Main

The mechanism(s) resulting in sudden infant death (SID) are still unknown. This is partially because it is still exceptionally difficult to obtain data on the pathophysiologic changes occurring immediately before these deaths. Reports on infants who died while attached to memory monitors showed that alarms were usually triggered by bradycardia but did not provide a systematic analysis of the cardiorespiratoryr changes occurring before death(1–3). In particular, the issue of gasping, a potentially very powerful autoresuscitative mechanism(4,5), was not addressed. We analyzed memory monitorr recordings obtained during SID to answer the following questions: 1) is there evidence that autoresuscitative mechanisms such as gasping occur before these deaths, 2) how quickly does heart rate fall, and 3) what is the precise temporal sequence between apnea, bradycardia, and/or gasping.

MATERIALS AND METHODS

Nine recordings from infants who had died suddenly and unexpectedly while attached to a cardiorespiratory memory monitor and who had a postmortem diagnosis of either sudden infant death syndrome (SIDS, n = 7) or "mild bronchopulmonary dysplasia" (BPD, n = 2) were analyzed. These were selected from a sample of 21 recordings of alleged SID obtained from the home monitoring programs of the authors or sent from various sources to the authors for review because of potential death during monitoring. Reasons for exclusion of the other 12 recordings were as follows: infant died with tracheostomy in place and no information available on patency of cannula (3 cases), infant not on monitor at the time of death (3 cases), no autopsy performed (2 cases), monitor switched off at a time when heart rate had fallen to 40 bpm, with the impedance signal being suggestive of international suffocation (6) (1 case), monitor only switched on during resuscitation (1 case), memory log only containing tabular data owing to memory overload (1 case), and memory printout showing severe recurrent heart rate decelerations and apneas, but infant surviving as determined during the course of this investigation (1 case). Some data from six of the nine recordings have already been published but with different methods of analysis(1–3).

Monitors used were cardiorespiratory monitors of the following brands: Corometrics (Wallingford, CT), Aequitron, Edentec (both now Nellcor Puritan Bennett, Pleasanton, CA), and Healthdyne (Marrietta, GA). Prealarm recording intervals ranged from 10 to 90 s. Heart rate alarms had been set at 60-80 bpm and apnea alarms at 20 s. Clinical information was obtained using a purpose-written questionnaire sent to each infant's pediatrician, by reviewing medical examiners' reports, and, in four cases, by interviewing parents and first responders. Only standard autopsies had been performed (i.e. autopsies had not concentrated on specific aspects such as subtle changes in brainstem tissue).

Recordings were analyzed for the following parameters: 1) time and cause of the first monitor alarm, 2) respiratory and heart rate at onset of recording, 3) respiratory pattern at onset of recording (regular or nonregular breathing(7), or gasping), 4) duration of gasping and number of gasps, and 5) interval between monitor alarm and a) first apneic pause of >20 s duration, b) first gasp, c) heart rate falling to <15 bpm, and d) onset of stimulation and/or resuscitation as identified by the occurrence of characteristic artifact patterns (2) in the recordings. Gasping was defined as the presence of a rapid inspiratory rise with a retarded expiratory phase preceded and followed by a cessation of breathing movements(5,8). The rapid inspiratory rise could be preceded by a brief expiratory movement (i.e. type 1 gasp according to Gozal et al.(8); Fig. 1). The difference between the heart rate immediately before the first gasp and the maximum heart rate in the first 30 s following it was also determined.

Figure 1
figure 1

Section from the memory monitor printout of patient 3. There are low amplitude, irregular breathing movements interspersed with apneic pauses during the first 2 min of recording. These are followed by 3 gasps (G), the first of which seems to have an initial expiratory component (type I gasp according to Ref.((8)). Heart rate stays initially at around 60 bpm, increases shortly to 100 bpm after the 2nd gasp, and falls rapidly thereafter. Stimulation followed by CPR (the latter not shown on this section) occurs 3.8 min after the monitor alarm but remains unsuccessful.

Full size image

RESULTS

Clinical data. All infants had been born preterm, and monitoring prescribed for complications related to prematurity in all but one infant, who had been monitored because of an apparent life-threatening event (ALTE) occurring at 7 wk of age (Table 1). No infant was a SIDS sibling, and none had had multiple ALTE or unusually severe or persistent episodes of apnea of prematurity (AOP) before discharge, although five had shown AOP beyond a postconceptional age of 34 wk. All infants had been clinically considered healthy on discharge from hospital, and none had required additional inspired oxygen at that time. No infant was reported to have been found with his head covered by bedding.

