KActa Paediatr86:873-80.1997

The febrile stress of routine vaccination does not increase central apnoeain normal infantsDM Tappin', RPK Ford1, KP Nelson'. B Price2. PM Macey3 and R Dove2Community Paetlintrir Unit, HealthUnk South'. Department of Medical Physics'. Christchurcli Hospital ami Department of Electrical anilElectronic Engineering . University of Canterbury. Christchurcli, New Zealand

Tappin DM. Ford RPK, Nelson KP. Price B, Macey PM. Dove R. The febrile stress of routine vaccinationdoes not increase central apnoea in normal infants. Acta P;cdialr 1997; 86:873-80. Stockholm. ISSN 0803-5253We tested the hypothesis that the febrile stress of routine vaccination would increase central apnoea innormal infants. Twenty-one normal infants had continuous overnight breathing and temperature recorded athome, before and after 58 routine vaccination episodes. Central apnoea. of at least 5 sec duration, wasdetected by computer algorithm and confirmed by human inspection. The longest recorded apnoea was16sec (;i = I) during 3629h of sleep. Overnight rectal temperature increased after vaccination (median0.52°C, 95% CI 0.40. 0.65). Apnoea density reduced on 46/53 vaccination nights (median -29%. 95% CI-20, -37) followed by an increase on subsequent nights (median +10%. 95% CI +1%.+21%). Overall,apnoea density was similar during the 3 nights preceding and 4 nights following vaccination (median + 1 %,95% CI +9.-6). The febrile stress of routine vaccination did not increase central apnoea in normal infants.Q Human infant, respiration physiology, sleep physiology. Sudden Infant Death aetiology, temperatureDM Tappin. Scottish Col Death Trust. Royal Hospital for Sick Children. Yorkhill. Glasgow G.I HSJ. UK

The febrile reaction produced by routine vaccination hasbeen well described (I). There has been concern that febrilestress may lead to increased apnoea (2). Apnoea followingvaccination has been reported among preterm graduates ofa neonatal intensive care unit (3). Many paediatricianswarn mothers who use apnoea monitors that more alarmepisodes may occur after a vaccination, although thisadvice is not based on scientific evidence. Alarm episodesmay be true alarms associated with prolonged centralapnoea, or false alarms perhaps due to displacement ofthe sensor associated with irritability or sweating. Thisstudy aimed to assess if routine vaccination increased thelength or frequency of central apnoea or of false apnoeaalarm episodes recorded in normal term infants at homeafter routine vaccinations at 6 weeks, 3 months and 5months of age.

Subjects and methodsEnrolmentEthics committee approval allowed continuous overnighttemperature and respiratory patterns to be recorded ininfants at home. Recruiting of normal infants was madeby repeated newspaper articles in the Christchurcli dailynewspaper asking women in late pregnancy to volunteertheir infants for temperature and respiration recordingduring the first 6 months after birth. The study researchnurse returned telephone calls and arranged a convenienttime to meet at the parents' home to demonstrate theHomeLog (4) equipment and recording procedures. Parents

were then asked to phone her when their baby was born ifthey still wanted to join the study. At the first recording at 2weeks of age the research nurse delivered the equipment andasked for written consent. She was available by telephone incase of equipment breakage or other problems. Parents alsoattended a course on infant resuscitation.

SubjectsBetween 1 September 1993 and 31 August 1994,21 infantshad at total of 431 nights of temperature and respirationdata recorded using four HomeLog systems (4). Thenumber of nights recorded for each infant ranged between7 and 30. mean 20.5. All infants (12M and 9F) were born atterm, had no neonatal problems and mean birthweightwas 3850g (range 3000-4300g). Twenty infants werebreastfed. Twenty infants were put down to sleep on theirback or side, one infant was placed prone to sleep. Allmothers were non-smokers. All but 1 of the 20 infantsthrived throughout their first 6 months. One infant whostarted on the 90th weight centile dropped to the 10th at 3months associated with failure of lactation. He startedformula feeds and soon returned to the 75th centile.

VaccinesAt 6 weeks, a combined adsorbed diphthcria-tetanus-pertussis (DTP) (Diptheria toxoid, Tetanus toxoid andPertussis-whole inactivated organisms, with adjuvantaluminium hydroxide, and preservative thiomersal) wasgiven as an intramuscular injection into one arm and.

