Factors for Pain in Patients With Different Grades of Knee Osteoarthritis

Transcription

1 Arthritis Care & Research Vol. 65, No. 5, May 2013, pp DOI /acr , American College of Rheumatology ORIGINAL ARTICLE Factors for ain in atients With Different Grades of Knee Osteoarthritis DIEUWKE SCHIHOF, 1 HANNEKE J. M. KERKHOF, 2 JURGEN DAMEN, 1 BIANCA M. DE KLERK, 1 ALBERT HOFMAN, 1 BART W. KOES, 1 JOYCE B. J. VAN MEURS, 2 AND SITA M. A. BIERMA-ZEINSTRA 1 Objective. Discordance between having pain and radiologic osteoarthritis (OA) is a well-established fact. It is suggested that this particularly applies to the less severe grades of OA. However, some people with a Kellgren/Lawrence (K/L) grade of 3 or 4 for OA are without pain. This study aimed to identify factors and differences in the factors associated with pain in persons with different grades of knee OA. Methods. We stratified the knees of more than 5,000 participants of a population-based cohort study, the Rotterdam Study, based on the grade of knee OA. Multivariate generalized estimating equation analysis was used to analyze the association with knee pain. We tested several factors not directly related to structural damage of the knee. Results. As expected, an increasing percentage of participants did not report pain with decreasing severity of knee OA: 25.8% for K/L grade 3 or 4 and 84.5% for no knee OA. Being a woman, having widespread pain, reporting general health symptoms, familial OA, and morning stiffness are factors for knee pain, but not specific for a particular grade of radiographic knee OA. Depression and hip OA showed significant interactions with the grade of OA being a factor for knee pain in knees without OA (K/L grade 0), but not in knees with OA. In addition, increasing age is protective for reporting pain in general. Conclusion. Several factors are associated with knee pain, but are not specific for a grade of radiographic knee OA. Two factors were associated with knee pain in the knee without signs of OA. INTRODUCTION Although pain is a symptom of osteoarthritis (OA) that is present in (almost) all classification criteria for OA, there is often discordance between reports of pain and radiologic OA (1 4). It is suggested that this discordance applies, in particular, to the less severe grades of knee OA and that pain is more common in more severe grades of OA (1,2). However, there are people with a Kellgren/Lawrence (K/L) grade of 3 or 4 for knee OA without any pain in the knee (1 4). For example, in an open population study of persons with K/L grade 2 in the knee, 29.9% had at some resented by Dr. Kerkhof in partial fulfillment of the requirements for a hd degree, Erasmus MC, Rotterdam, The Netherlands. Supported by the Netherlands Organisation for Scientific Research (project ). The Rotterdam Study was supported by the Netherlands Organisation for Scientific Research Investments ( , ), the Research Institute for Diseases in the Elderly ( ), the Netherlands Genomics Initiative/Netherlands Organisation for Scientific Research (project ), the Netherlands Consortium for Healthy Ageing, and the Erasmus Medical Center and Erasmus University Rotterdam. Dr. Kerkhof s work was supported by the European Commission framework 7 program TREAT-OA (grant ) and the Netherlands Genomics Initiative/Netherlands Organisation for Scientific Research (project ). time experienced pain, and with K/L grade 3 in the knee, 64.1% had at some time experienced pain (5). Explanations for the discordance between radiologic knee OA and pain in the knee include shortcomings in the measurement of pain, which can be influenced by recall or social/environmental factors (1 3), as well as shortcomings in the definition of radiographic OA and in the definition of pain (1). In addition, pain can be related to conditions other than OA, as well as to factors other than structural changes. For example, bone marrow lesions or joint effusion, which can be made visible on magnetic resonance imaging (MRI) (6), but also psychological factors (e.g., depression or anxiety), are suggested to be associated 1 Dieuwke Schiphof, hd, Jurgen Damen, MD, MSc, Bianca M. de Klerk, hd, Albert Hofman, hd, Bart W. Koes, hd, Sita M. A. Bierma-Zeinstra, hd: Erasmus MC, Rotterdam, The Netherlands; 2 Hanneke J. M. Kerkhof, MD, hd (current address: fizer, Capelle a/d Ijssel, The Netherlands), Joyce B. J. van Meurs, hd: Erasmus MC, Rotterdam, and Netherlands Consortium for Healthy Ageing, Leiden, The Netherlands. Address correspondence to Dieuwke Schiphof, hd, Erasmus MC, Department of General ractice, Room GK-1044, O Box 2040, 3000 CA Rotterdam, The Netherlands. d.schiphof@erasmusmc.nl. Submitted for publication March 26, 2012; accepted in revised form October 14,

