Psychometric Evaluation

The psychometric evaluation was guided by the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) [21] and a previous psychometric evaluation of the SwQoR [18]. Acceptability, which measures the clinical user friendliness, was assessed in terms of the successful response rate on postoperative days 1-7. Floor and ceiling effects (ie, the number of respondents who achieved the lowest or highest possible scores) were measured on days 1-7; it was considered a problem if more than 15% of the study population achieved the lowest or highest possible score [22]. Construct validity is the extent to which questionnaire scores are consistent with hypotheses, assuming that the questionnaire validly measures the construct being addressed. A correlation coefficient >0.4 was considered to be evidence of construct validity (ie, moderate to strong correlation). To analyze construct validity, a priori hypothesis testing was conducted, under the hypothesis that the SwQoR-LA, just as with the SwQoR [18], on day 1 would correlate positively with the duration of surgery, duration of stay at the postanesthesia care unit (PACU), and patient age. In addition, lower quality of recovery (ie, higher degree of postoperative symptoms) was not expected in women versus men, just as with the SwQoR [18]. Discriminant validity was tested on day 7, and it was expected that patients who underwent local anesthesia would have significantly lower scores on the symptoms related to general anesthesia that are not included in the SwQoR-LA: voice not sounding the same as usual, sore throat, sore mouth, having trouble breathing, muscle pain, trouble urinating, diarrhea, and feeling constipated. For example, sore throat and sore mouth are symptoms related to the endotracheal tub or laryngeal mask used under general anesthesia.

Reliability was assessed with (1) internal consistency, by measuring the average correlation between the SwQoR items on days 1-7, indicated by Cronbach α, and (2) split-half reliability, by measuring the correlation between randomly split segments of the SwQoR on days 1-7. Responsiveness, which was used to evaluate the SwQoR-LA’s sensitivity and ability to detect clinically important changes, was measured with (1) Cohen d effect size, calculated as average changes in scores from days 1 to 7, divided by the pooled SD of all measurements (where 0.2-0.5 indicates a small effect, 0.5-0.8 a moderate effect, and 0.8-1.2 a large effect) [23], and (2) mean changes over time and percent changes from baseline on days 1-7.