Abstract
Objective. We explored whether knee pain or a history of knee injury were associated with a knee injury in the following 12 months.
Methods. We conducted longitudinal knee-based analyses among knees in the Osteoarthritis Initiative. We included both knees of all participants who had at least 1 followup visit with complete data. Our first sets of exposures were knee pain (chronic knee symptoms and severity) at baseline, 12-month, 24-month, and 36-month visits. Another exposure was a history of injury that we defined as a self-reported injury at any time prior to baseline, 12-month, 24-month, or 36-month visit. The outcome was self-reported knee injury during the past year at 12-month, 24-month, 36-month, and 48-month visits. We evaluated the association between ipsilateral and contralateral knee pain or history of injury and a new knee injury within 12 months of the exposure using generalized linear mixed model for repeated binary outcomes.
Results. A knee with reported chronic knee symptoms or ipsilateral or contralateral history of an injury was more likely to experience a new knee injury in the following 12 months than a knee without chronic knee symptoms (OR 1.84, 95% CI 1.57–2.16) or prior injury (prior ipsilateral knee injury: OR 1.81, 95% CI 1.56–2.09. Prior contralateral knee injury: OR 1.43, 95% CI 1.23–1.66).
Conclusion. Knee pain and a history of injury are associated with new knee injuries. It may be beneficial for individuals with knee pain or a history of injury to participate in injury prevention programs.
Individuals with a history of knee injury are 3 to 6 times more likely to develop knee osteoarthritis (KOA)1,2 and are on average diagnosed with osteoarthritis (OA) about 10 years younger than individuals without a history of knee injury3. Knee injuries are also strongly associated with an accelerated form of KOA4. Understanding risk factors for knee injury may provide important insight into OA prevention. Despite extensive research assessing risk factors for knee injuries among young physically active individuals (under 35 yrs of age)5,6,7, it remains unclear which factors may predispose older adults with, or at risk for, KOA to experience a knee injury.
Individuals with knee pain are at greater risk of falls8,9,10,11, but it is unclear whether they are also at risk of knee injuries. Knee pain is associated with altered proprioception12 that may increase the risk of falls9,13. Hence, knee pain may be a risk factor for knee injuries among older adults.
A history of knee injury to the ipsilateral or contralateral knee may also be risk factors for new knee injuries14,15,16. Specifically, young physically active individuals with an anterior cruciate ligament (ACL) tear are more than 5 times as likely to sustain another ACL tear in either knee15. Despite a history of knee injury being one of the most important risk factors for knee injuries in young physically active individuals17,18, there is little evidence to determine whether it is also an important risk factor among older and more sedentary adults19,20.
The aim of our study was to assess whether knee pain or a history of knee injury increases the risk for a knee injury in the subsequent 12 months among older adults with, or at risk for, KOA. We hypothesized that chronic knee symptoms and greater pain severity predict injuries among individuals with, or at risk for, KOA. Further, we hypothesized that a history of ipsilateral or contralateral knee injury was associated with new injuries.
MATERIALS AND METHODS
To test for an association of knee pain and history of knee injury, with subsequent knee injuries, we conducted longitudinal knee-based analyses using data from the Osteoarthritis Initiative (OAI). We focused on knee-based analyses to test our hypotheses that knee pain or histories of knee injury in the ipsilateral or contralateral knee were important risk factors for new knee injuries. The OAI is a longitudinal observational study of KOA that occurred at 4 clinical sites in the United States: Memorial Hospital of Rhode Island, Ohio State University, the University of Maryland/Johns Hopkins University, and the University of Pittsburgh. The staff at the 4 clinical sites enrolled 4796 men and women (45 to 79 yrs of age) between February 2004 and May 2006. The OAI cohort was composed of 3 groups classified at baseline: (1) progression subcohort, including individuals with at least 1 knee with symptomatic radiographic KOA; (2) incidence subcohort, including individuals at risk for symptomatic radiographic KOA; and (3) nonexposed control subcohort that included individuals with no KOA and no risk factors for KOA. Detailed descriptions of the eligibility criteria for each subcohort and the OAI protocol are available at the OAI Website21. Institutional review boards at each OAI clinical site and the OAI coordinating center (University of California, San Francisco, USA) approved the OAI study.
