Data published in Annals of the Rheumatic Diseases indicate that frequent reports of knee pain are more strongly associated with radiographic changes of osteoarthritis (OA) than once-reported pain.
Researchers extracted patient data (n=2756) from the Multicentre Osteoarthritis Study and tested in a cross-sectional design the capacity of X-ray features to discriminate between patients with and without frequent knee pain or consistent frequent knee pain. Patients were asked 3 times in the 2 weeks prior to imaging if they had knee pain, aching, or stiffness during the majority of the previous 30 days. Consistent frequent knee pain was characterized by affirmative answers to each of the 3 questions at all 3 time points and frequent knee pain was defined as answering 1 question at each time point.
Baseline radiographs were obtained at the imaging visit. Researchers used 36 individual radiographic features to classify patient knee OA and determined the discriminative association of each feature with clinical pain scores. Researchers also calculated area under the curve (AUC) for both manual and automated grading models of OA.
Mean age of patients in the study cohort was 62.3±8 years, mean body mass index (BMI) was 30.7±5.9, and 60% were women. Of the radiographic features used to score OA, Kellgren and Lawrence grade, osteophytes, joint space narrowing, and sclerosis were the most discriminative. Chondrocalcinosis, cyst, and attrition and ossification of the patella-tendon were not found to be associated with pain scores. The AUCs for consistent pain were higher compared with lower pain grades. The best-performing classifier model used manual grading and accounted for patient gender and BMI, with an AUC for consistent pain of 73.9%. The automated and manual scoring models were found to perform in a similar fashion, and no improvement over manual grading alone was found for a combination of automated image analyses with manual grading.
These results suggest that repeated reports of knee pain are more likely to be detected by X-ray than pain reported less frequently. These results support the extension of osteoarthritic pain classification research to MRI features so that clinicians may best interpret pain perception in patients with OA.
Minciullo L, Parkes MJ, Felson DT, Cootes TF. Comparing image analysis approaches versus expert readers: the relation of knee radiograph features to knee pain [published online August 1, 2018]. Ann Rheum Dis. doi:10.1136/annrheumdis-2018-213492