Postoperative respiratory monitoring aims to detect. Pediatric Anesthesiology

Transcription

1 Pediatric Anesthesiology Section Editor: James A. DiNardo Comparison of Postoperative Respiratory Monitoring by Acoustic and Transthoracic Impedance Technologies in Pediatric Patients at Risk of Respiratory Depression Mario Patino, MD,* Megan Kalin, MS, CCRP, * Allison Griffin, MS, Abu Minhajuddin, PhD, Lili Ding, PhD, Timothy Williams, MS, Stacey Ishman, MD, MPH, Mohamed Mahmoud, MD,* C. Dean Kurth, MD,*# and Peter Szmuk, MD** BACKGROUND: In children, postoperative respiratory rate (RR) monitoring by transthoracic impedance (TI), capnography, and manual counting has limitations. The rainbow acoustic monitor (RAM) measures continuous RR noninvasively by a different methodology. Our primary aim was to compare the degree of agreement and accuracy of RR measurements as determined by RAM and TI to that of manual counting. Secondary aims include tolerance and analysis of alarm events. METHODS: Sixty-two children (2 16 years old) were admitted after tonsillectomy or receiving postoperative patient/parental-controlled analgesia. RR was measured at regular intervals by RAM, TI, and manual count. Each TI or RAM alarm resulted in a clinical evaluation to categorize as a true or false alarm. To assess accuracy and degree of agreement of RR measured by RAM or TI compared with manual counting, a Bland Altman analysis was utilized showing the average difference and the limits of agreement. Sensitivity and specificity of RR alarms by TI and RAM are presented. RESULTS: Fifty-eight posttonsillectomy children and 4 patient/parental-controlled analgesia users aged 6.5 ± 3.4 years and weighting 35.3 ± 22.7 kg (body mass index percentile 76.6 ± 30.8) were included. The average monitoring time per patient was 15.9 ± 4.8 hours. RAM was tolerated 87% of the total monitoring time. The manual RR count was significantly different from TI (P =.007) with an average difference ± SD of 1.39 ± 10.6 but were not significantly different from RAM (P =.81) with an average difference ± SD of 0.17 ± 6.8. The proportion of time when RR measurements differed by 4 breaths was 22% by TI and was 11% by RAM. Overall, 276 alarms were detected (mean alarms/patient = 4.5). The mean number of alarms per patient were 1.58 ± 2.49 and 2.87 ± 4.32 for RAM and TI, respectively. The mean number of false alarms was 0.18 ± 0.71 for RAM and 1.00 ± 2.78 for TI. The RAM was found to have 46.6% sensitivity (95% confidence interval [CI], ), 95.9% specificity (95% CI, ), 88.9% positive predictive value (95% CI, ), and 72.1% negative predictive value (95% CI, ), whereas the TI monitor had 68.5% sensitivity (95% CI, ), 72.0% specificity (95% CI, ), 59.0% positive (95% CI, ), and 79.5% negative predictive value (95% CI, ). CONCLUSIONS: In children at risk of postoperative respiratory depression, RR assessment by RAM was not different to manual counting. RAM was well tolerated, had a lower incidence of false alarms, and had better specificity and positive predictive value than TI. Rigorous evaluation of the negative predictive value is essential to determine the role of postoperative respiratory monitoring with RAM. (Anesth Analg 2017;124: ) Postoperative respiratory monitoring aims to detect changes in respiratory dynamics before life-threatening events occur. Patients at risk of respiratory complications include patients with obstructive sleep apnea (OSA), whose incidence of respiratory complications is as high as 27%, 1 and patients who are receiving IV patient/ parental-controlled analgesia (PCA). The American Society of Anesthesiologists recommendations for postanesthetic From the *Department of Anesthesiology, Cincinnati Children s Hospital Medical Center, Cincinnati, Ohio; Departments of Anesthesiology and Clinical Sciences, UT Southwestern and Children s Medical Center, Dallas, Texas; Division of Biostatistics and Epidemiology, Cincinnati, Ohio; University of Texas Southwestern and Children s Medical Center, Dallas, Texas; Division of Pediatric Otolaryngology, Cincinnati Children s Hospital Medical Center, Cincinnati, Ohio; #Department of Anesthesia and Pediatrics, University of Cincinnati College of Medicine, Cincinnati, Ohio; **Department of Anesthesiology, University of Texas Southwestern and Children s Medical Center, Dallas, Texas; and Dallas and Outcome Research Consortium, Cleveland, Ohio. Mario Patino, MD, is currently affiliated with Department of Pediatric Anesthesiology, Texas Children s Hospital, Department of Anesthesiology, Baylor College of Medicine, Houston, Texas. Copyright 2017 International Anesthesia Research Society DOI: /ANE C. Dean Kurth, MD, is currently affiliated with Department of Anesthesiology and Critical Care, Children s Hospital of Philadelphia, Philadelphia, Pennsylvania. Accepted for publication February 6, Funding: Masimo Corporation provided the equipment, sensors, and funds to support research coordinators efforts. No financial interest exists related to this project. The authors declare no conflicts of interest. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal s website ( Reprints will not be available from the authors. Address correspondence to Mario Patino, MD, Department of Anesthesiology, Perioperative and Pain Medicine, Texas Children s Hospital, 6621 Fannin St, A3300, Houston, TX Address to mario.patino@cchmc.org. June 2017 Volume 124 Number

2 Postoperatory Respiratory Monitoring respiratory monitoring include assessment of ventilation, respiratory rate (RR), and oxygen saturation (Spo 2 ). 2 These recommendations are particularly pertinent to the postanesthesia care unit (PACU), although these recommendations are relevant to at-risk patients in hospital after PACU. Postoperative assessment of ventilation and RR is usually performed by periodic clinical observation by nurses at bedside with pulse oximetry without continuous monitoring of ventilation. Monitoring RR alerts clinicians about the respiratory status of patients. 3 Given the serious patient safety risks and resultant morbidity and mortality seen after opioid administration, the Anesthesia Patient Safety Foundation advocates the respiratory monitoring strategy in patients receiving IV PCA therapy to include continuous monitoring of oxygenation and ventilation, especially if oxygen supplementation is administered to patients. 4 Despite the importance of monitoring ventilation continuously, the technologies to monitor ventilation have limitations and consequently variation exists in clinical monitoring practices between institutions. The most commonly used continuous respiratory monitors are transthoracic impedance (TI) and nasal capnography. TI uses chest electrodes that detect changes in electrical conductivity from chest movements, 5 but it is limited by the lack of direct airflow monitoring and high incidence of false alarms. 6 Nasal capnography has the advantage of recording end-tidal CO 2, but it is not well tolerated in children. 7 The rainbow acoustic monitor (RAM; Masimo Corporation, Irvine, CA) is a technology that continuously monitors RR through the detection of an acoustic signal from airflow (Figure 1). In a previous study of children after surgery in PACU, we observed that the RAM had similar RR accuracy as manual counting and capnography but better tolerance than capnography. 7 RAM is currently Food and Drug Administration-approved for continuous RR monitoring in adults and children with a weight greater than 10 kg. 8 In this study, we evaluated the degree of agreement, accuracy, and tolerance of the RAM and TI as compared with manual counting in postoperative children at risk for Figure 1. Recruitment diagram. respiratory depression during extended monitoring. As a secondary aim, we performed an analysis of the alarms recorded by each device. We hypothesized that RAM provides a better degree of agreement of RR monitoring than TI in comparison with manual counting with a decreased incidence of false alarms and an acceptable tolerance during the postoperative respiratory monitoring of children. METHODS This study was registered on October 1, 2014, with Clinical Trials.gov prior to patient enrollment (registration number NCT ), principal investigator Mario Patino. This is a prospective, pilot study to determine the feasibility of RAM monitoring in postoperative pediatric patients. This pilot data will allow for an outcomes study with a larger recruitment to evaluate postoperative respiratory monitoring with RAM. With IRB approval, we enrolled 62 patients at Children s Health: Children s Medical Centre Dallas (CMCD) and Cincinnati Children s Hospital Medical Center (CCHMC). We obtained written consent from parents and written assent from children 10 years and older (Figure 1). We included children from 2 to 16 years old with a weight greater than 10 