In our study, we included all knees among OAI participants who attended the baseline visit and at least 1 followup visit, whether or not they had KOA. Our first sets of exposures were knee pain (chronic knee symptoms and severity) at baseline, 12-month, 24-month, and 36-month visits. Another exposure was a history of injury that we defined as a self-reported injury at any time prior to baseline, 12-month, 24-month, or 36-month visit. The outcome was self-reported knee injuries that occurred during the subsequent year — assessed at 12-month, 24-month, 36-month, and 48-month OAI visits.
Knee pain
To determine whether knee pain was associated with subsequent knee injuries, we defined pain based on the presence of chronic knee symptoms or pain severity. At each OAI visit, participants answered “During the past 12 months, have you had any pain, aching, or stiffness in or around your right knee?” If the participant answered “yes”, then they were asked “During the past 12 months, have you had pain, aching, or stiffness in or around your right knee on most days for at least one month? By most days, we mean more than half the days of a month.” The same questions were asked for the left knee. We defined chronic knee symptoms as pain on most days of a month in the past 12 months22,23.
Knee pain severity was based on the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain subscale, a validated outcome measure for KOA24. The WOMAC pain subscale has 5 questions about pain in the last 7 days with walking, stairs, lying in bed, sitting or lying down, and standing. The WOMAC pain subscale was knee-specific and we used data from both knees. Each item is scored 0 (none) to 4 (extreme pain), and the subscale ranges from 0 to 20. However, across the annual OAI visits, 95% of knees had WOMAC pain scores between 0 and 9 and > 50% had a WOMAC pain score 0 or 1. Further, we detected a nonlinear association between WOMAC pain and injury. Because of the skewed pain scores and nonlinear association, we dichotomized knee severity for our primary analyses as no to little pain (0 to 2) or knee pain (≥ 3). The knee pain data are publicly available [Files: allclinical00 (version 0.2.2), allclinical01 (version 1.2.1), allclinical03 (3.2.1), allclinical05 (5.2.1)]21.
History of knee injuries
To determine whether a history of joint-specific knee injury was associated with subsequent knee injuries, we defined a history of knee injury based on self-reported data from the baseline, 12-, 24-, and 36-month OAI visits. At the OAI baseline visit, participants answered, “Have you ever injured your right knee badly enough to limit your ability to walk for at least two days?” At each followup visit, participants were asked “Since your last annual visit to the OAI clinic about 12 months ago, have you injured your right knee badly enough to limit your ability to walk for at least two days?” Similar questions were asked for the left knee. The injury data are publicly available [Files: allclinical00 (version 0.2.2) allclinical01 (version 1.2.1), allclinical03 (3.2.1), allclinical05 (5.2.1)]21.
New knee injuries
Our main outcome was self-reported knee injury at an OAI followup visit. This was defined based on the followup visit self-reported questions listed in the previous section. These data are publicly available [Files: allclinical01 (version 1.2.1), allclinical03 (3.2.1), allclinical05 (5.2.1), allclinical06 (version 6.2.1)]21.
Radiographic assessments
We conducted secondary analyses based on the absence or presence of radiographic KOA at the OAI baseline visit. Weight-bearing, bilateral, fixed-flexion, and posterior-anterior knee radiographs were obtained at the OAI baseline. Central readers scored the images for Kellgren-Lawrence (KL) grades (0 to 4). We defined the subset with radiographic KOA as individuals with both knees having KL grade ≥ 2. We defined the subset without radiographic KOA as individuals with both knees having KL grade 0 or 1 throughout the study period (baseline to 48 mos). We also defined a subset with unilateral KOA as individuals with no OA in the ipsilateral knee, but with contralateral radiographic KOA. The agreement for these readings (read-reread) was good [weighted κ (intrarater reliability) = 0.70 to 0.78]. These KL grades are publicly available [File: kXR_SQ_BU##_SAS (version 0.6, 1.6, 3.5, 5.5, 6.3)]21.
