Main

Despite the dramatic decline in the incidence of SIDS since the introduction of nonprone sleeping campaigns worldwide, it is still the leading cause of death during the postneonatal period (1–12 mo of age) (1). The incidence of SIDS has been found to be consistently higher in preterm and low birth weight infants, and this increase is inversely related to gestational age (24). The National Institute of Child Health and Human Development epidemiologic study found that infants born <37 wk and <33 wk gestation were 5 and 16 times, respectively, more likely to die of SIDS than age-matched controls (5). It has been estimated that approximately 20% of all SIDS cases occur in the preterm population (57).

Arousal from sleep is an important protective response to life-threatening stimuli, and one of the current leading hypotheses to explain SIDS is a failure of the infant to arouse (8). Supporting this idea is the finding that one of the few indicators in prospective studies that distinguished infants who subsequently died of SIDS from healthy infants was the occurrence of fewer body movements during sleep in the SIDS infants, indicating fewer episodes of spontaneous arousal (9, 10).

Previous longitudinal studies by our group in term infants have demonstrated a sleep state-related difference in arousal response to an external stimulus, with infants at 2–3 wk after birth and 2–3 mo of age being significantly more difficult to arouse in QS than in AS (11, 12). Arousability to the external stimulus paralleled the probability of spontaneous arousal from sleep, which was consistently lower in QS at both ages. The methodology used in these studies, a graded pulsatile puff of air applied alternately to the nares, was readily applicable in preterm infants. This study reports our findings in a similar longitudinal study in healthy premature infants contrasted with term infants matched for gestational age. The specific aims of the study were to investigate whether prematurity affected arousal responses and whether an increased arousal threshold, when compared with term infants, could be a factor in their increased risk for SIDS.

METHODS

Subjects.

Ethical approval for this project was granted by the Monash Medical Centre Human Ethics Committee. Nine premature infants (three boys and six girls) born at 31–35 wk (median, 33 wk) gestation with birth weights of 1591–2408 g (mean ± SEM, 2112 ± 120 g) were recruited from Newborn Services, Monash Medical Centre. Apgar scores were 5–9 (median, 8) at 1 min and 8–10 (median, 9) at 5 min, and all infants had required <2 d of assisted ventilation and their subsequent clinical course was uneventful. Maternal age ranged from 21 to 36 y (mean, 31 ± 1.7 y), and eight were nonsmokers; one mother had smoked in the first 3 mo of pregnancy. Infants were studied on three occasions, a preterm study at a mean postconceptional age of 36.7 ± 0.2 wk (range, 35–38 wk) and mean weight of 2441 ± 108 g (range, 2050–3160 g), a term study at a mean post-term age of 7.5 ± 2 d (range, 2–21 d) and mean weight of 3121 ± 484 g (range, 2540–3880 g), and a term study at 2–3 mo post-term, mean 65 ± 5 d (range, 48–75 d) and mean weight 5218 ± 961 g (range, 3850–5880 g). Five of the infants were exclusively breast-fed at the 40- to 42-wk study, and the remaining four were both breast-fed and bottle-fed. One infant was lost to follow-up at the second study. Control infants for this study were 22 healthy full-term infants (seven boys and 15 girls) recruited from the Moorabbin Birth Centre of the Monash Medical Centre (12). All infants were from a low-risk obstetric group born at term (median, 40 wk; range, 37–42 wk) with birth weights of 3578 ± 85 g (range, 3040–4240 g). Apgar scores were 4–10 (median, 8) at 1 min and 8–10 (median, 9) at 5 min. Maternal age ranged from 21 to 37 y (mean, 28 ± 1 y), and 15 mothers were nonsmokers, four had smoked during pregnancy, and information was unavailable for three mothers. Control infants were studied on two occasions, at a mean postnatal age of 13 d (range, 9–17 d) and 2–3 mo post-term (mean, 78 d; range, 56–98 d). At the time of the first study, 19 infants were breast-fed and two were bottle-fed.

All infants were studied at the Sleep Laboratory in the Department of Paediatrics, Monash Medical Centre, with daytime polysomnography recordings between 1000 h and 1600 h. Infants generally had both a morning and afternoon sleep interrupted by a midday feed. On arrival, the electrodes for the physiologic variables were attached to the infant while it fed, and when drowsy, the infant was placed in a bassinet under dim lighting and constant room temperature (22–23°C). The study did not begin until the infant was in a stable sleep state.

Recording methods.

Recordings were made on a Grass Polygraph model 78A 16-channel recorder (Grass Instrument Company, Quincy, MA) of EEG, electrooculogram (EOG), chin EMG, ECG, instantaneous HR, thoracic and abdominal breathing movements (Resp-ez Piezo-electric sensor, EPM Systems, Midlothian, VA), expired CO2 (CO2/O2 Analyser, Engstrom Eliza MC, Bromma, Sweden), and arterial blood oxygen saturation (Biox 3700e Pulse Oximeter, Ohmeda, Louisville, CO). Sleep state was assessed as either QS, AS, or indeterminate sleep using EEG, behavioral, HR, and breathing pattern criteria according to Anders et al. (13) and Curzi-Dascalova and Mirmiran (14).

Sleep diaries and sleep profiles.

Parents were asked to complete a sleep diary covering the 48-h period before each sleep study. Analysis of the daytime sleeping patterns and sleep duration for the period of the study was performed to see whether the arousal studies affected the usual sleeping pattern or sleep duration. Sleep epoch length during the sleep studies was determined for complete epochs of both AS and QS, with those epochs terminated by arousals induced by our stimulus being disregarded from the analysis. Sleep cycle length was determined as the time between the onset of successive epochs of AS; only complete epochs of AS and QS were included in this analysis.

Stimulus and arousal criteria.

A pulsatile air-jet (frequency of 3 Hz for 5 s) delivered to the nostrils of the infant was used to induce arousal in both AS and QS, and arousal thresholds were calculated in a similar manner to that described in Read et al. (11). This method was based on that of the double staircase method of Cornsweet (15). The stimulus was presented alternately to the left and right nostrils; if the infant failed to arouse, the air-jet pressure was increased and the stimulus again presented to that nostril. Whenever an arousal response occurred, the pressure was then decreased. The changes of the stimulus driving pressure ranged from 25 to 200 cm H2O, but were usually 100 cm H2O. Arousal threshold was calculated as the mean pressure between each arousal and nonarousal response. In determining whether a presentation elicited an arousal response, the three criteria as described previously (11) were used: change in ventilation pattern of more than two breaths, an observed behavioral response, and an HR acceleration of >10% above baseline. A fourth criterion, that of an increase in EMG activity, was used in the preterm infants. All these changes needed to occur within 7 s of the stimulus onset, allowing for the time delay to reach peak HR acceleration. The 10 s of recording immediately preceding the stimulus presentation was the baseline level used to assess the change in each variable. After each stimulus presentation, a score of 1 (response) or 0 (no response) was assigned for each variable and summed to obtain a total response score. A cut-off total response score of ≥3 was used as the designation of an arousal response in the preterm infants and ≥2 in the term infants. Of the criteria used in scoring an arousal, a behavioral response was present in 97% of those responses scored as an arousal, EMG activity was closely correlated with behavioral responses and occurred in 93%, ventilatory changes occurred in 85%, and tachycardias occurred in 75% of preterm infants. Results were similar in term infants, with behavioral responses occurring in 92% of the total arousal responses, ventilatory changes in 90%, and tachycardias in 77%.

Data analysis.

For each infant, AS and QS epochs were numbered in sequence, and the time of presentation of each stimulation from the onset of each sleep epoch was recorded. The site of presentation (left or right nostril), stimulus pressure (centimeters of water), and the researcher's designation of the response was noted on the chart recording at the time each stimulation was given.

Before each stimulus presentation, the new pressure was calibrated on the chart record. To measure the probability of spontaneous arousal from sleep, the number of arousals (using the same arousal criteria as for the stimulus arousals) that coincided with the 5-s calibration of the stimulus was measured. During this time, care was taken that the air-jet did not come in contact with the infant. The probability of spontaneous arousal was calculated as the number of spontaneous arousals occurring expressed as a percentage of the total number of stimulus calibrations.

Data were first tested using the Kolmogorov-Smirnov normality test and the Levene median test for equal variance. Arousal thresholds for left and right nostrils were compared using a one-way ANOVA. For both AS and QS, mean arousal threshold, sleep epoch length, and sleep cycle length were calculated for each infant and compared for each of the preterm and term studies using a two-way ANOVA for repeated measures. Comparisons between preterm and term infants at matched conceptional ages were made with two-way ANOVA. Comparisons of the spontaneous and test arousal probabilities were made using χ2 analysis. Relationships between arousal threshold in a given sleep state and the total time asleep for each sleep state, for each infant, were determined by linear regression analysis. Sleep durations recorded by parents in the sleep diaries were compared with those obtained during the sleep studies with paired t test. All values are expressed as mean ± SEM, and a p < 0.05 was considered significant.

RESULTS

There was no difference between the mean ages of the two groups of infants at the 2–3 wk or 2–3 mo studies.

Spontaneous arousal.

In preterm infants, the probability of spontaneous arousal from sleep was greater in AS than in QS in studies at 2–3 wk and 2–3 mo post-term (p < 0.001) (Table 1). Similarly, the probability of spontaneous arousal was greater in AS than QS (p < 0.001) in term infants at both ages studied (Table 1). In AS, preterm infants had significantly more spontaneous arousals at the 2- to 3-wk study than term infants (p < 0.01), conversely, at 2–3 mo, term infants had significantly more spontaneous arousals than preterm infants (p < 0.001). In term infants, the probability of spontaneous arousal was significantly greater in the 2- to 3-mo study than in the 2- to 3-wk study in both QS (p < 0.05) and AS (p < 0.001). In QS, the percentages of spontaneous arousals were not different between preterm and term infants at the 2- to 3-wk study; however, preterm infants had significantly fewer spontaneous arousals at 2–3 mo than term infants (p < 0.01).

Table 1 Summary of stimulus presentations and responses in preterm and term infants *p < 0.001 AS vs QS. †p < 0.01 preterm vs term. ‡p < 0.001 preterm vs term.

In both preterm and term infants, the probability of spontaneous arousal was significantly lower (p < 0.001) than the probability of arousal in response to the stimulus in both sleep states and in all studies (Table 1).

Arousal threshold.

No difference between arousal thresholds was found between nostrils in either premature infants (left nostril, 167 ± 7 cm H2O; right nostril, 168 ± 9 cm H2O, n = 9) or term infants (left nostril, 234 ± 19 cm H2O; right nostril, 226 ± 19 cm H2O, n = 22). Accordingly, the data for each nostril have been pooled for all subsequent analyses of threshold.

In preterm infants, there was no difference between arousal thresholds in AS and QS at 36 wk or at 2–3 wk post-term; however, at 2–3 mo, the arousal threshold in QS was significantly greater than in AS (p < 0.05) (Fig. 1). This was caused by a significant decrease in arousal thresholds in AS (p < 0.02) rather than any change in QS thresholds. In contrast, in term infants, there was a consistent state-related difference in arousal thresholds, with thresholds being significantly higher in QS than in AS at both 2–3 wk and 2–3 mo (p < 0.001) (Fig. 1). Also, in contrast to the preterm infants, in term infants, there was no significant decrease in AS arousal thresholds with age. When arousal thresholds were compared between preterm and term infants there was no difference in AS at either age, but in QS at 2–3 mo, thresholds were significantly lower in preterm infants (p < 0.05) (Fig. 1).

Figure 1
figure 1

Comparison of mean arousal thresholds in preterm (black bars) and term (gray bars) infants during AS (solid bars) and QS (hatched bars).

Arousal threshold is plotted against time within each sleep state in Figure 2 with regression lines for each preterm infant. When data from all preterm infants were combined, arousal threshold was positively correlated with the length of time the infant had been in a particular sleep state in AS at 36 wk and 2–3 wk (p < 0.001 and p < 0.01, respectively); however, preterm infants remained readily arousable over the entire AS sleep period at 2–3 mo. In QS, arousal threshold was not correlated with time in sleep state at 36 wk; however, at 2–3 wk and 2–3 mo, arousal threshold increased significantly (p < 0.01and p < 0.001, respectively) with time asleep.

Figure 2
figure 2

Regression lines plotted through the arousal thresholds for each preterm infant plotted against time in sleep state for AS 36 wk study (A), QS 36 wk study (B), AS 2–3 wk post-term study (C), QS 2–3 wk post-term study (D), AS 2–3 mo post-term study (E), and QS 2–3 mo post-term study (F).

Sleep profiles.

The sleep diary data for the preterm infants during the 48-h before each study are summarized in Table 2. Mean sleep duration during the period of each sleep study was not different to that recorded in the infant's home during the same period on the 2 d before each study.

Table 2 Comparison of sleep duration covering the period of the study averaged for the 48 h before the sleep study with that recorded during the sleep study for preterm infants

Sleep epoch durations are presented in Table 3. In preterm infants, the mean epoch length of AS fell with gestational age, and the length of epochs at 2–3 mo were significantly shorter than at 36 wk gestation (p < 0.05). There was no difference in the length of QS epochs with age. Term infants showed a similar pattern, with AS epoch lengths declining with age (p < 0.05) (Table 3). There was no difference between epoch lengths in either sleep state between term and preterm infants at either 2–3 wk or 2–3 mo. In both groups of infants, the percentage of the total time asleep spent in QS increased with increasing postnatal age whereas the percentage of time spent in AS decreased (p < 0.001). In term infants, cycle length was significantly shorter at 2–3 mo (p < 0.009); however, there was no difference in cycle lengths at either 2–3 wk or 2–3 mo between the two groups of infants.

Table 3 Mean epoch length, percentage of total time asleep spent in each sleep state, and cycle length for preterm and term infants *p < 0.05 36 wk vs 2–3 mo preterm infants, 2–3 wk vs 2–3 mo term infants. **p < 0.009 2–3 wk vs 2–3 mo term infants. †p < 0.001% time in QS. ‡p < 0.001% time in AS.

DISCUSSION

In this study of preterm infants, we have confirmed our previous findings in term infants that arousal threshold is altered by sleep state and, in addition, have demonstrated that arousability is also affected by gestational and postnatal age. In a previously published study, we demonstrated that term infants had a significantly higher arousal threshold in QS than in AS at both the 2- to 3-wk and 2- to 3-mo studies (p < 0.001) (11). In contrast, in this study, we have demonstrated that in preterm infants this state-related difference in arousability is maturationally delayed and does not appear until 2–3 mo post-term. In term infants, arousal thresholds in either state did not change with age; however, in preterm infants, arousal thresholds decreased in AS at 2–3 mo whereas thresholds in QS remained unchanged across the ages studied. Additionally, when arousal thresholds were compared between term and preterm infants, the preterm infants had significantly lower arousal thresholds in QS at 2–3 mo.

Although several previous studies have demonstrated that arousability is decreased in QS compared with AS in term infants (11, 12, 1619), no studies have compared preterm and term infants in a longitudinal study at matched postconceptional ages. The differences in arousability between term and preterm infants may be explained in terms of delayed maturation of sleep patterns in preterm infants. The postnatal maturation of sleep involves both the temporal organization of sleep and wake patterns and the development of sleep EEG characteristics. This maturation parallels the maturation of the CNS (20). Previous polysomnographic studies have demonstrated that significant differences exist in sleep patterns between full-term and preterm infants at matched postconceptional ages, when studied at term (2123). Sleep states appear poorly organized, with a lower correlation between electrophysiologic variables of sleep and behavioral variables in comparison to term infants at 40 wk, and authors have concluded that sleep in premature infants is immature (21, 22). Scher et al. (23) reported that sleep cycles were longer, trace-alternate patterns were more abundant in QS, and low-voltage irregular patterns were less abundant in AS in preterm infants. The authors concluded that these differences reflected altered brain maturation in the preterm infants. Previous studies have not, however, continued past term conceptional age. A less mature QS in the preterm infants in our study may explain the lack of a state-related difference in arousability at 2–3 wk post-term and a increased arousability in this state at 2–3 mo in comparison to term infants.

The finding that arousability was not depressed in preterm infants when compared with term infants was unexpected. The healthy preterm infants in this study were born between 31 and 35 wk gestation and were selected for their uncomplicated neonatal histories. The previously reported increased risk for SIDS in premature infants has been inversely correlated to gestational age (5); however, studies have not taken into account the neonatal histories of the infants who subsequently died (24). It is therefore possible that those preterm infants at most risk for SIDS are infants born at earlier gestational ages with more severe clinical problems requiring extensive medical intervention. Recent studies by our group in premature infants with neonatal histories of apnea of prematurity have found significantly elevated arousal thresholds in QS at 2–3 mo in comparison to the healthy premature infants reported in this paper (24).

AS epoch lengths declined as did percentage of time spent in AS with increasing gestational age, whereas percentage of time in QS increased in both groups of infants, as has been previously reported (25). In our study, we found no difference between preterm and term infants in the epoch lengths of either QS or AS or in the cycle lengths at either 2–3 wk or 2–3 mo of age. Cycle length declined between 2–3 wk and 2–3 mo in both groups, but this only reached significance in term infants. Previous work by Anders and Keener (25) supports our findings of no difference in percentage of time spent in AS and QS or in sleep cycle length between term and preterm infants.

Previous studies have also identified differences in spontaneous arousal from sleep between term and preterm infants. Scher et al. (23) found fewer and shorter arousals and fewer body movements in preterm infants. Our data extend these findings inasmuch as the infants in their study were only studied up to term conceptional age and not longitudinally and, in addition, were studied while sleeping in the prone position. Prone position is known to reduce spontaneous arousals from sleep in term infants (26), although the effects on preterm infants have not been investigated. Our study has also demonstrated that preterm infants had significantly fewer spontaneous arousals from sleep at 2–3 mo in both QS and AS than term infants. The probability of spontaneous arousal decreased with age in both QS and AS in preterm infants; however, arousal thresholds to our air-jet stimulus did not change with postnatal age in QS, whereas arousal thresholds in AS decreased. In term infants, we found no changes in arousability with age in either sleep state. In both preterm and term infants, we found that the probability of spontaneous arousal paralleled that of arousability to our external stimulus in that the probability of spontaneous arousal was highest in AS, the state associated with lower arousal thresholds. These findings suggest that the arousal pathways involved in both spontaneous arousal from sleep and our stimulus-induced arousals are similar.

One of the advantages of the air-jet stimulus is that it allows the continuous measurement of arousability across an entire sleep period. Previously, we demonstrated that term infants remained readily arousable across entire epochs of AS at both 2–3 wk and 2–3 mo whereas arousal thresholds increased with time spent in QS at 2–3 mo (11). We hypothesized that this may reflect the maturation of the EEG with the development of sleep spindles, which appear in QS at 2–3 mo of age and are known to exert inhibitory influences on the reticular formation, inhibiting arousal and maintaining sleep (21). Premature infants exhibited a similar pattern, with arousal threshold increasing with time in QS at 2–3 mo. Also, as was found in term infants, the variability in AS arousal thresholds was significantly less than in QS at 2–3 mo, with infants remaining readily arousable during the entire AS period. Inasmuch as arousability is lower in QS, an infant may be more vulnerable to arousal failure when exposed to a life-threatening stress in this state.

The possibility that preterm infants exhibit a delay in maturation may also be reflected in the finding that preterm infants die of SIDS at a later postnatal age, but a younger postconceptional age, than term infants (2). Malloy and Hoffman (4) demonstrated that premature infants born at 29–32 wk and 33–36 wk gestation had a postconceptional age at death because of SIDS of 47.3 ± 8.6 wk (mean ± SD) and 48.0 ± 8.3 wk, respectively, compared with 53.3 ± 8.5 wk for infants born at term. Infants in this study were born at 31–35 wk gestation. Thus, as in our study, correction for postconceptional age did not remove differences arising from preterm delivery.

In conclusion, this longitudinal study has demonstrated that arousal from sleep is not impaired in preterm infants compared with term infants at the age when SIDS incidence is highest.