kg who were either status after tonsillectomy with a diagnosis of OSA or obstructive breathing disorder that required postoperatory admission for continuous respiratory monitoring; or status after other surgery and receiving an opioid by patient- or PCA. Patients were excluded if skin abnormalities (rash, eczema, etc) were observed at the planned application site for the RAM sensor, if they had a planned admission to the intensive care unit, if they had a tracheostomy, or if they used a noninvasive ventilator support such as continuous positive airway pressure or bilevel positive airway pressure. Demographic data including age, weight, race, body mass index percentile by age, American Society of Anesthesiologists physical status, diagnosis, procedure, and comorbidities were collected. After arriving in the PACU or in the inpatient care unit, patients were monitored per routine (Spo 2, electrocardiogram, RR, temperature) using the GE Dash 3000/4000 (General Electric Company, Chicago, IL) at CCHMC or using the IntelliVue MP2 (Philips Inc, Andover, MA) at CMCD. In addition, an adhesive RAM bio-acoustic sensor (RAS 125, revision C; Masimo, Irvine, CA) and an adhesive, pediatric Spo 2 finger sensor (LNCS Pdtx; Masimo, Irvine, CA) were applied and connected to Pulse CO-Oximeter with RAM technology (Rad- 87, version , , and ; Masimo, Irvine, CA). The RAM sensor was applied to the lateral aspect of the neck anterior to the sternocleidomastoid muscle in an oblique way after cleaning and drying the skin (Figure 2). The Ramsay sedation score was documented at the moment of sensor placement. The placement of the sensor was most commonly done in patients with a Ramsay sedation scale from 2 to 6 to avoid early removal of the sensor. The RR was measured and recorded at the same time every 2 hours from the: (a) RAM monitor, (b) thoracic impedance (as part of the standard monitors), and (c) manual count by research personnel over 1 minute. In addition the presence of alarms and their evaluation as true or false was recorded during a fixed interval (15 minutes at CCHMC and 5 minutes at CMCD) ANESTHESIA & ANALGESIA

3 The description of the methods performed during this study as well as this manuscript adhere to the applicable Equator guidelines. Figure 2. Respiratory acoustic sensor placed in a patient s neck. To evaluate the degree of accuracy and agreement, RR recorded by RAM and TI were compared to manual counting. We assessed accuracy using the average of the pairs of RR measurements by RAM and TI monitor in comparison with manual counts. Bland Altman plots were created for the agreement in RR measurements of RAM and TI with manual counting. Also, we evaluated the frequency in which RR measured by RAM and TI were at least 4 breath per minute (bpm) greater than RR measured by manual counting as an assessment of clinically important disagreement. To evaluate tolerance, the total time of RAM and TI monitoring was recorded as well as the total time the RAM sensors were on the neck and measuring RR. Tolerance was defined as the ratio of total time sensor stayed in place over the total time expected of monitoring (either 24 hours or until the patient was discharged). Patients were also monitored by research personnel for 5 to 15 minutes every 2 hours during their stay at the PACU or patient room until their discharge. During these clinical observation periods, the presence of signal drop off for each device and the cause was categorized as patientrelated (lack of tolerance, agitation, disconnection to allow patient movement, etc), device-related (monitor failure, sensor malfunction), nurse-related (disconnection of the device to facilitate nursing care), or parental request (commonly when they perceived the child to be uncomfortable with the sensor or requested the data collection to be discontinued). For the purpose of the alarm analysis, we measured the frequency of true-positive alarms, true-negative alarms, false-positive alarms, and false-negative alarms to calculate sensitivity, specificity, positive predictive value, and negative predictive value. True-positive alarms were considered to be those that were triggered by a device and were verified by the clinician at the patient s bedside. False-positive alarms were considered to be those that were triggered by the device but were categorized as false alarms by the clinician at the bedside. True-negative alarms were recorded