Sleep Medicine 99 (2022) 41e48 Contents lists available at ScienceDirect Sleep Medicine journal homepage: www.elsevier.com/locate/sleep Original Article Nasal cannula use during polysomnography in children aged under three with suspected sleep apnea  Jurado a, c, d, *, Gabriel Sampol b, c, d, Manuel Quintana e, Odile Romero a, c, d, María Jose Roser Cambrodí a, c, d, Alex Ferre  a, c, Júlia Sampol b, c, d a Department of Clinical Neurophysiology, Hospital Universitari Vall d’Hebron, Universitat Auto noma de Barcelona, Passeig Vall d’Hebron 119-129, 08035, Barcelona, Spain b Department of Respiratory Care, Hospital Universitari Vall d’Hebron, Universitat Auto noma de Barcelona, Passeig Vall d’Hebron 119-129, 08035, Barcelona, Spain c noma de Barcelona, Passeig Vall d’Hebron 119-129, 08035, Barcelona, Multidisciplinary Sleep Unit, Hospital Universitari Vall d’Hebron, Universitat Auto Spain d n Biom Centro de Investigacio edica en Red Enfermedades Respiratorias (CIBERES), Instituto de la Salud Carlos III (ISCIIII), Avenida de Monforte de Lemos, 3- 5, 28029, Madrid, Spain e Department of Neurology, Hospital Universitari Vall d’Hebron, Universitat Autonoma de Barcelona, Passeig Vall d’Hebron 119-129, 08035, Barcelona, Spain a r t i c l e i n f o a b s t r a c t Article history: Objective: Early diagnosis of obstructive sleep apnea (OSA) in children is important. The use of a nasal Available online 19 July 2022 cannula as an airflow sensor during polysomnography has not been evaluated in younger children. The study aims to evaluate the use of nasal cannula in detecting respiratory events in children under three Keywords: with suspected OSA during daytime nap studies. Pediatric sleep apnea Methods: A total of 185 patients were prospectively included. Respiratory events were scored using nasal OSA cannula alone, thermistor alone, and both methods simultaneously as the airflow sensor. Agreement and Nasal cannula diagnostic accuracy were assessed. Thermistor Polysomnography Results: One hundred and seventy-two children were finally analyzed and 110 (64.0%) presented OSA. Total sleep time with an uninterpretable signal was longer with the nasal cannula than with the thermistor (17.8% vs 1.9%; p < 0.001), and was associated with poor sensor tolerance and adenotonsillar hypertrophy. In the estimation of the apnea-hypopnea index, the nasal cannula showed lower agreement than the thermistor with the joint use of the two sensors (intraclass correlation coefficient: 0.79 vs 0.996 with thermistor). Compared with the thermistor, the nasal cannula presented lower sensitivity for detecting OSA (82.7% vs 95.5%) and a lower negative predictive value (76.5% vs 92.4%). Overall, fewer children were diagnosed with severe OSA with the nasal cannula (19.8% vs 30.8% with the thermistor, and 32.6% with both). Conclusions: In children under the age of three, the ability of the nasal cannula to detect obstructive events was relatively low. Therefore, other non-invasive measurements for identifying respiratory events during sleep may be of additional value. © 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction Sleep disordered breathing is a continuum of severity with primary snoring at the mild end and obstructive sleep apnea (OSA) * Corresponding author. Multidisciplinary Sleep Unit, Department of Clinical at the severe end [1]. Obstructive sleep apnea (OSA) is increasingly noma de Neurophysiology, Hospital Universitari Vall d’Hebron , Universitat Auto recognized in children and adolescents, with an estimated preva- Barcelona, Passeig Vall d’Hebron 119-129, 08035, Barcelona, Spain. lence between 1% and 4% [2]. Untreated OSA in children is associ- E-mail addresses: mjjuradoluque@gmail.com (M.J. Jurado), gsampol@vhebron. ated with cognitive and behavioral deficits, cardiovascular and net (G. Sampol), maquintana@vhebron.net (M. Quintana), oromero@vhebron.net metabolic dysfunction, enuresis and impaired growth [3e5]. For (O. Romero), rcambrodi@vhebron.net (R. Cambrodí), aferre@vhebron.net ), jsampol@vhebron.net (J. Sampol). (A. Ferre these reasons, an early and accurate diagnosis and management of https://doi.org/10.1016/j.sleep.2022.07.009 1389-9457/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). M.J. Jurado, G. Sampol, M. Quintana et al. Sleep Medicine 99 (2022) 41e48 respiratory sleep disorder might be underestimated or pass un- Abbreviation list recognized in this age group, leading to a delay in its appropriate treatment. AASM American Academy of Sleep Medicine Based on these considerations, the aim of this study is to assess BMI Body mass index the adequacy of the NC for detecting respiratory events in younger CPAP Continuous positive airway pressure children. Specifically, we evaluated the use of NC and Th, both EEG Electroencephalogram channels separately and in combination for detecting respiratory events in EMG Submental electromyogram children under 3 years of age with suspected OSA. EOG Electrooculogram NA/SP Nasopharyngeal airway/soft palate ratio 2. Methods NC Nasal cannula NIMV Non-invasive mechanical ventilation 2.1. Study population OAHI Obstructive apnea-hypopnea index OSA Obstructive sleep apnea A prospective observational study was performed in 256 PSG Polysomnography consecutive patients under the age of three referred to our Sleep RERA Respiratory effort-related arousals Unit for suspected sleep disorder. Children presenting clinically RERAI Respiratory effort-related arousal index suggestive symptoms of OSA e snoring, witnessed apneas and SpO2 Peripheral oxygen saturation mouth breathing e were included in the study. Those who used TcCO2 Transcutaneous carbon dioxide oxygen, continuous positive airway pressure (CPAP) or non- Th Thermistor invasive mechanical ventilation (NIMV), and children with symp- TRT Total recording time toms suggestive of sleep disorder other than OSA were excluded. TST Total sleep time The study was approved by the Hospital's Ethical Committee, and written informed consent was obtained from the parents/legal caretaker of all participating children. OSA in children is of the utmost importance. 