Clinical data
Demographic, anthropometric, and other participant characteristics were acquired based on a standard protocol. We extracted age, body mass index (BMI), sex, self-reported strong prescription medication use, Physical Activity Scale for the Elderly, and responses regarding frequent knee-bending activities. For posthoc analyses, we adjusted for frequent knee-bending that we defined as a person who did any of the following tasks for ≥ 4 days per week in the 30 days prior to a study visit: (1) climbed up “10 or more flights of stairs during a single day”, (2) kneeled “30 minutes or more during a single day”, (3) squatted “30 minutes or more during a single day”, or (4) got “in and out of a squatting position 10 or more times during a single day”. The protocol and data are publicly available [Files: enrollees (version 20), allclinical## (version 0.2.2, 1.2.1, 3.2.1, 5.2.1, 6.2.1]21.
Statistical analyses
We evaluated the association between knee pain (chronic symptoms and severity) and a history of knee injury prior to the baseline, 12-month, 24-month, and 36-month OAI visits and a knee injury within 12 months by performing knee-based longitudinal analyses using a logistic regression with repeated measures (generalized linear mixed model) to adjust for correlations within person observations over time and between knees. We performed 2 primary analyses with new knee injuries being the outcome in each. The predictors in each analysis were from the annual visit prior to the outcome of self-reported knee injury status. For example, the analyses assessed whether knee pain at baseline would be associated with a self-reported injury between baseline and the 12-month OAI visit, when a participant reported an injury. The first primary analysis included the presence of chronic knee symptoms in the ipsilateral knee, the presence of chronic knee symptoms in the contralateral knee, the history of knee injury in the ipsilateral knee, and the history of knee injury in the contralateral knee. The second primary analysis was the same except we replaced chronic knee symptoms with knee pain severity in the ipsilateral and contralateral knees. Analyses were adjusted for sex, age, and BMI at each visit because they may be associated with knee pain25,26,27,28 and knee injury6,29,30,31,32,33. We did not censor knees because a knee could experience multiple injuries during the observation period. Each knee could contribute up to 4 observation periods.
As secondary analyses, we replicated our primary analyses in 2 subsets: (1) individuals with bilateral radiographic KOA, and (2) individuals without radiographic KOA. We also conducted secondary analyses by replicating our primary analyses stratified by sex since rates and types of knee injuries vary by sex6,32. Finally, we performed a series of posthoc analyses to validate the findings: (1) analyses among individuals with unilateral KOA in the contralateral knee; (2) replication of our analyses after adjusting for a history of knee pain (chronic knee symptoms or severe knee pain), self-reported strong prescription pain medication use (e.g., narcotics) on more than half the days in the month prior to a study visit, physical activity (Physical Activity Scale for the Elderly), or frequent knee bending; (3) person-based analyses in which the outcome was a new knee injury in either knee and the predictors were knee pain in either knee and a history of knee injury in either knee; and (4) analyses of knee pain severity with WOMAC knee pain entered as a continuous variable. We performed all analyses with SAS version 9.3. We considered 95% CI of OR that excluded 1 to be statistically significant.
RESULTS
For our primary analyses, we identified 8870 knees (4435 individuals) in the OAI that had data available at baseline and at 1 or more followup visits. This sample was primarily women and on average 61 years (SD 9) of age and overweight [BMI 28.6 kg/m2 (4.8)] at baseline: bilateral radiographic OA [ROA; age 63 yrs (9), BMI 30.4 kg/m2 (5.0)], no ROA [age 59 yrs (9), BMI 27.0 kg/m2 (4.4)], unilateral ROA [age 62 yrs (9), BMI 28.3 kg/m2 (4.2)]. About one-third of knees had reported chronic knee symptoms or had a WOMAC pain score ≥ 3 at baseline. For our secondary and posthoc analyses, we identified 2694 knees among 1443 eligible individuals with bilateral radiographic KOA at baseline, 2408 knees among 1347 individuals without radiographic KOA throughout the study period, and 653 knees with unilateral KOA in the contralateral knee. The baseline characteristics of these secondary samples are presented in Table 1. Within the full cohort, 875 new knee injuries occurred between OAI baseline and the 48-month followup visit among 772 knees (8.7%). Overall, 683 knees (7.7%) had 1 reported knee injury, 77 knees (0.9%) had 2 reported injuries, and 12 knees (0.1%) had 3 or 4 reported injuries during the observation period. Of the 875 new knee injuries, 335 new injuries occurred among those with radiographic KOA at baseline, 153 new injuries occurred among those without radiographic KOA, and 46 new injuries occurred among those with unilateral KOA.