when a monitoring technology did not alarm while another monitor alarmed and was found to be a false alarm or when neither monitor activated an alarm. False-negative alarms were recorded when an event was verified by the clinical observation but the monitor did not alarm (see Supplemental Digital Content, Statistical Analysis This was a prospective, observational study to determine the degree of agreement, accuracy, tolerance, and alarm analysis with the use of RAM and TI in comparison to manual counting. We assessed accuracy of RAM and TI monitors by assessing whether the mean differences of RR measurements by RAM or TI monitor compared to manual counts were different from 0 using a mixed effects model to account for within-patient correlation followed by pairwise comparisons between estimated mean RRs from RAM and TI monitors with that from the manual counts. The estimated mean RRs and their standard errors (SEs) were obtained using leastsquare means statement in Statistical Analysis System (SAS) the Mixed Procedure (PROC MIXED) that accounts for the correlated nature of the repeated measurements within the same subject over time. Data were summarized using mean (SD) for continuous variables and frequency and percent for categorical variables. To assess the degree of agreement of RR measurements by RAM and by TI to that of manual counting, the repeated measures Bland Altman method was used to account for the correlation within patient. The Bland Altman plot is a graphical display of the difference scores of 2 measurements against the mean for each subject. The bias d is measured by the average of the differences, and the upper and lower limits of agreement are computed as (d ± 1.96 SD) where SD is the standard deviation of the differences. This methodology proposed by Bland and Altman 9 accounted for variability in number of repeat observations and thus, no additional weighting was required. Based on the clinical adjudication of alarms as true or false during 5 to 15 minutes clinical observation every 2 hours, an analysis of sensitivity, specificity, positive predictive value, and negative predictive value was performed. Because each subject was followed up for a different number of periods, we computed the average number of alarms for each patient per period and multiplied it by 62 to get the weighted total number of alarms that we used to compute sensitivity, specificity, positive predictive value, and negative predictive value. The calculation of these variables has been done in a descriptive manner. A sample size of 34 was found to be sufficient to detect a difference of 1 bpm between an experimental method and the reference method of manual counting of RRs during clinical observations using a paired Student s t test assuming a SD of 2 bpm. Assuming a dropout rate of 20%, a sample size of 42 was deemed sufficient. However, in the end, 62 patients stratified by weight in 3 groups of 20 patients each (10 20 kg, kg, and >40 kg) were enrolled in the study. RESULTS Sixty-two patients were enrolled from March 2015 to August 2015 with 30 children recruited at Cincinnati Children s Hospital and 32 children recruited at Children s Medical Center Dallas (Figure 1). Fifty-eight children were status posttonsillectomy, and 4 patients were being treated with an IV PCA. The mean age, weight, and body mass index were 6.5 years old, 35.3 kg, and 76.6 percentile, respectively. Forty-three percent of the posttonsillectomy children had a June 2017 Volume 124 Number

4 Postoperatory Respiratory Monitoring polysomnography diagnosis of severe OSA; the other 57% had sleep-disordered breathing without a polysomnography (Table 1). The average monitoring time per patient was 15.9 ± 4.8 hours. The mean RR recorded with the TI was significantly higher than the mean manual count RR (21.3 bpm with a SE of 0.77 vs 19.6 bpm, SE 0.56, P =.01). The mean RR recorded with the RAM was found to be not different to the mean RR counted manually (19.7 bpm with a SE of 0.52 vs 19.6, with a SE of 0.56, P =.80). The average RRs difference between manual and RAM were 0.17 ± 6.81, whereas for TI and manual counting were 1.39 ± (Table 2). Bland Altman plots showing the mean difference and the limits of agreement are shown in Figure 3. These comparisons were weighted for the variable number of monitoring duration of each patient as well as adjusted for the correlation within patient. The RAM