To describe the association between chondral defects, bone marrow lesions, knee and hip radiographic osteoarthritis (OA), and knee pain.
To describe the association between chondral defects, bone marrow lesions, knee and hip radiographic osteoarthritis (OA), and knee pain.
Knee pain was assessed by the Western Ontario and McMaster Universities Osteoarthritis Index. T1- and T2-weighted fat saturation magnetic resonance imaging was performed on the right knee to assess chondral defects and subchondral bone marrow lesions. Radiography was performed on the right knee and hip and scored for radiographic OA. Body mass index (BMI) and knee extension strength were measured.
A total of 500 randomly selected men and women participated. The prevalence of knee pain was 48%. In multivariable analysis, prevalent knee pain was significantly associated with medial tibial chondral defects (odds ratio [OR] 2.32, 95% confidence interval [95% CI] 1.02–5.28 for grade 3 versus grade 2 or less; OR 4.93, 95% CI 1.07–22.7 for grade 4 versus grade 2 or less), bone marrow lesions (OR 1.44, 95% CI 1.04–2.00 per compartment), and hip joint space narrowing (OR 1.36, 95% CI 1.07–1.73 per unit), as well as greater BMI and lower knee extension strength. It was not significantly associated with radiographic knee OA. These variables were also associated with more severe knee pain. In addition, there was a dose response association between knee pain and number of sites having grade 3 or 4 chondral defects (OR 1.39, 95% CI 1.12–1.73 per site), with all subjects having knee pain if all compartments of the knee had these defects.
Knee pain in older adults is independently associated with both full and non–full-thickness medial tibial chondral defects, bone marrow lesions, greater BMI, and lower knee extension strength, but is not associated with radiographic knee OA. The association between radiographic hip OA and knee pain indicates that referred pain from the hip needs to be considered in unexplained knee pain.
Knee pain is an important clinical symptom and the major determinant of knee arthroplasty (1). Although the reported prevalence of knee pain varies according to case definition and age profile of subjects, it clearly increases with age (2–5) and will inevitably grow as the proportion of older persons in the population increases (5). However, the causes of knee pain remain uncertain. The correlation between radiographic osteoarthritis (OA) and pain is significant but modest (6–9); osteophytes are most consistently associated with knee pain (10–12), but reports on the association of joint space narrowing (JSN) with pain are inconsistent (10, 11, 13). However, JSN only indirectly assesses cartilage morphology, and may underestimate the importance of cartilage damage. Furthermore, the radiographic joint space consists not only of articular cartilage, but also other soft tissues such as menisci (14). Normal hyaline cartilage does not possess pain fibers, suggesting that articular cartilage cannot be the origin of knee pain. However, substance P nociceptive fibers have been found in abnormal cartilage such as erosion channels in horse OA (15), and superinduction of cyclooxygenase 2 (COX-2) and prostaglandins has been observed in OA-affected cartilage explants (16), suggesting that articular cartilage may directly produce pain.
In a study using magnetic resonance imaging (MRI), researchers found that subjects with full-thickness articular cartilage defects accompanied by adjacent subchondral cortical bone defects are more likely to have pain in the presence of knee OA (17). Recently, we reported that non–full-thickness chondral defects at distal femoral and patellar sites were significantly associated with self-reported knee pain in younger subjects (18). To date, there have been no data reported for older groups.
In addition, knee pain can also originate from other sources such as the synovial membrane, joint capsule, periarticular ligaments or muscle, periosteum, and subchondral bone because nociceptive fibers are present in these structures (19). This is evident from recent reports of significant association between knee pain and knee effusions, popliteal cysts, and synovial thickening (20). Subchondral bone marrow lesions have been reported to have an association with knee pain in individuals with radiographic knee OA (21). However, it is unclear whether this association is independent of cartilage damage, and whether it is relevant in a non-OA population. Also, more than half of individuals who report hip pain also report knee pain (2), implying that either pathology at both sites or unexplained knee pain may be referred from hip OA. This has not been formally evaluated.
The aim of this cross-sectional study was to investigate the association between knee pain and chondral defects, subchondral bone marrow lesions, and radiographic knee and hip OA in older male and female subjects.