© Scandinavian University Press 1997. ISSN 0803-5253

874 DM Tappin et at. ACTA P/fcDIATR *6 < IW|

concurrently. Hepatitis B subunit vaccine (HB-Vax II.MSD) was given intramuscularly into the other arm. Thiswas repeated at 3 months with the addition of oral liveattenuated poliomyelitis vaccine. At 5 months, DTP andoral polio vaccine were given.

EquipmentEach HomeLog system was based on a 286 IBM-PCcompatible laptop computer with built-in digital signalconversion capable of recording 16 input channels simultaneously directly onto the hard disk. Temperature probeswere based on a semiconductor and were validated locally(5). Temperature measurements were sampled at a rate of1 Hz, digitized and recorded. Regular two-point softwarecalibration was performed on each set of temperatureprobes against a mcrcury-in-glass reference thermometermarked to 0.1 °C Rectal temperature was recorded from aprobe 5 cm from the anal margin, sealed in a soft plasticfeeding tube. A second probe in this plastic tube 2 cm fromthe anal margin was used to indicate that the rectal probewas positioned correctly and had not slipped down.

Graseby MR 10 respiration monitors were used to measure abdominal movements. This signal was continuouslyrecorded at 10 Hz. Monitors were calibrated monthly toreduce false alarms. Parents were supplied with a clipboardto log nightly details which included time put down, sleeptimes, time of morning waking and the start and finish offeeds and nappy changes. Each clipboard had a digitalclock attached which was set to the HomeLog computerclock at the start of each recording period and remainedwithin 2min of the computer time over a 10-day period.Parents were asked to record the time of Graseby monitoralarm episodes, the condition of their infant and what theyfelt had caused the alarm, e.g. probe disconnected. DMTlooked at each of these episodes on the Graseby record,searching for long episodes of apnoea for 10 rnin before and10 min after the documented time. After each 3-7-nightperiod of recording, data was downloaded from the Home-Log PC to a 1 Gb optical drive storage facility via tapedrive. Each 7-night record required about 19 Mb of storagespace.

Apnoea density per hour of sleepBabyLog analysis used a peak sensing algorithm (6) tocalculate the length of each individual breath. Apnoea wasdefined by the Graseby signal as a flat region (B-C, Fig. 1)usually preceded by a decay curve (A-B. Fig. I). where thecombined length of the decay curve and the flat region wasat least 5 sec. Five seconds was chosen as being shortenough to include any significant apnoea. A computeralgorithm (written by PM) was used to detect apnoeas inall overnight records. This algorithm had been previouslyshown to have a sensitivity of 99% to detect apnoeas, but aspecificity of 71 %, giving 29% false positive apnoea detections when compared with three human experts (6). Therefore each "algorithm detected" apnoea was individually

Fig. I. Central apnoea and the Graseby respiratory wave form. This is atypical apnoea in a 30-scc sample of Graseby signal recorded on a Home-Log computer. Apnoea was defined by the Graseby signal as a flat region(B-C) usually preceded by a decay curve (A-B) where the combinedlength of the decay curve and the flat region was at least 5 sec.

inspected (by DMT) and false detections were discarded.The usual reason for false algorithm detections was smallamplitude signals which were easily distinguished fromtrue central apnoea.

Information from parental logs was transferred onto theBabyLog computer record. Midday was defined as the startof each "day" so that one whole night of recording would bein each "day". "Sleep" as oppose to "awake" was definedby parental logs. Of the time defined as "sleep" fromparental logs, 98.5% was consistent with sleep calculatedfrom respiratory variables (7).

For this analysis we cut each record into sequential 5-minperiods to examine associations between respiratory patternsand temperature variables. For each of these 5-min segmentsthe following variables were calculated: median breathlength, breath rate variability, median rectal temperature,"sleep" as marked on parental records, the number ofapnoeas 5 sec or longer, and the number of apnoeas 8 secor longer. The number and length of all apnoeas of 5 sec orgreater for each night were also recorded. The resulting files(size 0.7 Mb for each 60 Mb of raw data) were imported intothe SAS statistical package so that comparisons could bemade between multiple variables in this large data set.