2.2. Study design Nocturnal polysomnography (PSG) is the gold standard method for establishing the presence and severity of OSA in children of any All participants were evaluated by an expert in sleep medicine. age [3e5]. Polysomnographic manifestations differ in children and The evaluation protocol consisted of a clinical and sleep history adults with OSA, and specific criteria for staging sleep and scoring obtained from the parents/legal caretaker, a complete physical and respiratory events in children have been published [6,7]. Accurate ear, nose and throat examination, a lateral neck X-ray and an monitoring of airflow and respiratory effort is essential in order to attended PSG. The clinical data recorded were demographic vari- identify and quantify these respiratory events. ables, anthropometric measurements including weight, height, Pneumotachography via a snug-fitting mask is considered the body mass index (BMI) for sex/age z-score, tonsils and adenoid most accurate representation of airflow assessment. Nevertheless, grade, and co-morbidities including prematurity, obesity, gastro- it is not used for routine PSG due to the associated patient esophageal reflux, respiratory and neurologic co-morbidities. discomfort and its negative effect on sleep architecture. The rec- Obesity was defined if BMI z-score 2, using national reference ommended alternative is the use of non-invasive methods, namely values [26]. the oronasal thermal airflow sensor (Th) and the nasal pressure Adenoid hypertrophy was determined based on a lateral neck X- transducer using a nasal cannula (NC) [6]. Oronasal thermal sensors ray measuring the airway space immediately behind the upper part are considered adequate for the detection of apnea, but they un- of the soft palate (nasopharyngeal airway/soft palate (NA/SP) ratio) derestimate hypopnea because of their slow response time and according to the Cohen and Konak method [27]. Patients were marked non-linearity [8]. Studies in adults have shown that the NC graded as normal ¼ NA/SP ratio 1; mild-to-moderate has excellent agreement with a pneumotachograph [9e11] and hypertrophy ¼ NA/SP ratio 0.5e1; and severe hypertrophy ¼ NA/ intraesophageal pressure [12], and is more sensitive for the SP ratio <0.5. Tonsils were graded according to the Brodsky grading detection of apneas, hypopneas and respiratory effort-related scale [28] where 0 ¼ no tonsils visible, þ1¼ <25% of the oropharynx arousals (RERA) than Th, with reported differences ranging from occupied by the tonsils, þ2 ¼ 25e50% of the oropharynx occupied 30% to 50% [13e18]. by the tonsils, þ3 ¼ 50e75% of the oropharynx occupied by the However, few studies have assessed the validity of NC in chil- tonsils, and þ4¼ >75% of the oropharynx occupied by the tonsils. dren or compared its performance with that of Th [19e23]. Tonsils and adenoid grade variables were reduced as a single Consistent with research on adults, these few studies found NC to combined variable. Patients were classified in three groups: normal be more sensitive for detecting respiratory events than Th. A po- (NA/SP ratio 1 and tonsils grade 0 or þ1), mild-to-moderate hy- tential drawback to the use of NC in children is the possibility that pertrophy (NA/SP ratio 0.5e1 and/or tonsils grade þ2), and severe the airflow signal will be unreliable for a substantial amount of time hypertrophy (NA/SP ratio <0.5 and/or tonsils grade þ3 or þ4). during sleep. This failure of the NC signal has been attributed to displacement due to movement and intolerance of the NC in the 2.3. Polysomnography (PSG) nares, occlusion of the probe with nasal secretions, and mouth- breathing potentially related to the frequent presence of adeno- An attended PSG was performed in the sleep laboratory using an tonsillar hypertrophy and recurrent upper airway viral infections E-Series system (Compumedics Inc, Melbourne Abbotsford, [22,23]. All these problems are more likely in younger children, a Australia). The company of a parent/legal caretaker beside the child population with an increased risk for OSA, which can be severe was allowed, for greater patient comfort. Polysomnography was [24,25]. However, studies performed to date have included very few performed during a daytime nap from 09:00am to 02:30pm, after a cases of children younger than three years of age. Only Trang et al. night of partial sleep deprivation as described previously [29]. [23], in a study with a small sample of children, reported a notable Children were allowed to sleep in their preferred position. No presence of NC problems in this population. In this situation, the sedation was used. 42 M.J. Jurado, G. Sampol, M. Quintana et al. Sleep Medicine 99 (2022) 41e48 Monitoring included electroencephalogram channels (EEG), 2.5. Quality of flow tracings electrooculogram (EOG), intercostal and submental electromyo- gram (EMG), airflow, chest and abdominal movements using res- Quality of the NC and Th signals was visually assessed