Prevalence of Knee and Hip Osteoarthritis and the Appropriateness of Joint Replacement in an Older Population

Background  Relatively little is known about the prevalence of knee and hip osteoarthritis in the general population.

Methods  To estimate the prevalence of knee and hip osteoarthritis and the appropriateness of joint replacement in a general population of older individuals, the validated Knee and Hip OsteoArthritis Screening Questionnaire (KHOA-SQ) was sent to a random sample of individuals aged 60 to 90 years, stratified by age and sex, living in a single province in Spain. Respondents positive for knee or hip osteoarthritis on the KHOA-SQ were invited to be examined by an orthopedic surgeon. Diagnosis of knee or hip osteoarthritis was based on clinical and radiographic data. For respondents judged as having osteoarthritis, the appropriateness of knee or hip replacement was evaluated using published explicit criteria.

Results  Of 11 002 individuals contacted, 7577 completed the KHOA-SQ. The derived prevalence of hip osteoarthritis was approximately 7.4%. It was slightly higher in women (8.0%) than in men (6.7%) and tended to increase with age. The estimated appropriateness rate for hip replacement was 37.7% in men and 52.7% in women with osteoarthritis. The derived prevalence of knee osteoarthritis was 12.2%; it was significantly higher in women (14.9%) than in men (8.7%) and tended to increase with age. The estimated appropriateness rate for knee replacement was 11.8% in men and 17.9% in women with osteoarthritis.

Conclusions  Knee and hip osteoarthritis are highly prevalent diseases in the older population. The estimation of appropriateness for hip replacement seems to be significantly higher than that for knee replacement.

Osteoarthritis of the knee and hip are common conditions that tend to require frequent follow-up, medical therapy, and potentially expensive treatments, such as joint replacement surgery.¹ Nevertheless, relatively little is known about the prevalence of these pathologic conditions in the general population.² Estimating prevalence requires tools that allow investigators to screen a large sample drawn from the general population to identify individuals most likely to have the disease.³ Currently, investigators interested in determining the prevalence of knee or hip osteoarthritis must rely on medical histories, radiographs,^4,5 or descriptions of symptoms.⁶ Given that no single symptom can identify patients with knee or hip osteoarthritis,⁷ the most relevant symptoms are often combined to construct an algorithm that ideally has high specificity and reasonable sensitivity.³ Individuals identified as likely candidates using such screening tools must then undergo clinical examination by an experienced physician and radiography of the joint for a definitive diagnosis to be made.

We used the Knee and Hip OsteoArthritis Screening Questionnaire (KHOA-SQ)⁸ to screen a sample of elderly individuals (aged ≥60 years) in the general population. The first goal was to estimate the prevalence of hip and knee osteoarthritis. The second goal was to estimate the proportion of patients who may have an appropriate indication for knee and hip replacement according to explicit criteria previously developed using the RAND appropriateness method.^9,10

The study was conducted between April 1, 2002, and December 31, 2003, in the province of Bizkaia (Basque Country), a predominantly urban region in the north of Spain. Bizkaia has 1 125 000 inhabitants, 23.6% of whom are 60 years or older.

To recruit participants from the general population, we used the registry of the Basque Department of Health. This registry includes all people covered by the National Health System, which, in Basque Country, includes almost 100% of the population. We performed a stratified random sampling by sex and age (60-69, 70-79, and 80-90 years).

Inclusion criteria included being 60 years or older, residing in Bizkaia, being able to complete the questionnaire and to give consent to participate in the study, and being able to attend an outpatient clinic. We excluded individuals who did not have a valid postal address or telephone number; who had severe psychiatric, sensorial, or physical illness; who had language problems that made it difficult to complete the questionnaire; and who were unable to attend the outpatient clinic.

Based on a prevalence of 10% for knee osteoarthritis and of 5% for hip osteoarthritis observed in previous studies^11-17 for patients 60 years or older, and using α = .05, 1 − β = .80, and an error in the prevalence of osteoarthritis of less than ±1.5%, we estimated that we would need to recruit more than 7000 individuals. Factoring in an estimated 10% exclusion rate and a participation rate of at least 70%, we planned to include at least 11 000 individuals in this study.

