Validation of a screening questionnaire for hip and knee osteoarthritis in old people

Osteoarthritis (OA) of the hip and knee are common conditions that, in many cases, necessitate frequent follow-up, medical therapy, and potentially costly treatments, such as joint replacement surgery [1]. Nevertheless, relatively little is known about the prevalence of these pathologies in the general population [2]. Estimating prevalence requires tools that allow investigators to screen a large sample drawn from the general population to identify those individuals most likely to have the disease [3]. This can be a blood test, imaging modality, or other simple assay. None of these are yet available for hip or knee OA, although the search is underway for such tools [4, 5]. Currently, investigators interested in determining the prevalence of hip or knee OA must rely on medical histories or descriptions of symptoms to identify individuals most likely to have the disease. Given that no single symptom can identify patients with hip or knee OA [6], the most relevant ones are often combined to construct an algorithm that ideally have high specificity and reasonable sensitivity [7]. Individuals identified by such screening tools as likely candidates must then undergo further evaluation so a definitive diagnosis can be made. Only the clinical evaluation of a patient with symptoms of knee or hip OA by an experienced physician and an x-ray of the joint can provide a definitive diagnosis.

We developed the Knee and Hip OsteoArthritis Screening Questionnaire (KHOA-SQ) as a possible tool for gauging the prevalence of hip and knee OA in a general population study.

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

KHOA-SQ development

The KHOA-SQ was created to be a short, quickly completed questionnaire that included all relevant variables that alone or in combination would indicate, with a high sensitivity and specificity, possible OA of the knee or hip. To create the KHOA-SQ, we reviewed the available literature to determine if similar tools existed [3, 9–12] and selected from prior studies the variables most likely to identify patients with knee or hip OA.

The resulting questionnaire had 28 questions in three sections: 11 pertaining to knee osteoarthritis, 11 to hip osteoarthritis, and 6 common or general questions about both joints related to previous fractures, future interventions, and sociodemographic data. We conducted a pilot study with 400 patients undergoing urography or barium enema examinations to test which combination of questions offered the best balance of sensitivity and specificity for hip OA. The algorithms selected are presented in Figure 1. For the hip screening algorithm (algorithm #1, hip), we included 4 of the 11 original questions : "During the last 12 months, have you often had pain in one or both your hips for one month or more?"; "Has a doctor ever told you that you have osteoarthritis in one or both of your hips?"; "Do you have a prosthesis in one or both of your hips?"; and, "During the last 12 months, have you had frequent limitations or difficulties walking more than 4 blocks (500 m) because of pain or stiffness in one or both of your hips?"

For the knee screening algorithm (algorithm #1, knee), we also included 4 of the original 11 questions: "During the last 12 months, have you often had pain in one or both your knees for one month or more?"; "Has a doctor ever told you that you have osteoarthritis in one or both of your knees?"; "During the last 12 months, have you had any limitations rising from a chair or toilet because of pain or stiffness in one or both of your knees?"; and "During the last 12 months, have you had stiffness in one or both your knees for one month or more?" We based our election on previous studies [13].

In both cases, we created algorithms based on the previous questions to achieve the best possible sensitivity, while preserving reasonable specificity and positive predictive value (PPV). PPV positive was defined as OA real cases among those screening positive. Negative predictive value (NPV) is defined as no OA real cases among those screening negative. Sensitivity was defined as percentage of OA cases who screened positive, specificity as percentage of non-OA subjects who screened negative and accuracy as the percentage of screening results, both positive and negative, that were correct.

Of the 11,002 randomly selected residents, 852 were excluded. Of the 10,150 finally included in the study, 2,326 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 or other individual (Figure 2). The final participation rate was 74.65%.

Of the final sample of 1,464, including those positive and negative to the KHOA-SQ, from whom we have information about the definite diagnosis, 934 were women and 530 men; 840 were between the ages of 60 and 69, 525 were between 70 and 79, and 99 were aged 80 years or older.

We analyzed all relevant individual variables included in the KHOA-SQ to identify associations with a final diagnosis of OA. For knee OA, the questions with the highest specificity were " Has a doctor ever told you that you have osteoarthritis in one or both of your knees?" (99.5%), "Have you had any surgical intervention in one or both of your knees" (94.1%), and " During the last 12 months, have you had stiffness in one or both your knees for one month or more?" (67.9%). The questions with the highest sensitivity were "During the last 12 months, have you often had pain in one or both your knees for one month or more?" (94.8%), and " During the last 12 months, have you had any limitations going up (85.4%) or down (88%) stairs because of pain or stiffness in one or both of your knees?" (Table 2).