Apnoea density (apnoeas per hour of sleep) was calculated from the number and length of apnoeas per night, andthe length of time spent in "sleep".

Recording scheduleWc set out to record each infant over four prc-determinedperiods, and at the time of any parent-determined intercurrent illness during the first 6 months of life. The predetermined periods were: (i) at 2 weeks of age for 3 nights(n = 15 infants, 44 nights recorded); (ii) at the first vaccination episode at 6 weeks of age for 7 nights (3 nightsbefore the vaccination, the night of the vaccination and 3nights afterwards); (iii) at the second vaccination episode at3 months of age for 7 nights; and (iv) at the third vaccination episode at 5 months of age for 7 nights. Twenty-sixnights associated with 10 episodes of intercurrent illnesswere recorded from 6 infants. From 431 nights, a total of

ACTA P/EDIATR 86 (1997) Central apnoea ami vaccination 875

3851 h of respiratory recording was made of which 3629were designated by parents as "sleep".

Vaccination recordings madeTwo infants were put forward by their parents after 6 weeksof age and one after 3 months, therefore two infants wererecorded only at the 3- and 5-month vaccinations and one atthe 5-month vaccination only. Another infant did not reach5 months of age by the end of the study period in August1994 and was only recorded at 6 weeks and 3 months ofage. Seventeen infants were recorded at all three vaccinations, therefore parents connected the equipment to theirinfants for a total of 58 vaccination episodes. For 36/58(62%) vaccination episodes good recordings were made foreach of the 7 nights. For one of these vaccination episodes,nights 6 and 7 following vaccination (20h of "sleep")were unfortunately deleted during analysis (the longestapnoea on these 2 nights was 12 sec, n = I). For five(9%) vaccination episodes day I was not recorded due toall machines being already in use (6 days recorded foreach); all these vaccination episodes were included in theanalysis. Table 1 shows the remaining 17 vaccinationepisodes with the reasons for missing nights and if theywere included in the analysis. Records were inspectedusing BabyLog software (8) and bad sectors removed.Bad sectors of respiratory signal were only seen during

two vaccination episodes associated with partial disconnection of the lead between the respiratory monitor and thecomputer (Table I, infants 13 and 14). Complete disconnection occurred during a further vaccination episode(Table 1. infant 10). During two further vaccination episodes, repeated alarms occurred, in one no respiratoryrecord was recorded for half of night 3. night 4. and night5 (Table 1. infant 11), during the second, the infant wasonly connected to the respiratory monitor for 3.5 h on thenight of vaccination (Table 1, infant 12). The apnoeadensity during this recording may not have been representative of the complete night of sleep, due to repeated alarmepisodes. Fifty-three of 58 vaccination episodes weretherefore included in the analysis comparing apnoea density on the three nights before vaccination, with the apnoeadensity on the night of vaccination, and the apnoea densityon the three subsequent nights. No other respiratory recordswere excluded.

Short overnight recordsThe distributions of length of overnight respiratory recordings were similar in shape comparing 156 pre-vaccinationnights, 54 vaccination nights, and 149 subsequent nights.All were slightly left skewed towards shorter length ofrecord. No significant difference was seen between themedian length of overnight records: prc-vaccination, 9.3 h

Table I. Reason for no recording of individual nights for 17 vaccination episodes from 14 infants.

Night 1 N i g h t 2 N i g h t 3

Prc-vaccination nights

Night 4 Night 5 Night 6 Night 7

Infant nights following vaccination

12

YY

YY

YParents out

YY

YY

YY

Computer fullY

3 Y Y Y Y Y Y Parents out4 Y Y Child upset Y Y Y Y5 Y Y Child asleep Y Y Y Y6 Parent wishes Y Y Y Parent wishes6 As above Y Y Y Y Y As above6 As above Y Y Y Y Y As above7 As above Y Y Y Y As above78

YNo machine

YY

YTemp sensor broken

YY

Moving houseDisconnectionmonitor from

9 Y Y Y Y As above YcomputerY

10 Y Y Disconnectionmonitor from

Y Y

I I Y YcomputerRepeated alarmsparentsdiscontinued

Y Y

12 Y Y Y As Above Y Y Y13 Y Y Partial

disconnectionrespiratorymonitoroutput plug to

14 Y Y YcomputerAs above

Infants 10-14 were not included in analysis for this vaccination episode.

