RESULTS
In the VAS validation group, 21 individuals responded, of whom 3 were excluded for not submitting a second VAS score. A slight majority of the 18 participants was female (n = 10 [56%]) and median (IQR) age was 71 (55; 76) years. The most common etiology of facial palsy was parotid gland tumor (n = 8 [28%]), followed by acoustic neuroma (n = 2 [11%]). Median time since diagnosis of facial palsy (IQR) was 1.8 (0.8; 8.7) years. A median time of 7 days elapsed between the responses. Median (IQR) VAS score was 60 (23; 74) at T1 and 61 (17; 73) at T2. There was a moderate, positive correlation between the VAS score and FaCE total score that was statistically significant (rs= 0.561, p < 0.001), indicating acceptable validity. Test-retest reliability was high, with an intraclass correlation coefficient of 0.91 (95% CI = 0.78-0.97).
A total of 80 participants were included in the linear regression analysis; of the 130 eligible patients, 14 were not interested in participating, 28 did not respond to the invitation for the following measurement, and eight respondents submitted incomplete responses. Forty-one participants were male (51%), and median (IQR) age was 63 (51; 73) years (Table 1). Median (IQR) duration of facial palsy was 17.1 (9.6; 33.1) years, and the most common etiology was acoustic neuroma (n = 22 [27%]). Median (IQR) FaCE total and VAS scores were 51.7 (38.3; 62.9) and 70 (52; 93), respectively.
The multivariate linear regression analysis of the FaCE subscale scores on the FaCE total score resulted in regression coefficients that were exactly proportional to the number of questions contained in the subscales (Table 2). In the second regression analysis, social function and facial comfort contributed significantly to the VAS score (Table 3). Social function showed a higher regression coefficient than in the first regression analysis (β = 0.456; +0.189 compared to the original coefficient), as did facial comfort (β = 0.334; +0.134 compared to original coefficient). All other subscales were non-significant contributors to the VAS score. The explained variance in the model was 50.4% (R2 = 0.504). The mean (range) values for tolerance and VIF were 0.85 (0.756-0.912) and 1.18 (1.096-1.323) respectively, indicating no multicollinearity.