2 696 Schiphof et al Significance & Innovations Being a woman, having widespread pain, reporting general health symptoms, familial osteoarthritis (OA), and morning stiffness are factors for knee pain, but are not specific for a grade of radiographic knee OA. Depression and hip OA are factors for knee pain in knees without any signs of knee OA. with pain (7,8). Additional factors that may be related to the experience of pain include various comorbidities and/or inferior general health status (1), as well as sex differences and education level (9,10). It is unclear why some people experience no pain at all when they have an established joint pathology. Little information is available on the factors that relate to knee pain and whether these factors differ between the radiologic grades of knee OA. These factors might be different in the presence and absence of certain joint diseases such as OA. Factors that relate to knee pain could be possible targets for treatment, as well as for future research. Therefore, the aim of this study was to identify factors, and differences in these factors, that are associated with (the absence of) pain in people with different grades of radiologic knee OA in a population-based cohort. ATIENTS AND METHODS Study population. The data used in this study were baseline data obtained from the Rotterdam Study, a populationbased prospective cohort study of men and women (ages 55 years) in which the incidence and risk factors for chronic disabling diseases were investigated (11,12). All 10,275 inhabitants in that age group in a district in Rotterdam were invited for a baseline examination between August 1990 and June 1993; of these, 7,983 participated. The Medical Ethics Committee of Erasmus University Medical Center approved the study. All participants gave written consent and were extensively interviewed at home by trained interviewers. Furthermore, participants were asked to visit the research center for radiographs and other medical examinations. Of the 7,983 participants, 6,450 visited the research center for the baseline measurements. Radiographic assessment. Radiographs of the knees were taken at the research center at 70 kv, a focus of 1.8 mm 2, and focus-to-film distance of 120 cm, using High Resolution G cm film (Fujifilm Medical Systems). Radiographs of the knees were weight-bearing anteroposterior radiographs. A random set of 5,647 baseline radiographs of the knees was scored for OA using the K/L scale (13). Five trained readers evaluated the radiographs of the knee. Interrater reliability of the K/L score was good ( 0.71). All readers were unaware of the participants clinical status. ossible knee OA was defined as K/L grade 1 in either knee (osteophytic lipping or one osteophyte). Mild (or definite) OA was defined as K/L grade 2 in either knee (at least 2 definite osteophytes and possible joint space narrowing [JSN]). Moderate to severe knee OA was defined as a K/L score of 3 in either knee (at least 2 definite osteophytes and definite JSN). The 40 knees that had a total or partial knee replacement or an osteotomy were excluded from the analysis. Unilateral knee OA was defined as having a K/L grade of 2 in one of the knees, and bilaterality was defined as having a K/L grade of 2 in both knees. Additional standard anteroposterior radiographs of the hips and hands were taken. Radiographs of the hips were weight bearing. Hip OA is defined as having a K/L grade of 2 in one or both hips. Hand OA is defined as having a K/L grade of 2 in one the following joints: first carpometacarpal joint, proximal interphalangeal joints, distal interphalangeal joints, interphalangeal joint, metacarpophalangeal joints, or trapezioscaphoid joint (14). Outcome. At baseline, trained interviewers conducted an extensive standardized home interview. ain was determined to be present based on the answers (yes/no) to the following question: Have you had knee pain in the last month? In addition, the same question was asked about knee pain during the last 5 years. If one of these questions was answered with yes for the knee, we classified the knee with pain, separately, for the left and the right knees. Factors. Factors were selected based on evidence from data described in the literature on factors that might influence pain with radiologic OA (5,9,10,13,15), and the factors of interest should be available in the data of the Rotterdam Study from the interview and the visit at the research center. No MRIs of the knees were made in these participants at that time; therefore, no data on soft tissue lesions or other structures specifically visible on MRI were available. Height and weight were measured and body mass