876 DM Tappin et al. ACTA Prt-DIATR 86 (1997)

(lower quartilc 7.2, upper quartile 10.5). on the night ofvaccination 9.3 h (lower quartilc 7.6, upper quartilc 10.3)or the subsequent nights 9.5 h (lower quartile 7.9, upperquartile 10.6). To be sure that short overnight recordingsdid not bias the results, apnoea density analysis was repeatedexcluding nights shorter than the lower quartile of overnightrecord length. This analysis included 38 discrete vaccinationepisodes in which the apnoea density during sleep on pre-vaccination and vaccination nights could be compared.

Change in apnoea densityApnoea density varied markedly between infants (Fig. 5)and decreased with infant age. Each of the 53 vaccinationepisodes was therefore taken as a discrete event. Calculationof apnoea density was made for the 3 nights before vaccination and compared to: (i) the apnoea density on the night ofvaccination; (ii) the apnoea density on the subsequent nights;and (iii) (he apnoea density after vaccination (the night ofvaccination plus the subsequent nights) for each vaccinationepisode. Comparison was made by calculating the percentage change in apnoea density for each vaccination episode[separately for considerations (i). (ii). and (iii)j.

Central apnoea analysis using missed breathsTo allow for different breath rates on the night of vaccination compared to control nights, the length of all apnoeas 5sec and greater was also categorized by the number ofbreaths missed. An apnoea of l-missed-breath in lengthwas defined as a respiratory pause whose length was twicethe median breath length for the 5-min period in which itoccurred (3-missed-breaths = 4 breath lengths).

Sleep stateWe have previously reported from this study assignment ofsleep state to 5-min periods of respiratory record (7) usingthe method described previously. This was performed inperiods assigned by parents as "sleep".

Statistical analysisThe statistical package SAS was used to manipulate thislarge datasct. Minitab for Windows, version 10, was usedfor further statistical analysis. The significance of meantemperature rise between control nights and the night ofvaccination was assessed using a 2-sample /-test for eachvaccination episode. Assessment of the significance ofchanges from prc-vaccination nights in: temperature,respiratory rate, and apnoea density (separately for calculations I, 2 and 3), for all 53 individual vaccination episodes, was performed using the Wilcoxon Signed RankTest as the data were not normally distributed.

ResultsPrevalence of central apnoeasThe longest apnoea recorded during 3629 h of parent

designated "sleep" was 16 sec (n = I). Two hundredand ninety-six of 429 nights (69%) had at least one centralapnoea of 10 sec or above (/i = 1754).

Rectal temperature and vaccinationA significant increase in temperature was seen after 48/53vaccinations when the mean temperature on the three nightsbefore vaccination was compared with the night of vaccination, with no significant change after 5 (Fig. 2). The mediantemperature increase for all infants on the night of vaccination compared with the preceding control nights was 0.52°C(95% CI 0.40, 0.65 p < 0.0001) (range 0.00-1.99°C). Theoverall difference in mean temperature between control andvaccination nights was 0.58°C (95% CI 0.56, 0.60). Thetemperature usually returned to the pre-vaccination level bynight 5. which followed the vaccination night (Fig. 2). Asignificant fall in mean overnight rectal temperature was seenbetween the 6-week and the 3-month vaccinations (Fig. 2).

Breath rate and vaccinationA significant increase in breath rate was seen after vaccination. The median increase was 9.5% (95% CI 6.5. 12.5,p < 0.0001, range -11% to +41%).

Apnoea length and vaccinationA reduction in apnoea length was seen on the night ofvaccination (median 4%, 95% CI 2%, 5%, p < 0.001).

Apnoea density and vaccinationApnoea frequency and length generally decreased on thenight of vaccination associated with an increase in rectaltemperature and respiratory rate. Figure 3 shows apnoeafrequency and length, (in seconds and missed-breaths), forall periods recorded pre-vaccination, on the nights of vaccination, and during the subsequent nights. An individualexample is shown from a 3-month-old infant in Fig. 4. Eachof the 7 nights recorded are shown for temperature (on theupper trace) and apnoea occurrences and apnoea length(lower trace, each apnoea is a vertical line whose lengthcorresponds to the length of the apnoea). There was a largeincrease in rectal temperature during the night of vaccination (mean 1.99°C). Apnoea episodes were less frequentand shorter on the night of vaccination.