This study was conducted as part of the Tasmanian Older Adult Cohort Study, an ongoing, prospective, population-based study aimed at identifying the environmental, genetic, and biochemical factors associated with the development and progression of OA at multiple sites (hand, knee, hip, and spine). The study was approved by the Southern Tasmanian Health and Medical Human Research Ethics Committee, and written informed consent was obtained from all participants. Subjects between ages 50 and 79 years were selected randomly from the roll of electors in southern Tasmania (population 229,000), a comprehensive population listing, with an equal number of men and women. Subjects were excluded if they had contraindication for MRI, e.g., metal sutures, presence of shrapnel, iron filling in the eye, and claustrophobia. Institutionalized persons were also excluded. A stratified simple random sampling with replacement was utilized and subjects were selected in batches of 60 (30 men and 30 women). The current study consisted of the first 500 participants who completed the interview, MRI scans, and radiography by October 2003. A total of 99% of participants were white.
Knee pain was assessed by the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), a self-administered questionnaire (22). Subjects completed and returned the questionnaire on the day the interview and radiography were conducted. MRI scan was performed after this. Five categories of pain (walking on flat surface, going up/down stairs, pain at night, sitting/lying, and standing upright) were assessed separately with a 10-point scale from 0 (no pain) to 9 (most severe pain). Each score was then summed to create a total pain score (range 0–45). With no a priori reason to categorize pain, prevalent knee pain was defined as a total score >1. Assessment of knee pain was not knee specific.
Height was measured to the nearest 0.1 cm (with shoes, socks, and headgear removed) using a stadiometer. Weight was measured to the nearest 0.1 kg (with shoes, socks, and bulky clothing removed) using a single pair of electronic scales (Seca Delta Model 707, Bradford, MA) that were calibrated using a known weight at the beginning of each clinic. Body mass index (BMI; kg/m2) was calculated as weight/height2. Knee extension strength in the right leg was measured by a pocket balance (Stamina, Munich, Germany). Subjects were instructed in the technique prior to testing. There were 2 attempts and the greatest force was recorded.
An MRI scan of the right knee was performed. Knees were imaged in the sagittal plane on a 1.5T whole-body magnetic resonance unit (Picker, Cleveland, OH) using a commercial transmit-receive extremity coil. The following image sequence was used: 1) a T1-weighted fat saturation 3-dimensional gradient recall acquisition in the steady state, flip angle 30 degrees, repetition time 31 msec, echo time 6.71 msec, field of view 16 cm, 60 partitions, 512 × 512–pixel matrix, acquisition time 5 minutes 58 seconds, 1 acquisition; sagittal images were obtained at a partition thickness of 1.5 mm without between-slice gap; and 2) a T2-weighted fat saturation 2-dimensional fast spin echo, flip angle 90 degrees, repetition time 3,067 msec, echo time 112 msec, field of view 16 cm/15 partitions, 228 × 256–pixel matrix; sagittal images were obtained at a partition thickness of 4 mm with a between-slices gap of 0.5–1.0 mm.
Chondral defects were assessed on the T1-weighted MR images and scored with a modification of a previous classification system (23) at medial tibial, medial femoral, lateral tibial, lateral femoral, and patellar sites as follows: grade 0 = normal cartilage; grade 1 = focal blistering and intracartilaginous low-signal intensity area with an intact surface; grade 2 = irregularities on the surface or basal layer and loss of thickness <50%; grade 3 = deep ulceration with loss of thickness >50%; and grade 4 = full-thickness chondral wear with exposure of subchondral bone. We found that cartilage surface in some images was still regular but cartilage adjacent to subchondral bone became irregular, so we included these changes in the classification system. A cartilage defect also had to be present in at least 2 consecutive slices. The cartilage was considered to be normal if the band of intermediate signal intensity had a uniform thickness. The highest score was used if >1 defect was present on the same site. One observer (GZ) scored the MRI blinded to knee pain score. Intraobserver repeatability measured by intraclass correlation coefficient (ICC) was assessed in 50 subjects with an interval of at least 1 week between the 2 measurements. ICCs were 0.92, 0.80, 0.93, 0.95, and 0.94 for the chondral defects score at the medial femur, lateral femur, medial tibia, lateral tibia, and patella, respectively.