The KHOA-SQ was created to be a short, quickly completed questionnaire suitable for mailing. It includes relevant variables that alone or in combination indicate possible osteoarthritis of the knee or hip.⁸ To create the KHOA-SQ, we reviewed the available literature^3,12-16 and selected from previous studies those variables most likely to identify patients with knee or hip osteoarthritis. The algorithms selected (Figure 1) and the validation study were presented in a previous article.⁸

Letters were sent to 11 002 randomly selected residents inviting them to participate in the study. The letter presented the goals of the study, invited the recipient to participate, and asked for informed consent. It also included the KHOA-SQ. A reminder letter was sent to those who had not replied after 15 days. After 30 days, a copy of the questionnaire was again sent to those who had not responded. Finally, we contacted by telephone individuals who had not responded within 45 days of the initial mailing.

An algorithm⁸ was applied to the results of the KHOA-SQ to identify individuals likely to have knee or hip osteoarthritis. These individuals were contacted by mail (up to 3 times, if necessary) or telephone to invite them to be examined for osteoarthritis at 1 of 3 hospitals (Hospital Galdakao-Usansolo, Hospital de Cruces, and Hospital de Basurto) chosen to provide ready access for potential participants. Three orthopedic surgeons examined participants at any of the hospitals' outpatient clinics. These surgeons, who were blinded to the specific goals of the study, were trained by the research team before beginning the study, were asked to conduct each examination according to a protocol described in a manual, and were provided with standardized questionnaires that requested sociodemographic data, comorbidities, symptoms (pain, stiffness, and functional limitations), and signs related to knee or hip osteoarthritis. Information from a full clinical examination of the knees, hips, and lower back was also recorded. If the clinical examination findings were suggestive of knee or hip disease and the patient had not undergone radiography in the preceding 6 months, he or she was invited to have a radiograph of the affected joint(s) or to provide any previously completed radiographs. All the radiographs were evaluated using the Ahlbäck¹⁸ scale for knee osteoarthritis and the Kellgren and Lawrence¹⁹ scale for hip osteoarthritis. Each orthopedic surgeon provided a final diagnosis regarding the presence or absence of osteoarthritis based on the presence of symptoms and radiographic alterations as defined by the American College of Rheumatology²⁰ and the American Rheumatism Association.²¹

We also classified individuals as having knee or hip osteoarthritis if they had symptoms and radiographic evidence of osteoarthritis in either of the hip or knee articulations previously recorded in their medical records. Respondents who underwent orthopedic examination also completed the Western Ontario and McMaster Universities Osteoarthritis Index,²² which has also been validated in Spain.²³

We applied previously developed^9,10 and validated²⁴ appropriateness criteria for total knee or hip replacement to people judged as having osteoarthritis by any of the orthopedic surgeons and for whom data regarding all the variables for the algorithm were available. We considered the appropriateness rate for joint replacement of the knee or hip separately. Judgment of the appropriateness of hip replacement was based on a patient's report of joint symptoms (pain and functional limitation) and on clinical variables such as previously performed nonsurgical treatments, age, and comorbidities, all categorized as in published hip appropriateness criteria.¹⁰ Judgment of the appropriateness of knee replacement was based on a patient's age, previously performed surgical interventions, anatomical location, symptoms and functional impairment, joint mobility/stability, and radiographic evidence of the lesion (based on the Ahlbäck classification,¹⁸ again all categorized as in published knee appropriateness criteria).⁹ The present study was approved by the research committees of the participating hospitals. All data were kept confidential.

Disease prevalence was estimated separately for the knee and hip using 2 different methods, both stratified by age (in 10-year periods) and sex. For what we will call the derived prevalence, a respondent was considered positive for osteoarthritis if osteoarthritis was confirmed by means of clinical examination by an orthopedic surgeon and radiographic evaluation, if the medical record included confirmation of osteoarthritis, or if he or she had a hip prosthesis not placed owing to fracture. The number of positive cases of osteoarthritis was then divided by the number of individuals in the appropriate age- and sex-specific stratum. For what we will call the modeled prevalence, we used a method similar to that used by Frankel et al¹² based on statistical modeling using the KHOA-SQ. The number of individuals with a positive diagnosis of osteoarthritis was estimated separately for those with examination findings of positive or negative using the osteoarthritis screening questionnaire by using the number of positive findings of osteoarthritis in people examined by an orthopedic surgeon in each group. These estimates were smoothed by fitting a quadratic model across age and sex groups in appropriate logistic regression models to derive age- and sex-specific estimates. Estimates for the population number of cases in each age- and sex-specific stratum were based on extrapolation from the examined group to the overall group, stratified by positive or negative osteoarthritis results from the screening questionnaire. Then, all the estimated positive cases of osteoarthritis were summed and divided between the number of people in each age- and sex-specific stratum.

