6.4||289||4.3 ± 3.2||0.007|
|Bilateral hip OA||189||68 (36)||288||99 (34.4)||0.7|
|Associated symptomatic knee OA||62||98||0.5|
|Left||24 (38.7)||33 (33.7)|
|Right||30 (48.4)||45 (45.9)|
|Left and right||8 (12.9)||20 (20.4)|
|Absent||4 (2.1)||1 (0.34)|
|Mild||15 (8)||5 (1.7)|
|Moderate||101 (53.7)||81 (27.6)|
|Severe||68 (36.2)||206 (70.3)|
|Absent||32 (16.9)||32 (11.1)|
|Mild||65 (34.4)||81 (28.2)|
|Moderate||65 (34.4)||104 (36.2)|
|Severe||27 (14.3)||70 (24.4)|
|Localization of minimal joint narrowing||94||148||0.01|
|Global||29 (30.8)||49 (33.1)|
|Superolateral||22 (23.4)||58 (39.2)|
|Superomedial||28 (29.8)||22 (14.9)|
|Superior||8 (8.5)||14 (9.5)|
|Absence||7 (7.4)||5 (3.4)|
|OARSI joint space narrowing grade||98||154||0.002|
|0||8 (8.2)||5 (3.2)|
|1||31 (31.6)||26 (16.9)|
|2||30 (30.6)||46 (29.9)|
|3||29 (29.6)||77 (50)|
|OARSI osteophyte grade||96||153||0.9|
|0||10 (10.4)||15 (9.8)|
|1||23 (24)||31 (20.3)|
|2||34 (35.4)||56 (36.6)|
|3||29 (30.2)||51 (33.3)|
|0||3 (3)||2 (1.3)|
|1||7 (6.9)||9 (5.8)|
|2||43 (42.6)||33 (21.1)|
|3||25 (24.7)||55 (35.3)|
|4||23 (22.8)||57 (36.5)|
|WOMAC pain subscale score||189||47.7 ± 13.6||288||55.9 ± 14.8||<0.0001|
|WOMAC function subscale score||174||49.6 ± 16||238||58.7 ± 15.3||<0.0001|
|Quality of life|
|SF-12 physical subscale||186||32 ± 6.4||284||28.5 ± 5.9||<0.0001|
|SF-12 mental subscale||186||45.1 ± 10||284||43.8 ± 10.4||0.2|
|New Zealand score||170||44.5 ± 14.8||258||52.7 ± 15.9||<0.0001|
|Patient's global assessment||190||5.8 ± 1.5||294||6.6 ± 1.5||<0.0001|
|Analgesics||170||156 (91.8)||275||264 (96)||0.06|
|NSAIDs||157||115 (73.2)||264||189 (71.6)||0.7|
|Slow-acting drugs||162||130 (80.2)||244||173 (70.9)||0.03|
|Rehabilitation||134||37 (27.6)||210||49 (23.3)||0.4|
|Walking sticks||135||41 (30.4)||222||93 (41.9)||0.03|
In contrast, age, sex, BMI, work status, general comorbidities, the presence of bilateral hip OA or symptomatic knee OA, the SF-12 mental score, and treatment with analgesics, NSAIDs, or rehabilitation were not related to the decision of surgery. There was, however, a trend toward a relationship between the surgeon's decision and age, the presence of severe cardiovascular disease, and treatment with analgesics.
Because centralized radiograph scoring was not available for a significant proportion of patients, 2 multivariate analyses were performed using different structural variables. In the first analysis, the structural variable was the participating physician's global assessment of index hip OA joint space narrowing and osteophytes, and in the second analysis, the centralized lecture data (Kellgren/Lawrence and OARSI scores for joint space width and osteophytes) were used. However, the results of both analyses were quite similar: the variables related to surgeons' decisions were the presence or absence of severe cardiovascular disease, the SF-12 physical subscale score, and the amount of joint space narrowing (physicians' global and OARSI score for joint space) (Table 5).