index (BMI; kg/m 2 ) was calculated. Data on widespread pain were collected from the 2 pain questions mentioned above, following the definition of Wolfe et al, i.e., pain is considered to be widespread when all of the following have been present for at least 3 months: 1) pain in the left side of the body, 2) pain in the right side of the body, 3) pain above the waist, 4) pain below the waist, and 5) in addition, axial skeletal pain (cervical spine or anterior chest or thoracic spine or lower back) (15). Answers to the question of the highest attained education were categorized into low, medium, or college level. The lower level indicates primary, lower general, and lower vocational education; the medium level indicates intermediate general, higher general, and intermediate vocational education; and the college level indicates higher vocational and university education. Answers to the questions in the interview were used to assess general health symptoms and comorbidities, such as feeling depressed and having had a heart attack. High blood pressure was defined as having a systolic blood pressure 160 mm Hg, having a diastolic blood pressure

3 ain Determinants for Knee OA 697 Table 1. revalence of pain per grade of knee OA (n 11,022 knees)* Knees without pain, no. (%) Knees with pain, no. (%) K/L grade 0 5,662 (84.5) 1,037 (15.5) K/L grade 1 2,128 (79.0) 564 (21.0) K/L grade (60.4) 540 (39.6) K/L grade 3 or 4 69 (25.8) 198 (74.2) *OA osteoarthritis; K/L Kellgren/Lawrence. 100 mm Hg, or the use of antihypertensive medication. Nonfasting serum samples were collected at the research center at baseline. The samples were immediately put on ice and processed within 30 minutes. A prevalent diabetes mellitus case was defined as receiving anti diabetes mellitus medication or having had a nonfasting or postload glucose concentration of 11.1 mmoles/liter. Data on familial OA were obtained from the participants answers to the question as to whether their parents or siblings have had OA in one or more of their joints. The answers to the questions Do you suffer from stiffness in your legs when you wake up in the morning? and How long was this stiffness persisting? were combined to define the presence of morning stiffness. The presence of morning stiffness was defined as having morning stiffness in the legs for less than 30 minutes. Statistical analysis. For the association between the factors and pain in the knee with a grade of OA, we used multivariate logistic generalized estimating equation (GEE) analysis. Odds ratios (ORs) with 95% confidence intervals (95% CIs) show the association between the factors and pain. A adjusts for the correlations between the right and left knees of the same person. First, we analyzed the factors for knee pain independent from the K/L grades. In addition, we analyzed the factors in each K/L grade (K/L grade 0, 1, 2, or 3 or 4) separately. To evaluate if the significantly associated factors were specific for a grade of OA (interaction), we used a standard normal approximation for z, which was calculated as z 1 2 / ((se 1 ) 2 (se 2 ) 2 ), where 1 the log odds of the factor (se 1 the standard error of this factor) for the specific grade of radiographic knee OA in which this factor is significantly associated with knee pain, and 2 the log odds (se 2 the standard error) of the same factor for the other grades of radiographic knee OA. A 2-sided test with a significance level of 0.05 was used. There is a significant difference in the factor for the specific grade if z is less than 1.96 or if z is 1.96 (16). The factors were on the patient level and not on the knee Table 2. Associations between knee pain and the study factors* (n 7,900 knees [71.7%]), articipant characteristics Age, mean SD years , ( ) BMI, mean SD kg/m , ( ) Women 3,191 (57.7) 5, ( ) Widespread pain 525 (9.5) 5, ( ) Education 5,466 Low 1,995 (36.5) 1.14 ( ) Medium 2,866 (52.4) 1.19 ( ) High 605 (11.1) Ref. Health symptoms 2,607 (47.2) 5, ( ) Feeling depressed 1,809 (33.2) 5, ( ) Diabetes mellitus 539 (9.8) 5, ( ) High blood pressure 1,536 (28.0) 5, ( ) Ever had a heart attack 478 (8.7) 5, ( ) Hip OA (K/L grade 2) 518 (9.8) 5, ( ) Hand OA (K/L grade 2) 3,135 (62.9) 4, ( ) Familial OA 1,069 (20.6) 5, ( ) Morning stiffness 1,023 (20.0) 5, ( ) Knees 11,022 K/L grade 0 6,699 (60.8) Ref. K/L grade 1 2,692 (24.4) 1.42 ( ) K/L grade 2 1,364 (12.4) 3.04 ( ) K/L grade 3 or (2.4) ( ) * Values are the number (percentage) unless otherwise indicated. GEE generalized estimating equation; OR odds ratio; 95% CI 95% confidence interval; BMI body mass index; OA osteoarthritis; K/L Kellgren/ Lawrence. Statistically significant association between the factor and knee pain ( 0.05).