Descriptive baseline characteristics of the study knees in primary and secondary analyses. Values are n (%).
In our primary analyses, we found that knees with reported chronic knee symptoms were 1.84 times as likely to experience an ipsilateral knee injury in the following 12 months (Table 2) compared with knees without chronic knee symptoms. Similarly, knees with greater knee pain severity (WOMAC pain score ≥ 3) were almost twice as likely to have an ipsilateral knee injury in the following 12 months (Table 3). Contralateral knee pain (chronic knee symptoms or pain severity) was not associated with a knee injury in the following 12 months. Knees with a history of knee injury in the ipsilateral or contralateral knee were more likely to have a new knee injury (history of ipsilateral knee injury OR 1.81, history of contralateral knee injury OR 1.43; Table 2). Similar findings were observed when adjusting for knee pain severity instead of chronic knee symptoms (Table 3).
Chronic knee symptoms and history of injury predict a new knee injury within 12 months.
Knee pain severity and history of injury predict a new knee injury within 12 months.
These findings were supported in our secondary and posthoc analyses. One exception was that a history of a contralateral knee injury was not related with a new knee injury among those without ROA at baseline. Analyses among 653 participants with unilateral KOA were limited because only 46 injuries occurred. In unadjusted analyses, we found that knees without ROA — but with contralateral ROA — were more likely to have a new knee injury if they had chronic knee symptoms in either knee, severe knee pain in the ipsilateral knee, and a history of knee injury in either knee (Supplementary Table 1 and Supplementary Table 2, available online at jrheum.org).
We conducted several posthoc analyses by adjusting for potential confounders among the full cohort and they supported the primary results (Supplementary Table 3 and Supplementary Table 4, available online at jrheum.org). The person-based analyses supported the primary results that an individual with knee pain [chronic (OR 1.69) or severe (OR 1.62)] or a history of injury (OR 1.88) was more likely to have a new knee injury in either knee (Supplementary Table 5 and Supplementary Table 6, available online at jrheum.org). Finally, WOMAC knee pain as a continuous predictor (scale 0 to 20) was associated with a new knee injury (OR 1.11, per 1 point increase in WOMAC pain score; Supplementary Table 7 available online at jrheum.org).
DISCUSSION
Despite knee injuries being associated with a greater risk for early-onset OA3 and accelerated KOA4, very little is known about possible risk factors for injuries among adults over 45 years of age. We found that among knees with or without KOA, a history of knee injury and ipsilateral knee pain are associated with new knee injuries. These findings were consistent when we defined pain based on severity or the presence of chronic knee symptoms.
Because knee pain, irrespective of the presence of ROA, is associated with a greater risk of injuries, it may be beneficial to determine whether aggressive treatment of knee pain is an important intervention to prevent future knee injuries. If so, then it may be advisable for clinicians to manage knee pain, even in the absence of ROA, based on KOA treatment guidelines. Most OA treatment guidelines34 advocate managing knee symptoms with a pharmacological intervention, as well as exercise programs, which could include goals for pain modification and injury/fall prevention.
These exercise programs could also be beneficial if applied to individuals with a history of knee injury. We found that a history of knee injury, in either knee, was associated with a new knee injury. This was consistent among knees with and without KOA; however, a history of contralateral knee injury was not associated with new injuries among knees without radiographic KOA. A history of knee injury could be related to bilaterally altered biomechanics35,36,37,38 or neuromuscular control39,40,41, which predisposes the ipsilateral or contralateral knee to a new injury. For example, individuals with a history of an ACL injury have altered biomechanics and neuromuscular control in the injured and uninjured knees, both of which could predispose the knees to a future ACL injury37,39. This may explain why an injury in either knee could increase the risk of a new injury. Further, knee pain in either knee was associated with a new injury in the unadjusted analyses. However, when we combine all the variables into the model, contralateral knee pain is no longer associated with new knee injuries. This could be because ipsilateral knee pain or a history of injury in either knee is more relevant when assessing risk of a new knee injury. Hence, knees with a history of knee injury, with or without OA, may benefit from exercise programs that incorporate neuromuscular training with a goal of preventing injuries or falls.