Subchondral bone marrow lesions were assessed on the T2-weighted MR images and defined as discrete areas of increased signal adjacent to the subcortical bone at the lateral, medial femur and/or tibia. Each bone marrow lesion was scored on the basis of lesion size, e.g., a lesion was scored as grade 1 if it was only present on 1 slice, grade 2 if present on 2 consecutive slices, or grade 3 if present on >3 consecutive slices. The highest score was used if >1 lesion were present on the same site. Prevalent bone marrow lesions were defined as a total score >1. One observer (GZ) scored the bone marrow lesions blinded to knee pain score. Intraobserver repeatability was assessed in 50 subjects with at least a 1-week interval between 2 readings with ICCs of 0.89, 0.96, 0.94, and 1.00 for the bone marrow lesions scores at lateral tibia, lateral femur, medial tibia, and medial femur, respectively.
A standing anteroposterior semiflexed view of the right knee was performed in all subjects. Radiographs were then assessed using the atlas developed by Altman et al (24). Each of the following was assessed: medial JSN (score range 0–3), lateral JSN (score range 0–3), medial femoral osteophytes (score range 0–3), medial tibial osteophytes (score range 0–3), lateral femoral osteophytes (score range 0–3), lateral tibial osteophytes (score range 0–3), medial femoral sclerosis (score range 0–3), medial tibial sclerosis (score range 0–3), lateral femoral sclerosis (score range 0–3), and lateral tibial sclerosis (score range 0–3). Each score was determined by consensus of 2 readers (VS and HC) who simultaneously assessed the radiograph with immediate reference to the atlas and were blinded to the subject's status of knee pain. Intraobserver repeatability was assessed in 40 subjects with an ICC of 0.65–0.85.
Weight-bearing anteroposterior pelvic radiographs with both feet in 10° internal rotation were obtained and then assessed in the same manner. Radiographic features of axial JSN; superior JSN; and osteophytes, sclerosis, and lucency at the acetabulum and femoral head were graded on a 4-point scale (range 0–3), where 0 indicates no disease and 3 indicates most severe disease. Each score was determined by consensus of 2 readers (VS and HC) who simultaneously assessed the radiograph with immediate reference to the atlas, blinded to the subject's status of knee pain. Intraobserver repeatability was assessed in 40 subjects with ICCs of 0.60–0.87.
Comparisons between subjects with and without knee pain were made by unpaired t-test, Mann-Whitney U test, or chi-square test (as appropriate). Preliminary analysis suggested there was no difference in prevalence of knee pain between subjects with lower chondral defect scores (grade 0, 1, and 2); therefore, these were combined into one group for further analysis. No dose response association between bone marrow lesions at each site and prevalent knee pain was detected, and the number of sites with any bone marrow lesions was used for the prevalent pain analysis. With the WOMAC pain score dichotomized as 0 (score 0) or 1 (score ≥1), logistic regression modeling was utilized to estimate the prevalence odds of reported knee pain and study factors.
For the analysis of the association between pain severity and study factors, 2 approaches were utilized. First, subjects with more severe pain were identified and defined as having a WOMAC pain score ≥4, which was the median of the total WOMAC pain score in subjects with a score ≥1. The comparison was then made between subjects with more severe pain and those without pain by logistic regression modeling. Second, excluding the subjects with a pain score of 0, linear regression modeling was used to estimate the associations between the zero-skewness logarithmic transformation of the total pain score and the same study factors. Excluding the 261 (52%) of 500 subjects without reported pain (score = 0) was necessary because the residuals were heavily skewed. The approach was conservative, but there was less risk of Type I error. We considered but rejected the alternative of categorizing the pain scores and analyzing them as an ordered categorical measure. We had 2 reasons for rejecting this approach: first, there is no consensus in the literature on how to categorize the WOMAC scores, and second, we found that the models we might have used did not fit the categorized data particularly well.
