Evaluation of measures used for diagnosis of obstructive sleep apnea in children

BACKGROUND: In children, sleep-related airway obstruction by large tonsils and adenoids can cause obstructive sleep apnea (OSA). OSA may lead to poor growth, developmental delay, behaviour or learning problems. Recent evidence also suggests that children with OSA may develop cardiovascular complications, the mechanisms perhaps involving hypoxemia, the autonomic nervous system, apneas, and arousals. Surgical removal of tonsils and adenoids (adenotonsillectomy (T&A)) usually cures pediatric OSA. To diagnose OSA at all levels of severity, polysomnography is currently the best approach. The McGill Oximetry Score (MOS) is a validated measure based on nocturnal pulse oximetry. An abnormal MOS has a 97% positive predictive value at detecting moderate-severe OSA. Because the MOS was devised by measuring frequency of desaturations (<90%) and numbers of clusters of desaturations, it is not accurate at detecting OSA in children who do not have such drops in oxygen saturation. Accordingly, other measures applicable to a wider spectrum of children should be assessed. These measures should be simpler, less cumbersome, cheaper, and more accessible than polysomnography.OBJECTIVES: To study alternative approaches that may be used to identify moderate-severe OSA in children, two studies were conducted. We examined one subjective measure - the OSA-18 parent questionnaire - and two objective measures - pulse rate and pulse rate variability. For the OSA-18 study, the goal was to determine whether it would accurately detect children with moderate-severe OSA as indicated by an abnormal MOS. For the pulse rate and pulse rate variability study, the goal was to determine if either or both would decrease after treatment with T&A for children with moderate-severe OSA.METHODS: For the OSA-18 study, we used a cross-sectional design that included children 1-18 years old referred to a pediatric sleep laboratory for evaluation of suspected OSA. Alongside data from the OSA-18, we analyzed demographic and medical data (from a parent questionnaire) and information regarding adenotonsillar hypertrophy. We estimated sensitivity, specificity, positive and negative predictive values as well as receiver operating curves of the OSA-18 in detecting an abnormal MOS. We also conducted univariate and multivariate logistic regression analyses, using the MOS as the dependent variable and the OSA-18 score and others (age, gender, comorbidities, race) as independent variables. For the second study, we used a retrospective before-after design to compare pulse rate and pulse rate variability as measured by nocturnal pulse oximetry pre- and post-T&A of otherwise healthy children 1-18 years old with moderate-to-severe OSA.RESULTS: For the OSA-18 study, we studied 334 children (58% male, mean age 4.6 +/- 2.2 years). The OSA-18 had a sensitivity of 40% and a negative predictive value of 73% for detecting an abnormal MOS. In addition, the area under the receiver operating curve was 0.611. While controlling for other independent variables in the logistic regression model, for each unit increase in the OSA-18 Score, the odds of having an abnormal MOS were increased by 2%. However, for each increase in age of 1 year, the odds of having an abnormal MOS were decreased by 17%. In the pulse rate and pulse rate variability study, 25 subjects (88% male; mean age 4.3 +/- 3.6 years) were enrolled. Following T&A, pulse rate and pulse rate variability decreased in 21 of 25 and 23 of 25 children, respectively. Mean pulse rate dropped from 99.7+/-11.2 to 90.1+/-10.7 bpm, p<0.001; age-standardized pulse rate (z-score) from 0.8 (0.4, 1.5) to 0.4 (0, 0.9), p=0.04). Pulse rate variability, as measured by the standard deviation of the pulse rate decreased from 10.3 +/- 2.1 to 8.2 +/- 1.6 bpm, p<0.001. As well, OSA symptomatology, parental concern about breathing during sleep and the MOS all improved.CONCLUSIONS: Based on the first study we conclude that among children referred to a sleep laboratory, the OSA-18 does not accurately detect which children will have an abnormal MOS. The OSA-18 should not be used in place of objective testing to identify moderate-severe OSA in children. However, from the second study we conclude that measures of the autonomic nervous system such as pulse rate and pulse rate variability, as measured by pulse oximetry, decreased following surgical treatment of moderate-severe OSA. The results of this study potentially serve as important data for further work that would determine the accuracy of pulse rate and pulse rate variability measures and their diagnostic usefulness for OSA at all levels of severity