and was piratory inductive plethysmography, arterial oxygen saturation by defined as uninterpretable when no airflow signal was recorded for pulse oximetry (SpO2), transcutaneous carbon dioxide (TcCO2) 30 s while respiratory motion signals and SpO2 signal and values (TCM4, Radiometer, Copenhagen, Denmark), snoring by micro- remained unchanged. Time spent with an uninterpretable airflow phone, electrocardiography, body position and simultaneous video signal with Th and NC was expressed as a percentage of TST. recording. The airflow signal was simultaneously monitored using an oronasal thermistor (Th) (Protech Services Inc, Murrysville, PA, 2.6. Statistical analysis USA) and a pediatric nasal cannula (NC) (Salter Labs, Arvin, Cali- fornia; USA) connected to an AC pressure transducer with filter Data analysis was carried out using the software IBM SPSS Sta- settings recommended by the AASM [6], low-frequency filter 0.1Hz tistics 26.0 (SPSS Inc., Chicago, IL, USA). Continuous variables were and high-frequency filter 15Hz for Th, and 0.03Hz and 100Hz expressed as median (IQR) unless otherwise indicated, while cat- respectively for NC. egorical variables were reported as absolute numbers and per- centages. As most quantitative variables were not normally distributed, non-parametric tests were applied in all comparisons. Differences between NC and Th in airflow sensor repositioning were evaluated using the McNemar test. Comparison of sleep time 2.4. Scoring criteria and data analysis with uninterpretable signal between NC and Th was performed using the Wilcoxon test. Factors associated with an uninterpretable Sleep stages, arousals and respiratory events were scored ac- signal 25% of TST were evaluated using Pearson's chi-square or cording to standard criteria [6,7]. Respiratory events were manually Fisher's exact test for categorical variables and the linear-trend chi- scored in 2-min epochs using three different respiratory montages square test for ordinal variables. A logistic regression analysis was for each patient based on which airflow signal was displayed on the performed to obtain variables independently associated with a computer screen: NC þ Th, Th alone and NC alone. This procedure longer uninterpretable signal time. The number of respiratory was performed by the same expert in sleep medicine (MJJ) blinded events detected using the different methods was compared using to the results of each particular participant. The three analyses for Friedman's test followed by a pairwise post-hoc analysis with each patient were scored in random order and were performed on Bonferroni correction. Differences in OSA severity were assessed three different days, at least seven days apart. Obstructive apnea using the McNemar-Bowker test. was defined as the cessation of airflow (<10% of baseline level) for The agreement between respiratory events obtained with Th, at least the duration of two breaths associated with respiratory NC and both sensors was evaluated using the intraclass correlation effort. Central apnea was defined as the cessation of airflow (<10% coefficient (ICC). The diagnostic accuracy of Th and NC was evalu- of baseline level) with absent inspiratory effort for at least the ated by obtaining receiver operating characteristic (ROC) curves duration of two breaths and followed by an arousal or an oxygen and by calculating the sensitivity, specificity, positive and negative desaturation 3% or for at least 20 s in the absence of any associ- predictive values, and likelihood ratios for different cut-off points of ated arousal or oxygen desaturation events. Mixed apnea was the OAHI. A p-value <0.05 was considered statistically significant. recorded if apnea criteria were met for at least the duration of two breaths with absent respiratory effort during one portion of the 3. Results event and the presence of inspiratory effort in another portion. Hypopnea was defined as a decrease in airflow 30%, for the A flow-chart of the study is shown in Fig. 1. Of the 185 children duration of at least two breaths and associated with an arousal or included in the study, ten (5.5%) did not tolerate the placement of 3% oxygen desaturation. Hypopnea was scored as obstructive if the NC from the beginning of the PSG recording, and one of them snoring, increased inspiratory flattening of the NC signal or an (0.5%) did not tolerate the Th (p ¼ 0.04). In three patients the PSG associated thoracoabdominal paradox were present during the could not be assessed due to technical problems. Thus, 172 PSG event, and as a central character when none of them were present. records included both flow sensors and were available for analysis. Although the NC is the sensor of choice for detecting hypopneas The characteristics of the sample finally included are shown in and the Th for apneas, in case of failure of the sensor of choice we Table 1, and the main results of the PSG in Table 2. The prevalence of use the other flow sensor as an alternative as has been recom- OSA was 64.0%, and the proportions of children with mild, mod- mended. Respiratory effort-related arousals (RERA) were defined as erate, and severe OSA were 23.3%, 8.1%, and 32.6% respectively. a sequence of breaths lasting 2 breaths characterized by increasing respiratory effort, flattening of the inspiratory portion of 3.1. Flow signal quality and interpretability the NC waveform or snoring, that did not meet criteria for an apnea or hypopnea and lead to an arousal from sleep. During the sleep study, the intervention of the technician