Stata's fracpoly procedure (Stata, College Station, TX) was utilized in each type of modeling to check the appropriate scale of covariates. The predictor for a study factor was expressed on a linear scale only if no nonlinear transform significantly improved model fit. A P value less than 0.05 (2-tailed) or a 95% confidence interval not including the null point was considered statistically significant. All statistical analyses were performed on Intercooled Stata 8.2 for windows (Stata).
A total of 500 subjects (248 men, 252 women) with a mean age of 63 years were included in this study. The characteristics of the study population are presented in Table 1. The prevalence of knee pain was 48%. Most subjects with knee pain reported mild pain, with 88% having a WOMAC total pain score <8 (possible score range 0–45). Women were more likely to report knee pain than men. There was a significant difference in weight, BMI, and knee extension strength between subjects with and without knee pain. The prevalence of grade 2 or higher chondral defects was higher at all sites except for medial tibia in subjects with knee pain compared with those without knee pain, but this difference was small and not statistically significant. However, the difference was more pronounced for severe chondral defects (defined as grade 3 or higher) and was statistically significant except for the lateral femoral site. Prevalence of bone marrow lesions, knee JSN and osteophytes, and hip JSN was also significantly higher in subjects with knee pain.
|Characteristic||No knee pain n = 261||Knee pain n = 239||P|
|Age, mean ± SD years||63.0 ± 7.4||62.7 ± 7.1||0.70|
|Height, mean ± SD cm||167.8 ± 9.0||166.5 ± 9.0||0.10|
|Weight, mean ± SD kg||75.9 ± 13.6||79.8 ± 16.1||< 0.01|
|BMI, mean ± SD kg/m2||26.9 ± 4.1||28.7 ± 5.3||< 0.001|
|Knee extension strength, mean ± SD kg||32.4 ± 10.6||28.4 ± 11.6||< 0.001|
|Total chondral defect score (possible range||8.49 ± 3.1||9.36± 3.9||0.01|
|0–20), mean ± SD (range)||(1–17)||(1–20)|
|Any lateral femoral chondral defect†||43||45||0.61|
|Any medial femoral chondral defect†||73||74||0.91|
|Any lateral tibial chondral defect†||61||68||0.09|
|Any medial tibial chondral defect†||83||83||0.87|
|Any patellar chondral defect†||58||62||0.35|
|Severe lateral femoral chondral defect‡||4||7||0.11|
|Severe medial femoral chondral defect‡||12||21||< 0.01|
|Severe lateral tibial chondral defect‡||30||42||< 0.01|
|Severe medial tibial chondral defect‡||7||18||< 0.001|
|Severe patellar chondral defect‡||37||50||< 0.01|
|Any bone marrow lesion§||28||41||< 0.01|
|Total radiographic knee OA score (possible||0.9 ± 1.3||1.7 ± 2.3||< 0.001|
|range 0–30), mean ± SD (range)||(0–10)||(0–14)|
|Any knee JSN§||53||62||0.05|
|Any knee osteophyte§||6||13||< 0.01|
|Any knee sclerosis§||6||7||0.49|
|Total radiographic hip OA score (possible||0.8 ± 1.3||1.2 ± 1.8||< 0.01|
|range 0–24), mean ± SD (range)||(0–8)||(0–11)|
|Any hip JSN§||28||42||0.001|
|Any hip osteophyte§||17||18||0.73|
|Any hip sclerosis§||2||2||0.65|
There was a significant increase in the prevalence of knee pain with increasing chondral defects from grade 2 or less up to grade 4 at all knee sites with the exception of the lateral tibial site (Figure 1). The results of multivariable analysis of the association between prevalence odds of knee pain and study factors are presented in Table 2. Knee pain was statistically significantly and independently associated with greater BMI, lower knee extension strength, number of sites with bone marrow lesions, medial tibial chondral defects, and hip JSN. These significant associations persisted after further adjustment for knee osteophytes, which was not statistically significant in the final model (P = 0.51). Age was borderline significant and negatively associated with prevalent knee pain. Knee JSN was not significantly associated with prevalent knee pain (P = 0.07) after adjustment for age, sex, BMI, knee extension strength, bone marrow lesions, hip JSN, and knee osteophytes.
