Sleep apnea-hypopnea syndrome (SAHS) in childhood is a very common disease affecting 2%–4% of children between 2 and 6 years of age.1,2 Tonsil and adenoid hypertrophy, craniofacial malformations, obesity, neurological diseases, such as infantile cerebral palsy, neuromuscular diseases and gastroesophageal reflux are some of the risk factors most commonly encountered in childhood.2 Pediatric SAHS is associated with significant morbidity mainly involving the central nervous system, causing neurocognitive (memory, general intelligence, executive function, etc.) and behavioral disorders, and affecting the cardiovascular system, leading to autosomal dysfunction, cardiac arrhythmias, arterial hypertension,3 ventricular remodelling,4 and endothelial involvement,5,6 and producing endocrine metabolic comorbidities.7,8

Chronic diseases such as SAHS can alter a patient's sense of well-being by affecting their psychosocial environment. These changes can be measured and quantified by way of health-related quality of life (HRQoL) questionnaires. Specific validated questionnaires are required for these purposes, as these tools are more sensitive to alterations following therapeutic interventions, and can be used to discern changes.9–11 Subjective perception of improvement after treatment (particularly when continuous positive airway pressure is required) is associated with improved patient adherence.12 A recent metaanalysis found that children with SAHS have a poorer health status than healthy children.13 As in adults, generic questionnaires have been used in this population, for example the Glasgow Children's Benefit Inventory,14 the Youth Quality of Life Instrument15 and others.13

The first questionnaires aimed at children and teenagers have only recently been developed. This population has particular features that differ from adults, so particular attention must be paid to evaluating the psychometric properties of these instruments.16 In childhood SAHS, questionnaires have so far evaluated HRQoL from the parents’ or caregivers’ point of view. Moreover, they have all been designed for administration in an English-speaking cultural context. If they are to be used in Spain, they must be translated and adapted to our culture.17–20 This must be done in a stepwise fashion: first the concept being measured must be shown to exist in the culture to which the questionnaire is being adapted, then the questionnaire must be translated from the original language to that of the target population, the metrics of the adapted questionnaire must be assessed, and its psychometric properties (validity and reliability) must be measured.21

Very few specific questionnaires have been designed for childhood SAHS, but the OSA-18, published in 2000 by Franco et al.,22 has been used in several studies for evaluating post-surgical changes, and has been proven to be reliable and sensitive to post-treatment changes.23–25

While many tools have been developed to measure sleep disturbances and disorders in adults, no specific validated questionnaire in Spanish for childhood SAHS has yet been published. Consequently, the aim of our study was to analyze the reliability and validity of the Spanish version of the HRQoL OSA-18 questionnaire for use in the initial evaluation and follow-up of children with SAHS.

The general guidelines of the International Test Commission26 and the guidelines proposed for adapting HRQoL questionnaires27 were followed to adapt the OSA-18. The adaptation process was based on the method of translation-backtranslation by professionals and a pilot study with patients at various stages.28

Sixty consecutive patients were included, and all parents agreed to participate. In total, 45 boys and 15 girls were evaluated, aged 6±3 years, with a BMI of 18±4kg/m2 and mean neck circumference of 28±5cm. Brodsky index was 0 in 7%, <25% in 12%, 25%–50% in 27%, 50%–75% in 45%, and >75% in 6%. Mean AHI after PSG was 12±7h−1.

Translation difficulty was greater than 5 in 3 items (11%) and 1 domain (4%). After backtranslation, 2 items (7%) were considered type B and 1 was considered type C (4%). Questions surrounding equivalence (B and C) and expressions that were equivalent but lacked natural fluency or were grammatically incorrect were discussed by researchers and translators in 2 meetings. The final wording was agreed by consensus and included in the second version. The final version achieved scores >7 in all items and domains. The new version was administered to the parents of 10 children aged 2–14 years with a diagnosis of SAHS on PSG, and agreement was reached on the definitive version of the questionnaire (Table 1).

