Background
Knee pain is one of the widespread, disturbing joint symptoms in the older population worldwide, and osteoarthritis (OA) is one of the most common causes of the symptoms. It has been estimated that more than 30% of the general population aged ≥ 50 years suffer from OA of the knee joint [
1,
2]. A recent report showed that disability-adjusted life years for knee OA reduced by 2.4% from 2006 to 2016 even after standardization for age, which is much more than the reduction from rheumatoid arthritis or low back and neck pain [
3]. Therefore, the pathophysiology and etiology of knee pain have attracted increasing attention and preventive measures have been vigorously pursued, especially in developed countries.
The prevalence of the disease has been estimated using mainly X-rays because of their availability and reliability worldwide. However, this requires an accessible X-ray device, radiological exposure, and a reliable evaluator. It is also well known that a difference exists between the degree of joint destruction judged by X-ray and the actual symptoms in the knee joint; there is a certain percentage of the population whose knees show radiographic OA but who have negligible symptoms, and vice versa. One report even showed that over a period of 20 years the percentage of people who had knee symptoms had increased despite a decrease in the percentage of those with radiographic knee OA [
4]. Therefore, it is essential to consider the symptom rather than the radiographic degree of joint destruction when preventive strategies are considered and planned.
The identification of prognostic and risk factors for progression of knee pain and/or clinical knee OA has attracted intensive study, and several such factors have been reported. A meta-analysis showed that age, ethnicity, body mass index (BMI), baseline OA severity, and joint effusion were strongly linked to progression [
5]. In addition, several reports have shown that poor mental health is associated with worsening of symptoms [
6]. Moreover, the relationship between pain in different parts of the body has also gradually gained attention. Especially, knee pain and low back pain are two of the most frequent, unanimous pain/disabilities in the elder population. It is highly conceivable that one can affect the other. However, the association between knee and low back pain has not been thoroughly investigated. The entire spectrum of risk factors for this association remains ambiguous.
To detect those at risk of knee pain, several scores and formulae have been proposed [
7‐
9]. However, few of these are applicable to people who do not yet have knee pain but who are likely to develop symptoms later. Awareness of factors that are applicable to these individuals is crucial for developing a formula to predict the development of knee pain.
Study objectives
The aims of this study were to investigate the association between knee and low back pain/disabilities and to develop a predictive score that enables the identification of those who are likely to develop new knee pain within a period of 5 years. We selected participants aged ≥ 50 years because it has been shown that the prevalence of knee OA increases dramatically over the age of 50 [
1,
2].
Discussion
This longitudinal study of 4638 participants (78.2% of the possible participants) in the general population showed that older age, female sex, higher BMI, weight increase, lower mental health score, and higher low back pain/disability score were significant risk factors for developing new knee pain in people aged ≥ 50 years. We developed a predictive score that showed that the risk of developing new knee pain within 5 years ranged from 11.0 to 63.2% depending on the total score. This is the first study to show the effect of low back pain/disability and other risk factors on the risk of developing new knee pain and to develop a reliable, easy-to-use predictive score.
In general, the association between knee and low back pain/disabilities has not been well studied. Muraki et al. reported that knee pain and low back pain were significantly associated with the magnitude of quality of life loss in 1369 women aged ≥ 40 years in the general population [
20]. However, they did not show a direct association between the two symptoms nor analyze the predictive value of low back pain. We previously reported that combined knee and low back pain additively strengthened the correlation with sleep problems, but a direct association between the two types of pain was not shown [
10]. The current study clearly illustrates this association, because in univariate analysis, the presence of low back pain/disability scored at just 1 point increased the risk of new knee pain 1.6 times, which was a greater effect than female sex or weight increase and a similar effect of higher BMI, three of the known risk factors. Furthermore, a previous report of musculoskeletal pain showed that knee pain had poorer outcomes compared with low back pain, indicating that it was a constant burden in the daily life of older people [
21].
One of possible pathophysiological mechanisms of this association is that osteoarthritic pathology can affect any joints or body parts in the older population, especially, load-bearing organs such as the knee and the lumbar vertebrae. From a clinical point of view, it is not exactly known how preceding low back pain/disability can predict new knee pain, but it is conceivable that one tends to affect the other by worsening the load-bearing burden of the other and/or by loosening the balance of the body when walking and even standing. Indeed, it was shown that the number of painful sites outside the knee, including low back pain, independently predicted knee cartilage volume loss without knee OA [
22,
23], which indicates the crucial association between musculoskeletal pain at different sites. Pain is one of the central issues in the management of knee OA [
24‐
26], and this association should be investigated in future studies.
