Abstract
Introduction:
The Oxford Knee Score (OKS) has been designed for patients with knee osteoarthritis. It uses 12 questions to give an objective patient reported outcome measure. It is one of the most used questionnaires in the orthopaedic community. Each question has the same numeric value, 0-4, giving a maximal number of 48. The OKS has been observed to have a significant ceiling effect, especially when distinguishing slight postoperative differences between implant and alignment philosophies.
Hypothesis:
We hypothesized that each question in fact does not have the same weight for each patient, and that it depends significantly on the patient’s sociodemographic data and lifestyle.
Methods:
In this ethics-board approved prospective study, we included patients coming to a specialist outpatient knee clinic. Each patient filled out 3 questionnaires, in this consecutive order: demographic data (age, gender, height, weight, BMI) as well as data pertaining to the OKS, to which the patient was initially blinded for (stairs at home, self-use of a car, use of walking aids, working, previous surgery). Then the patient filled out the standard OKS. Finally, the patient gave a note on the importance of each of the question’s topics, using a 5-point Likert scale. Linear regression models were created to predict the weight of each of the questions.
Results:
We included 77 patients, median age 66 (IQR 18) of which were 40 women and 37 men. Median BMI was 27.7 (IQR 8.2). As for the OKS-related demographic data, 89.6% of patients had stairs, 70.1% were still driving a car, 23.4% were using walking aids, 33.7% were still working and 72.7% previous knee surgery.
According to the Likert scale, the questions with the highest weight were about washing, night pain, and stability (all median 5 [IQR 1]). The questions with the lowest weight were about walking a longer distance (median 4 [2]), getting up after sitting (median 4 [1]), and kneeling (median 4 [1]). The difference between high-weighted and low-weighted questions was significant (p<0.001). A weak correlation between the recorded score and recorded question weight was observed for the question on stability (p=0.036, r=0.239) and a reverse correlation for the question on shopping (p=0.021, r=-0.263). In the regression models, question on pain was predicted by higher age (p=0.009, B = - 0.02), question on transport was predicted by the patients’ self-use of a car (p=0.008, B=0.557), question on walking for a longer period by height (p=0.02, B 0.167), weight (p=0.33, B=-0.155) and BMI (p=0.032, B=0.427), question on standing up after eating was predicted by female gender (p=0.024; B=0.634) and working (p=0.019; B=0.306), question on night pain by female gender (p=0.049, B=0.255) and self-use of a car (p<0.001, B=0.525), and the question on stability by self-use of a car (p=0.041; B=0.434).
Conclusion:
The study demonstrates that each patient’s lifestyle significantly influences the importance of each of the OKS questions. This is, however, not reflected in the OKS, which in turn gives each question the same numeric value. Patients who still work and drive by themselves have given a higher value on questions that reflect these actions. This might be one of the reasons of the ceiling effect of OKS.