was The obstructive apneaehypopnea index (OAHI) was calculated frequently required for the repositioning of both the NC and the Th based on the number of obstructive and mixed apneas and hypo- (45.3% of patients and 40.1% respectively, p ¼ 0.306). The number of pneas per hour of total sleep time (TST). OSA was diagnosed with times the sensor was repositioned was 0 (0e2) for the NC and 0 (0e1) OAHI 1, and was classified as mild when OAHI 1e4.9, moderate for the Th (p ¼ 0.018). Repositioning more than twice was required in when OAHI 5e9.9 and severe when OAHI 10, according to in- 9.9% of patients in the case of NC and in 4.7% of patients in the case of ternational guidelines [3e5]. Th (p ¼ 0.022). In most cases, the repositioning of both the NC and the Airflow sensors were repositioned if displacement from their Th was due to a poor tolerance of the sensor with voluntary position resulted in absent or poor quality signal. The NC was displacement of the airflow sensor during a period of awakening replaced if the probe was blocked by excessive nasal secretions. (36.5% vs 29.1% respectively, p ¼ 0.047), and to a lesser extent to the These interventions during sleep were performed mainly during displacement of the sensor due to movements during sleep (4.7% vs periods of slow wave sleep in order to minimize patient distur- 11.0%, p ¼ 0.027). Furthermore, NC required repositioning or changing bances or during spontaneous awakenings. due to secretion obstruction in three patients (4.1%). 43 M.J. Jurado, G. Sampol, M. Quintana et al. Sleep Medicine 99 (2022) 41e48 Fig. 1. Flow chart of the study. OSA: obstructive sleep apnea; NREM: non-rapid eye movement; RLS: restless legs syndrome; RMD: rhythmic movement disorder; NIMV: non-invasive mechanical ventilation; PSG: polysomnography; Th: thermistor; NC: nasal cannula. Table 1 Table 2 Demographic and clinical characteristics of the study population (n ¼ 172). Polysomnographic measures (n ¼ 172). Variable Variable; median (IQR) or n (%) Age [years]; mean (SD) 2.3 (1.2) TRT [min] 262.5 (228.2e317.6) Sex [M/F]; n (%) 103 (60)/69 (40) TST [min] 184.3 (146.0e239.1) Ethnicity; n (%) Sleep efficiency [%] 79.3 (64.0e89.0) White 140 (81.4) Stage NREM [%TST] 82.9 (78.8e88.7) Hispanic 27 (15.7) Stage NREM 1 [%TST] 9.4 (5.8e13.9) Black 4 (2.3) Stage NREM 2 [%TST] 40.0 (31.4e47.5) Asian 1 (0.6) Stage NREM 3 [%TST] 32.9 (26.8e40.2) Prematurity; n (%) 29 (16.9) Stage REM [%TST] 17.0 (11.3e21.3) BMI [Kg/m2]; mean (SD) 16.5 (2.5) Patients without stage REM 11 (6.4) BMI z-score; mean (SD) 0.1 (1.7) Sleep cycles [No.] 3.0 (2.0e4.0) Obesity; n (%) 21 (12.2) Arousal index [No./h] 15.4 (10.9e25.1) Gastroesophageal reflux; n (%) 21 (12.2) Respiratory arousal index, [No./h] 3.3 (0.7e12.7) Tonsils and adenoid grade; n (%) AHI [events/h] 4.9 (1.8e21.1) No hypertrophy 42 (24.4) OAHI [events/h] 4.0 (0.3e17.0) Mild-moderate 44 (25.6) AI [events/h] 1.4 (0.3e3.4) Severe 86 (50.0) HI [events/h] 2.2 (0.3e13.9) Respiratory co-morbidities; n (%) 99 (57.6) RERAI [events/h] 0.0 (0.0e0.4) Recurrent upper respiratory infection 93 (54.1) SpO2 mean [%] 97.0 (96.0e98.0) Laryngomalacia 3 (1.7) SpO2 nadir [%] 89.0 (85.2e93.0) Bronchopulmonary dysplasia 2 (1.2) CT90 [%] 0.0 (0.0e0.3) Asthma 1 (0.6) ODI3 [No./h] 2.6 (0.9e8.8) Neurologic co-morbidities; n (%) 46 (26.7) TcCO2 [mmHg] 38.0 (36.0e40.0) Developmental delay 28 (16.3) TcCO2>50 [%TST] 0.0 (0.0e0.2) Epilepsy 4 (2.3) TRT: total recording time; TST: total sleep time; NREM: non-rapid eye movement; Genetic conditions 36 (20.9) REM: rapid eye movement; AHI: apnea-hypopnea index; OAHI: obstructive apnea- Neuromuscular disorders 7 (4.1) hypopnea index; AI: apnea index; HI: hypopnea index; RERAI: respiratory effort- Metabolic disorders 2 (1.2) related arousal index; SpO2: peripheral oxygen saturation; CT90: percentage of Neurodevelopmental disorder 10 (5.8) total sleep time with saturation under 90%; ODI3: 3% oxygen desaturation index; Congenital hindbrain abnormalities 9 (5.2) TcCO2: basal transcutaneous carbon dioxide; TcCO2>50: percentage of total sleep M/F: Male/Female; BMI: body mass index; BMI z-score: BMI standard deviation; SD: time with transcutaneous carbon dioxide value exceeding 50mmHg; IQR: inter- standard deviation. quartile range. 44 M.J. Jurado, G. Sampol, M. Quintana et al. Sleep Medicine 99 (2022) 41e48 Overall, the time with interpretable flow signal was shorter with poorer tolerance of the sensor and the poor quality signal associ- the NC. Percentage of total recording time (TRT) with uninter- ated with the presence of adenotonsillar hypertrophy. There is a pretable signal was 21.8% (8.7e55.2) with NC and 5.8% (1.9e17) rationale for the current practice of using both sensors because one with Th (p < 0.001). Differences were also observed when consid- may not work for periods during the study; however, our results ering the percentage of TST, 17.8% (5.3e49.7) with NC vs 1.9% show some limitations associated with the use of NC in this age (0e8.5) with Th (p < 0.001); NREM sleep time, 15.9% (3.5e44.5) vs group. 1.4% (0e8.5) (p < 0.001); and REM sleep time, 6.2% (0e34.2) vs 0% In our patients, the use of NC as the sole flow sensor led to an (0e0.5) (p < 0.001). The signal was uninterpretable 25% of TST in underdiagnosis of OSA and an underestimation of its severity. 