Hot and cold knees: exploring differences in patella skin temperature in patients with patellofemoral pain

Objectives: To investigate the distribution of patella skin temperature (Tsk) measurements and to explore the presence of temperature subgroups in patellofemoral pain (PFP) patients. Design: Cross-sectional observational study design. Participants: One dataset of 58 healthy participants and 232 PFP patients from three different datasets. Main outcome measures: Patella skin temperature, measured by physiotherapists using a low cost hand held digital thermometer. The distribution of patella skin temperature was assessed and compared across datasets. To objectively determine the clinically meaningful number of subgroups, we used the average silhouette method. Finite mixture models were then used to examine the presence of PFP temperature subgroups. Receiver operating characteristic curves were used to estimate optimal patella Tsk thresholds for allocation of participants into the identified subgroups. Results: In contrast to healthy participants, the patella skin temperature had an obvious bimodal distribution with wide dispersion present across all three PFP datasets. The fitted finite mixture model suggested three temperature subgroups (cold, normal and hot) that had been recommended by the average silhouette method with discrimination cut-off thresholds for subgroup membership based on receiver operating curve analysis of Cold =