For hip OA, the questions with the highest specificity were " Do you have a prosthesis in one or both of your hips?" (99.3%), " Have you ever had any surgical intervention in one or both of your hips?" (98.9%), and "During the last 12 months, have you often had pain in one or both your hips for one month or more?" (78.5%) (Table 3). The questions with the highest sensitivity were "During the last 12 months, have you often had pain in one or both your hips for one month or more?" (87.1%), followed by " During the last 12 months, have you had any limitations on your stockings or shoes because of pain or stiffness in one or both of your hips?" (72.8%).

In relation to the study of the correlation of our KHOA-SQ with the WOMAC, the results were as follows. For people who, according to the KHOA-SQ, did not seem to have problems of OA in anyone of their hips the mean score in the 3 WOMAC domains (pain, stiffness, and functional limitation) measured by the WOMAC were, respectively, 25.1 (18), 23.9 (22.2) and 25 (19.8), whereas for those that, according to the KHOA-SQ, may suffer of OA the means were 38.5 (20.5), 37.7 (24.6) and 40 (20.1) respectively in the WOMAC domains. In the case of the knee, the means were 19.1 (15.8), 21. 4 (22.3), and 23.2 (21.7) for those who did not seem to have OA problems; whereas in the case of those who were suggestive to suffer OA problems the means were, respectively, 36.5 (19.7), 35 (25), and 38.2 (20.8). In all the cases, the differences between those which were positive for OA, according to our KHOA-SQ, and those who were not, were statistically significant with p < 0.001 in all the conducted comparisons.

Since the construction of our negative rows for the calculations of the sensitivity and specificity came from different sources (those evaluated as negatives and that had reported negative both of the screening algorithms (hip/knee) and those evaluated as negatives and that had reported negative to one of the screening algorithms (hip or knee)) we studied the presence of some bias. In both joints, the analysis between those evaluated as negatives and that had reported negative both of the screening algorithms (hip/knee) versus those who were negative to the explored joint but not the other on relation to the final diagnosis of osteoarthritis showed not statistically significant differences in both joints for those groups.

We examined whether alternative algorithms could achieve better sensitivity or specificity. None of those evaluated (algorithms #2 and 3), offered improved sensitivity or specificity. We also used the CART analysis to select additional algorithm trees for knee or hip OA. In the case of hip OA, the single question "During the last 12 months, have you often had pain in one or both your hips for one month or more?" had a reasonable good sensitivity (87.1%) though a relatively low specificity (54.9%). A simple algorithm that included only two questions about the presence of pain and a diagnosis of OA by a physician) provided similar sensitivity (87.1%) and specificity (54.5%) (Figure 3). The CART knee algorithm provided with a better specificity (62.8%) and worse sensitivity (82.1%) than our algorithm (Algorithm #1), and included 4 variables in a more complex algorithm.

In the absence of a simple screening test for knee or hip OA [20], determining the prevalence in a general population is difficult. Basically, the symptoms referred by the patient are the key issue to screening. We developed a screening questionnaire for identifying individuals with knee or hip OA with high sensitivity and moderate specificity.

Accurate data on the prevalence and severity of knee and hip OA in the general population are needed for a variety of reasons. In many cases, patients present with severe pain and functional limitation of the affected joint, which often translates into a diminished quality of life [21]. Accurate prevalence data can help health care systems allocate resources for the costly medical and surgical care, as well as rehabilitation programs, needed to manage and treat patients with knee and hip OA [13, 22, 23].

Our study is not the first to attempt to ascertain the symptoms most likely to identify individuals with knee or hip OA [3, 9–12, 24]. However, to the best of our knowledge it is the first to evaluate different algorithms of symptoms for both joints together. Several simple or complex algorithms have been developed to screen for knee OA [3, 10, 12, 24] or hip OA [9, 11] in the general population. In fact, we based the selection of variables for our KHOA-SQ on findings of these studies. In the case of hip OA, prior investigators had suggested "During the last 12 months, have you often had pain in one or both your hips for one month or more?" as the unique question to identify individuals with this condition [9]. We found that this single question has good sensitivity, though low specificity. Screening for knee OA appears to be more complex, as we and others [3] have shown. The question "During the last 12 months, have you often had pain in one or both your knees for one month or more?" was not nearly as sensitive as was the corresponding question about hip pain. A selection of the most predictive questions by CART analysis provided a stronger algorithm, though a more complex one that was far from perfect.