|Factors||Step 1†||Step 2‡|
|Age, years||0.97 (0.94–1.00)||0.97 (0.94–1.00)|
|Sex, female versus male||0.82 (0.49–1.38)||0.84 (0.50–1.42)|
|BMI, kg/m2||1.08 (1.03–1.14)||1.08 (1.03–1.13)|
|Knee extension strength, kg||0.96 (0.94–0.98)||0.96 (0.94–0.98)|
|Bone marrow lesion (per site)||1.45 (1.05–2.01)||1.44 (1.04–2.00)|
|Lateral femoral chondral defects|
|Grade 3 versus grade 2 or less||0.92 (0.31–2.72)||0.90 (0.30–2.69)|
|Grade 4 versus grade 2 or less||1.81 (0.25–13.32)||1.42 (0.16–12.60)|
|Medial femoral chondral defects|
|Grade 3 versus grade 2 or less||1.27 (0.67–2.39)||1.24 (0.66–2.36)|
|Grade 4 versus grade 2 or less||0.60 (0.14–2.53)||0.56 (0.13–2.39)|
|Lateral tibial chondral defects|
|Grade 3 versus grade 2 or less||1.61 (0.96–2.73)||1.64 (0.97–2.76)|
|Grade 4 versus grade 2 or less||0.87 (0.46–1.62)||0.84 (0.45–1.59)|
|Medial tibial chondral defects|
|Grade 3 versus grade 2 or less||2.36 (1.05–5.34)||2.32 (1.02–5.28)|
|Grade 4 versus grade 2 or less||5.45 (1.22–24.34)||4.93 (1.07–22.74)|
|Patellar chondral defects|
|Grade 3 versus grade 2 or less||1.24 (0.69–2.25)||1.25 (0.69–2.27)|
|Grade 4 versus grade 2 or less||1.52 (0.94–2.44)||1.53 (0.95–2.46)|
|Hip JSN (per grade)||1.35 (1.06–1.71)||1.36 (1.07–1.73)|
Figure 2 documents a significant association between prevalent knee pain and the number of compartments with grade 3 or higher chondral defects. The prevalence of knee pain increased with increasing numbers of compartments with defects; all subjects had pain if all 5 compartments had these defects. Similarly, prevalence of knee pain increased markedly with increasing hip JSN total score (Figure 3).
The results of multivariable analysis of the association between the study factors and more severe knee pain are presented in Table 3. Similar to prevalent knee pain, more severe knee pain was statistically significantly and independently associated with BMI, knee extension strength, number of sites with bone marrow lesions, medial tibial chondral defects, and hip JSN. These significant associations persisted even after further adjustment for knee osteophytes, with the exception being medial tibial chondral defects whose association with more severe knee pain became borderline (P = 0.09). Knee osteophytes was not statistically significant in the final model (P = 0.24).
|Factor||Step 1||Step 2|
|Age, years||0.97 (0.94–1.01)||0.97 (0.94–1.01)|
|Sex, female versus male||0.62 (0.33–1.15)||0.65 (0.35–1.22)|
|BMI, kg/m2||1.14 (1.08–1.20)||1.14 (1.08–1.20)|
|Knee extension strength, kg||0.95 (0.92–0.97)||0.95 (0.92–0.98)|
|Bone marrow lesion, compartment||1.72 (1.17–2.52)||1.66 (1.12–2.45)|
|Lateral femoral chondral defects, grade||1.41 (0.62–3.21)||1.14 (0.45–2.90)|
|Medial femoral chondral defects, grade||0.97 (0.52–1.80)||0.95 (0.51–1.77)|
|Lateral tibial chondral defects, grade||0.94 (0.66–1.32)||0.91 (0.64–1.30)|
|Medial tibial chondral defects, grade||2.04 (1.06–3.95)||1.82 (0.91–3.65)|
|Patellar chondral defects, grade||1.11 (0.83–1.48)||1.12 (0.84–1.49)|
|Hip JSN, grade||1.36 (1.08–1.71)||1.38 (1.09–1.74)|
In linear regression analyses that excluded subjects with a WOMAC pain score of 0, severity of knee pain was significantly and independently associated with greater BMI and hip JSN, with 5.2% and 2.5% of the variation in the WOMAC pain score explained by BMI and hip JSN, respectively. The associations for knee extension strength and medial tibial chondral defects were in the direction expected from the prevalence odds analysis; however, with 52% of subjects excluded from this analysis, none of these associations were statistically significant.