40.7% of the PSG studies with the NC, compared to 8.7% with Th These findings contrast with those of multiple studies in adults, (p < 0.005). Age 1 year and the presence of severe adenotonsillar which have reported a greater sensitivity with NC for identifying hypertrophy were the only clinical variables associated with a respiratory events than with Th [13e18,30]. Studies in children longer time with an uninterpretable NC signal (Table 3), and in a have also found a good correlation of respiratory events identified logistic regression model only severe adenotonsillar hypertrophy by NC with the increase in respiratory effort identified by esopha- was associated with non-interpretability  25% of the TST, with an geal manometry [19,23], as well as a greater sensitivity than Th for odds ratio of 2.01 (95% CI: 1.29e3.14) (p < 0.002). the identification of respiratory events [19,23]. However, these studies evaluated small samples and included very few children 3.2. Detection of respiratory events under three years of age [19,21e23] or none at all [20]. Additionally, in their comparisons of NC and Th some studies have included only Table 4 summarizes the overall values of respiratory events polysomnographic tracings with a good quality signal for both obtained using the three methods, NC, Th and the combination sensors [20]. In our study, in contrast, we assessed a wide series of NC þ Th. NC detected fewer respiratory events due to periods patients with frequent neurologic and respiratory co-morbidities without an interpretable signal (Fig. 2). In contrast, Th identified a and a wide range of OAHI values, which reflect the challenging similar number of events to the NC þ Th combination. We did not situations encountered at a sleep unit in this age range. We found detect central hypopneas in our patients. Of the 110 children with that NC performed worse than Th in detecting upper airway OSA, the use of Th alone allowed diagnosis of 105 (95.5%), while the obstructive events because its potential greater sensitivity in NC alone diagnosed 91 (82.7%) (p < 0.001). detecting these events was outweighed by a greater available time Table 5 shows the intraclass correlation coefficients between of flow signal with Th. NC, Th and the combination of the two. Th presented greater reli- The frequent loss of flow signal during unsupervised home PSG ability than NC in estimating OAHI. in children has already been reported in a population-based study, Table 6 shows the sensitivity, specificity, predictive values, in which only 42% of records showing an interpretable flow signal likelihood ratio values and the area under the ROC curve (AUC) of both NC and Th for more than 47% of TST [20]. Similarly, Marcus obtained with each sensor separately for different OAHI cut-off et al. [31], also using unattended home PSG in children aged 5e12 points. The diagnostic capacity of Th was higher than that of NC, years without suspected OSA, identified the NC as the most prob- with higher values of sensitivity and AUC. The lower sensitivity of lematic sensor with only 67% of patients with more than 75% of the the NC at the different cut-off points of the OAHI affected the study with a satisfactoy NC signal; however, the combination of NC classification of OSA severity, with a lower proportion of severe and Th provided the presence of some flow signal for more than cases (19.8% using the NC alone vs 30.8% using the Th alone, 75% of the night in 96% of the subjects. With an attended PSG, 40% p < 0.001)), and a higher proportion of non-OSA cases (47.1% vs of our patients presented an uninterpretable NC signal for more 38.4%, respectively (p < 0.001)). than 25% of TST, while with Th this occurred in only 8% of cases. These findings are consistent with a previous study in 30 children with suspected OSA in whom problems with the flow signal were 4. Discussion mainly observed with the NC and were more frequent in the sub- group under three years of age [23]. In this study in children under the age of three with suspected We identified poor sensor tolerance and adenotonsillar hyper- OSA, the capacity of NC to detect respiratory events was lower than trophy as factors related to the presence of an uninterpretable NC that of Th, due to the presence of a longer sleep time with an un- signal. During PSG, both NC and, to a lesser extent Th, required interpretable signal. The factors underlying this limitation were the Table 3 Clinical variables and sleep time with uninterpretable signal for nasal cannula (NC). TST with uninterpretable NC signal <25% (n ¼ 102) 25% (n ¼ 70) P Value Age [years]; n (%) 0.003 <1 year 25 (24.5) 5 (7.1) 1 year 77 (75.5) 65 (92.9) Obesity; n (%) 13 (12.9) 8 (11.4) 0.818 Recurrent upper respiratory infection; n (%) 49 (48.0) 44 (62.9) 0.055 Gastroesophageal reflux; n (%) 15 (14.7) 6 (8.6) 0.227 Prematurity; n (%) 18 (17.6) 11 (15.7) 0.739 Adenotonsillar grade; n (%) <0.001 Normal 34 (33.3) 8 (11.4) Mild-to-moderate hypertrophy 31 (30.4) 13 (18.6) Severe hypertrophy 37 (36.3) 49 (70.0) Developmental delay; n (%) 19 (18.6) 13 (18.6) 0.993 Genetic conditions; n (%) 13 (12.7) 12 (17.1) 0.421 TST: total sleep time; OSA: obstructive sleep apnea; NC: nasal cannula. 