4 698 Schiphof et al Table 3. Association with knee pain in knees with K/L grade 0* (n 4,855 knees [72.5%]), articipant characteristics Age, mean SD years , ( ) BMI, mean SD kg/m , ( ) Women 2,019 (53.2) 3, ( ) Widespread pain 331 (8.7) 3, ( ) Education 3,749 Low 1,297 (34.6) 1.24 ( ) Medium 1,995 (53.2) 1.24 ( ) High 457 (12.2) Ref. Health symptoms 1,749 (46.1) 3, ( ) Feeling depressed 1,212 (32.4) 3, ( ) Diabetes mellitus 342 (9.0) 3, ( ) High blood pressure 996 (26.4) 3, ( ) Ever had a heart attack 334 (8.9) 3, ( ) Hip OA (K/L grade 2) 292 (8.0) 3, ( ) Hand OA (K/L grade 2) 2,009 (58.1) 3, ( ) Bilaterality, no OA 3,603 (95.0) 3,792 Ref. Unilateral OA 189 (5.0) 3, ( ) Familial OA 728 (20.4) 3, ( ) Morning stiffness 640 (18.2) 3, ( ) * Values are the number (percentage) unless otherwise indicated. K/L Kellgren/Lawrence; GEE generalized estimating equation; OR odds ratio; 95% CI 95% confidence interval; BMI body mass index; OA osteoarthritis. Statistically significant association between the factor and knee pain ( 0.05). level, i.e., for the separate grade analyses, it is possible and highly likely that a person is included in the analysis with one knee only. For example, in the analyses of K/L grade 1, only the knees with K/L grade 1 are in the analyses, but this group might include participants with K/L grade 2, 3, or 4 in the other knee (and grade 0 is also possible). This is indicated by the unilateral or bilateral knee OA factor. The software package SSS, version 17, was used for all analyses. RESULTS There were 5,527 participants with complete data on K/L score, pain, age, sex, and BMI. A total of 11,022 knees were used for the analysis, which means that 32 participants were included with one knee; the other knee was excluded due to a total or partial knee replacement or an osteotomy. Table 1 shows the proportion of painful knees in the different grades of knee OA. Due to the small number of knees with K/L grade 3 or 4, these 2 grades were taken together in the analysis. Regarding the knees with OA, a considerable proportion (25.8%) with moderate to severe OA was not painful during the last 5 years. The number of individuals without knee pain increased with decreasing grade of joint damage. More than half of the knees (60.4%) with K/L grade 2 were pain free, with an increasing number of pain-free knees with K/L grade 0. Table 2 shows the associations between the factors and pain in the knee, including the K/L grade. Irrespective of the K/L grade, the following are significantly associated with knee pain: being a woman (OR 1.48, 95% CI ), having widespread pain (OR 4.35, 95% CI ), having general health symptoms (OR 1.53, 95% CI ), feeling depressed (OR 1.19, 95% CI ), ever having a heart attack (OR 1.55, 95% CI ), having hand OA (OR 1.22, 95% CI ), familial OA (OR 1.36, 95% CI ), and morning stiffness (OR 1.96, 95% CI ). Older age (OR 0.98, 95% CI ) is a protective factor for knee pain, irrespective of the grade of knee OA. Increasing ORs are shown with increasing grade of K/L (K/L grade 1: OR 1.42, 95% CI ; K/L grade 2: OR 3.04, 95% CI ; and K/L grade 3 or 4: OR 13.91, 95% CI ). Tables 3 6 show the associations between the factors and knee pain with increasing grades of K/L. Being a woman, having widespread pain, having health symptoms, and morning stiffness are factors associated with knee pain in all K/L grades up to grade 2. Familial OA is a factor that is significantly associated with knee pain in K/L grade 0 and 1; in K/L grade 2, familial OA is only borderline significant. In K/L grade 3 or 4, only widespread pain is still significant. In K/L grade 0, feeling depressed (OR 1.35, 95% CI ), ever having a heart attack (OR 1.47, 95% CI ), and hip OA (OR 1.44, 95% CI ) are significantly associated with knee pain. Depression and hip OA showed a significant interaction (z and z 2.21, respectively) with grades of knee OA (K/L grade 0 versus K/L grade 1 and higher). Ever having a heart attack and unilateral knee OA showed no interaction with the grades of knee OA (z for