|No. of patients||OR (95% CI)||P|
|No severe cardiovascular comorbidity||373||2.2 (1.1–4.3)||0.025|
|Hip OA joint space narrowing according to physician||373|
|None or mild||1|
|SF-12 physical subscale (0–100, worst to best)||373|
|No severe cardiovascular comorbidity||196||3.1 (1.3–7.2)||0.025|
|OARSI joint space narrowing score||196|
|0 and 1||1|
|SF-12 physical subscale (0–100, worst to best)||196|
The 3 variables selected in our final model (joint narrowing, SF-12 physical subscale, and severe cardiovascular disease) were also selected in 99%, 76%, and 66%, respectively, of the 1,000 bootstrap samples to which logistic regression was applied. Other variables appeared in <37% of the models (data not shown).
A large majority (98%) of patients in whom surgery was indicated agreed to undergo the operation. The expectations from the results of surgery were usually high (significant improvement of mobility, pain, and function in 79.2%, 88.6%, and 80.9%, respectively; minor improvement in 2.1%, 0.7%, and 0.7%, respectively).
In this observational study, patients who were referred to surgeons to evaluate the indication for THA in hip OA presented with severe disease. Numerous factors were related to the surgeons' decisions, in particular clinical severity and structural degradation. In multivariate analysis, the variables independently related to the surgeons' decisions were quality of life, amount of joint space narrowing, and cardiovascular comorbidity.
This study is situated in the context of current discussions on the right time to indicate THA and attempts to establish universally accepted indication criteria for THA. In this context, it is very important to evaluate and weigh the criteria that really influence a surgeon's decision to recommend or not recommend THA. Some surveys have made it possible to collect information. However, it is not possible to assume that surgeons use the same criteria in theory and in clinical practice. Thus, the data obtained from the surveys must be confirmed by observational real-life studies, such as the present study.
There are some limitations and strengths to this study. One limitation is that it did not allow us to evaluate intersurgeon variability in the factors influencing the decision to perform THA. A second limitation is that the study was performed in only 1 country, and it cannot be assumed that similar results would have been obtained elsewhere. However, the characteristics of the patients for whom surgery was indicated were similar to those observed in other countries (8, 17), suggesting that surgeons' decision processes are similar in France and other countries. Third, the study did not allow collection of data on characteristics of the surgeons. The fourth limitation is the response rate (26.56% of GPs and 33% of rheumatologists). Finally, this study did not allow us to evaluate certain potential predictive factors, such as the patients' willingness to undergo surgery. A major strength, as stated above, is that the present study was a prospective, observational, real-life study, rather than a survey. The second strength is the high proportion (87%) of included patients for whom the followup questionnaire, and thus the surgeon's decision, was available. The third strength is the collection of scores from treating physicians and centralized radiograph scoring. Centralized scoring by trained observers using validated scores theoretically produces more valid and reliable results. However, in large observational studies, it can lead to a significant proportion of missing data. In the present study, the results of the multivariate analyses using centralized scoring were similar to those of the treating physicians. This similarity suggests that the obtained results are valid.
In a North American study, a questionnaire was sent to orthopedic surgeons concerning the usual indications for THA (10). Although there were wide variations among surgeons, most of them required at least severe pain daily, rest pain several days per week, transfer pain several days per week, and significant functional limitation (10). In a recent European study, a questionnaire was sent to orthopedic surgeons and referring physicians asking what different domains were important in their decision to recommend THA (11). Overall, rest pain, pain with activity, and walking distance were the most important criteria. The present real-life study confirms such results, since pain, function, and quality of life were associated with surgery in univariate analysis. In multivariate analysis, the only clinical parameter significantly contributing to indication for THA was quality of life. This result suggests that surgeons take into account not only pain and function, but more particularly the impact of pain and functional impairment on daily living and quality of life, which differs from one patient to another. The result might also suggest that generic quality of life questionnaires, or specific questionnaires including at least 1 domain assessing quality of life, should be used in studies that aim to establish criteria for THA indication.