Age- and sex-specific incidence rates were calculated using the method of Leske et al,²⁵ which uses the increase in age-specific prevalence between consecutive age bands to calculate incidence considering the presence of competing mortality. The Basque Institute of Statistics provided sex- and age-specific mortality rates for 2003.²⁶ Prevalence and incidence rates are shown per 10 000 individuals.

All effects were considered significant at P < .05, unless otherwise noted. Statistical analyses were performed using SAS for Windows statistical software, version 8.2.²⁷

A total of 11 002 residents were randomly selected from the registry. Of these residents, 852 were excluded because they did not meet the selection criteria. Of the remaining 10 150 individuals, 2326 did not return the KHOA-SQ after 3 mailed requests and a telephone contact, 177 explicitly refused to participate, and 70 provided information about a relative. A total of 7577 individuals answered the KHOA-SQ, for a final response rate of 74.7% (Figure 2). Their general characteristics and answers to the main questions of the KHOA-SQ are listed in Table 1.

A total of 3168 respondents (41.8%) had KHOA-SQ results indicative of knee or hip osteoarthritis or both, and 4409 respondents did not (Table 2). A total of 1115 of the 2538 respondents with results indicative of knee osteoarthritis were examined by an orthopedic surgeon. Of these respondents, 710 (63.7%) were diagnosed as having knee osteoarthritis. Regarding hip osteoarthritis, 772 of the 1708 respondents whose KHOA-SQ results were indicative of hip osteoarthritis were clinically examined. Of these respondents, 339 (43.9%) were diagnosed as having hip osteoarthritis. There were no statistically significant differences in the presence of knee or hip symptoms on the KHOA-SQ between respondents who agreed to undergo an orthopedic examination and those who did not.

Suspicion of osteoarthritis, based on KHOA-SQ results, was higher in women than in men in each age group and generally increased with age, except in those aged 80 to 89 years. Suspicion of knee osteoarthritis was more common than hip osteoarthritis; both were exceeded by suspicion of osteoarthritis in both joints.

In respondents examined by an orthopedic surgeon, a diagnosis of hip osteoarthritis was more common in women than in men and somewhat more common in the right extremity than in the left, although it was generally recorded in both joints (Table 3). Knee osteoarthritis was much more common in both extremities than was hip osteoarthritis. When it occurred in a single joint, the right knee was more commonly affected than the left. Differences in body mass index (calculated as weight in kilograms divided by height in meters squared), weight, and obesity, based on body mass index, were statistically significant between those diagnosed as positive and those diagnosed as negative for knee osteoarthritis (for positive knee osteoarthritis: mean [SD] body mass index, 28.37 [4.29]; mean obesity, 31.5%; for negative knee osteoarthritis: mean [SD] body mass index, 26.94 [3.81]; mean obesity, 16.0%), whereas no statistically significant differences existed for the hip in any of these variables.

Age- and sex-specific prevalence rates for hip and knee osteoarthritis are detailed in Table 4. The derived prevalence of hip osteoarthritis (7.4%) was slightly higher in women (8.0%) than in men (6.7%). It increased with age, except in women aged 80 to 89 years. The modeled prevalence also increased with age in both sexes. Similar patterns were observed for knee osteoarthritis (12.2% prevalence), for which women had a higher prevalence (14.9%) than men (8.7%). Incidence data were estimated by age and sex.

In all survey respondents examined by an orthopedic surgeon, osteoarthritis was diagnosed more often in the knee (1076 joints affected) than in the hip (603 joints affected). We, then, estimated the rates of appropriate hip replacement or knee replacement for both joints, by sex and age (Table 5). The appropriateness rate for hip replacement was 37.7% in men and 52.7% in women. The appropriateness rate for knee replacement was 11.8% in men and 17.9% in women. The appropriateness rate for hip or knee replacement increased with age, except in men aged 70 to 79 years. Inference to the whole sample indicates that, globally, the rate of appropriate hip replacement would be approximately 3.6% (2.5% for men and 4.2% for women) and the rate of appropriate knee replacement would be approximately 2.0% (1.0% for men and 2.7% for women), both increasing with age.