Numerous reports have identified several risk factors for knee pain or knee OA. A meta-analysis showed strong evidence for age, ethnicity, BMI, co-morbidity count, joint effusion, and baseline severity as risk factors [
5]. The results of the present study support those of the meta-analysis, identifying age and BMI as strong risk factors. In contrast, smoking and alcohol consumption were not significant risk factors in our study, which is also consistent with previous reports [
12,
27]. We did not find any significant differences in risk between people with different levels of daily activity, although the current consensus would be that exercise and an active daily life contribute to reducing the possibility of knee OA [
26]. A possible explanation for this difference is that the current study used a simple question to evaluate activity with the responses sedentary, moderately active, or active, and could not define how each participant lived their daily life and how much they moved or exercised. More detailed collection of data may detect differences.
Several meta-analyses and reviews have shown that metabolic syndrome and dyslipidemia are risk factors for knee OA [
28‐
30], and a large-scale study has also shown that OA is a significant risk factor for cardiovascular diseases [
31]. Therefore, we decided to include the ABI, which is a reliable, objective measurement of peripheral artery disease [
32]. Contrary to our expectations, no significant association between the development of knee symptoms and ABI was apparent, possibly because ABI alone is not sufficient to predict new knee symptoms, or because knee symptoms, unlike radiographic OA, may not be directly related to vascular manifestations such as arteriosclerosis. In addition, a previous report showed that hsCRP was strongly associated with all definitions of radiographic OA [
33]. However, that study also showed that the association was not independent of BMI, and our data support the notion that hsCRP is not an independent risk factor for new knee symptoms. Similarly, a previous study showed that the radiographic features of OA are associated with bone mineral density of the lumbar spine and femoral neck [
34]. However, NTX and CTX, two reliable biomarkers of osteoporosis, failed to predict new knee symptoms. It is reasonable to assume that these current osteoporosis biomarkers alone, or possibly osteoporosis itself, are insufficient to predict the development of knee symptoms. It may be necessary to collect more detailed information about osteoporosis to identify any contribution to the development of knee symptoms.
The production of models to predict the development of radiographic knee OA or knee symptoms has been vigorously pursued. However, most previous studies report only the odds ratios of certain risk factors or the results of statistical models such as Cox proportional hazards models, and the process for selecting people at risk remains ambiguous. Kerkhof et al. reported a predictive model for knee OA incidence including clinical, genetic, and biochemical risk factors [
8]. However, gene analysis requires reliable access to a competent analytical department and is not suitable for screening of people at risk. Zhang et al. reported a simple predictive score using age, BMI, and scores defined relative to an index person [
7], but that score also requires analysis of knee radiography, which necessitates radiological exposure and a reliable evaluator. Fernandes et al. recently reported a useful, simple predictive model using only self-reported predictors without any imaging studies or laboratory data [
9]; however, their calculation is rather complex and requires certain stratagems to obtain an individual risk. The current study shows that a self-reported score without any invasive tests can be sufficient to select people at risk with a desirable probability. The actual potential of the developed score should be verified in the future.
Limitations of the study
Nevertheless, this study involves some unavoidable limitations. First, the origin of knee symptoms was not confirmed by any method. Knee symptoms may be confused with lower leg pain originating from back ailments, although the JKOM questionnaire is designed to elicit knee-specific symptoms. Second, we did not collect any imaging data, which would increase the reliability of the score. However, the purpose of the current study was to develop a reliable score that did not require any invasive measurements, and any imaging studies should be used in a different setting. Third, our analyses were performed with a predetermined set of data. We cannot exclude the existence of other factors that could contribute to the prediction of new knee symptoms, including injury history, educational level, and metabolic syndrome. Fourth, the threshold of the score determining the need for appropriate intervention is unclear, and this should be extensively studied in a future longitudinal, proof-of-concept study. Finally and importantly, prediction score constructed by regression analysis tends to overfit the data which the score was derived from, especially when using stepwise selection. Therefore, the performance of our score needs to be confirmed by external patient data.
Acknowledgements
We are grateful to Dr. Yoshihiko Kotoura for his tremendous help regarding clinical measurements, Nagahama City Office and the Zeroji Club, a non-profit organization, for their assistance in conducting this study. We thank Drs. T. Tsuboyama, T. Ikezoe, N. Ichihashi, S. Kuriyama, S. Nakamura, and N. Taniguchi (Kyoto University Graduate School of Medicine) for their valuable technical assistance and thoughtful discussion.
The Nagahama Study group executive committee is composed of the following individuals: Yasuharu Tabara, Takahisa Kawaguchi, Kazuya Setoh, Yoshimitsu Takahashi, Shinji Kosugi, Takeo Nakayama, and Fumihiko Matsuda from Center for Genomic Medicine, Kyoto University Graduate School of Medicine (Ya.T, T.K., K.S., F.M.); Department of Health Informatics (Yo.T, T.N.), Department of Medical Ethics and Medical Genetics (S.K.), Kyoto University School of Public Health.