Apnoea density for prc-vaccination and vaccinationnights, comparing each discrete vaccination episode, isillustrated in Fig. 5. Central apnoea density is on a logarithmic scale which illustrates the wide variation betweeninfants. A reduction in central apnoea density was seen onthe night of 46/53 vaccination episodes. This reductionwas seen for all apnoeas of 5 sec or greater in length(median -29%, 95% CI -20, -37. p < 0.001) (Fig. 5)and in apnoeas 8 sec or greater (median -44%, 95% CI-33, -55, p < 0.001). When the dataset was limited to 38vaccination episodes with pre-vaccination and vaccinationnights greater in length than the lower quartile. the

ACTA IMiDIATRX6< 1997) Central apnoea and vaccination 877

39 -

Fig. 2. Mean overnight rectal temperature lor 21 normal Infants at 2 weeks, o weeks. 3 months and 5 months. Mean overnight rectal temperature for eachinfant covering each series of nights at 2 weeks (3 control nights only). 6 weeks (7 nights: 3 before, the night of vaccination, and the subsequent 3 nights).3 months 17 nights) and 5 months (7 nights). The black lines describe the median of the mean temperatures at each age. The error bars on mediantemperature at 6 weeks indicate that the 5 month median was outside the 95% confidence interval for the median temperature for infants at 6 week- "I age.The largest change in median temperature occurred between 6 weeks and 3 months of age. little change occurred during the next 8 weeks to 5 months ofage. (A) Peak indicates an infant who was given a vaccination during a preceding febrile episode, the mean temperature on the 3 days prior to vaccinationwas the same as the mean temperature on the night of vaccination.

reduction in apnoea density was almost identical (apnoeas5 sec and greater, median -30%, 95% CI -23. -38.P < ().()() I: apnoeas 8 sec and greater, median -47%, 95%CI -36. -59. p< 0.001).

When the 3 nights prior to vaccination were compared withthe 3 subsequent nights (not including the vaccination night),there was an increase in apnoea density, for till apnoeas 5 secor greater (median +10%. 95% CI +1 %, +21 %, p = 0.02) andfor apnoeas ol S sec or greater (median+7''i. 95''; (I S'i.+22%. p = 0.35). but the latter did not reach statistical significance. When the 3 nights prior to vaccination were compared with all 4 nights following vaccination (including thevaccination night), no significant increase in apnoea densitywas present, for all apnoeas 5 sec or greater (median +1%.95% CI +9. -6. /j = 0.8). or for apnoeas of 8 sec or greater(median -10%. 95% CI -19%, +!%./> = 0.06).

Apnoea density categorized as breaths missedWhen apnoea length was measured using the missed

breath method, for each of 53 discrete vaccination episodes, no significant reduction in apnoea density on thenight of vaccination was seen for apnoeas 3-misscd-brcathsor more in length (median -5%, 95% CI -33, +15.p = 0.61).

Sleep stale, apnoea density and vaccinationThe apnoea density of all central apnoeas of 5 sec andgreater was 9.1 /It during Rapid Eye Movement I REM I sleepand 6.4/h during i/iiiet sleep. Vaccination did not disturbthe proportion of time spent awake reported by parents(6% pre-vaccination. 4% vaccination nights. 5% subsequent nights, respectively), awake from Harper sleepstate, not noted by parents (2%. 2%. 2%), in REM (45%.46%, 46%), in quiet (33%, 32%, 33%) or in anundetermined sleep state (14%, 16%, 14%).

Parents recorded 120 episodes of waking for feeds andnappy changes pre-vaccination. 42 during vaccinationnights and I2S on subsequent nights.

878 DM Tappin el al. ACTA PiEDIATR 86 (1997)

s.>■ocCD3cr2

30

25-

20

15

10

Q.sCD

5-

£ S 0to0)ocQ.re

c0)U

□ Seriesl■ Series2■ Series3

f D J ,9 1 0 1 1 1 2 1 3 1 4 1 5 1 6

Length in seconds

3 4 5 6 7Length in 'missed-breaths'

Fig. .'. The average frequency and length of respiratory pauses for all vaccination episodes: pre-vaccination, on the night of vaccination, anil on followingnights measured in sec and "missed-breaths". The frequency distribution of all apnoeas greater than 5 sec is shown in series I for pre-vaccination nights(;r - I So), the nights of vaccination (ii — 54) in series 2. and the subsequent nights (it = I49) in series 3. The length of apnoeas is shown in sec. anil inmissed-bieaths which takes account of change in respiratory rale. The longest apnoea recorded (n = I) was 16 sec (6.2 missed-breaths) on a subsequentnight 7 recorded from a 6-week-old infant.

Eal.se alarm signals

Patents reported 190 alarm signals none of which werepreceded by a central apnoea of 20 sec or greater, thereforetill were false alarms. The frequency of false alarms wasless on the night of vaccination than before, and loweston the subsequent nights. A small amplitude signal wasseen at the time of 98 false alarm episodes. Three ofthese were associated with apnoeas of length 13, 13 and15 sec, respectively. This type of alarm diminished aspatents became more experienced at better positioningof the Graseby abdominal movement sensor. Thirty-four false alarm episodes occurred during periods ofquiet breathing when the recorded signal was of normalamplitude and regular breaths were present for 10millon either side of the alarm signal. Thirty-eight episodeswere due to displacement of the sensor, regular breathswere present until the probe became displaced.Seventeen episodes, with repeated unexplained falsealarms, were presumed to be due to poor calibrationof the Graseby monitor as the signal quality was good.no apnoeas were present and when the monitor wasreplaced, alarm episodes disappeared. Three episodeswith continuous false alarms were cured by a replacementbattery.

DiscussionThis study has shown that routine vaccination does notincrease central apnoea density in normal infants, and falsealarm episodes were less frequent after vaccination.

The subjects were a parent selected group of 21 volunteer infants. Only one infant (4%) was placed to sleep pronewhich is a similar proportion to a recent survey of childcare practices in Christchurcli (9). None of these motherssmoked during pregnancy, in contrast to a recent surveyusing cord blood colinine which showed that 22% ofChristchurcli mothers were actively smoking at the endof pregnancy (10). Ninety-percent of infants (n = 19) wereexclusively breastfed at 6 weeks of age. 86% (n = 18) at 3months and 38% (/i = 8) at 5 months. This level of breastfeeding is the same as a random sample of Christchurcliinfants from 1992 (9). With regard to these factors, the infantsstudied were representative of normal Christchurcli infants.apart from maternal smoking during pregnancy.

Using the parental home as a laboratory and parents astechnicians was successful. Their enthusiasm, and abilityto keep detailed records was very important. The equipment needed to be robust and simple as parents had to insertthe rectal temperature probe and connect the Grasebyrespiration sensor to their infant on each of 7 nights

ACTA P.EniATRS6 (1997) Cen t ra l apnoea and vacc ina t ion 879

V .r/

39.Pi

39.0"J

38.5'J

38.0"-

37.5*-

37.0*

36.5°-

36 .0

RECTAL TEMPERATURE

§o 1 0 4

4 /VM

vaccinat ion

2 0 ] C E N T R A L A P N O E A

IS-

3 4

Time (days)

Fig. -I- A 7-day record from a 3-monlh-old infant. The upper record showsa mean temperature increase of 1,99°C from pre-vaccination control nightsto the night of vaccination. The lower record show s all apnoeas of 5 see orgreater on the same time-scale. Each apnoea is shown as a vertical linewith height indicating the duration of the apnoea.

around a vaccination episode. Parents found effectivepositioning of the respiratory sensor far more difficultthan locating the rectal temperature probe, which once inplace caused few problems. It would have been helpful tohave recorded cardiac signals to assess heart rate variability, and used a nasal thermister to establish obstructiveapnoea. The feasibility and ethics of recording this extradata in the parental home on normal infants, even if robustequipment were available, may be in doubt. We concentrated on recording a few channels of robust data and thendeveloped methods of analysis.

