In this population based, middle-aged cohort a majority had early disease (i.e. joint pain with no or minor radiographic changes) at baseline. Despite this, 70% had bilateral radiographic changes 12 years later. Of those with unilateral knee osteoarthritis at baseline, 80% developed bilateral disease over 12 years. Osteoarthritis may have an asymmetrical onset but it has a tendency to affect both joints with time.
We have previously suggested that knee pain is the first sign of knee osteoarthritis , and the results from the present study add to this message that early knee osteoarthritis, without any known previous injuries to the anterior cruciate ligament or the meniscus, in most cases develops into a bilateral disease. We are unable to comment on the cause for the pattern of osteoarthritis progression that we have observed in this paper, as this would require much more complicated biological and biomechanical methods, however at present such studies are not available with the length of follow up achieved in our cohort.
As patients with joint injury were excluded from this cohort, it would be interesting to study a ‘unilateral joint injury’ cohort in the future, to see if those individuals who developed unilateral disease as a result of injury also progressed towards bilateral disease in the same way. Whilst it would be easy to assume that our results are solely due to genetic pre-disposition, changes in gait, mechanical loading and behaviour also need to be considered as causes for bilateral disease development . A comparative cohort study which assessed those with knee injury at baseline and those without a known cause would be of interest to understand this question in more detail.
Medial compartment osteoarthritis was the most common finding at 12 years, and this is very much expected. However, it was interesting to find that even amongst patients with lateral compartment changes in one knee, medial compartment disease was more common in the other knee. Numbers were too low to examine statistically and would need to be confirmed in larger cohorts, but this may be an interesting sub-group to study in the future.
In a previous study with 2 year follow-up, 34% of patients with unilateral disease subsequently developed disease in the contra-lateral knee (with osteoarthritis defined as KL ≥ 2) which corresponds reasonably well with our longer-term results . It may be argued that the contra-lateral knee is not necessarily free of disease in patients with unilateral radiographic disease using standard radiographic protocols . However, the development of incident radiographic osteoarthritis in this joint does demonstrate significant progression of the disease process and therefore remains a relevant end point.
It can be seen from Figure 3 that bilateral knee osteoarthritis becomes much more common as the disease severity increases. This is no surprise, but it is important to note that this happens at an early stage in the disease process, as those with KL = 2 in one knee were very likely to have disease present on the other side. Preventive interventions should therefore be instituted early in order to avoid or postpone symptomatic and radiographic knee osteoarthritis on the contra lateral side.
The variable relationship between pain and radiographic disease is well documented [13, 23, 24]. However, more severe radiographic osteoarthritis is associated with increased frequency and intensity of pain, as well as reduced physical function [24, 25]. The major focus on radiographic disease in this study is therefore clinically relevant, as bilateral structural disease progression would be expected to result in increased pain and functional decline over time. Pain is not necessarily the most important symptom to the patients, but the impact of pain on physical function. It has previously been shown that impaired physical function has more impact on help seeking than pain severity . Although all patients had chronic pain at inclusion, a larger sample size would be required to reliably examine the relationship between laterality, severity and the presence or absence of pain at follow-up.
The lack of association between gender, age or obesity and bilateral osteoarthritis needs to be interpreted with caution, since the current study did not have sufficient numbers to separate out the effects of these factors properly. Whilst the sample size used in this study was good overall, some of the numbers in individual cells were relatively small and this may have been a factor in the lack of a significant result. This deficiency is emphasised by the width of the confidence intervals for the odds ratios, which were large. However, the odds ratios were close to 1 and differences in means were small, which suggests that even if these factors were significant in a large study, they may not have clinically relevant effects. The higher odds ratio (OR = 1.49) for bilateral pain at baseline suggests that this may be a risk factor for bilateral disease. Larger studies would be required to analyse this in more detail.
The primary time-point of interest was the 12 year follow-up. The loss to follow up at 5 years (18/143) was relatively small and the results between baseline and 12 year follow up for those 18 patients were representative of the results for the whole population. We therefore believe that it is unlikely that the loss of those patients at the 5 year time point had a significant effect on the 5 year findings.
As this was a population based cohort that would traditionally be considered at low risk for osteoarthritis, with a high proportion of males, low levels of obesity and a low age range compared to most osteoarthritis population studies, the high incidence of bilateral radiographic disease after 12 years was striking. Knee pain cannot be considered innocent, even in a young and relatively healthy group of patients.
The gender distribution is not typical of a cohort of patients with established osteoarthritis. Rather, the gender distribution was representative of the community that the patients were recruited from at the start of the study .
The reason for the relatively high number of males may lie in the age group of the patients. The gender distribution in this study is consistent with other studies which have examined middle-aged subjects with knee osteoarthritis, as the proportion of males with knee osteoarthritis tends to be higher in younger age groups [27, 28]. We believe that the findings are generalisable to middle-aged patients who present with knee pain in the community, which was the purpose of the study when it was initiated.
Therefore the weaknesses of the study are the relatively small sample size, particularly when subgroups were considered or in the logical regression results. The logistic regression analysis was performed on both patients with no disease at baseline and those with unilateral disease at baseline and we accept that these may be two distinct phenotypes which progress differently towards bilateral disease, although we have seen no evidence to either confirm or refute this assertion. The lack of association between bilateral disease and age, gender, BMI and pain should therefore be considered with some caution.
Certainly, more information would have been gained from yearly radiographs compared to the two time points of follow-up in the study. Whilst this might have been preferable scientifically, it would require a much large research budget, and compliance would be a challenge, as well as the ethical issues of radiation doses and inconvenience to the participants. Yearly radiographic examination might not have influenced the overall conclusion, namely that bilateral osteoarthritis is common in this population over a prolonged period of time.
The strengths of the study were that this was a prospectively collected, community based dataset with radiographic data collected over a long time period and at two time points. We believe that numbers were more than adequate to draw the primary conclusion of the paper, that bilateral disease is a common outcome in patients who present with knee pain and particularly in those who already have evidence of osteoarthritis in one knee.
A recent analysis of a large longitudinal cohort of patients at high risk of osteoarthritis found that bilateral knee pain was an independent risk factor for poor physical function, even when pain severity was accounted for . The authors of this paper speculated that this may be due to the loss of a ‘good limb’ to compensate during functional activities. Given the high frequency of bilateral disease after 12 years in our study, it is likely that the development of disease in a second joint is a significant cause of additional disability in the population.
Studies have demonstrated that the biomechanics of the unaffected limb are not normal in patients with unilateral knee osteoarthritis, and also that gait asymmetries exist in patients with unilateral osteoarthritis that subsequently change when a patient develops bilateral disease [7, 11]. The treatment of the patient may therefore change depending on whether they have unilateral or bilateral disease. It is important to note from our findings that the majority of patients with unilateral disease would potentially benefit from interventions aimed at preventing disease in the other, apparently normal joint. Techniques such as wedged insoles, neuromuscular exercises and gait retraining may be appropriate and further research is warranted to examine ways to protect the ‘joint at risk’.
These findings have implications for clinical practice as well as future research in osteoarthritis. Patients can be given a prognosis which may help motivate them to comply with treatments and address symptoms in the other knee appropriately. Clinicians should be aware that the presence of osteoarthritis in one knee is likely to herald a process that also involves the other side in the future. Radiographic changes develop slowly and for patients with chronic knee pain “radiographically healthy knees” may not be as healthy.