45 M.J. Jurado, G. Sampol, M. Quintana et al. Sleep Medicine 99 (2022) 41e48 Table 4 Respiratory events identified using the three methods. NC þ Th NC Th P Value OAHI [events/h]; median (IQR) 2.5 (0.3e16.8) 1.4 (0.0e7.8) 1.9 (0.1e15.1) <0.001 AI [events/h]; median (IQR) 1.4 (0.3e3.4) 1.0 (0.0e2.9) 1.5 (0.3e3.4) <0.001 HI [events/h]; median (IQR) 2.2 (0.3e13.9) 1.0 (0.0e5.8) 1.7 (0.1e11.6) <0.001 OAHI: obstructive apnea-hypopnea index; AI: apnea index; HI: hypopnea index; Th: thermistor; NC: nasal cannula; IQR: interquartile range. Fig. 2. Polysomnographic views of the behaviour of nasal cannula and thermistor during hypopneas in children under the age of three with obstructive sleep apnea. A) Obstructive events detected by nasal cannula and thermistor; B) Obstructive hypopnea detected by nasal cannula. It could be missed if relying only in thermistor; C) Period without nasal cannula signal due to displacement of the sensor by the patient. Only hypopneas detected by the less sensitive thermistor are considered. frequent intervention by the sleep technician for their reposition- without waking the child. It may also be the case that the NC signal, ing due to voluntary or involuntary displacement by the patient. dependent on the changes in nasal pressure, is more likely to be lost Previously, studying the two sensors in a series of ten children with due to displacement or blockage by secretions than the Th signal, a mean age of five years Verginis and cols [22] reported that the NC which is sensitive to the temperature of the nasal and mouth flow. was more bothersome and difficult to reposition during sleep As in other series of patients of a similar age with suspected OSA, a 46 M.J. Jurado, G. Sampol, M. Quintana et al. Sleep Medicine 99 (2022) 41e48 Table 5 age range evaluated in our work. Other possible consequences of Intermeasurement agreement using nasal cannula (NC), thermistor (Th) and both the underestimation of OSA would be an increase in respiratory (NC þ Th) in estimation of OAHI. complications as well as a higher residual OAHI associated with ICC 95% CI adenotonsillectomy as a treatment for OSA, which have been re- Between NC þ Th and Th 0.995 0.993e0.997 ported to be more frequent in children under three years of age Between NC þ Th and NC 0.618 0.482e0.718 [39,40]. Between NC and Th 0.609 0.485e0.705 Our study has some limitations. First, we used a daytime nap, an ICC: intraclass correlation coefficient; CI: confidence interval; Th: thermistor; NC: option previously applied successfully in children [29,39e41]. The nasal cannula. aim of studying children under three years of age in this way in our sleep unit is motivated by our intention to reduce the long waiting list. This procedure have limited the sleep time in our patients, as high percentage of our participants had neurological and respira- previously described with the same methodology by Trang et al. tory co-morbidities that may have affected in some degree the studying flow signals in children <12 months [23]. In addition, to problems related to flow sensors tolerance. maximize sleep time in our patients, 10 children who did not We also identified the presence of severe adenotonsillar hy- tolerate the NC at the beginning of the study were excluded from pertrophy as a factor associated with poor NC signal. Adenotonsillar the analysis without trying to reposition the NC when the child is hypertrophy is known to be a risk factor for the development of already asleep as we usually do in overnight sleep studies, reducing OSA in children [32], particularly in predominantly non-obese the studied sample. We believe that these facts did not affect the populations such as ours and in children of similar age; its influ- main results of our study; however, we cannot rule out that over- ence is lower in older children, who have a proportionally larger night studies, with longer TST and more REM sleep [42e44], could upper airway and in whom the role of obesity in the development lead to different results. Second, we did not use intraesophageal of OSA is more relevant [24,33e35]. In adults, mouth breathing due pressure sensors to detect respiratory effort, or a pneumotacho- to nasal obstruction has been associated with a poor NC signal but graph to detect airflow; their use would have provided us with an with an unchanged Th signal [15,16]. However, the influence of accurate OAHI value of our patients, but they are intrusive pro- adenotonsillar hypertrophy on the flow signal during PSG has not cedures that may cause sleep fragmentation and are thus unsuit- been previously studied. Supporting our findings, it is known that able in routine PSG studies, especially in children. Third, to avoid a adenotonsillar hypertrophy, frequently associated with recurrent potential interaction between both sensors when measuring upper airway infections, and the presence of neurological syn- changes in airflow, an alternative design of our study would have dromes or anatomical alterations related to OSA have been asso- been to compare both sensors in 2 studies performed with each ciated with mouth breathing in children under 3 years of age [24]. sensor separately. However, our methodology using both sensors Although the combined use of the NC and the Th during PSG is simultaneously is the usual one in clinical practice and we believe the accepted gold standard, our findings suggest a possible un- our results reflect a problem associated with this practice. And derdiagnosis and underestimation of OSA in children under three fourth, all children studied were referred due to suspicion of OSA years of age because some obstructive events would remain un- and frequently presented neurological and respiratory co- detected due to the lack of NC signal and the lower sensitivity of Th morbidities, making sleep studies more difficult. This implies that to detect obstructive hypopnea. It can be speculated that this might our results can not be extrapolated to asymptomatic children. be associated with an increase in the neurocognitive, cardiovas- In conclusion, in children under the age of three, the ability