We considered disabling the alarm on the Grasebymonitors, so that we could be sure of recording all longapnoeas. undisturbed by parents. However, this was felt tobe unwise, as in the unlikely event of a life-threateningepisode, the parents would not be alerted. False alarmepisodes associated with poor sensor placement alsohelped ensure that good quality signals were recorded.The false alarms caused parents concern, but reassurancewas given by DMT who studied the respiratory signal for10 minutes before and after the time logged from digitalclocks on parental clipboards. Apnoea was present prior to3 alarm episodes associated with a small signal. The lengthof these apnoea episodes should not in themselves havetriggered the alarm as all three were less than 20 sec duration.We presume that the monitor missed a breath before or afterthe apnoea. The overall distribution of apnoeas 5 sec orgreater, was unlikely to have been influenced by alarm

1000

3E

§ 7-I X■a oS eP

pre-6 week

pro-3 month

vaccinationnight

3 month

pre-5 month

vaccinationnight

5 month

pro-vaccinatlon and the night ot vaccination

Fig. 5. Central apnoea density pre-vaccination and on the night of V aecmation at 6 weeks, 3 months and 5 months for individual vaccination episodes. Theapnoea density is drawn on a logarithmic scale showing the large variation in apnoea density between the 21 infants. The thick black lines betweencontrol and vaccination join the median apnoea density at each age. the lighter grey lines join the pre- and vaccination night apnoea density for individualvaccination episodes. The error bars indicate the 95% confidence intervals of the median apnoea density. A significant reduction in the median apnoeadensity was present between pre- and vaccination nights: at 6 weeks a reduction of 17.1 apnoeas per night (') h of "sleep") (95% CI 6.3, 35.5), at 3 monthsa reduction of 17.1 apnoeas per night (95% CI 4.0. 29.2) and at 5 months a reduction of 16.2 apnoeas per night (95% CI 7.7. 25.2).

880 DM Tappin el id. ACTA P/F.DIATR 86 (1997)

signals, because only three apnoeas were present prior to thealarm episodes.

Following vaccination, a significant rise in rectal temperature usually occurred. Figure 2 shows that the rectal temperature had usually returned to the pre-vaccination level within36 h of the vaccination event. Wailoo previously reported theextent of temperature rise following vaccination (12). Ourstudy shows the febrile reaction to be short-lived. The average overnight temperature fell with infant age, before andafter vaccination. The major drop occurred between 6 weeksand 3 months of age as previously described (12).

Central apnoea longer than 16 sec was not present. Thislength of respiratory pause is similar to that found in otherstudies (11). The Graseby apnoea alarm setting of a 20-secrespiratory pause prior to the alarm signal is thereforeappropriate. If the 10-sec setting were used, parentswould have been disturbed, on multiple occasions, ontwo-thirds of nights.

Figure 5 illustrates the wide variation in apnoea density(logarithmic scale) between infants at each age, and the fallin apnoea density with age. Each vaccination episode wastherefore treated as a discrete event, with the 3 nightsbefore each vaccination being the control period for thenight of vaccination and the subsequent nights. The percentage changes in apnoea density for each of 53 vaccination episodes did not form a normal distribution, thereforenon-parametric statistical methods were used to assess theoverall change in apnoea density. Apnoeas of 8 sec orlonger, showed a greater reduction on the night of vaccination than apnoeas of 5 sec or longer. A compensatoryincrease occurred during the subsequent nights but overall,no increase in apnoea density was seen on the 4 nightsfollowing vaccination (including the vaccination night).

What explanations can be put forward for the reduction incentral apnoea density on the night of vaccination? Could thisresult be artefact due to less time spent asleep. By analysingapnoea density (apnoeas/h of sleep), less time spent in"sleep" should not be a problem. Parent-documented"sleep" agreed well with sleep defined from respiratoryvariables. Infant waking episodes documented by parentswere not increased on the night of vaccination, so periodsasleep should not have been shorter on the night of vaccination. No significant difference was seen in median length ofrecords on the night of vaccination compared with pre-vaccination or subsequent nights. To be sure that shorternights in some vaccination episodes did not explain thedecrease in apnoea density, a further analysis was performedwith all nights shorter than the lower quartile removed. Thedecrease in apnoea density was almost identical in the analysis of 38 remaining discrete vaccination episodes. Weconclude that the reduction in central apnoea density on thenight of vaccination was real and not artefact.

