The development of a health-related quality-of-life instrument for young people with narcolepsy: NARQoL-21.

BACKGROUND: Narcolepsy is a lifelong sleep disorder with a prevalence of between 0.03% and 0.06% and onset at around puberty. It is associated with psychiatric comorbidities and cognitive difficulties. No valid and reliable condition-specific health-related quality-of-life (HrQoL) instrument has been developed for this population. METHODS: A questionnaire based on four mixed-gender age-defined focus group discussions and a patient panel analysis was administered to young people with narcolepsy and a control group. External reliability was measured by a test-retest procedure and internal reliability was measured using Cronbach's alpha. Convergent validity with the KIDSCREEN-10 index was assessed using with intraclass correlation coefficients (ICC) and receiver operating characteristic (ROC) curves. Factor analysis techniques were used to identify suitable items and confirm the factor structure. Baseline values were assessed for convergent validity, ceiling effects, agreement and sensitivity. Comparison with KIDSCREEN-10 was made on the basis of area under the curve (AUC). RESULTS: One hundred young people with narcolepsy and 95 control subjects returned questionnaires. The factor structure revealed two main factors with five domains and 21 questions, which was confirmed with confirmatory factor analysis. The domains of the NARQoL-21 showed good independence while the floor and ceiling effects were acceptable. The external reliability (0.928), convergent validity (rs = 0.769) and internal consistency (Cronbach's alpha = 0.886) were excellent. A Bland-Altman plot revealed some proportional bias. Good discriminant validity was detected for control/patient (Cohen's d = 2.114). ROC analysis showed significantly better AUC for NARQoL-21 (0.939) than KIDSCREEN (0.877). A cut-off score equivalent to KIDSCREEN-10 for suboptimal HrQoL which maximized sensitivity (84%) and specificity (92%) was found at NARQoL-21 score below 42. CONCLUSIONS: Establishing the validity of a disease-specific HrQoL instrument in a population of people with a rare condition poses significant challenges. The mixed-methods approach adopted here has resulted in a questionnaire of 21 items with good discrimination and convergent validity, and excellent internal and external reliability, allowing precise and stable measurements. The cut-off score can be useful to identify patients with very poor HrQoL and thus improve the design of treatment options. Further testing in a longitudinal cohort is recommended in order to establish responsiveness.