of cular and metabolic effects that have been associated with child- NC to detect obstructive episodes during sleep was lower than hood OSA in recent years [1e4]. Although respiratory polygraphy is previously reported in older children and adults. This is related to a not recommended in children <2 years [5] and may be considered poor tolerance of the sensor and the frequent presence of adeno- an alternative if PSG is not available in children >2 years [4], our tonsillar hypertrophy in these patients, and may lead to an un- results may be especially noteworthy if we use simplified diag- derestimation of OSA due to undetected events due to the absence nostic methods that are available for home studies and that use NC of NC signal and less sensitivity of Th in detecting obstructive as the sole flow sensor [36]. The clinical value of analyzing sleep hypopneas. Therefore, other non-invasive measurements to iden- studies in these patients considering exclusively the time with a tify respiratory events during sleep may be of additional value. correct signal from both sensors should also be evaluated. In addition, the accuracy of home respiratory polygraphy using NC Funding together with Th [37] or respiratory inductance plethysmography [38] for the diagnosis of OSA should be specifically assessed in the This research did not receive any specific grant from funding Table 6 Diagnostic accuracy of nasal cannula (NC) and thermistor (Th) for detecting the different grades of OAHI. NC Th OAHI 1 OAHI 5 OAHI 10 OAHI 1 OAHI 5 OAHI 10 n (%) 91 (52.9) 52 (30.2) 34 (19.8) 106 (61.6) 68 (39.5) 53 (30.8) Sensitivity (%) 82.7 74.3 60.7 95.5 97.1 94.6 Specificity (%) 100 100 100 98.4 100 100 PPV (%) 100 100 100 99.1 100 100 NPV (%) 76.5 85.0 84.1 92.4 98.1 97.5 LRþ ∞ ∞ ∞ 59 ∞ ∞ LR- 0.17 0.26 0.39 0.05 0.03 0.05 AUC (95%CI) 0.909 0.875 0.863 0.985 0.999 0.998 (0.864e0.955) (0.808e0.942) (0.787e0.939) (0.967e1) (0.997e1) (0.995e1) OAHI: obstructive apnea-hypopnea index; PPV: positive predictive value; NPV: negative predictive value; LRþ: positive likelihood ratio; LR-: negative likelihood ratio; AUC: area under the ROC curve; Th: thermistor; NC: nasal cannula. 47 M.J. Jurado, G. Sampol, M. Quintana et al. Sleep Medicine 99 (2022) 41e48 agencies in the public, commercial, or not-for-profit sectors. [17] Teichtahl H, Cunnington D, Cherry G, Wang D. Scoring polysomnography respiratory events: the utility of nasal pressure and oro-nasal thermal sensor recordings. Sleep Med 2003;4:419e25. CRediT authorship contribution statement [18] Ballester E, Badia JR, Herna ndez L, Farre R, Navajas D, Montserrat JM. Nasal prongs in the detection of sleep-related disordered breathing in the sleep María Jose Jurado: Conceptualization, Formal analysis, Data apnoea/hypopnoea syndrome. Eur Respir J 1998;11:880e3. [19] Serebrisky D, Cordero R, Mandeli J, Lamm C. Assessment of inspiratory flow curation, Methodology, Visualization, Writing - original draft, limitation in children with sleep-disordered breathing by a nasal cannula Writing - review & editing. Gabriel Sampol: Conceptualization, pressure transducer system. Pediatr Pulmonol 2002;33:380e7. Formal analysis, Data curation, Methodology, Supervision, Visuali- [20] Budhiraja R, Goodwin JL, Parthasarathy S, Quan SF. Comparison of nasal pressure transducer and thermistor for detection of respiratory events during zation, Writing - original draft, Writing - review & editing. Manuel polysomnography in children. Sleep 2005;28:1117e21. Quintana: Formal analysis, Writing - review & editing. Odile [21] Griffiths A, Maul J, Wilson A, Stick S. Improved detection of obstructive events Romero: Formal analysis, Visualization, Writing-review, and edit- in childhood sleep apnoea with the use of the nasal cannula and the differ- entiated sum signal. J Sleep Res 2005;14:431e6. ing. Roser Cambrodí: Formal analysis, Visualization, Writing- [22] Verginis N, Davey MJ, Horne RSC. Scoring respiratory events in paediatric review, and editing. Alex Ferre : Formal analysis, Visualization, patients: evaluation of nasal pressure and thermistor recordings separately Writing-review, and editing. Júlia Sampol: Formal analysis, Visu- and in combination. Sleep Med 2010;11:400e5. [23] Trang H, Leske V, Gaultier C. Use of nasal cannula for detecting sleep apneas alization, Writing - review & editing. and hypopneas in infants and children. Am J Respir Crit Care Med 2002;166: 464e8. Declaration of competing interest [24] Katz ES, Mitchell RB, D'Ambrosio CM. Obstructive sleep apnea in infants. Am J Respir Crit Care Med 2012;185:805e16. [25] Don DM, Geller KA, Koempel JA, Ward SD. Age specific differences in pediatric The authors have no disclosures or financial conflicts of interest. obstructive sleep apnea. Int J Pediatr Otorhinolaryngol 2009;73:1025e8. [26] Carrascosa A, Ferna ndez J, Ferra ndez A, Lo pez Siguero J, Lopez D, Sanchez E. Estudios espan ~ oles de Crecimiento 2010. Rev Esp Endocrinol Pediatr Acknowledgments 2011;2(Suppl):53e8. [27] Cohen D, Konak S. The evaluation of radiographs of the nasopharynx. Clin We are extremely grateful to all the families who took part in Otolaryngol Allied Sci 1985;10:73e8. this study. We also thank our sleep technicians and nurses, for their [28] Brodsky L. Modern assessment of tonsils and adenoids. Pediatr Clin 1989;36: 1551e69. assistance in performing the sleep studies. [29] Marcus CL, Keens TG, Ward SD. Comparison of nap and overnight poly- somnography in children. Pediatr Pulmonol 1992;13:16e21. References [30] BaHammam A. Comparison of nasal prong pressure and thermistor mea- surements for detecting respiratory events during sleep. Respiration 2004;71: 385e90. [1] Benjafield AV, Ayas NT, Eastwood PR, et al. Estimation of the global prevalence [31] Marcus CL, Traylor J, Biggs SN, et al. Feasibility of comprehensive, unattended and burden of obstructive sleep apnoea: a literature-based analysis. Lancet ambulatory polysomnography in school-aged children. J Clin Sleep Med Respir Med 2019;7:687e98. 