Klaske van Sluis

50 3.2. Methods (VHI-10). The EQ-5D-5L is a validated patient-report questionnaire that as- sesses a patient’s current health related QoL [22]. It consists of 5 dimensions: mobility, self-care, daily activities, pain/discomfort and anxiety/depression. The final continuous outcome ranges from 0 to 1, a higher score indicates better health related QoL [22]. Scores were interpreted with the Dutch country specific reference values [23]. The VHI-10 assesses experienced voice handi- cap [24]. VHI-10 includes ten questions covering three sub-themes: functional, physical and emotional. The total VHI-10 continuous outcome is a score rang- ing from 0 to 40, a higher score indicates a greater handicap. Scores above 11 are considered as abnormal [20, 24–26]. The VHI-10 is a widely used and validated questionnaire, although not specifically validated for use after total laryngectomy. 3.2.2 Demographics and oncological history Demographics and oncological history were collected during the first and second assessment. Demographic variables included sex and age at time of surgery. On- cological history included tumour site, T-stage, and N-stage as defined by the pathologist post-surgery, timing of (chemo)radiotherapy, and primary, func- tional or salvage indication for total laryngectomy. Surgery specific data in- cluded neck dissection, neopharyngeal reconstruction, tongue base resection,myotomy of the upper esophageal sphincter, primary voice prosthesis placement, and secondary voice prosthesis placement. In the finalizing phase of the study the variables were checked with information retrieved from the local data desk. 3.2.3 Acoustic analysis Segmentation, acoustic analysis, and obtaining AVQI scores is performed using Praat [27]. The main outcome measure Acoustic Voice Quality Index (AVQI) requires recordings of a sustained vowel /a/ and a read aloud text [5, 6]. Sus- tained /a/ sounds of at least three seconds were used. If no single realization of 3 seconds was available, realizations were concatenated. From the read aloud text or read aloud sentences four seconds of connected speech was used. If these included long pauses, these were removed. The AVQI algorithm includes the cepstral peak prominence, harmonics-to-noise ratio, shimmer local, shim- mer local dB, as well as the slope and tilt of the regression line through the long-term average spectrum. When incorporated into Praat, the analysis script estimates an AVQI score, which ranges from 0 to 10. A lower score indicates a better voice quality, >2.95 is the cut-off point, scores above are indicated as distorted. Participants who were unable to produce voice post-surgery but did perform the assessment were rated with an AVQI score of 10.