5 ain Determinants for Knee OA 699 Table 4. Associations with knee pain in knees with K/L grade 1* (n 1,924 knees [71.5%]), articipant characteristics Age, mean SD years , ( ) BMI, mean SD kg/m , ( ) Women 1,162 (60.7) 1, ( ) Widespread pain 184 (9.6) 1, ( ) Education 1,898 Low 723 (38.1) 0.93 ( ) Medium 994 (52.4) 1.09 ( ) High 181 (9.5) Ref. Health symptoms 908 (47.4) 1, ( ) Feeling depressed 631 (33.2) 1, ( ) Diabetes mellitus 196 (10.2) 1, ( ) High blood pressure 559 (29.4) 1, ( ) Ever had a heart attack 165 (8.7) 1, ( ) Hip OA (K/L grade 2) 202 (10.9) 1, ( ) Hand OA (K/L grade 2) 1,191 (68.7) 1, ( ) Bilaterality, no OA 1,478 (77.3) 1,911 Unilateral OA 433 (22.7) 1, ( ) Familial OA 367 (20.4) 1, ( ) Morning stiffness 350 (19.8) 1, ( ) * Values are the number (percentage) unless otherwise indicated. K/L Kellgren/Lawrence; GEE generalized estimating equation; OR odds ratio; 95% CI 95% confidence interval; BMI body mass index; OA osteoarthritis. Statistically significant association between the factor and knee pain ( 0.05). ever having a heart attack and z 0.21 and z 1.62 for unilateral knee OA, respectively). DISCUSSION Widespread pain, health symptoms, being a woman, familial OA, and morning stiffness are factors for knee pain, but are not specific for a particular grade of radiographic knee OA. Age is a nonspecific protective factor for knee pain. Feeling depressed and having hip OA are factors for knee pain in knees without signs of radiographic knee OA (K/L grade 0). The nonspecific protective association of age is not rare. Gagliese and Melzack showed that older people have significantly lower total scores on pain questionnaires and choose fewer words to describe their pain than younger people (17). Furthermore, in such a relatively old cohort, the principle of survival of the fittest might play a role, certainly in combination with a potential health-based selection bias. The participants who are less mobile with more severe disease and pain might not have participated due to problems getting to the research center. Having hip OA can give referred pain in the knee, which is highly likely in participants with knee pain and no signs of knee OA (16). Having a family member with OA may raise awareness about joint pain; in addition, genetic predisposition and/or social/environmental factors can influence the feeling and reporting of pain. Morning stiffness is associated with knee pain in all grades of OA, including knees without radiographic OA. In the group with moderate to severe radiographic knee OA, the association was only borderline significant, probably due to a (statistical) power problem. One limitation regarding our definition of morning stiffness is that it is unknown whether the stiffness occurred in the knee, the hip, or elsewhere in the legs. In addition, patients were not asked on which side the stiffness was present. Therefore, the association would probably have been higher had morning stiffness been measured in a knee-specific and side-specific way. Morning stiffness is one of the symptoms already included in the clinical diagnosis of OA (18). Besides pain, it is seen as another symptomatic expression of OA. Apart from the abovementioned limitations, morning stiffness is associated with knee pain in all grades of radiographic knee OA, even in K/L grade 0. This could indicate that, together with knee pain, morning stiffness is an important symptom of early knee OA. Having OA in the knee other than the knee in the analysis is a significant factor for knee pain in K/L grade 1, although it is not specific for the knees with this grade of radiographic OA. Bilateral knee OA or having knee OA might be a nonspecific factor for knee pain. Another study found that participants with bilateral knee OA had significantly more pain than participants with unilateral knee OA (19). Hand OA was significantly associated with knee pain, irrespective of the grade of knee OA. However, this finding needs to be interpreted with caution because there is no consensus about the definition of hand OA (14,20). In the present study, the definition used is having OA in one or