Table 1 Clinical data
Full size table

Recording results. Primary cause of monitor alarm was bradycardia in all but two infants, with heart rate falling relatively slowly during the first 30 s of recording (median, -10 bpm; range -60 to +10) (Table 2). The time for heart rate subsequently to fall to <15 bpm ranged from 1.4 to 25 min. In four infants, the heart rate was recorded together with the ECG. In these infants, the ECG invariably showed either sinus or nodal bradycardia but no indication of heart block or ventricular tachycardia.

Table 2 Results from analysis of recordings
Full size table

Apnea of >20 s duration began within 0.3 to 13.7 min after the first alarm in five infants and within 7 to 20 s before this alarm in three infants (in the remaining infants, stimulation occurred before the development of any apnea). Breathing movements were nonregular in four infants and regular in two infants (the remaining infants were already gasping, see below). No recording showed a pattern suggestive of intentional suffocation(6).

Gasping was already present at the time of the first monitor alarm in three infants (Fig. 2) and occurred within 2.7 min after it in a further four infants (Fig. 3). In contrast, one infant only began to gasp 13 min after the first monitor alarm. The duration of gasping was variable, lasting between 3 s (i.e. 1 gasp) and 11 min in those five infants in whom it was not interrupted by CPR. All but one infant had gasps that did not have an initial expiratory component, i.e. type 2 or "late" gasps according to Gozal et al.(8). Heart rate following the first gasp increased from 70 to 140 bpm in one infant (Fig. 2) and remained unchanged in the other infants. The two infants who did not gasp or had only 1 gasp recorded were also the only ones in whom an absence of pleural or visceral petechiae was noted at postmortem (Table 1).

Figure 2
figure 2

Section from the memory monitor printout of patient 8. There are 11 gasps that progressively decrease in amplitude. There is an increase in heart rate from 72 to 140 bpm during the first 20 s of recording, followed by several smaller increases in heart rate, which seem to occur in response to the gasps and are also decreasing in amplitude over time.

Full size image
Figure 3
figure 3

Section from the memory monitor printout of patient 9. There are irregular breathing movements accompanied by a decrease in heart rate from 150 to 30 bpm during the first 45 s of recording. These are followed by a cessation of breathing movements for 2.4 min, after which a small gasp with a rapid inspiratory but retarded expiratory movement occurs. Thereafter, respiration ceases altogether.

Full size image

CPR was given to three infants within 4, 21, and 228 s of the monitor alarm; however, two of these were already gasping at that time. CPR had no effect on the progressive fall in heart rate. Reasons stated by caregivers of the remaining infants for not giving prompt stimulation or CPR were "slept through alarms," "was not at home at time of alarm," and "did not hear alarm."

DISCUSSION

This study was undertaken to determine cardiorespiratory patterns during sudden infant deaths. We found that the primary cause of monitor alarm was bradycardia in all but two infants and that seven infants (88%) were already gasping at or shortly after this alarm, including those where apnea had been the primary cause of monitor alarm.

Were the respiratory movements that we interpreted as gasps correctly identified as such? Gasping is distinguished from other respiratory patterns by a slower frequency, a higher amplitude and a more prolonged expiratory phase(9). At least two of these conditions (frequency and expiratory duration) also apply to the breaths interpreted by us as gasps; amplitude is difficult to quantify in recordings of chest wall impedance. Moreover, shape and rate of the respiratory waveforms were similar to those recorded during observed episodes of gasping in infants(10,11); they also resembled gasps observed and recorded in various animal species(4,9,12). Also, we are not aware of any other respiratory pattern in human infants that would produce waveforms that are even vaguely similar to gasping. Breathing against an obstructed airway occurs at a much faster rate(6), and motion-induced artifacts, e.g. reflecting seizure activity, do not occur with such regularity regarding rhythm and shape(13); in mice, they also produce respiratory waveforms that are quite dissimilar from that of gasps(12). We, therefore, feel reasonably confident that the waveforms interpreted in this study as gasps were indeed gasps, although we acknowledge that there is at present no scientific proof for this.

Gasping is a strong indicator of hypoxemia. In various animal species gasping does not occur unless Pao2 has fallen to <5-15 mm Hg and is only elicited by hypoxemia but not hypercarbia or acidosis(4,9,12,14). The decrease in heart rate observed in this study is also suggestive of hypoxemia(15,16). Hence, although oxygenation was not measured, there is strong evidence that at least seven of the nine infants in this study were already severely hypoxemic around the time of their first monitor alarm. This hypoxemia was apparently not caused by a prolonged cessation of breathing movements, as there was no apnea alarm before the development of gasping.