measurements were at least 4 bpm different from the manual counts only 11% (95% confidence interval [CI], 8% 14%) of the observation periods while TI Table 2. Average Difference and SD Between Measurement Methods Used to Determine RR and the Proportions of Time That the Difference in RR Measurements Are 4 bpm When Comparing RAM and TI With Manual Counting Manual Versus RAM Manual Versus TI Average difference 0.17 (6.81) 1.39 (10.63) 95% CI, 0.77 to % CI, 2.34 to 0.44 Proportion of 11% 22% difference 4 bpm 95% CI, % CI, Abbreviations: bpm, breaths per minute; CI, confidence intervals; RAM, rainbow acoustic monitor; RR, respiratory rates; TI, transthoracic impedance. Table 1. Demographics Continuous Variables Total Sample (n = 62) Mean (SD) Age (y) 6.58 (3.40) Weight (kg) (22.77) BMI (6.34) BMI percentile 76.6 ± 30.8 Sleep study AHI (12.39) Sleep study lowest SpO (9.83) Sleep study peak EtCO (9.17) Ramsay 2.94 (1.57) Categorical Variables n (%) Female gender 29 (46.77) Male gender 33 (53.23) Race Caucasian 37 (59.68) African American 14 (22.58) Other 11 (17.74) Hispanic ethnicity 23 (37.10) Enrollment group kg 20 (32.26) kg 21 (33.87) 40+ kg 21 (33.87) Surgical procedure T 2 (3.23) T & A 43 (69.35) T & A with other procedure 13 (20.97) Other with PCA 4 (6.45) ASA 1 3 (4.84) 2 40 (64.52) 3 19 (30.65) Sleep apnea Diagnosed 27 (43.55) Symptoms 30 (48.39) None 5 (8.06) Participated in sleep study 30 (48.39) Sleep study diagnoses Normal 3 (10.00) Mild 4 (13.33) Moderate 7 (23.33) Severe 13 (43.33) Other 3 (10.00) Abbreviations: A, adenoidectomy; AHI, apnea/hypopnea index; ASA, American Society of Anesthesiologists; BMI, body mass index; EtCO 2, end tidal carbon dioxide; PCA, patient/parental-controlled analgesia; SpO 2, oxygen saturation; T, tonsillectomy. Figure 3. Bland Altman plots for agreement between RAM and TI RR with manual counts RR. RAM indicates rainbow acoustic monitor; RR, respiratory rates; TI, transthoracic impedance. measurements were at least 4 bpm different from manual counts of RR for 22% (95% CI, 18% 27%) of the observation periods. The tolerance by RAM was 87% with an SE of 4.26% of the total expected monitoring time. A total of 276 alarms was detected with a mean of 4.5 alarms per patient during the study, yielding an average of 1 alarm every 4 hours per patient. Of the alarms, 98 (mean = 1.58 ± 2.49 per patient) were from the RAM and 178 alarms (mean = of 2.87 ± 4.32 per patient) were from the TI. The number of false alarms was significantly greater by the TI than by RAM (Table 4). Sensitivity, specificity, as well as positive and negative predictive values of the 2 experimental methods were reported in Table 4. The specificity and positive predictive value were greater for RAM, whereas the sensitivity and negative predictive value were not different for RAM and TI ANESTHESIA & ANALGESIA

5 Table 3. Alarm Analysis of RR Monitoring With RAM and With TI Respiratory Monitoring Number of Alarms Mean Number of Alarms per Patient Number of False Alarms Mean of False Alarms per Patient Number and Proportion of Patients With at Least One Alarm Number and Proportion of Patients With at Least One False Alarm Mean Number of Alarms per 2- Hour Episode Mean Number of False Alarms per 2-Hour Episode RAM ± ± (0.58) 6 (0.1) 0.19 ± ± 0.25 TI ± ± (0.69) 19 (0.31) 0.35 ± ± 0.99 Abbreviations: RAM, rainbow acoustic monitor; RR, respiratory rates; TI, transthoracic impedance. Table 4. Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value of RAM and TI Monitoring Sensitivity Specificity Positive Predictive Value Negative Predictive Value RAM 46.6 % CI, wn: TI 68.5% CI, wn: % CI, wn: % CI, wn: % CI, wn: % CI, wn: Abbreviations: RAM, rainbow acoustic monitor; TI, transthoracic impedance; wn, weighted number of alarms. 72.1% CI, wn: % CI, wn: DISCUSSION In this prospective observational study of postoperatory RR monitoring in children, we found that the RR measurements with RAM were not significantly different to the manual counting method, and the degree of agreement decreased with RR measurements from TI in comparison with manual counting. Moreover, an analysis of the monitoring time for which there was a difference in RR greater than 4 bpm among RAM and TI versus manual counting showed that this discrepancy was twice as common for the TI than for the RAM. The difference of 4 or more bpm was selected a priori as we felt this to be a clinically significant level of difference for RR in children. Previous studies in adults have demonstrated similar good accuracy of RR measurements with RAM when compared to capnography. 