Injury prevention programs are commonly advocated for young physically active individuals because these programs improve performance, reduce the risk of injuries, and may subsequently reduce the risk of OA. This last point is based on consistent evidence that knee injuries increase the risk of KOA1,2,3. We previously demonstrated that among adults over 45 years of age, knee injuries are associated with incident KOA, particularly accelerated KOA4. Based on the current analyses, we found that knee pain and a history of knee injury increase the risk of a knee injury among adults with and without ROA. This supports a hypothesis that some knees may experience a dangerous cycle of injuries, pain, and structural progression (Figure 1). Prior to the onset of KOA (accelerated or common OA), ipsilateral knee pain and a history of injury may be risk factors for a new injury that could trigger the onset of accelerated or common KOA that are associated with knee symptoms. Once a patient has OA, we found that a history of knee injury or ipsilateral knee pain that could be attributable to OA are still risk factors for a new knee injury. This new injury could exacerbate OA progression and increase knee symptoms. We should strive to avoid this vicious cycle. Therefore, among older adults who report knee pain or a history of knee injury, we should explore injury/fall prevention strategies.
A vicious cycle of injury and accelerated knee osteoarthritis (OA).
Despite evidence that knee pain and a history of knee injury are associated with a greater risk for knee injuries, there are a few key limitations. First, the current study could not identify the underlying cause of knee pain (e.g., referred hip pain, patellofemoral lesions, tendinopathy, preradiographic degenerative changes). Further, it is unclear whether there was knee pain immediately prior to the new injury. Despite these limitations, we observed an association between knee pain and knee injuries. Interestingly, we found similar associations between pain and injuries among those with and without ROA. Prior research has indicated that knee pain was associated with falls8,9,10,13, even when most (> 80%) of the individuals with knee pain had no clinician-diagnosed KOA9. This highlights the importance of knee pain as a risk factor for knee injuries irrespective of whether KOA was present. Some sources of knee pain (e.g., patellofemoral pathology, ligamentous instability, effusion) may be more likely to alter proprioception12 or increase the risk of subsequent injuries than other sources of pain. Future research that clarifies the source of the knee pain may help us understand the possible link between knee pain and subsequent injuries, and ultimately lead to better injury prevention strategies.
It would also be beneficial to understand which injuries increase the risk of subsequent injuries and which new injuries are occurring. We could not address these important questions with existing data within the OAI. Despite not knowing the source of the knee pain or the type of injury, our study represents an important first step in understanding the association between knee pain, history of knee injury, and new knee injuries. Future studies could also pursue this research in population-based cohorts, but these findings will likely be confirmed because they complement the existing literature about the association between knee pain and falls8,9,10,13 and findings among younger, physically active individuals14,15,16.
We found that self-reported knee pain and a history of knee injury are associated with a greater risk for a new knee injury, with or without KOA. Hence, some patients may experience a dangerous cycle of knee injuries, pain, and structural progression. Because knee injuries are a key risk factor for KOA, it may be ideal to test the benefit of implementing injury prevention strategies among adults with knee pain or a history of knee injuries. This may entail treatment strategies to reduce knee pain, as well as exercise programs that aim to promote neuromuscular control and reduce the risk of injury/falls8,9,10.
ONLINE SUPPLEMENT
Supplementary data for this article are available online at jrheum.org.
Footnotes
Analyses supported by grants from the US National Institutes of Health (NIH; Eaton: 268201000020C-1-0-1 and Driban: 1R01AR065977-01A1). The Osteoarthritis Initiative (OAI) is a public-private partnership consisting of 5 contracts (N01-AR-2-2258, N01-AR-2-2259, N01-AR-2-2260, N01-AR-2-2261, N01-AR-2-2262) funded by the NIH, a branch of the US Department of Health and Human Services, and conducted by the OAI Study Investigators. Private funding partners include Merck Research Laboratories, Novartis Pharmaceuticals Corporation, GlaxoSmithKline, and Pfizer Inc. Private sector funding for the OAI is managed by the Foundation for the National Institutes of Health. Also supported in part with resources at the Veterans Affairs (VA) HSR&D Center for Innovations in Quality, Effectiveness and Safety (#CIN 13-413) at the Michael E. DeBakey VA Medical Center, Houston, Texas, USA.
- Accepted for publication April 20, 2015.
