A cross-sectional analysis of facial palsy-related quality of life in 125 patients: Comparing linear, quadratic and cubic regression analyses

Introduction Facial function correlates with quality of life in facial palsy. Previous studies have examined a linear relationship; based on clinical experience, we hypothesize a curved regression (i.e. quadratic or cubic) will be more fitting to show the correlation between quality of life and facial function. Methods We compared the fit of a linear regression model between Sunnybrook scores (facial function) and FaCE and FDI scores (quality of life) to a quadratic and cubic regression model in 125 patients cross-sectionally. Results A total of 125 patients were included, 53.6% female with a mean (standard deviation) age of 56.6 (16.7) and a median (interquartile range) duration of palsy of 6.6 (1.5; 18.3) years. The quadratic regression proved a significant improvement over a linear regression analysis in the model using the FaCE total score (linear R-2 =.346, quadratic R-2 = .378, p = .033) and the FDI physical score (linear R-2 = .245, quadratic R-2 =.276, p = .034). The cubic regression analysis was no significant improvement over a quadratic regression. Discussion The relationship between facial function and quality of life in facial palsy is not linear meaning that there is a lot of variation in QoL in cases with severe and moderate facial impairment. This is most applicable to patients suffering from post-paralysis synkinesis, proving the highly individually experienced burden of synkinesis. As the relationship is not linear it should not be included as such in future research studies.