The influence of structural degradation in the indication for THA is controversial. According to certain studies, the relationship between radiographic findings and pain and function is weak (18–20). It has been suggested that the preoperative radiographic grade does not influence the clinical outcome 1 year after THA (20). Consequently, it has been proposed that symptoms, and not the degree of radiographic change, should provide the indication for surgery (6, 20). The amount of structural degradation has no or little impact in numerous criteria and/or recommendations on indications for THA (3, 21–23). Thus, the present results, which demonstrate that the amount of joint space narrowing is a major and independent factor influencing the surgeons' decision, could be considered as quite surprising. In fact, these results partly confirm that in real life, surgeons act in accordance with the opinions they express in surveys. More than 70% of the North American surgeons thought that joint space narrowing should be at least greater than 50% (10). In Europe, although orthopedic surgeons ranked structural degradation on radiographs as less important than pain and functional impairment in the THA decision process, and although there were wide variations among surgeons' opinions, the majority considered that significant structural degradation would be an appropriate factor to indicate THA (11). Considering the literature presented above, one could wonder why structural degradation influences surgeons' decisions. Several hypotheses can be put forward. The first is that the literature presents a wide range of results. The relationship between pain and structure is higher in patients with severe radiographic changes than in patients with mild/moderate radiographic changes (18), and is higher in longitudinal studies compared with transversal studies (24). The structural degradation on radiographs is an independent predictor of subsequent radiologic deterioration (25), and especially of subsequent THA or the likelihood of being placed on a waiting list for THA (26, 27). The second potential explanation is the differential diagnosis. Some surgeons might have considered that pain and functional impairment were certainly due to hip OA in patients with severe joint space narrowing, but might be due, at least in part, to another disease in those with mild joint space narrowing. As a third possible explanation, the surgeons might have considered that an additional or complementary medical treatment was less likely to be efficient and/or, because the clinical course of hip OA is sometimes heterogeneous, that a spontaneous clinical improvement would be less likely observed in patients with severe structural degradation. Whatever the explanations, they have to be evaluated and understood in further studies, because it is critical to know whether hip joint structural degradation, particularly joint narrowing, should or should not be included as an independent domain in the criteria and/or recommendations on the indication for THA.
Another interesting point of the present study is that it evaluated the importance of some factors not directly related to disease severity, such as age, sex, and comorbidities, in the decision-making process. The results suggest that general comorbidities, and in particular cardiovascular comorbidities, influence the decision, rather than age and other inferior limb joint disorders. In other words, in patients with a good or acceptable global health status, most surgeons do not consider older age as being a contraindication for surgery. Such information confirms what was previously stated in a survey (10), whereas in a recently published prospective cohort study, participants in the youngest and oldest cohorts were less likely to undergo total joint replacement for hip and knee OA than those in the middle-age group (63–81 years) (28). BMI did not influence the decision, whereas obesity is frequently considered as a factor that would sway the decision against surgery (10). However, the patients included in the present study rarely presented with severe obesity.
Major variations in annual THA rates have recently been reported among developed countries, with France having one of the highest rates (1). As stated above, in the present study, the characteristics of the patients for whom surgery was decided were similar to those observed in other countries (8, 17). Thus, the present results suggest that differences in THA rates between France and other countries are not due to intercountry differences in the perception of the level of severity at which THA would be appropriate. Other factors, such as access to surgery, patients' willingness or expectations, patient-physician interaction, or health service policy, might be involved.
Dr. Maillefert 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. Maillefert, Nizard, Ravaud.
Acquisition of data. Maillefert, Cadet, Berdah, Ravaud.
Analysis and interpretation of data. Maillefert, Roy, Nizard, Ravaud.
Manuscript preparation. Maillefert, Roy, Cadet, Berdah, Ravaud.
Statistical analysis. Maillefert, Roy, Ravaud.
Negma-Lerads laboratory was involved in the data collection, but did not interfere with the analysis or interpretation of the data.