This study suggests that both prevalent and more severe knee pain in older adults are independently associated with full and non–full-thickness chondral defects at the medial tibial plateau, bone marrow lesions, hip JSN, greater BMI, and lower knee extension strength, but not with osteophytes. Consistent with our previous report in younger subjects (18), we found a significant and independent association between prevalent knee pain and chondral defects in this older adult sample. However, in contrast to our previous study in which we demonstrated that grade 2 defects were associated with increasing prevalence of knee pain, we only detected the association for more severe chondral defects. This may be due to the low prevalence of grade 0 and 1 chondral defects in this sample compared with younger age groups. Indeed, the prevalence of grade 2 chondral defects was higher at all sites in subjects with knee pain than in those without pain, although the difference was not statistically significant. This association was most marked at the medial tibial plateau, which again contrasts with our previous findings of femoral and patellar sites in younger subjects, suggesting possible site specificity between younger and older age groups. Furthermore, we demonstrated that medial tibial chondral defects were also significantly associated with more severe pain, although this significance became weak after adjustment for knee osteophytes. This is most likely due to the sample size reduction in the analysis because the odds ratios were similar in magnitude in both forms of analysis. Alternatively, a possible threshold effect of chondral defects on knee pain may occur. Importantly, we also demonstrated an additive association between the number of sites with chondral defects and knee pain, indicating the importance of chondral defects at all sites independent of radiographic knee OA and other factors we measured. These findings extend those of a previous report in which only subjects with full-thickness articular cartilage defects accompanied by adjacent subchondral cortical bone defects were more likely to have pain in the presence of knee OA (17). The apparent discrepancy between our results and those of the previous study may be due to sample size considerations and/or having only female subjects that were younger than our study participants (17). However, the variation in results between studies indicates the need for further studies. Severe chondral defects were more prevalent in the lateral than medial tibial compartment in our sample, which is in direct contrast to the distribution of OA. The reason for this is currently unclear.
The mechanism for an association between knee pain and chondral defects remains unknown. Loss of articular cartilage leads to a decrease in the protection of the underlying bone and an increase in physical stresses transmitted to the subchondral bone, resulting in subchondral bone structure changes such as subchondral bone sclerosis and bone marrow lesions, which may cause knee pain. However, the association was independent of radiographic knee OA and bone marrow lesions, suggesting that damaged articular cartilage can directly lead to pain (15, 16). Substance P, which is involved in transmission of pain, has been found in abnormal cartilage such as eroded channels in OA-affected metacarpophalangeal articulations in horses (15), and human OA-affected cartilage in ex vivo conditions shows superinduction of prostaglandins, which amplify pain signals, due to upregulation of COX-2 (16), supporting the latter hypothesis.
Bone marrow lesions were common in this random sample and comparable with previous studies, which is surprising given the much lower prevalence of radiographic OA (17, 21). The presence of bone marrow lesions was strongly associated with prevalent knee pain as well as more severe pain, consistent with and expanding those previous reports (17, 21). In addition, we documented an additive effect of knee compartments with presence of bone marrow lesions on knee pain, indicating the importance of bone marrow lesions in all compartments. Furthermore, we demonstrated that the strong association between bone marrow lesions and knee pain was independent of chondral defects and radiographic knee OA, expanding the findings of reports in which the association was confined to subjects affected with OA (21) and to those with full-thickness chondral defects (17), and directly linking bone marrow lesions to pain even though the underlying histopathology remains uncertain.
