Objective The American Academy of Pediatrics recommends all infants born at <37 weeks gestation spend a period of observation in a car seat prior to hospital discharge to assess for apnoea, bradycardia or oxygen desaturation. The most recent Cochrane review suggested further studies to determine if the infant car seat challenge (ICSC) accurately predicts the risk of clinically adverse events. We reviewed our experience with the ICSC and the polysomnogram (PSG) to determine if the ICSC accurately predicts the risk of adverse events when compared with the PSG.
Study design Retrospective chart review of all infants in our institution who had an ICSC and a PSG between January 2005 and December 2008.
Result 785 infants had ICSCs. In addition, 313 infants (56.6%) had an abnormal PSG, even though the vast majority, 158 (88.3%), passed their ICSC. There were no significant differences in gestational age at birth, birth weight, chronological age at study or postmenstrual age at study between infants who either passed or failed the ICSC with those who passed or failed the PSG. The sensitivity of the ICSC was 0.11 and specificity was 0.96. The positive predictive value of the ICSC was 0.77 and the negative predictive value was 0.45.
Conclusions The ICSC has a low negative predictive value (0.45) when compared with the PSG as a reference standard for identifying adverse cardiorespiratory events. Although less time consuming and cumbersome than extended polysomnography, the ICSC is not a reliable substitute.
Statistics from Altmetric.com
What is already known on this topic
Preterm and infants at risk are screened for cardiorespiratory instability in car seats prior to hospital discharge.
The most recent Cochrane review suggested further studies to determine if the car seat challenge accurately predicted the risk of clinically adverse events.
What this study adds
Many infants who pass their car seat challenge have significant cardiorespiratory instability when studied for a longer period of time with a polysomnogram.
There was no significant difference among the demographic factors examined that differentiated those who would either pass or fail the studies.
In 1974, the American Academy of Pediatrics (AAP) first recommended the universal use of car seats for infants,1 and in 1980 the ‘First Ride’ programme encouraged the universal use of car seats for newborns at the time of hospital discharge.2 Following several studies that documented premature infants experiencing oxygen desaturation in car seats,2–6 the AAP, in 1991, recommended all infants born at less than 37 weeks gestation have a period of observation in a car seat to monitor for apnoea, bradycardia or oxygen desaturation.7
Subsequent studies have questioned the reliability of the car seat test. DeGrazia8 found that in a cohort of 49 premature infants who had two infant car seat challenges (ICSC), 8% passed the first challenge but failed the second, while 6% failed the first challenge but passed the second. DeGrazia et al9 later evaluated 80 infants and found that neither weight nor age at initial or repeated ICSC predicted passing the test. Pilley and McGuire10 in their Cochrane review of the ICSC noted that they were unable to find any randomised controlled trials that fulfilled their eligibility criteria, and concluded that further studies were needed to determine whether the ICSC accurately predicted the risk of clinically significant adverse events in preterm infants.
While the ICSC accurately records physiological data including apnoea, bradycardia and oxygen desaturation, it does so for a very limited time period, currently 90–120 min.11 The infant polysomnogram (PSG) is a continuous recording, commonly 12–24 h, of multiple aspects of the infant's cardiorespiratory status. The study most commonly includes the heart rate and respiratory pattern by impedance, and may also include oxygen saturation, nasal air flow, oesophageal pH and other physiological data. Since our unit's standard practice is to obtain a PSG on all infants less than 35 weeks gestation at birth prior to their hospital discharge, we felt that we would be able to examine a large number of infants to determine if the data captured in a 60 min ICSC, which was the standard at the time of our study, were similar to that collected in a 24 h PSG.
The objective of this study was to review our recent experience with ICSC and compare those results with a PSG in a diagnostic accuracy test with the PSG as the reference standard. We also looked for demographic factors that might identify infants at risk for failing each test.
We performed a retrospective chart review of all infants in both the term nursery and the Neonatal Intensive Care Unit who had an ICSC between January 05 and December 08. The setting was a university-affiliated hospital that served a minority, predominantly publicly insured population with approximately 2800 births per year. ICSC was performed within 24 h of hospital discharge on all infants born at <37 weeks gestation or those infants who are small for gestational age. The ICSC was performed for 60 min which was consistent with the AAP recommendations at the time of ‘a period of observation in a car safety seat before hospital discharge’.12
The infants were positioned in their car seats by their nurses. Heart rate and respiratory rate were monitored by impedance and O2 saturation was measured by oximetry. The infant and the monitoring equipment were visually monitored by nursing with vital signs documented every 15 min. The test was performed at the convenience of nursing and no attempts were made to correlate the test with the infant's sleep state.