Central apnoea or respiratory pauses were more commonduring REM than quiet sleep, but the proportion of timespent in quiet and REM sleep was unaffected by vaccination. The reduction in apnoea density on the night ofvaccination cannot therefore be explained by a reductionin the proportion of time spent in REM sleep.

Would the overall increase in respiratory rate on the nightof vaccination have affected the apnoea density? Breaths wereshorter on the night of vaccination. If the rise in temperatureon the night of vaccination was associated with a rise inmetabolic rate, then shorter respiratory pauses may havebeen equivalent, in terms of carbon dioxide build-up oroxygen lack, to longer respiratory pauses prior to vaccination.To examine this further, apnoeas were categorized as thenumber of breaths missed (13). The reduction in apnoeadensity from control to vaccination nights was no longersignificant. Much of the reduction in apnoea density maytherefore be explained by a shorter breath length on the nightof vaccination.

In well normal infants vaccination was associated withboth a rise in temperature and a reduction in central apnoeadensity, on the night of vaccination. Graseby respiratorymonitor false alarm episodes did not occur more frequentlyafter vaccination.Acknowledgments.—The Canterbury Cot Death Fellowship supportedDr D. M. Tappin; the Scottish Cot Death Trust supported Mrs B. Slade.Research Nurse, the National Child Health Research Fund supported MissKerne Nelson, Riostatistician. The Graseby monitors were loaned fromPhilips (New Zealand).

References1. Raw son D. Petersen SA, Wailoo MP. Rectal temperature of normal

babies after first diphtheria, pertussis, and tetanus immunisation. ArchDis Child 1990; 65: 1305-7

2. I'erlsiein PH. Edwards NK. Sutherland JM. Apnea in prematureinfants and incubator air-temperature changes. N Engl J Med 1970;282:461-6

3. Botham SJ. Isaacs D. Incidence of apnoea and bradycardia in preterminfants following triple antigen immunization. J Pacdiatr Child Health1994; 30: 533-5

4. Ford RP. Brown PJ. Dove RA. Tuffncll CS. Macey PM. HomeLog:long-term recording of infant temperature, respiratory and cardiac-signals in the home environment. J Pacdiatr Child Health 1992; 28Suppl l:S26-S32

5. Brown J, Dove R. Price B, Fong S. Ford R. Continuous multiplelocation body temperature measurement on infants. Aust Phys EngSci Med 1990; 13: 185-7

6. Macey PM. Ford RPK. Brown PJ. Larkin J. Fright WR, Garden KL.Apnoea detection: human performance and reliability of a computeralgorithm. Acta Paedialr 1995; 84: 1103-7

7. Tappin DM. Ford RPK. Nelson KP. el al. Breathing, sleep stale, andrectal temperature oscillations. Arch Dis Child 1996: 74: 427-31

8. Dove R. Brown J. Fright R. Tuffnell C. Ford R. Computer polygraphssystem for infants at risk for Sudden Infant Death Syndrome (SIDS).Aust Phys Eng Sci Med 1990; 13: 188-91

9. Ford RPK. Cowan S. Wild CJ. Effect of Col Dcalh Prevention messagesin Canterbury. NZ Practice Nurse J 1993; September: 78-80

10. Tappin DM, Ford RPK. Wild CJ. Smoking at the end of pregnancymeasured by cord blood cotinine assay. NZ Med J 1995; 108: 108-9

11. Richards JM. Alexander JR. Shincboumc GA, deSwiet M, Wilson AJ.Southall DP. Sequential 22-hour profiles of breathing patterns andheart rale in 110 full-term infants during their first 6 months of life.Pediatrics 1984;74:763-77

12. Lodcmorc MR. Petersen SA. Wailoo MP. Factors affecting thedevelopment of night time temperature rhythms. Arch Dis Child1992; 67: 1259-61

13. Franks CI. Johnston DM. Brown BH. Non-invasive home monitoringof respiratory patterns in infants. Develop Med Child Neurol 1977;19: 748-56

Received Oct. 14, 1996. Accepted in revised form Apr. 16, 1997