10,11 Ramsay et al 11 also found RAM more sensitive to detect periods of apnea than capnography in adults. The use of RAM during moderate sedation was associated with less episodes of desaturation and fewer alarm events. 12 RAM has also shown to provide accurate RR measurements in patients receiving intravenous anesthesia and in patients breathing spontaneously with the use of a laryngeal mask airway. 13,14 In a previous study of children recovering in the PACU and monitored for a mean of 90 minutes, we found the RAM to have similar accuracy and precision when compared to capnography and found the RAM to be better tolerated (97.5% vs 62.5%). 7 We noted that our previous study had limited generalizability due to the fact that capnography is not routinely used for postoperative respiratory monitoring secondary to issues with poor compliance and cost. As compared to our previous report, the current study design allowed for a prolonged monitoring time (16 hours vs 1.5 hours) and we compared the RAM with TI, which is commonly used for patients at high risk for respiratory depression in our institutions. Our study included 2 groups of patients known to be at risk for postoperative respiratory complications, namely children with OSA and those receiving intravenous PCA. We purposely did not limit this study to the evaluation of a single-risk population since our aim was to evaluate the accuracy and tolerance of the RAM monitoring and evaluation in a broader population strengthens our results. We used the manual counting of the RR to compare to RR measured by RAM and TI as this is a standard of care that is very accurate and reliable recognizing that it is limited by the fact that it is not continuous. When we stratified our patients by weight, we showed that the RAM technology works similarly across the weight range and by inference should be effective for children both young and old. Tolerance to the RAM sensor was acceptable with children maintaining the sensor in place for the duration of their intended monitoring time (>15 hours) in 87%. This is comparable to the 95% tolerance that we found in our previous study that employed a shorter monitoring time (average 1.5 hour). 7 An analysis of the alarms showed significantly more total alarms and false alarms with the use of the TI than with the use of the RAM. This is critical as the potential of alarm fatigue becomes an increasing concern among clinical providers. It is important to consider that our data collection was performed during the first 5 to 15 minutes of every 2-hour episode. In other words, we analyzed approximately 1/8 of the total monitoring time (approximately 2 hours per patient with an average of 15.9 hours of monitoring per patient). If we assume a similar number of alarms was recorded during the rest of the monitoring time, but there was no additional collection of data at the bedside, we would have an incidence of false alarms of 0.8 per hour per patient with the RAM and 4.0 per hour per patient with TI. The specificity and positive predictive value were also significantly better for the RAM than for the TI, without a difference seen in the sensitivity or negative predictive value. Limitations to our study include the nonuniform recruitment from 2 different populations (posttonsillectomy patients and patients receiving IV PCA). Most of the recruited patients were posttonsillectomy patients (93%). This was not intentional but likely because of the fact that tonsillectomy is a commonly performed procedure that is seen more commonly in our practice settings that is management of children with IV PCAs. Considering that our principal aim was to evaluate the degree of agreement of the RAM and TI in comparison to manual counting, without consideration of the population evaluated, this issue did not seem to affect our analysis. Also, since Food and Drug June 2017 Volume 124 Number

