Data source

We conducted this single-centre, prospective cohort study at the University Hospital of Angers, France from 15 August 2021 to 13 April 2022. Written consent was not requested; however, all patients were informed and accepted information collection as mandated by French law.11 This study was reviewed by a French ethics committee (Comité de Protection des Personnes Ile de France VI; registration ID: 21.02487.003521) and was registered with ClinicalTrials.gov (NCT04845763; first submitted 11 April 2021). The study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

Study population

Eligible patients were identified during the presurgery anesthesia consultation. Participants met the following inclusion criteria: 18 yr or older, admitted for urgent surgery (desired surgical procedure time < 72 hr), capable of completing the questionnaire independently or with assistance, French-speaking, and willing to participate in the study. We excluded patients with significant psychiatric or neurologic disorders impeding cooperation in questionnaire completion, patients under legal wardship or guardianship, patients admitted for cardiac or obstetric surgery (Cesarean delivery), patients admitted for revision surgery, and patients previously included in the study during prior admissions.

Available data

A validated French version of the QoR-15 score (FQoR-15) was used.12 We relied on the FQoR-15 to ensure thoroughness in gathering data, as this version was already commonly used at our hospital centre. Participants completed the questionnaire at three timepoints: before surgery (baseline, H0) and at 24 (H24) and 48 (H48) hr after surgery. Patients completed the questionnaire independently, or with an assessor’s assistance if required. For postoperative evaluations, phone interviews were conducted if the patient had been discharged home or to another facility (follow-up care and rehabilitation department). We collected several characteristics at inclusion, such as demographic information (age, weight, height, sex), the American Society of Anesthesiologists Physical Status score, comorbidities, trauma status, surgery type, and emergency level according to the Timing of Acute Care Surgery (TACS) classification.13 The time taken to complete the questionnaire was noted by the patient or a medical staff member. Perioperative information at H24, such as the executed surgical procedure and its duration, the surgical outcome risk tool (SORT) score indicating the procedure’s severity,14 and anesthesia type, were also recorded.

Opioid use in the last 24 hr (including consumption in the postanesthesia care unit for the H24 evaluation), significant complications as per the postoperative morbidity survey (POMS) classification evaluating nine domains: pulmonary, infectious, renal, gastrointestinal, cardiovascular, neurologic, hematologic, wound, and pain),15 and hospitalization status were recorded at H24 and H48. Total length of stay and postoperative complications were documented at discharge or three months after surgery if the patient remained hospitalized. We also conducted a three-month assessment of quality of life via phone interview, using the EQ-5D-3L questionnaire and EQ-VAS (vertical visual analog scale scored from 0 to 100, 0 being “worst” and 100 being “best imaginable” health). Data on the number of days spent at home during the first 30 and 90 days after surgery (DAH30 and DAH90, respectively) were also collected. The patient’s vital status was confirmed, or the date of death was recorded.

The psychometric assessment

We conducted a psychometric study to confirm the validity and reliability of the FQoR-15 questionnaire within the context of emergency surgery.16,17

Content validity examines whether the questionnaire items effectively encapsulate the Quality of Recovery concept. The QoR-15 questionnaire items have already been validated in a postoperative context. We applied them to a distinct target population, namely patients undergoing emergency surgery. As we did not modify these items, we did not specifically re-evaluate this validity in our study.

Internal consistency refers to the degree to which the items capture the Quality of Recovery construct. We also assessed the unidimensionality of the questionnaire.

Convergent validity is the association between the questionnaire and a “gold standard.” We compared the QoR-15 with a general state visual analogue scale, which yields a score between 0 (“very impaired health”) and 10 (“excellent health”). The question used was: “How would you rate your overall health over the past 24 hours?” We assessed this at H0, H24, and H48.

Construct validity indicates the score’s suitability concerning theoretical alterations related to quality of recovery in the context of emergency surgery. To test this, we formulated several hypotheses, with over 75% requiring confirmation. We hypothesized that there would be a gender variation in the score (higher scores for men), a positive correlation with age (higher scores in younger individuals), a negative correlation with the emergency level of surgery (according to TACS classification), a negative correlation with surgical risk of the procedure (according to the SORT score), a negative correlation with the occurrence of postoperative complications (according to the POMS classification), a negative correlation with morphine use, and a negative correlation with length of stay.

Reproducibility is determined through a “test-retest” comparison. It suggests that repeated tests on stable individuals yield similar results. Two measurements, conducted 24 hr after surgery and separated by 30 min to one hour, were performed to assess response consistency. Agreement pertains to absolute measurement error, while reliability indicates the extent to which patients can be differentiated from one another, despite potential measurement error.

Responsiveness reflects the ability of a questionnaire to detect clinically relevant changes over time.

We considered floor or ceiling effects if more than 15% of respondents achieved the lowest or highest possible score.

Acceptability and feasibility are measures of user-friendliness, such as the patient recruitment rate, the total participation rate in the three time frames (H0, H24, and H48), and the time taken to complete the questionnaire.

The minimum clinical difference (MCD) and clinically significant difference (CSD) denote the smallest differences that need to be perceived in the total QoR-15 score to identify a minimal or substantial change in a patient's recovery status. At H24 and H48, patients evaluate their recovery over the past 24 hr on a seven-item Likert scale ranging from “much worse” to “much better.” The mean difference in QoR-15 score between “same” and “slightly better” was used to ascertain the MCD, based on the anchor-based method.18 To determine the CSD, patients were asked, “Do you believe you have had a good recovery?” at H24 and H48.19

Sample size

At present, there are no established guidelines for calculating sample size in the context of a psychometric assessment study.16 An acceptable limit was identified as 300 patients.20 Accounting for a potential 20% loss to follow-up or instances of missing data, we targeted a sample size of 375 patients for the assessment of QoR-15 in the context of emergency surgery.

Statistical analysis

We present data as mean (standard deviation [SD]) or median [interquartile range] for quantitative variables. Qualitative variables are represented by the number of patients and the percentage (%). To compare continuous variables, Student’s t tests were used for normally distributed variables and Wilcoxon tests were used for non-normally distributed variables. Associations between quantitative variables were measured using Spearman correlation coefficients.21 An interitem correlation matrix is proposed and composed of Spearman correlation coefficients. Internal consistency was measured using the Cronbach α coefficient. The objective was to obtain a value between 0.70 and 0.90.22 We explored the number of dimensions of the questionnaire by the total percentage variance explained by the first factor. Test-retest reliability was measured using the agreement intraclass correlation coefficient. A value of 0.70 is usually recommended as a minimum standard for reliability.22 The test-retest provided an estimate of the standard error of measurement (SEM), including systematic differences (i.e., SEM agreement). Responsiveness was quantified using the Cohen effect size (average change score divided by the SD at baseline) and standardized response mean (change scores divided by the SD of the change scores). The final MCD corresponded to the average of the MCD obtained by the distribution methods (corresponding to 1.96 × SEM, a larger variation than the random variation at 5% of uncertainty) and anchoring methods (the difference in mean score values between the “same” and “slightly better” status). The 95% confidence intervals (95% CIs) were obtained by bootstrapping (adjusted bootstrap percentile method, with 1,000 bootstrap replicates, using the R “boot” library).23,24 We rejected the null hypothesis if the P value was < 0.05. To control the false discovery rate, we adjusted the P values obtained during the analysis using the Benjamini–Hochberg method. We performed all statistical analyses using R software version 4.1.3 (R Foundation for Statistical Computing, Vienna, Austria).