Studies on the prevalence of knee and hip OA have been conducted in all parts of the world, including the UK [9–12, 24], the US [3, 25], Canada [13], Europe (specifically Sweden, Denmark, Iceland, Netherlands, and Italy) [26–30], Australia [31], China [32, 33], and Saudi Arabia [34]. In most of these studies, screening – when it was conducted – used a simple question such as "Do you have pain in a particular joint?" followed by a complete examination of those reporting such pain by an orthopedic surgeon, as was done in our study. In contrast, we employed more questions in a more complex algorithm to screen for individuals with knee or hip OA.

The main limitation of our study is missing data. Despite three mailings and a telephone contact, one-quarter of the study population did not complete or return the screening questionnaire. Even so, our response rate (74.65%) was similar to or better than response rates achieved in previous studies. Among the respondents identified by the KHOA-SQ as being positive for knee or hip OA, 55–56% did not attend an appointment for evaluation by one of the orthopedic surgeons involved in the study. We aimed to make these appointments as convenient as possible by asking the respondent to choose an opportune day and time, and the surgeons saw patients in clinics that were accessible by public transportation and that offered a wide geographic range. Even so, some of the respondents would have had to travel up to 50 kilometers to be evaluated. Those without a relative or friend to help with transportation may have found this a difficult task. It is possible that respondents who did not come for evaluation were more likely to have mobility problems due to OA. This fact may have also biased our estimations of sensitivity and specificity, though, as we have shown, no differences were found in the rate of symptoms among attendants and non attendants.

An additional problem is the composition of the group of negative result to our screening algorithm that was evaluated by the orthopaedics surgeons. It contained people who gave negative for both joints and people who gave negative to the symptoms of the evaluated joint but not to the other. This may have biased our results. Nevertheless, the rates of those diagnosed finally as osteoarthritis between both groups for each joint were not different statistically.

Another different and important issue is the presence of a spectrum bias in our study. This comes from the different rate from negative/positive test between the data found in the field work (test positive 33% for knee and 22. 5% for hip- Figure 2) and the configuration of the 2 × 2 table for the estimation of sensitivity and the specificity. The number of included for the positive/negative test rows for the estimation of sensitivity and the specificity did not respect those percentages (76.2/23.8 for the knee and 52.7/47.3% for the hip respectively). This do have implications on the estimation of the sensitivity and the specificity, being, possible that the real sensitivity is lower and higher the specificity.

The performance of the OA evaluations by three different three orthopedic surgeons may have been another limitation of the study. Differential assessments by each of the surgeons could have biased the results. However, all three were trained before the study, were asked to conduct each examination according to a prepared manual, and completed a standardized questionnaire to gather identical information for each patient. For making a definitive diagnosis of knee or hip OA, we followed criteria established by several scientific societies [35, 36], based on the patient's self-reported symptoms and evaluation of an x-ray. In addition the surgeons were blinded as to the study goals.

Finally, our study population included only individuals aged 60 and older, thus our results have no generalizability to a younger population.

The screening algorithms we defined have, in general, acceptable sensitivity but poor specificity. Their low positive predictive value and the high rate of false positives have some consequences. With the current estimations of the sensitivities and specificities of our screening algorithm false positives may outweighing true positives. If this happens, it may be an important overload for any sanitary service or may require furtherscreening to avoid unnecessary exams. In any case, our real rate of positive tests in this field work was of 33 of and a 22.5% (knee and hip). From the health system perspective, use of such algorithms would necessitate the evaluation of several patients without knee or hip OA for each one definitively diagnosed with these conditions, something usual on screening tools. From the patient's point of view, a positive screening test could be stressful, but given that knee and hip OA are usually not potentially fatal conditions, the stress may not be as serious as it would be for a positive test for lung cancer, for instance.

The development of a screening algorithm for knee or hip OA that provides adequate sensitivity and specificity is an important goal for the management of these conditions [3]. Our algorithm, which with the current results seem to be sensitive but not highly specific for knee or hip OA, can be used in screening studies. At the present time, screening the general population for knee and hip OA must still depend on assessment of individuals' self-reported symptoms followed by clinical evaluation. But given the importance of learning more about the real prevalence and impact of knee and hip OA, even imperfect tools such as the KHOA-SQ have a place in epidemiological studies. Nevertheless, an important issue for the future would be to test the KHOA-SQ in other settings with complete data for those positive and negative to the test.