6 Postoperatory Respiratory Monitoring Administration-approved RAM sensor only for children above 10 kg, our data cannot be extrapolated to infants. We also recognize that with the involvement of multiple comparisons, the actual type I error rate may be higher than 0.05 with an elevated chance of the results being false-positive. In addition, our analysis of the alarms in terms of sensitivity, specificity, positive predictive value, and negative predictive value is of limited interpretation given that the collection of the data on the case report form was not specifically designed to obtain this information. However, our calculation of true-positive alarms and false-positive alarms is very reliable given the evaluation of patients at the bedside during the activation of an alarm for either device. Therefore, the calculation of the positive predictive value should be very accurate. The calculation of true-negatives and false-negatives required some assumptions given than these data were indirectly obtained from the collection report form. We recognize the importance of further studies calculating the negative predictive value in a more rigorous way and feel it is critical in terms of safety for a respiratory monitoring technology to provide an acceptable negative predictive value to avoid the use of a technology that omits the detection of respiratory events. CONCLUSIONS In summary, our study found a significant degree of agreement between RR counted manually and RAM measurements of the RR with an acceptable tolerance in pediatric patients. TI measurements of the RR had a lower degree of agreement with manual counting. RAM had a significantly lower incidence of false alarms than TI with good specificity and positive predictive value. Future studies require an accurate determination of the negative predictive value of respiratory monitoring technology to better establish their role and safety. E DISCLOSURES Name: Mario Patino, MD. Name: Megan Kalin, MS, CCRP. Name: Allison Griffin, MS. Name: Abu Minhajuddin, PhD. Name: Lili Ding, PhD. study, and analyze the data. Name: Timothy Williams, MS. Contribution: This author helped design the study and conduct the study. Name: Stacey Ishman, MD, MPH. Name: Mohamed Mahmoud, MD. Name: C. Dean Kurth, MD. Name: Peter Szmuk, MD. This manuscript was handled by: James A. DiNardo, MD, FAAP. REFERENCES 1. Leong AC, Davis JP. Morbidity after adenotonsillectomy for paediatric obstructive sleep apnoea syndrome: waking up to a pragmatic approach. J Laryngol Otol. 2007;121: Apfelbaum JL, Silverstein JH, Chung FF, et al. Practice guidelines for postanesthetic care: an updated report by the American Society of Anesthesiologists Task Force on Postanesthetic Care. Anesthesiology. 2013;118: Cretikos MA, Bellomo R, Hillman K, Chen J, Finfer S, Flabouris A. Respiratory rate: the neglected vital sign. Med J Aust. 2008;188: Weinger M, Lee LA. No patient shall be harmed by opioidinduced respiratory depression. APSF Newsletter. 2011;26:21, Al-Khalidi FQ, Saatchi R, Burke D, Elphick H, Tan S. Respiration rate monitoring methods: a review. Pediatr Pulmonol. 2011;46: Nassi N, Piumelli R, Lombardi E, Landini L, Donzelli G, de Martino M. Comparison between pulse oximetry and transthoracic impedance alarm traces during home monitoring. Arch Dis Child. 2008;93: Patino M, Redford DT, Quigley TW, Mahmoud M, Kurth CD, Szmuk P. Accuracy of acoustic respiration rate monitoring in pediatric patients. Paediatr Anaesth. 2013;23: Food and Drug Administration (k) Summary: K ed. UFaD Administration, pp Available at: fda.gov/cdrh_docs/pdf12/k Accessed April 6, Bland JM, Altman DG. Agreement between methods of measurement with multiple observations per individual. J Biopharm Stat. 2007;17: Mimoz O, Benard T, Gaucher A, Frasca D, Debaene B. Accuracy of respiratory rate monitoring using a non-invasive acoustic method after general anaesthesia. Br J Anaesth. 2012;108: Ramsay MA, Usman M, Lagow E, Mendoza M, Untalan E, De Vol E. The accuracy, precision and reliability of measuring ventilatory rate and detecting ventilatory pause by rainbow acoustic monitoring and capnometry. Anesth Analg. 2013;117: Applegate RL II Lenart J, Malkin M, et al. Advanced monitoring is associated with fewer alarm events during planned moderate procedure-related sedation: a 2-part pilot trial. Anesth Analg. 2016;122: Ouchi K, Fujiwara S, Sugiyama K. Acoustic method respiratory rate monitoring is useful in patients under intravenous anesthesia. J Clin Monit Comput. 2017;31: Atkins JH, Mandel JE. Performance of Masimo rainbow acoustic monitoring for tracking changing respiratory rates under laryngeal mask airway general anesthesia for surgical procedures in the operating room: a prospective observational study. Anesth Analg. 2014;119: ANESTHESIA & ANALGESIA