Internal consistency or reliability analyzed using Cronbach's alpha was 0.91, demonstrating excellent reliability for the overall questionnaire and for each domain, with no significant variation after exclusion of each item from a certain domain, or after exclusion of 1 domain from the overall score (Table 2). Correlations between domains were significant except in the case of emotional aspects, although the overall score correlated with all domains. Items showed a normal distribution, with no ceiling effect in any domain (maximum 10% of patients in D5), and no floor effect, except in D3 (30% of patients). When the overall questionnaire was analyzed, both the floor and the ceiling effects were 0%. The value of the overall questionnaire correlated significantly with all domains. The construct validity determined by factor analysis of principal components with varimax rotation and forcing a structure of 5 components corresponding to the 5 questionnaire dimensions is shown in Table 3 (showing saturations>0.5 only). We found that the structure of the Spanish version is practically the same as that of the original questionnaire. Items are grouped similarly and saturate in the components to which they belong. Only 1 exception was observed: the item “poor attention span”, which in the original questionnaire belongs to the domain of daytime problems, saturates better in this questionnaire in the emotional domain, which is understandable given the characteristics of the item in question and the domain. In terms of concurrent validity, the overall analysis shows good correlation (0.2–0.45) between most of the variables and their correlating domains and between the domains and the variables they measure. Individually, D1 shows correlations with snoring, apneas, and choking of 0.32–0.46; D2 with nasal obstruction and nasal symptoms of 0.2–0.28; D3 shows the poorest correlations (e.g., with hyperactivity or shyness) but correlates well with school performance (0.35); D4 correlates with school performance and excessive sleepiness (0.2–0.3); and D5 correlates significantly with symptoms and school performance (0.24–0.35). When predictive validity was evaluated, the questionnaire differentiated well between levels of severity as measured by the Mallampati score (ANOVA P=0.002) and AHI (ANOVA P=0.006) (Figs. 1 and 2). Sensitivity to change was excellent, both overall (P<0.001) and in each domain (P<0.001) (Table 4). Test–retest repeatability was also excellent, both overall and in each domain (Table 5).

Fig. 1.

Predictive validity. Bar chart showing the total number of patient in each group. The lower part shows the Mallampati score (ANOVA, Bonferroni 0.002); the upper part, mean OSA-18 results by groups.

(0.06MB).

Fig. 2.

Predictive validity. Bar chart showing the total number of patient in each group. The lower part shows severity measured by Apnea-Hypopnea Index (ANOVA Bonferroni 0.006). The upper part shows mean OSA-18 by groups.

(0.07MB).

The concept of health in children and teenagers does not only consider physical, psychological and social aspects, but also their ability to perform age-appropriate activities. In the case of children and teenagers, the dimensions that are generally explored are associated with their ability to perform activities of daily living (mobility and personal care), cognitive acquisition (memory, ability to concentrate and learn), emotions (positive and negative), self-perception, interpersonal relationships (with friends and family members), and relationships in their immediate environment (family cohesion, social support). Most HRQoL instruments for the pediatric population are designed on the psychometric model, measuring the ability of the individual to distinguish between stimuli of different intensity and recording the responses on a scale (usually a Likert-type scale).

Some units use the Pediatric Sleep Questionnaire, developed by the University of Michigan, to screen patients with childhood SAHS, to obviate the need for a PSG, or as a symptom questionnaire. The validated version in Spanish does not appear to be a true reflection of the original32,33 in terms of its daytime behavior and sleepiness dimension, which is unquestionably of the utmost importance in childhood SAHS.

The OSA-18 questionnaire has been adapted to other languages,34–36 and has demonstrated its utility in childhood SAHS. Some authors have used it as a screening tool to determine if it can substitute the PSG, but the questionnaire was not designed for this purpose.37

The OSA-18 has excellent internal consistency. Its predictive validity is good-to-excellent, although it would be improved if the severity groups were more balanced in number (there is a trend towards significant SAHS severity), and the concurrent validity is also good. The emotional domain is the most problematic in terms of psychometric characteristics. The construct is practically identical to that of the original questionnaire. Moreover, this questionnaire has been seen to be a very good predictive tool when compared with the Mallampati score and with SAHS severity measured by AHI.

In conclusion, the results suggest that the Spanish version of the OSA-18 questionnaire is equivalent to the original and is suitable for use in Spanish-speaking countries (Appendix A)

Funded in part by a grant from the Valencian Foundation of Pulmonology (2013) and a SEPAR 2014 Grant.

The authors state that they have no conflict of interests.