6 700 Schiphof et al Table 5. Associations with knee pain in knees with K/L grade 2* (n 940 knees [68.9%]), articipant characteristics Age, mean SD years , ( ) BMI, mean SD kg/m , ( ) Women 760 (74.4) 1, ( ) Widespread pain 134 (13.1) 1, ( ) Education 1,016 Low 445 (43.8) 1.07 ( ) Medium 499 (49.1) 1.17 ( ) High 72 (7.1) Ref. Health symptoms 497 (48.6) 1, ( ) Feeling depressed 348 (34.7) 1, ( ) Diabetes mellitus 122 (11.9) 1, ( ) High blood pressure 329 (32.5) 1, ( ) Ever had a heart attack 90 (8.9) 1, ( ) Hip OA (K/L grade 2) 145 (15.2) ( ) Hand OA (K/L grade 2) 719 (80.2) ( ) No OA 1,022 Unilateral OA 582 (56.9) 1,022 Bilateral OA 440 (43.1) 1, ( ) Familial OA 214 (22.1) ( ) Morning stiffness 248 (26.0) ( ) * Values are the number (percentage) unless otherwise indicated. K/L Kellgren/Lawrence; GEE generalized estimating equation; OR odds ratio; 95% CI 95% confidence interval; BMI body mass index; OA osteoarthritis. Statistically significant association between the factor and knee pain ( 0.05). more joints of the hand, which results in a large number of participants diagnosed with hand OA. It is the definition most used in research (14), but might well overestimate the number of participants with clinical hand OA. Widespread pain is associated with knee pain in all grades of knee OA. In the case of widespread pain, the central nervous system is involved and sensitization mechanisms are present (21). That might explain why, in the present study, widespread pain was not specific for any stage of knee OA. Knee pain is a part of the widespread pain definition, but if we defined widespread pain without knee pain, the same results were found, only with somewhat lower associations (data not shown). In the present study, a high BMI was not associated with knee pain. This is in contrast with, for example, results from Nguyen et al (22). Explanations for this contrasting result might be a difference in analysis; we adjusted for more factors and we did not categorize BMI. Furthermore, there might be a difference in the study population as to fat mass distribution. Tanamas et al (23) showed that android fat mass is strongly associated with foot pain in contrast to gynoid fat mass. In the present study and in the study by Nguyen et al, this fat mass distribution is not taken into account. We defined knee pain as having had pain in the knee in the last month and/or in the last 5 years. This definition covers a considerable period of time for having knee pain. By separating these 2 time points, we may be able to observe differences in factors for knee pain, depending on different definitions of pain. For example, when pain was defined as having pain in the knee in the last month, no significant association was found for age, hand and hip OA, and familial OA. K/L grade 1 was no longer significantly associated with knee pain, whereas K/L grades 2 and 3 or 4 were still significantly associated (data not shown). When pain was defined as having knee pain in the last 5 years, the results were very similar to those of the combined pain definition (data not shown). We used a pain definition that is used widely in open populationbased cohorts for defining pain. The use of the Western Ontario and McMaster Universities Osteoarthritis Index questionnaire might have provided more information. In the absence of this questionnaire, we were not able to investigate pain severity or different dimensions of pain such as weight-bearing pain and non weight-bearing pain. Feeling depressed is associated with knee pain in knees without signs of knee OA. Knee pain or pain in general can lead to a depressed mood, which in turn can increase pain and lead to a further increased feeling of depression. In addition, depression is associated with a reduced pain threshold (24,25). None of the tested factors was specific for a grade of radiographic knee OA, except feeling depressed and having hip OA and K/L grade 0 for knee OA. This is important for treatment; patients with knee pain without structural pathologies in the knee might have a differential diagnosis

7 ain Determinants for Knee OA 701 Table 6. Association with knee pain in knees with K/L grade 3 or 4* (n 172 knees [64.4%]), articipant characteristics Age, mean SD years ( ) BMI, mean SD kg/m ( ) Women 156 (73.2) ( ) Widespread pain 36 (16.9) ( ) Education 210 Low 97 (46.2) 1.90 ( ) Medium 92 (43.8) 0.86 ( ) High 21 (10.0) Ref. Health symptoms 105 (49.3) ( ) Feeling depressed 76 (36.0) ( ) Diabetes mellitus 32 (15.0) ( ) High blood pressure 72 (34.3) ( ) Ever had a heart attack 20 (9.5) ( ) Hip OA (K/L grade 2) 43 (22.1) ( ) Hand OA (K/L grade 2) 162 (87.1) ( ) No OA 213 Unilateral OA 60 (28.2) 213 Ref. Bilateral OA 153 (71.8) ( ) Familial OA 49 (24.7) ( ) Morning