To our knowledge, this study provides the first estimates of the prevalence of knee and hip osteoarthritis in Spain in adults 60 years and older. Equally important, it provides the first estimates of the rates of appropriate knee or hip replacement in the general population based on previously developed explicit criteria according to the RAND method.^9,28

Studies of the prevalence of knee or hip osteoarthritis have been conducted worldwide,²⁹ including the United Kingdom,^12-15 Spain,³⁰ other parts of Europe,^17,31-33 Canada,¹⁶ the United States,^4,34 and China.^35,36 Given the different populations studied and the different methods used, the results of these studies are not always in line with each other and with the present results. In this study, as with some others, screening in the general population was performed to identify individuals with possible osteoarthritis symptoms; diagnosis was then based on further clinical examination and suspicion of osteoarthritis based on radiographic evidence.^12,13 Other studies^4,37 reversed this process, using radiographic identification first, followed by symptom presence. In the case of hip osteoarthritis, some studies used radiographic examination of individuals undergoing colon radiography,³² but such prevalence estimated is much higher than in studies that required the presence of symptoms as a requirement as well.

The derived prevalence estimates we obtained were similar to those seen in other studies of similar age ranges.¹⁷ They also demonstrated relationships between hip osteoarthritis³² and knee osteoarthritis^4,33,35,37 and between women and men—similar rates for hip osteoarthritis but higher rates for knee osteoarthritis in women⁶—that were similar to those seen elsewhere. However, the present estimate of the prevalence of hip osteoarthritis was higher than that seen in some other studies.^33,36,38 The present results match those of the only other epidemiologic study of knee osteoarthritis in Spain,³⁰ although its setting, age range, sample size, and method for considering a case as osteoarthritis differed from ours.

Determination of the appropriateness of various interventions is a relatively new research endeavor. A few other research teams have attempted to do this for hip and knee replacement, usually using tools that were not specifically developed for this purpose. Hawker and colleagues¹⁶ used the Western Ontario and McMaster Universities Osteoarthritis Index, the University of Leeds team^14,15 used the Lequesne scale, and other researchers^12,13 used the New Zealand score and the National Institutes of Health method. In contrast, we used the RAND method for determining the appropriateness of hip or knee replacement. This well-known and validated method provides a tool that is capable of classifying each individual as appropriate or not for joint replacement based on his or her specific clinical characteristics. These results indicate that the appropriateness rate for hip replacement is substantially higher than that for knee replacement (47.9% vs 16.0% in those with osteoarthritis) and that it increases with age. It seems to be higher for women than for men, even after accounting for the different prevalence rates. Although similar results have been reported by other investigators for knee replacement¹³ or both joints,³⁹ our estimate of appropriateness for hip replacement is higher than those seen elsewhere.¹² This may be due, in part, to our use of the explicit RAND criteria. It is also possible that the expert panels that developed the appropriateness criteria may have been more positively inclined to intervention for hip osteoarthritis, for which the benefits of joint replacement are clearer than for knee osteoarthritis, for which the results are good but not nearly as spectacular as they are for hip replacement. In any case, the appropriateness of surgery while meeting the criteria stated might vary in the orthopedist's mind depending on the health care system in the country where the study is performed.

This study has several strengths. It is based on data from a large population of randomly selected older individuals. It used a validated screening tool (the KHOA-SQ).⁸ It used methods based on the most relevant studies performed to date in this field for estimating prevalence rates by age and sex groups. Perhaps the most unique feature was the ability to estimate the appropriateness of total knee or hip replacement in those classified as having knee or hip osteoarthritis by using the well-validated RAND appropriateness method.

The main limitation of this study is missing data. Despite 3 mailings and a telephone contact, one-quarter of the study population did not complete or return the screening questionnaire. Even so, the response rate (74.7%) was similar to or better than response rates achieved in other studies. Among respondents identified by the KHOA-SQ as being positive for knee or hip osteoarthritis, 54.8% of those with suspected hip osteoarthritis and 56.1% of those with suspected knee osteoarthritis did not attend an appointment for examination by a study orthopedic surgeon, although we tried to make these appointments convenient by asking the respondent to choose an opportune day and time at 1 of 3 clinics accessible by public transportation that covered a wide geographic range. Even so, some respondents would have had to travel up to 50 km to be examined. For individuals without ready access to transportation, this would have created a difficult barrier to participation. We cannot discount the possibility that older respondents who declined to be examined were more likely to have had mobility problems due to osteoarthritis or that some decliners were healthy and not interested in such an orthopedic review, which would lead to an underestimation or overestimation of disease prevalence. Nevertheless, the rate of missing data in this study was similar to that seen in other studies in which comparable strategies were used.^12,13

Performance of the osteoarthritis examinations by 3 different orthopedic surgeons may have been another limitation of the study. However, as explained in the “Data Collection” subsection of the “Methods” section, all 3 were trained before beginning the study, were asked to conduct each examination according to a set protocol, and completed a standardized questionnaire.