Criteria for failure included apnoea greater than 20 s, bradycardia less than 80 beats per minute (BPM) or O2 desaturation to less than 85%. Infants who failed the test were retested after adjustment of position in the car seat. Infants who failed the ICSC twice were discharged in a car bed.
Per hospital protocol, infants less than 35 weeks gestation at birth had a 12–24 h PSG performed within 48 h of hospital discharge. This study was a two channel study that recorded heart rate and respiratory pattern by impedance, a three channel study that also included oxygen saturation or a five channel study that added airflow by nasal thermistor and evidence of gastro-oesophageal reflux by oesophageal pH probe placed at the level of the carina and confirmed by radiograph. Overall, 171 infants had two channel studies, 98 infants had three channel studies and 44 infants had five channel studies.
The infant was placed on the monitoring equipment by nursing and respiratory, and the infant's routine care was resumed. Nurses would note feeding times and any significant clinical events on a paper log. Significant O2 desaturation or bradycardia was a criterion for failure irrespective of any relationship to feeding.
Two channel studies were performed with Healthdyne SmartMonitor Model #970SE-10. Three channel studies were performed with a Respironics SmartMonitor 2PS. Oesophageal pH was monitored with SandhillSmartMeter Model #H91-1000. Nasal Airflow was measured with the Respironics SmartRecorder using a three bead strip. Criteria for failing the PSG included prolonged apnoea greater than 20 s, bradycardia less than 80 BPM for at least 3–5 s, significant oxygen desaturation to less than 85% for at least 3–5 s or oesophageal pH <4 for greater than 4% of the study.
The two and three channel studies were downloaded into the Synergy E program (Phillips Medical) which scanned them for all events with apnoea greater than 15 s, bradycardia less than 80 BPM, tachycardia greater than 250 BPM and O2 saturation less than 85%. The program would then print out each event with 15 s of data before and after each event. The five channel studies were downloaded into the Synergy S program (Phillips Medical) for viewing, and were scored manually by one of two respiratory therapists.
The decision to perform each test was strictly based on gestational age and size criteria listed above. The results of one test did not impact the decision to perform the other test. In addition, no infants with either congenital heart disease or any condition involving hypotonia which might have impacted the results of the study were included.
Demographic data were collected on all infants in the study and stratified by whether the ICSC was passed and the results of the PSG in those infants who had that test performed. For those infants who failed their PSG, the reason for failure and the infant's subsequent management (home monitor and any medications prescribed) were detailed.
The sensitivity, specificity, and positive and negative predictive values of the ICSC were compared with the PSG. Statistical analyses were performed using Student t test for continuous data and Fisher's exact test for categorical data. As this was a diagnostic test accuracy study with the PSG as the reference standard, we adhered to the Standards for Reporting of Diagnostic Accuracy (STARD) guidelines 13 and presented data in a format suggested by those guidelines. The Institutional Review Board (IRB) of the Albert Einstein Healthcare Network reviewed this study and determined that IRB approval was not necessary.
During the time period of this study, 785 infants in our institution had an ICSC. Of these infants 313 additionally had a PSG performed. In all, 177 of these infants (56.5%) failed the PSG, but in no case did an infant fail the PSG due to oxygen desaturation alone. Of the 785 who had an ICSC, 43 (5.5%) failed their initial challenge while only three infants (0.4%) failed a second ICSC. Among the 313 infants who had both a PSG and an ICSC, 26 (8.3%) failed their initial ICSC while 2 (0.6%) failed a repeat ICSC (figure 1).
Numerous comparisons were made between the various subgroups of infants who had both an ICSC and a PSG. When infants who failed the PSG were compared with those who passed the ICSC, there was no significant difference between the groups for gestational age (p=0.56), birth weight (p=0.52), age at study (p=0.09) or postmenstrual age (PMA) at study (p=0.48). When infants who passed the PSG were compared with those who failed the ICSC, the only difference was that those who failed the ICSC were significantly older chronologically (27.1±33.8 vs 14.8±14.1 days; p=0.001). However, there was no difference between the two groups in PMA at study (35.9±2.0 vs 35.2±1.4 weeks; p=0.56). Infants who passed both the PSG and the ICSC were compared with those who failed both tests (table 1).
The sensitivity and specificity of the ICSC as compared with the PSG for determining the physiological stability of these infants were calculated based on a 2×2 table of those who passed and failed the ICSC cross tabulated with those who passed and failed the PSG (table 2). Compared with the PSG, the sensitivity of the ICSC for determining clinically significant events was 0.11 and the specificity was 0.96. The positive predictive value of the ICSC compared with the pneumogram was 0.77 and the negative predictive value was 0.45.
In 2007, Elder et al14 published a study that compared the results of the ICSC to polysomnography in a group of 20 premature infants. They found that the ICSC does not accurately detect all adverse events during sleep in the car seat. We have demonstrated similar results in a much larger cohort of infants. Of our cohort of 313 infants, 177 (56.5%) failed the PSG; however, 158 of these 177 infants (89.3%) had passed the ICSC. In addition, there was no significant difference between any of the groups as regards PMA at the time of study, which could identify those infants likely to fail either the ICSC or the PSG. Among the entire cohort there was no difference in PMA between those who passed the ICSC and those who failed. Although infants who passed their PSG were of older gestational age at birth, they were the same PMA at the time of study as those who failed their PSG. Finally there was no difference in PMA at the time of study between those who passed the ICSC but failed the PSG and those who failed both studies.
Among our cohort of 785 infants who had an ICSC over the course of 4 years, only 43 infants (5.5%) failed their initial study. This is similar to the results of DeGrazia8 who found that 6% of a cohort of 49 premature infants failed their initial ICSC. There was no significant difference in gestational age at birth (34.4±2.7 vs 33.5±3.5 weeks; p=0.1) or birth weight (2.25±0.6 vs 2.1±0.7 kg; p=0.3) between the infants who passed their initial ICSC and those who failed the initial challenge. This is also similar to the findings of DeGrazia et al9 who attempted to determine if it were possible to use weight or age as predictors of passing the ICSC. They evaluated 50 infants who had failed their initial ICSC and compared them with 30 infants who had passed the ICSC. Unfortunately, despite the use of multivariate analysis and logistic regression analysis of variables such as gender, gestational age at birth, corrected gestational age at testing, birth weight and weight at testing, DeGrazia et al were not able to find any factors that accurately predicted passage of the ICSC.
Pilley and McGuire,15 in a review of the ICSC, noted that there were a number of issues with the ICSC itself. Primary among them was the fact that there were limited data on the variability of the test. Infants who were awake at the time of the test would be expected to be more likely to pass as there would be fewer apnoeic pauses with desaturation while awake than while asleep. They noted that there was little evidence that the episodes of desaturation, apnoea and bradycardia among those who fail the test were of clinical importance. They worried about the effect that failing the test might have on parental concern about their infants’ health and the cost implications of the test. Finally they posited that there were insufficient data to determine if the test improved clinically important outcomes for preterm infants. Last, in their Cochrane review of the ICSC, Pilley and McGuire10 searched for randomised controlled trials that evaluated whether predischarge monitoring in an infant car safety seat prevented morbidity and mortality in preterm infants. They were unable to find any randomised controlled trials that fulfilled their eligibility criteria, and concluded that further studies were needed to determine whether the ICSC accurately predicted the risk of clinically significant adverse events in preterm infants. Our results that demonstrated many more significant cardiorespiratory events among those infants who were evaluated for the longer period of the pneumogram support some of the concerns mentioned by Pilley.10
The infant PSG is a continuous recording, commonly 12–24 h, of multiple aspects of the infant's cardiorespiratory status. The study most commonly includes at least the heart rate and respiratory pattern by impedance, and may also include oxygen saturation, nasal air flow, oesophageal pH and other physiological data. Currently some centres perform a PSG prior to discharge to determine if an infant is experiencing significant apnoea of prematurity.16 A home cardiorespiratory monitor is then prescribed for an abnormal PSG.17 This care plan is, to say the least, controversial. Other centres simply document the presence of significant apnoea, bradycardia and oxygen desaturation clinically.18 Once an infant has gone a variable number of days without clinically documented events, he or she is discharged without a monitor. However, several investigators19–21 have shown that clinical documentation alone will miss a number of clinically significant events. Whatever one's position on the PSG and monitor debate, there is no question that the PSG does record valid physiological data. Our findings are not surprising to anyone who has reviewed infant PSGs. It is quite common that even for infants with multiple episodes of clinically significant events, these events are not spread evenly throughout the course of the 12–24 h study. Infants with significant disturbances of their cardiorespiratory stability will often have many periods of several hours where no clinically significant events occur. It is also common that there may be only one or two significant episodes of apnoea, bradycardia or oxygen desaturation during a 24 h period.22 As Pilley and McGuire15 opined, in a test over a relatively short period of time such as the ICSC, many extrinsic factors such as timing of feeds and sleep–wake patterns would be expected to contribute to the low sensitivity of the ICSC that we found (0.11) compared with the longer period of testing with the PSG.
In its current report on the safe transportation of preterm infants at hospital discharge,11 the AAP recommends a period of observation in a car bed similar to the ICSC for those infants who fail the ICSC and are being discharged in a car bed. We would expect that there would be a similar poor sensitivity of an infant car bed study of 90–120 min in duration when compared with the PSG as there was for the ICSC. Both Kinane et al23 and Cerar et al24 noted lower saturation levels in infants studied in a car bed as compared with their hospital cribs. Since the PSG is performed while the infant is in his or her hospital crib and lasts for a much greater time period, we would expect similar findings.
The limitations of our study include the retrospective nature of the study. In addition, while our 12–24 h PSG provided more physiological data than the 1 h ICSC, it is possible that more significant events would have been documented with an even longer period of monitoring as shown by Razi et al.20 In contradistinction to our study, which was performed under standard clinical conditions in a busy Neonatal Intensive Care Unit with event recording monitors and the infant's nurse responsible for documenting events, Elder et al14 were able to gather much more data through the use of a dedicated paediatric sleep technician. They monitored the EEG, electrooculogram and diaphragmatic effort in addition to heart rate, respiratory rate, nasal air flow and pressure and oesophageal pH. Unfortunately, this level of physiological data is unavailable in most units on a routine basis. Also it could be argued that the number of channels we analysed in our PSGs were not appropriate as the reference standard for a diagnostic accuracy test of the ICSC. However, for a busy clinical unit where three to four PSGs are often performed at the same time, this was the only practical reference standard available to us. Another limitation was the fact that the ICSC was not run contemporaneously with the PSG. However, this should not have had a large impact on our results as the ICSC was usually run within 24 h of the completion of the PSG. Last, the infant's position during the PSG is predominantly supine throughout the majority of the study, while the infant is in a semiupright position during the ICSC. Several studies have shown that there are significant changes in cardiorespiratory stability when infants are placed in car seats.3 ,5 ,6 However, this should have increased the number of infants who failed the ICSC relative to the pneumogram.
We have shown that the ICSC is not as effective at identifying those preterm infants with clinically significant events such as apnoea, bradycardia or oxygen desaturation when compared with the PSG. Consideration should be given to significantly increasing the length of time of either car seat or car bed testing, beyond the current 90–120 min recommendation11 prior to hospital discharge as this may better capture those infants with significant cardiorespiratory events. Alternatively, finding a test that could accurately predict those infants at risk for clinically significant events would be preferable.
The authors would like to thank John Hurley, RRT and John Turchi, RRT for performing the technical analysis of the infant car seat challenges and the pneumograms. The authors would also like to thank Matilde Irigoyen, MD, and Allan Arbeter, MD, for their invaluable editorial advice.
Contributors All of the authors listed were involved in the conception and design of the research, and all were involved in the acquisition of data, its analysis and interpretation. DLS was primarily responsible for drafting and revising the manuscript; however, all authors reviewed the drafts, suggested changes and approved the final copy. DLS, as the guarantor, is responsible for the overall content of the manuscript.
Competing interests None.
Ethics approval The Institutional Review Board of the Albert Einstein Healthcare Network.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.