stiffness 76 (37.8) ( ) * Values are the number (percentage) unless otherwise indicated. K/L Kellgren/Lawrence; GEE generalized estimating equation; OR odds ratio; 95% CI 95% confidence interval; BMI body mass index; OA osteoarthritis. Statistically significant association between the factor and knee pain ( 0.05). of hip OA or feeling depressed, which both need specific subsequent management. A limitation of the present study is the scarce amount of information available about the daily activities of the participants. Furthermore, definitions for the comorbidities are somewhat limited; e.g., feeling depressed is not based on a standardized questionnaire. Hawker et al (2011) described that painful knee OA determines disability and fatigue, which in turn leads to a depressed mood and therefore a worsening of pain (26). It would be interesting to see what factors are associated with pain at followup; however, this is not possible because at followup the pain in the knee was not specified for the left and right knees separately. The result might give the suggestion that with severity of radiographic knee OA, fewer factors are associated with pain. Having widespread pain is the only factor associated with knee pain in participants with severe knee OA. Having morning stiffness is borderline associated with having pain in these participants with severe knee OA. This association is not significant, although it shows the same estimate for the association. The same applies to general health symptoms. The lacking significance is due to a power problem; there are only 213 participants with severe knee OA. This small proportion could be due to a potential health-based selection bias in the population. The participants had to be mobile enough to visit the research center. Therefore, participants with more severe symptoms (pain and severe OA) were most likely not included. This might have resulted in an underestimation of the associations. In addition, there is also the lack of information about other structures, such as bone marrow lesions and effusion. A recent systematic review about how knee abnormalities visualized on MRI explained knee pain in knee OA concluded that knee pain in OA is associated with bone marrow lesions and effusion (6). The authors also concluded that the level of evidence was moderate; more research is needed to explore the features that may indicate the origin of pain. Therefore, future research has to focus on factors such as extensive imaging factors, pain sensitivity, and other psychosocial factors in addition to the associated factors found in the present study. The associated factors of the present study need more thorough research to find out what role they play in the pain mechanisms of OA. In conclusion, being a woman, having widespread pain, morning stiffness, familial OA, and reporting general health symptoms are factors for knee pain, but are not specific for a particular grade of radiographic knee OA. Depression and hip OA were factors for knee pain only in those knees without any sign of OA. Morning stiffness is a factor for knee pain, even in knees without radiographic signs of knee OA. This could indicate that morning stiffness in combination with knee pain is a symptom of OA in early knee OA; this certainly warrants further investigation.

8 702 Schiphof et al ACKNOWLEDGMENTS The authors are very grateful to all of the participants and staff of the Rotterdam Study. They would like to thank Dr. A. van Vaalen, Dr. M. Kool, Dr. D. A. C. Sluiter, Dr. E. Odding, Dr. A.. Bergink, Dr. M. Reijman, Dr. S. Dahaghin, Dr. V. Kahlmann, and Dr. E. S. Stille for scoring the radiographs, and E. Oei for assistance. AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Schiphof 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 conception and design. Schiphof, Hofman, Bierma- Zeinstra. Acquisition of data. Kerkhof, Hofman, van Meurs, Bierma- Zeinstra. Analysis and interpretation of data. Schiphof, Kerkhof, Damen, de Klerk, Koes, Bierma-Zeinstra. REFERENCES 1. Bedson J, Croft R. The discordance between clinical and radiographic knee osteoarthritis: a systematic search and summary of the literature. BMC Musculoskelet Disord 2008;9: Duncan R, eat G, Thomas E, Hay E, McCall I, Croft. Symptoms and radiographic osteoarthritis: not as discordant as they are made out to be? Ann Rheum Dis 2007;66: Creamer, Hochberg MC. Why does osteoarthritis of the knee hurt sometimes? Br J Rheumatol 1997;36: Wluka AE. Remember the Titanic: what we know of knee osteoarthritis is but the tip of the iceberg. J Rheumatol 2006; 33: Lethbridge-Cejku M, Scott WW Jr, Reichle R, Ettinger WH, Zonderman A, Costa, et al. Association of radiographic features of osteoarthritis of the knee with knee pain: data from the Baltimore Longitudinal Study of Aging. Arthritis Care Res 1995;8: Yusuf E, Kortekaas MC, Watt I, Huizinga TW, Kloppenburg M. Do knee abnormalities visualised on MRI explain knee pain in knee osteoarthritis? A systematic review. Ann Rheum Dis 2011;70: Creamer, Hochberg MC. The relationship between psychosocial variables and pain reporting in osteoarthritis of the knee. Arthritis Care Res 1998;11: Hopman-Rock M, Odding E, Hofman A, Kraaimaat FW, Bijlsma JW. hysical and psychosocial disability in elderly subjects in relation to pain in the hip and/or knee. J Rheumatol 1996;23: Hannan MT, Anderson JJ, incus T, Felson DT. Educational attainment and osteoarthritis: differential associations with radiographic changes and symptom reporting. J Clin Epidemiol 1992;45: Leveille SG, Zhang Y, McMullen W, Kelly-Hayes M, Felson DT. Sex differences in musculoskeletal pain in older adults. ain 2005;116: Hofman A, Grobbee DE, de Jong T, van den Ouweland FA. Determinants of disease and disability in the elderly: the Rotterdam Elderly Study. Eur J Epidemiol 1991;7: Hofman A, van Duijn CM, Franco OH, Ikram MA, Janssen HL, Klaver CC, et al. The Rotterdam Study: 2012 objectives and design update. Eur J Epidemiol 2011;26: Kellgren JH, Lawrence JS. Radiological assessment of osteoarthrosis. Ann Rheum Dis 1957;16: Marshall M, Dziedzic KS, van der Windt DA, Hay EM. A systematic search and narrative review of radiographic definitions of hand osteoarthritis in population-based studies. Osteoarthritis Cartilage 2008;16: Wolfe F, Smythe HA, Yunus MB, Bennett RM, Bombardier C, Goldenberg DL, et al. The American College of Rheumatology 1990 criteria for the classification of fibromyalgia: report of the Multicenter Criteria Committee. Arthritis Rheum 1990;33: Dahaghin S, Bierma-Zeinstra SM, Reijman M, ols HA, Hazes JM, Koes BW. Does hand osteoarthritis predict future hip or knee osteoarthritis? Arthritis Rheum 2005;52: Gagliese L, Melzack R. Age-related differences in the qualities but not the intensity of chronic pain. ain 2003;104: Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al. Development of criteria for the classification and reporting of osteoarthritis: classification of osteoarthritis of the knee. Arthritis Rheum 1986;29: Adegoke BO, Babatunde FO, Oyeyemi AL. ain, balance, self-reported function and physical function in individuals with knee osteoarthritis. hysiother Theory ract 2012;28: Kerkhof HJ, Meulenbelt I, Akune T, Arden NK, Aromaa A, Bierma-Zeinstra SM, et al. Recommendations for standardization and phenotype definitions in genetic studies of osteoarthritis: the TREAT-OA consortium. Osteoarthritis Cartilage 2011;19: Imamura M, Imamura ST, Kaziyama HH, Targino RA, Hsing WT, de Souza L, et al. Impact of nervous system hyperalgesia on pain, disability, and quality of life in patients with knee osteoarthritis: a controlled analysis. Arthritis Rheum 2008;59: Nguyen US, Zhang Y, Zhu Y, Niu J, Zhang B, Felson DT. Increasing prevalence of knee pain and symptomatic knee osteoarthritis: survey and cohort data. Ann Intern Med 2011; 155: Tanamas SK, Wluka AE, Berry, Menz HB, Strauss BJ, Davies-Tuck M, et al. Relationship between obesity and foot pain and its association with fat mass, fat distribution, and muscle mass. Arthritis Care Res (Hoboken) 2012;64: Chiu YH, Silman AJ, Macfarlane GJ, Ray D, Gupta A, Dickens C, et al. oor sleep and depression are independently associated with a reduced pain threshold: results of a population based study. ain 2005;115: Marazziti D, Castrogiovanni, Rossi A, Rosa C, Ghione S, Di Muro A, et al. ain threshold is reduced in depression. Int J Neuropsychopharmacol 1998;1: Hawker GA, Gignac MA, Badley E, Davis AM, French MR, Li Y, et al. A longitudinal study to explain the pain-depression link in older adults with osteoarthritis. Arthritis Care Res (Hoboken) 2011;63: