Objective. Most neonatologists include an apnea-free period in the criteria for the discharge of preterm infants. However, the length of time one should wait after the cessation of apnea before sending an infant home without a monitor is debated. We undertook this study in an attempt to define a minimal and safe observation period between the time of the last apnea episode and discharge.
Methods. We reasoned that in infants with idiopathic apnea of prematurity, the intervals between days on which apnea occurs gradually increase until some point at which clinically significant apnea ceases. Therefore, knowledge about the intervals between days on which apnea occurred just before the last apnea would provide a reasonable estimate of the minimal safe observation interval between the last apnea and discharge. We reviewed the charts of 266 infants born in 1993 and 1994 at ≤32 weeks' gestational age or weighing ≤1500 g at birth from two institutions to determine the intervals between the day on which the last apnea occurred and the previous two days on which apnea occurred. One hundred seventy-five infants were excluded because they never experienced apnea, or data about the last apnea was missing, or they were on xanthines during the period encompassing the last 3 apnea days, or they weighed <1500 g or were <34 weeks' postmenstrual age at the time of the last apnea. Of the 91 remaining infants, gestational age at birth, birth weight, 1- and 5-minute Apgar scores, and discharge weight were not different between the two institutions. For each infant we determined the longest of the intervals between the 2 days on which apnea occurred previous to the day of the last apnea (MAXINT for maximum interval). The infants were then ordered by MAXINT and, starting at the longest MAXINT, the medical records of each infant were carefully examined for other conditions known to be associated with apnea (eg, recovering from anesthesia, sepsis, chronic lung disease, and so forth). The minimal safe observation period was then defined as the longest MAXINT in which there was at least 1 infant with no other explanation for the apnea other than prematurity.
Results. The median duration of the intervals between the 2 days on which apnea occurred previous to the day on which the last apnea occurred were 3.0 and 2.0 days and the median duration of the MAXINT was 4.0 days. On careful examination of the charts, it was determined that each of 13 infants with a MAXINT preceding the day on which the last apnea occurred of greater than 8 days had some other condition that might result in apnea, including residual lung disease, sepsis, surgery, and so forth. In contrast, among the group of infants with a MAXINT of ≤8 days, at least 1 infant at each MAXINT (eg, 1 to 8) had significant apnea with no other explanation other than prematurity.
Conclusions. We conclude that otherwise healthy preterm infants continue to have apneas separated by as many as 8 days before the last apnea before discharge. Conversely, infants with longer apnea intervals often have identifiable risk factors other than apnea of prematurity.
A;-4q large percentage of premature infants born at <32 weeks' gestation continue to have significant apnea until near the time of discharge.1-4 Most neonatologists include in their discharge criteria a 5 to 10 day apnea-free period after discontinuing xanthine therapy before sending a premature infant home without a home monitor (Table 1). Although the appropriate duration of this apnea-free period is debated, there is little or no data, except perhaps years of anecdotal experience, that support either the concept of an apnea-free period, or any particular duration.
Recently, shortening the length of stay has become a popular strategy for reducing hospital inpatient costs. Significant theoretical savings have been calculated when systematic attempts have been made to shorten the length of stay.5 It seemed reasonable, therefore, to hypothesize that the duration of the apnea-free period required before discharge would influence the timing of discharge and thus the length of stay.
The major purpose of this investigation was to determine whether we could safely shorten the duration of the apnea-free period currently incorporated into the discharge criteria of preterm infants at two major academic medical centers, the University of Virginia Medical Center (UVA), and the Dartmouth-Hitchcock Medical Center (DHMC). The current practices at the two institutions are consistent with the practice of many neonatologists (Table 1). Xanthines are discontinued and the infants are observed before discharge. Currently, an apnea-free period of 7 days at DHMC and 10 days at UVA is required before sending the infant home. Infants are rarely sent home on xanthines. Occasionally, infants are sent home with monitors. Most of these are infants who require oxygen and/or ventilatory support. In rare instances, an infant with persistent mild apnea with no associated oxygen desaturation or bradycardia, is discharged with a home monitor.
We reasoned that as the respiratory control system matures, the frequency of clinically significant apnea6,7 would not follow a smooth course with a gradual increase in the time between apneas. Rather, we anticipated that intervals between apneic events would increase until at some point during development, a threshold would be reached after which clinically significant apnea would cease. We therefore hypothesized that the information on the apnea-free intervals between the days on which apnea occurred just before the last apnea would help define the minimum safe apnea-free period. We assumed that information in the medical records would identify significant clinical apnea events and that our current guidelines for required apnea-free periods (10 days at UVA and 7 days at DHMC) were safe.
We focused on a population in which the incidence of apnea would likely influence the timing of discharge. We reviewed the medical records of 266 infants born in 1993 and 1994, who were ≤32-weeks' gestational age or who weighed ≤1500 g at birth, and who were discharged home from our facilities. We did not review the charts of infants who died or who were transferred to other facilities. A total of 175 infants were excluded because: 1) there were no documented episodes of apnea (n = 77); 2) there were no data on the last apnea (n = 21); or 3) they were still receiving xanthine therapy during the period of study (n = 34). In addition, infants who were <34-weeks postmenstrual age (PMA) or weighed <1500 g at the time of their last apnea (n = 43) were excluded because apnea was not likely to influence the timing of discharge. The final data set consisted of 91 infants, 36 from UVA and 55 from DHMC.
We collected information on birth, admission, and discharge; the hospital course; selected other diagnoses; feeding; and apnea. Apnea information included the timing of the last 3 days on which apnea occurred, the severity of the apnea, xanthine use, and the intervals between selected significant events and the last 3 days on which apnea occurred. The timing of most major events were standardized by expressing their occurrence in terms of PMA.
Apnea information was taken from the apnea record kept at the bedside. The definition of clinically significant apnea was similar for the two institutions. To be considered a clinically significant event, the apnea duration was required to be >15 to 20 seconds, or the duration could be shorter if accompanied by bradycardia or color change.6,7 At each institution, a scoring system was used to record whether stimulation was required, whether there was bradycardia (defined as heart rate <80 beats/min), and whether there was color change (pallor, cyanosis, or an oxygen saturation of <80%) associated with the apnea event. At DHMC the apnea alarm was routinely set to 20 seconds, whereas at UVA, the alarm was set at 15 seconds. At both institutions, all clinically significant apneas as defined above were considered important when determining discharge timing.
The apnea record at each institution was examined and only those events meeting the above definition were included. Each significant apnea event was then assigned a descriptive score consisting of 1 to 4 numbers. A 1 was assigned if stimulation was required, a 2 if apnea duration was longer than 15 seconds (UVA) or 20 seconds (DHMC), a 3 if there was a color change or oxygen saturation was <80%, and a 4 if heart rate fell to <80 beats per minute. Because in most cases, nurses responded to the heart rate alarm rather than to an apnea alarm, the exact duration of most apneas was unknown. Thus, in most cases, a 2 was scored only when the nurse responded to the apnea alarm (as opposed to the heart rate alarm), or rarely, if the duration of the apnea was actually measured on the monitor screen. If pallor, cyanosis, and/or bradycardia occurred without an apparent apnea, the event was not included in the analysis.
The method used to determine the timing of the last apnea and the preceding apnea-free intervals is shown in Fig1. The intervals were determined between the days on which apnea occurred rather than between individual apnea events. The day on which the last apnea occurred was first defined. The 2 previous days on which at least one apnea occurred were then determined and the intervals between the days calculated. In cases in which there were more than one apnea on any given day, the most severe apnea was characterized.
Determination of the Shortest Safe Apnea-free Period
We reasoned that the intervals between the days on which apnea occurred before the last apnea would provide information on the appropriate time to wait after the last apnea before discharge. For example, if the previous intervals were all significantly less than the current required apnea-free period (in our case 7 or 10 days), one might reasonably consider shortening the apnea-free period required before discharge. Discharge considerations are confounded, however, by conditions, other than prematurity, that might also precipitate an apnea, such as an intervening infection or surgical procedure. Because we also wanted to err on the side of safety, we took the following approach to the analysis and interpretation of the duration of the apnea-free intervals before the day on which the last apnea occurred. For each infant, the longer of the two apnea-free intervals before the last apnea [maximum interval or (MAXINT)] was determined. Infants were then arranged according to their MAXINT, in descending order. Starting with the infants with the longest MAXINT, the medical records of each infant were carefully reviewed to determine whether there were any other conditions that might precipitate an apnea. Finally, we defined the shortest safe apnea-free period as the longest MAXINT in which at least 1 infant experienced a clinically significant apnea with no explanation, other than apnea of prematurity.
Data Reduction and Analysis
In some cases, subpopulations with certain characteristics were compared with those without those characteristics. In these analyses, a one-way analysis of variance (ANOVA) was attempted. If it was determined that the populations were not normally distributed or had dissimilar variances, a Kruskal-Wallace ANOVA on ranks was performed. Comparisons for the time period were performed with a one-way ANOVA for repeated measures or a Friedman One-Way Repeated Measures ANOVA on Ranks. Correlations were determined with a Pearson Product Moment analysis, and in some cases, linear regressions and other univariate and multivariate analyses were performed. When the timing of various events were compared, they were standardized by expressing each event in terms of maturity or PMA.
The Patient Population
Table 2 shows the general characteristics of the study population. It should be emphasized that the group of infants we studied represents a specialized subpopulation of infants in which apnea would likely influence the timing of discharge. Therefore, it may differ from a general neonatal intensive care unit (NICU) population in several respects. In our study population, brain hemorrhage occurred in 22% of infants. In those infants with brain hemorrhage, 35% of bleeding was classified as Grade I, 15% as Grade II, 30% as Grade III, and 20% as Grade IV. Hydrocephalus was diagnosed in 14.3% of infants. At 28 days of age, 66.7% of infants required supplemental oxygen. When the infants cared for at the two institutions were compared, there were no differences in mean or median gestational age at birth, birth weight, the 1- and 5-minute Apgar scores, the weight at discharge, or the number of days on continuous positive airway pressure (CPAP). However, 38% of the DHMC infants were ≤27 weeks' gestation compared with 25% of the UVA infants (P < .03; Fisher's exact test). In addition, the number of ventilator days (P = .002) and oxygen days (P = .021) were greater for the infants from DHMC compared with those from UVA. In addition, infants from DHMC were slightly more mature at the time of discharge as measured by PMA compared with the infants from UVA (median PMA at discharge, 38.7 weeks versus 40.0 weeks for UVA and DHMC infants, respectively; P = .014).
The timing of the occurrences of apnea and the intervals between days on which apnea occurred are shown in Table3. When all of the representative apneas were analyzed, 43.9% of the apneas were greater than 15 seconds in duration, 49% required stimulation, 58% were associated with a color change or an oxygen saturation <80%, and 87% were associated with bradycardia. Twenty-nine percent of this select group of infants were discharged on home monitors. Most of these were infants who had residual lung disease and who were still receiving supplemental oxygen at the time of their last apnea (23.4% of all infants). The remaining 6% were infants who had persistent mild apnea and who had met other criteria for discharge. In some of these cases, monitors were requested by parents.
Relationship of Important Clinical Events or Population Characteristics to the Timing of Discharge
The timing of selected clinical events during the course of hospitalization with respect to discharge is shown in Fig2. Because we selected a population in which the timing of discharge was likely to be influenced by the incidence of apnea, it was not surprising that the PMA at the time of discharge was highly correlated with the PMA of the last apnea day (R = 0.861; P < .001). Using bivariate analyses, other factors including the PMA when oxygen was discontinued, the PMA at the time of weaning from xanthines, PMA when weaned from ventilatory support, and the PMA when full oral (PO) feedings were achieved were also correlated with the timing of discharge. Further multivariate analyses showed that the PMA on the day of the last apnea, and the PMA when oxygen was discontinued had the greatest impact on the timing of discharge. These two variables were also highly correlated with each other. However, additional analyses showed that in this subpopulation of infants, there is a clear impact of the timing of the last apnea on the timing of discharge, independent of the existence of residual lung disease.
The Impact of the Timing of Other Significant Events on the PMA of the Last Apnea
Figure 3 shows that PMA age at the time last apnea was inversely correlated with gestational age at birth. When the infants were grouped by quartiles (closed circles in Fig 3), an ANOVA showed an overall significant relationship. Infants born at less than or equal to the median of 28 weeks' gestation were more mature (38.7 ± 0.3 weeks PMA) at the time of their last apnea, as measured by PMA, compared with infants who were >28 weeks' gestation at birth (36.3 ± 0.4 weeks PMA; P < .001). In addition, 36% of infants born at ≤28 weeks' gestation experienced apnea beyond 40 weeks PMA, compared with only 2.4% of infants born at >28 weeks' gestation (P < .001; Fisher's exact test).
Determination of the Shortest Safe Apnea-free Period
The distribution of the maximum apnea-free interval before the day on which the last apnea occurred (MAXINT) is shown in Fig4. Careful analysis of the medical records of infants starting with the longest MAXINT revealed that, in every case, infants with a MAXINT of greater than 8 days experienced conditions that might precipitate an apnea. One infant with a MAXINT of 8 days had no explanation for the apnea other than apnea of prematurity. Similarly, in each group of infants with MAXINTs <8 days, 1 or more had no other explanation for their apnea. Table 4 lists the conditions identified for the 17 infants with MAXINT >8 days. More than one confounding condition was identified in 5 infants. Infants with MAXINTs >8 days had more severe apnea (P = .029), and were more mature when full PO feeding was first achieved (P = .028) compared with infants with MAXINTs ≤8 days.
Much has been written on the incidence, etiology, and treatment of apnea of prematurity.1-4 Relatively little attention, however, has been paid to defining criteria for discharge of premature infants who have experienced significant apnea. Despite the lack of a clear rationale, except for perhaps several years of experience and some common sense, a quasi standard has evolved, to which most neonatologists adhere, that requires that infants be apnea free for a certain period of time before discharge.3,8,9 A recent survey of 252 neonatologists attending a recent large national meeting showed that 89% of those surveyed required an apnea-free observation period before discharge (Table 1). Of those that required an apnea-free period of observation, 86% observed infants for ≥5 days, with 75% requiring a 5- to 7-day observation period. Presumably, this provides an appropriate margin of safety, decreasing the probability that significant apnea will occur at home. The origin of these particular durations is unclear. A careful review of the literature has failed to reveal any previous scientific data to support any specific interval.
General Study Design
Notwithstanding our underlying bias that some period of observation before discharge is consistent with good common sense, and prompted by pressures to shorten length of stay and because of the distinct lack of information in the literature, we designed this retrospective study in an attempt to better define the minimum safe duration of an apnea-free period before discharge. We reasoned that as the respiratory control system matures, the frequency of clinically significant apnea6,7 would not follow a smooth course with a gradual increase in the time between apneas. Rather, we anticipated that at some point during development, a threshold would be reached after which clinically significant apnea would cease.
We decided therefore to examine the apnea-free days between apneas that occurred before the day on which the last apnea occurred. We reasoned that the duration of these apnea-free intervals would help determine the minimum interval. For example, if these intervals were all considerably shorter than our current criteria (7 days for DHMC and 10 days for UVA), then we would feel comfortable about decreasing our current apnea-free interval criteria. In our experience, apnea does not always occur as an isolated event. Often several apneas occur on any given day. In the current study, multiple apnea events occurred on 42% of the days on which apnea occurred. Also, it is clear that all apneas recorded as being clinically significant are not simply a developmental phenomenon, but may be precipitated by many other factors, including changes in environmental temperature, handling, anesthesia, head positioning, stretching, stooling, and many well known pathological conditions.1,2,10 Normally, all of these factors are taken into consideration by the caretakers when deciding which events to count.
We made two major assumptions in the design of this study: 1) the medical record would accurately reflect the number and description of clinically significant apnea, and 2) our current criteria of being apnea free for 7 to 10 days was safe. One could legitimately question both of these assumptions. There is ample evidence that not all apneas are recorded by clinical observers, compared with using electronic monitoring techniques.11,12 However, in the majority of cases, infants are discharged based on criteria involving clinical observers, rather than electronic methods. We submit, therefore, for the purposes of this study, using clinical observations is appropriate, and even preferred. To our knowledge, there have been no studies confirming the safety of current apnea discharge criteria. In our experience, however, it is rare for premature infants to be readmitted to the hospital because of apnea not associated with other conditions (respiratory syncytial virus, and so forth). Therefore, it is unlikely that infants in our populations experienced clinically significant apnea that was not described in this report.
We purposely limited our study population to a group of premature infants in which apnea might influence the length of stay. We therefore chose to include only those infants who were ≤32-weeks' gestational age or ≤1500 g birth weight and who were discharged home from our nurseries. We also excluded those infants who had not reached 34 weeks' PMA age or who had not attained a weight of 1500 g at the time of their last apnea. We reasoned that few if any infants would be discharged home at <34 weeks or <1500 g and therefore thought that in these cases apnea would not influence the timing of discharge as much as other factors such as adequate temperature regulation or the attainment of full PO feedings. Finally, we eliminated any infants who were still on xanthines during the period starting with the second to last apnea day, because, except in rare situations, it is our general policy not to discharge infants who are still receiving xanthines. In addition, we did not review the records of a large number of infants who either did not have apnea or were discharged back to community hospitals. Therefore, the results of this study cannot be generalized to apply to the entire NICU population, but only to the group of infants in whom apnea would likely influence the timing of discharge.
There were some differences between the DHMC and UVA subpopulations. Although the mean or median gestational age, birth weight, and condition in the delivery room were comparable, DHMC infants seemed to be more ill and to require more support and ended up going home slightly later than UVA infants. We doubt that this reflects major differences in practices, but is more likely a result of differences introduced by the selection process (eg, exclusion of back-transported infants).
Relationship of Other Events to the Last Apnea and Discharge
Others have reported that the timing of discharge of infants in the NICU is related to the maturity at birth and the degree of illness during the hospitalization.13,14 In our study population, which included those infants for which apnea would likely influence the timing of discharge, the PMA on the day on which the last apnea occurred was, as expected, the variable most highly correlated with the day of discharge. Multivariate analyses showed that the apparent relationship between the PMA when oxygen therapy was discontinued and the timing of discharge was largely dependent on the relationship between residual lung disease and the timing of the last apnea. In our select population, maturity at birth was only poorly correlated to the timing of discharge. This was most likely because of the large percentage of very immature infants in our study population.
Contrary to what one might expect, we found that infants ≤28 weeks' gestation at birth were more likely to stop having apnea at a later PMA than those infants >28 weeks' gestation at birth. These findings are consistent with those of Eichenwald et al,15 who found that 36% of infants <24 weeks' gestation and 11% of infants <28 weeks' gestation continued to have significant apnea beyond 39 weeks' PMA. In addition, these authors also reported that infants remained in the hospital on average 16 days awaiting resolution of self-resolving apnea and completion of an apnea-free period. In both studies, the impact of maturity at birth on the timing of the cessation of apnea, however, was not completely separated from the possible influence of chronic lung disease. In our population, infants were discharged on average 18 days (median, 18.2 days) after reaching full PO feedings and at a median of 11 days after the last apnea day. In contrast to the population of infants studied by Eichenwald et al,15 the achievement of full PO feedings occurred after the last apnea day in 50% of our infants and 47% required stimulation to recover from their apnea.
Our results are also consistent with those reported by Poets et al,16 who reported that 68% of preterm infants who were otherwise ready for discharge experienced a prolonged decrease in oxygen saturation (≤80% for ≥4 seconds) associated with apnea. The frequency of these events decreased significantly with increasing gestational age. Although the infants studied were reported as ready for discharge, there was no description of any discharge criteria related to the respiratory pauses and hypoxemia.
Cessation of apnea of prematurity may be a reflection of brainstem development as was suggested by the relationship between the cessation of apnea and maturity of the auditory evoked brainstem potentials reported by Henderson-Smart et al.17 Our data suggest that the nervous system of infants who are very immature at birth may not develop at the same rate as those born more mature. Supporting this hypothesis, a recent report by Hüppi et al,18described more immature patterns of gray-white matter distribution, myelination, and neurobehavior in premature infants when they reached term PMA compared with infants born at term. These observations and our data argue strongly against the advisability of using a fixed weight or PMA as a criteria for discharge. The infant born at 25 weeks' gestation, for example, who reaches 1800 g, may well be less mature and at higher risk than another 1800-g infant born at 32 weeks' gestation.
Although not extensively analyzed, in the DHMC population the PMA when full PO feedings were achieved also seemed to be correlated with the PMA on the day on which the last apnea occurred. It has also been reported that infants with feeding tubes experience more apnea than those without feeding tubes.19,20 It is interesting to speculate that removal of feeding tubes at the time that full PO feedings are achieved may decrease airway obstruction and thus perhaps decrease the frequency of apneas.
The Minimum Safe Apnea-free Period
Our results indicate that infants continue to have significant ideopathic apnea separated by as many as eight days up until the time of discharge. Of those infants who had MAXINTs <8 days, 82% of the apneic events on the last apnea day were associated with bradycardia and 49% with significant color change or oxygen saturations <80%. In addition, 46% of the apneic events were severe enough to require stimulation. Although there was no clear linear relationship between the duration of the maximum apnea-free interval and the need for stimulation, 30% of the apneas in infants with MAXINTs of 5 to 7 days required stimulation, whereas 60% of apneas in infants with MAXINTs of 2 to 4 days required stimulation (P = .035, Fisher's exact test).
Implications for Length of Stay Issues
Our study indicates that current guidelines for observation of infants after the cessation of clinically significant apnea before discharge are reasonable and confirm the wisdom of a long established practice. Simply shortening the apnea-free period of observation to shorten the length of stay would, in our populations, be associated with an increasing probability of significant apnea occurring at home, and could not be recommended. In addition, the assumption that infants should be free of dangerous apnea after some particular PMA or weight does not seem to be warranted. Our data indicate that infants born very prematurely (≤28 weeks' gestation) may continue to have significant apnea beyond term, as defined by 40 weeks' PMA.
In summary, this analysis has confirmed that maintaining preterm infants with a history of apnea in the hospital until they are apnea-free for a prescribed period of time, is a reasonable clinical practice. Such infants are likely to have recurrent events, many of which are associated with major instability and require intervention. Conversely, infants who have been apnea-free for 8 days or more are unlikely to have another apnea, unless they have some other risk factor known to be associated with apnea (eg, surgery, oxygen dependence, sepsis, and so forth). Our data also indicate that infants who continue to have apnea after reaching term are more likely to have been born at an earlier gestational age. Apnea of prematurity is a developmental phenomenon and a certain degree of maturity needs to be achieved before it is no longer a major risk to an infant's life.17The cessation of significant apnea may represent a reasonable developmental marker that can be determined with some degree of accuracy and is closely associated with the resolution of other developmental problems associated with prematurity such as the achievement of full PO feeding. Therefore, attempts to reduce lengths of stay by diminishing the required apnea-free period are not likely to be successful and could result in a life-threatening event at home.
- Received September 20, 1996.
- Accepted March 10, 1997.
Reprint requests to (R.A.D.) Department of Pediatrics, Dartmouth Medical School, Dartmouth-Hitchcock Medical Center, Lebanon, NH 03756.
- UVA =
- University of Virginia •
- DHMC =
- Dartmouth-Hitchcock Medical Center •
- PMA =
- postmenstrual age •
- MAXINT =
- maximum interval •
- ANOVA =
- analysis of variance •
- NICU =
- neonatal intensive care unit •
- CPAP =
- continuous positive airway pressure
- ↵Henderson-Smart DJ. Apnea of prematurity. In: Beckerman RC, Brouillette RT, Hunt CE, eds. Respiratory Control Disorders in Infants and Children. Baltimore, MD: Williams & Wilkins; 1992:161–177
- Nelson NM
- ↵US Department of Health and Human Services. Infantile Apnea and Home Monitoring: Report of a Consensus Development Conference. Washington, DC: National Institutes of Health; 1987. NIH Publication No. 87–2905
- Kattwinkel J
- ↵Klaus MH, Fanaroff AA. Care of the High Risk Neonate. 4th ed. Philadelphia, PA: WB Saunders Co; 1993:252
- Long JG,
- Philip AGS,
- Lucey JF
- Southall DP,
- Levitt GA,
- Richards JM,
- et al.
- Muttitt SC,
- Finer NN,
- Tierney AJ,
- Rossmann J
- Powell PJ,
- Powell CVE,
- Hollis S,
- Robinson MJ
- Pearlman SA,
- Stachecki S,
- Aussprung HL,
- Raval N
- Eichenwald EC,
- Aina AJ,
- Stark AR
- Poets CF,
- Stebbens VA,
- Alexander JR,
- Arrowsmith WA,
- Salfield SAW,
- Southall DP
- Van Someren V,
- Linnett SJ,
- Stothers JK,
- Sullivan PG
- Copyright © 1997 American Academy of Pediatrics