What then was the cause of the presumed hypoxemia, ultimately resulting in death? The short prealarm recording interval (10-90 s) of the monitors used did not allow us to answer this question. Nevertheless, some conclusions can be inferred. First, in none of the recordings was there any indication of vigorous breathing movements indicating struggle against external airway obstruction(6). Hence, if such struggle had occurred, it must have ceased at least 10-90 s before heart rate falling to below the alarm threshold (60-80 bpm). However, as none of the infants had a history of frequent ALTE or multiple SIDS in one family(17), we are inclined not to believe that these deaths were the result of infanticide. Second, there was no indication of a primary disorder of cardiac conduction causing these infants' deaths. This was clearly evident in those infants in whom the ECG was recorded but is also suggested by the relatively slow decrease in heart rate occurring in those infants in whom it was not. Other mechanisms potentially resulting in death, e.g. rebreathing of expired CO2, prolonged hypoxemia in conjunction with a blunted chemoreceptor response [as observed in infants with BPD(18)], or hypoventilation, cannot be commented on as neither oxygenation nor CO2 or minute ventilation was monitored.

Why was gasping not effective as an autoresuscitative mechanism? One striking finding is that gasping duration was relatively short (median, 4 min) and that it did not, with one exception (Fig. 2), have a marked effect on heart rate. Gasping duration in asphyxiated rats decreased from 33 min in 2-d-old to 1 min in 25-d-old rats(8). Correspondingly, all neonatal mice but only 19% of 16-23-d-old mice were able to resuscitate themselves from asphyxia by gasping when they were put back into room air immediately before the first gasp(19). Also, successful gasping in both piglets and mice is associated with a rapid increase in heart rate, whereas the absence of such an increase in heart rate seemed indicative of autoresuscitation failure(4,20).

The main purpose of gasping is to restore tissue oxygenation. Gasping will remain ineffective if there is circulatory failure, i.e. if the increased oxygen levels in the pulmonary circulation do not reach the brainstem. A significant fall in blood pressure during apneic/bradycardic episodes was reported in neonates(21) and in individual infants who had life-threatening events or died suddenly while on recording devices(22). That circulatory failure may be involved in some cases of SID is also supported by the finding of reduced muscarinic and kainate receptor binding in the arcuate nucleus(23), which could reflect a critical reduction in the ability to compensate to challenges that provoke hypotension(22).

Gasping will also remain ineffective if there is ongoing airway obstruction, or an alveolar-capillary block, e.g. owing to severe pulmonary edema or airway collapse. In mice, ongoing airway obstruction during gasping produced an expiratory wave form that was nearly a mirror image of the inspiratory limb(12); such a pattern was not observed in this study. Pulmonary edema is known to occur during asphyxia and is a characteristic finding in SID [whereas widespread airway collapse (atelectasis) is not], but the extent of it is difficult to quantify, and there was no specific information on this issue available. Thus, both circulatory failure and an alveolar-capillary block remain possible explanations for the gasping failure observed in this study.

In two infants, there was an absence of petechial hemorrhages noted at postmortem. These were the only infants who did not gasp at all or who showed just one rather weak gasp (Fig. 3). Although numbers are insufficient to draw conclusions, it is interesting that Brouardel found already in 1897 that petechiae are an indication that gasping has occurred before death(24).

A concern generated by this study is whether cardiorespiratory monitors alarm early enough. This is because seven of the eight infants in whom this could be analyzed were already gasping at or shortly after the monitor alarm, and two died despite CPR commencing within 21 s of the alarm. These findings shed further doubt on the ability of cardiorespiratory monitors to detect potentially life-threatening hypoxemia early enough in some patients.

The major limitation of this study is that recordings were not recruited in a systematic way, i.e. it had a purely descriptive design. Extrapolation from our data to sudden death in other preterm and particularly term infants may thus be misleading, and our observations can only serve to formulate hypotheses, not scientific conclusions. Nevertheless, the consistency with which gasping occurred is striking. Other limitations include the short prerecording intervals, which precluded any information on arousal as the other major autoresuscitative mechanism, and the fact that information on two-thirds of these patients has already been published(1–3). Gasping, however, was not an issue addressed in these reports.

In summary, this study has shown that gasping seems to be a common feature during sudden infant death. The fact that it occurred either before or shortly after the first monitor alarm in 88% of patients in this study suggests that for some patients cardiorespiratory monitors only alarm when the infant is already severely hypoxemic. The mechanism(s) eliciting this hypoxemia remain unknown.