A modest but significant correlation between radiographic knee OA and symptoms has been reported previously (6–9). Interestingly, in the current study, the significant association between knee pain and radiographic knee OA including JSN and osteophytes became nonsignificant after adjustment for other factors, including chondral defects, bone marrow lesions, and radiographic hip OA, suggesting the correlation is mediated by other factors. Therefore, these factors may be more important for knee pain.
A recent report (2) of a strong coexistence of knee and hip pain suggests that either pathology at both sites or unexplained knee pain may be referred from hip OA. In this study, we demonstrated a strong association between prevalence and severity of knee pain and radiographic hip OA, particularly with JSN, which is postulated as the best index for the presence of radiographic hip OA (25). Given the cross-sectional nature of our data, we cannot comment on a causal relationship between radiographic hip OA and knee pain. However, the relationship is biologically plausible and it is unlikely that the association is mediated by unmeasured factors in the knee such as effusions or synovitis. Furthermore, the significance persisted after adjustment for other factors, including radiographic knee OA, and was of a dose response nature, suggesting that a substantial component of unexplained knee pain is referred from hip OA as has long been recognized in clinical practice.
Similar to other reports (26, 27), we also demonstrated a strong association between greater BMI and knee pain. The prevalence and severity of knee pain increases with increasing BMI. The reason for this association remains elusive, but it is most likely due to repetitive application of increased axial loading at the knee joint (28). In this study, the association was independent of other factors. Similarly, we demonstrated a strong negative association between knee pain and knee extension strength, consistent with other studies (29–31).
There are a number of potential limitations to the current study. First, the reported prevalence of knee pain varies with case definitions, the composition of the study samples, and the methods used (1–3, 12). We chose a conservative definition of knee pain, and this contributed to the high prevalence of knee pain in this sample. There are no other comparative Australian prevalence studies with which to compare our results, but we also had a high prevalence in a younger sample (18). Alternatively, given the response rate of 57%, it is possible that there was selection bias in the sample as compared with the general population. Subjects with arthritis or pain may be more likely to participate, which may contribute to the high prevalence of knee pain and other study factors. In addition, more healthy subjects may participate. However, neither of these is likely to affect the associations between the factors we measured and pain because subjects were unaware of their results at the time of completing the questionnaire and for the reasons outlined by Miettinen (32).
Second, misclassification in the assessment of MRI indices is possible, but we had high reproducibility of the assessment techniques, suggesting that this is not a major concern. Third, because chondral defects and bone marrow lesions were assessed on different MRI images with different slice thicknesses, it is difficult to assess whether those bone marrow lesions were adjacent to the chondral defects. Against this was the observation that the significant associations between both these factors and pain were independent. Fourth, the reproducibility for radiographs was good rather than excellent, which may contribute to a weakening of associations. Fifth, administration of the WOMAC was not knee specific. This may not be that important because persons reporting knee pain are more likely to have bilateral knee pain (2). However, the most likely effect of this is a weakening of associations because there is likely to be an imperfect correlation between structural change in both knees; therefore, the results may actually be stronger than we reported. Sixth, the findings might be mediated by factors not assessed in the study such as synovitis and malalignment, which could be associated with both knee pain and chondral defects. However, it remains unclear whether synovitis and malalignment are causes of chondral defects or vice versa. Furthermore, the dose response relationships we reported in the current study suggest that this is less likely. Nevertheless, independent studies are needed to confirm the findings. Lastly, the study is cross-sectional in design and any causal relationship should be corroborated in future longitudinal studies.
In conclusion, our results suggest that knee pain is independently associated with both full and non–full-thickness medial tibial chondral defects, bone marrow lesions, greater BMI, and lower knee extension strength, but not with radiographic knee OA, expanding our understanding of knee pain in older adults. Furthermore, the likely biologic association between knee pain and hip JSN indicates that referred pain from the hip needs to be considered in unexplained knee pain.
A special thanks goes to the subjects who made this study possible. The role of Catrina Boon and Pip Boon in collecting the data is gratefully acknowledged. We would like to thank Dr. Stephen Quinn for statistical support and Mr. Martin Rush who performed the MRI scans, and Ms. Lesa Hornsey who conducted the x-ray measures.