2014;10(8):913e8. [2] Lumeng JC, Chervin RD. Epidemiology of pediatric obstructive sleep apnea. [32] Gulotta G, Iannella G, Vicini C, et al. Risk factors for obstructive sleep apnea Proc Am Thorac Soc 2008;5:242e52. syndrome in children: state of the art. Int J Environ Res Publ Health 2019;16: [3] Marcus CL, Brooks LJ, Draper KA, et al. American Academy of Pediatrics. 3235. Diagnosis and management of childhood obstructive sleep apnea syndrome. [33] Selvadurai S, Voutsas G, Propst EJ, Wolter NE, Narang I. Obstructive sleep Pediatrics 2012;130:e714e55. apnea in children aged 3 years and younger: rate and risk factors. Paediatr [4] Kaditis AG, Alonso Alvarez ML, Boudewyns A, et al. Obstructive sleep disor- Child Health 2020;25:432e8. dered breathing in 2- to 18-year-old children: diagnosis and management. Eur [34] Dayyat E, Kheirandish-Gozal L, Sans Capdevila O, Maarafeya MMA, Gozal D. Respir J 2016;47:69e94. Obstructive sleep apnea in children: relative contributions of body mass index [5] Kaditis AG, Alonso Alvarez ML, Boudewyns A, et al. ERS statement on and adenotonsillar hypertrophy. Chest 2009;136:137e44. obstructive sleep disordered breathing in 1- to 23-month-old children. Eur [35] Kang KT, Chou CH, Weng WC, Lee PL, Hsu WC. Associations between ade- Respir J 2017;50:1700985. notonsillar hypertrophy, age, and obesity in children with obstructive sleep [6] Berry RB, Brooks R, Gamaldo C, et al. AASM scoring manual updates for 2017. apnea. PLoS One 2013;8:1e8. Version 2.4. J Clin Sleep Med 2017;13:665e6. novas C, Molina V, Sancho-Chust JN, Van ~ es S, Pastor E, Martinez- [36] Chiner E, Ca [7] Anders T, Emde R, Parmelee A. A manual of standarized terminology, tech- Garcia MA. Home respiratory polygraphy is useful in the diagnosis of child- niques and criteria for scoring of states of sleep and wakefulness in newborn hood obstructive sleep apnea syndrome. J Clin Med 2020;9(7):2067. infants. Los Angeles, CA: UCLA Brain Information Service/BRI Publications  [37] Alonso-Alvarez ML, Teran-Santos J, Ordax-Carbajo E, Cordero-Guevara JA, Office, NINDS Neurological Information Network; 1971. Navazo-Egüia AI, Kheirandish-Gozal L, Gozal D. Reliability of home respiratory [8] Farre R, Montserrat JM, Rotger M, Ballester E, Navajas D. Accuracy of therm- polygraphy for the diagnosis of sleep apnea in children. Chest 2015;147(4): istors and thermocouples as flow-measuring devices for detecting hypo- 1020e8. pnoeas. Eur Respir J 1998;11:179e82. [38] Gudnadottir G, Hafsten L, Redfors S, Ellegård E, Hellgren J. Respiratory pol- [9] Heitman SJ, Atkar RS, Hajduk EA, Wanner RAFW. Validation of nasal pressure ygraphy in children with sleep-disordered breathing. J Sleep Res 2019;28(6): for the identification of apneas/hypopneas during sleep. Am J Respir Crit Care e12856. Med 2002;166:386e91. [39] Nosetti L, Zaffanello M, De Bernardi F, et al. Age and upper airway obstruction: [10] Thurnheer R, Xie X, Bloch KE. Accuracy of nasal cannula pressure recordings a challenge to the clinical approach in pediatric patients. Int J Environ Res Publ for assessment of ventilation during sleep. Am J Respir Crit Care Med Health 2020;17:3531. 2001;164:1914e9. [40] Aurora RN, Zak RS, Karippot A, et al. American Academy of Sleep Medicine. [11] Hosselet JJ, Norman RG, Ayappa I, Rapoport DM. Detection of flow limitation Practice parameters for the respiratory indications for polysomnography in with a nasal cannula/pressure transducer system. Am J Respir Crit Care Med children. Sleep 2011;34:379e88. 1998;157:1461e7. [41] Saeed MM, Keens TG, Stabile MW, Bolokowicz J, Davidson Ward SL. Should [12] Ayappa I, Norman RG, Krieger AC, Rosen A, O’malley RL, Rapoport DM. Non- children with suspected obstructive sleep apnea syndrome and normal nap invasive detection of respiratory effort-related arousals (RERAs) by a nasal sleep studies have overnight sleep studies? Chest 2000;118:360e5. cannula/pressure transducer system. Sleep 2000;23:763e71. [42] Scholle S, Wiater A, Scholle HC. Normative values of polysomnographic pa- [13] Herna ndez L, Ballester E, Farre  R, et al. Performance of nasal prongs in sleep rameters in childhood and adolescence: cardiorespiratory parameters. Sleep studies : spectrum of flow-related events. Chest 2001;119:442e50. Med 2011;12(10):988e96. [14] Norman RG, Ahmed MM, Walsleben JA, Rapoport DM. Detection of respiratory [43] Selvadurai S, Voutsas G, Propst EJ, Wolter NE, Narang I. Obstructive sleep events during NPSG: nasal cannula/pressure sensor versus thermistor. Sleep apnea in children aged 3 years and younger: rate and risk factors. Paediatr 1997;20:1175e84. Child Health 2019 Jul 10;25(7):432e8. [15] Serie s F, Marc I. Nasal pressure recording in the diagnosis of sleep apnoea [44] Rayasam S, Johnson R, Lenahan D, Abijay C, Mitchell RB. Obstructive sleep hypopnoea syndrome. Thorax 1999;54:506e10. apnea in children under 3 Years of age. Laryngoscope 2021 Sep;131(9): [16] Epstein MD, Chicoine SA, Hanumara RC. Detection of upper airway resistance E2603e8. syndrome using a nasal cannula/pressure transducer. Chest 2000;117: 1073e7. 48