The derived prevalence estimates were based on positive cases identified by an orthopedic surgeon or by medical records; individuals for whom we had no information about osteoarthritis were categorized as not having osteoarthritis. Thus, it is possible that we underestimated the prevalence of osteoarthritis. If so, the derived estimates could be considered as lower limits for the true prevalence. Moreover, any underestimation would be more evident in the oldest group because they were the least likely to consent to being examined by an orthopedic surgeon, probably because of accessibility problems to the hospital or their health status. Estimates of prevalence rates based on statistical modeling were much higher than the derived prevalence rates. This is because the modeled prevalence rates were based on estimates obtained from clinically examined individuals that were then extrapolated to those who did not undergo clinical examination. Thus, in the modeling scenario, many individuals were classified as positive who would likely not have been classified as positive based on the study criteria. Differences between the derived and modeled prevalence rates were highest for the group aged 80 to 89 years, which also had the highest rate of declining clinical examination. We tried to minimize the effect of overestimating the prevalence rates with statistical modeling, stratifying the estimation by the osteoarthritis screening questionnaire. However, based on previous studies,^14,15 we believe that estimates of osteoarthritis prevalence based on statistical modeling are artificial and overestimate the true prevalence. A similar problem exists with the incidence estimates. Consequently, we view the derived and modeled prevalence and incidence estimates for knee and hip osteoarthritis as lower and upper limits of the true values.

Finally, although the knee and hip appropriateness criteria have been validated in this setting and are based on recommendations of other international scientific societies or of National Institutes of Health Consensus Development Panels,^40,41 there could be variations in those criteria from country to country.

In conclusion, this study suggests that in an aged population, the prevalence of hip osteoarthritis is approximately 7.4%, slightly higher in women than in men; hip replacement is appropriate in 37.7% of men and in 52.7% of women with osteoarthritis. The prevalence of knee osteoarthritis is approximately 12.2%, higher in women than in men; however, knee replacement seems to be appropriate in only 11.8% of men and 17.9% of women with osteoarthritis. New methods⁴² may improve the estimation of the prevalence and incidence of osteoarthritis. Until then, data from this study provide a solid source of information for clinical and health services providers charged with estimating health needs and planning preventive measures for this growing health problem and its costly medical interventions.

Correspondence: José M. Quintana, MD, PhD, Unidad de Investigación, Hospital de Galdakao, Barrio Labeaga s/n, 48960 Galdakao, Bizkaia, Spain (josemaria.quintanalopez@osakidetza.net).

Accepted for Publication: January 27, 2008.

Author Contributions:Study concept and design: Quintana and Escobar. Acquisition of data: Azkarate, Goenaga, and Lafuente. Analysis and interpretation of data: Quintana, Arostegui, and Escobar. Drafting of the manuscript: Quintana, Arostegui, and Escobar. Critical revision of the manuscript for important intellectual content: Quintana, Arostegui, Escobar, Azkarate, Goenaga, and Lafuente. Statistical analysis: Arostegui. Obtained funding: Quintana and Escobar. Administrative, technical, and material support: Quintana and Lafuente. Study supervision: Quintana, Escobar, Goenaga, and Lafuente.

Financial Disclosure: None reported.

Funding/Support: This study was partially supported by grants from the Fondo de Investigación Sanitaria (01/1619) and the Department of Health of the Basque Country.

Additional Contributions: We are grateful to the individuals who generously participated in this study and for the support of the staff members and medical records sections of the Hospital de Cruces, Hospital Galdakao-Usansolo, and Hospital de Basurto. The Department of Health of the Basque Government provided some of the data necessary to perform this study, Patrick J. Skerrett provided editorial assistance, and Olga Espinazo, MD, Iñigo Herrera, MD, and Rodrigo Gonzalez, MD, contributed to the orthopedic examination of patients, data entry, and data retrieval.