Statistics from Altmetric.com
We read with interest the article by Riddle et al1 entitled ‘Associations between Statin use and changes in pain, function and structural progression: a longitudinal study of persons with knee osteoarthritis’, where the authors used data on a subset of 2207 subjects included in the Osteoarthritis Initiative with radiological suspected or confirmed knee osteoarthritis (OA). Their article was partly based on the article by Clockaerts et al,2 where we report an association between statin use, and incidence and progression of radiological knee OA in the Ommoord cohort of the Rotterdam study. Riddle et al also investigated the association between pain and function scores, progression of radiological knee OA and statin use, and found no associations.
Their data may indicate that there is indeed no association between statin use and knee OA, and that our results were due to an unexpected bias or chance. As discussed in the report by Clockaerts et al, we have tried to minimise these options. Riddle et al mention to have used similar methodology, but we believe that some important differences exist in study design that could explain their results.
In the Riddle report, data concerning statin use and its frequency, duration and compliance was based on self-report by patients. It is known that this type of registration is not very precise, and may lead to indication bias, that could bias the results upwards. In the Rotterdam study, we had the opportunity to use a digitalised pharmacy database, where every single medication prescription actually collected by the patient, was precisely registered. This is important because in an observational cohort, despite correction for confounding cofactors as much as possible, the association between statin users and the development of OA might be overshadowed by confounding by indication. In other words: statin users might have a profile that makes them at increased risk for OA. We were able to look at the differences in association between the actual doses where we looked at patients who were taking statins, but less than 120 days of the study period of 6 years, and less than 50% of the daily recommended dose. In this group of ‘minimal users’, we indeed observed no lower incidence and progression of radiological knee OA (OR 1.17, 95% CI 0.50 to 2.71, p=0.72), while we clearly showed the protective relationship in those who took statins more regularly, and more than 50% of recommended doses.
Another important difference with the study by Riddle et al is that we included people with a Kellgren and Lawrence score of 0. Although Riddle et al pose that statins should be investigated further in people with end-stage knee OA, we believe that the opposite may be true. Considering the theoretical effects of statins on OA (lowering of serum lipid levels, lowering local and systemic inflammation, and prevention of atherosclerosis), it is more plausible that statins could be useful in the early stages of OA or even before initiation of the disease process. We therefore recommend analysing the association also in patients with Kellgren and Lawrence scores of 0 at baseline (incidence).
In the study of Clockaerts et al, evolution towards a prosthesis was also included, although this was mentioned otherwise in the study by Riddle et al.1
In the Riddle report, there are no data concerning the occurrence of muscle pain in the study population.1 This is an important side effect of statins,3 which should be taken into account when considering pain scores for OA.
Finally, it was interesting to notice that there was a difference in body mass index between both study populations. Although this was corrected for in both statistical analyses, it may indicate that subgroup analyses may be needed to see in what type of OA patients statins may be beneficial.
Taken together, methodological issues or differences in included patients may explain the observed results. We look forward to future studies that confirm or exclude a potential association between statin use and knee OA. Together with the present available studies, they could provide sufficient data to identify a subgroup of patients with OA, that should be included in a well designed, randomised, controlled trial.
Funding All authors disclose any financial and personal relationships with other people or organisations that could inappropriately influence this work.
Competing interests None.
Provenance and peer review Not commissioned; internally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.