- Split View
-
Views
-
Cite
Cite
Yu Sun Bin, Peter A Cistulli, Christine L Roberts, Jane B Ford, Childhood Health and Educational Outcomes Associated With Maternal Sleep Apnea: A Population Record-Linkage Study, Sleep, Volume 40, Issue 11, November 2017, zsx158, https://doi.org/10.1093/sleep/zsx158
- Share Icon Share
Abstract
Sleep apnea in pregnancy is known to adversely affect birth outcomes. Whether in utero exposure to maternal sleep apnea is associated with long-term childhood consequences is unclear.
Population-based longitudinal study of singleton infants born during 2002–2012 was conducted using linked birth, hospital, death, developmental, and educational records from New South Wales, Australia. Maternal sleep apnea during pregnancy was identified from hospital records. Outcomes were mortality and hospitalizations up to age 6, developmental vulnerability in the first year of school (aged 5–6 years), and performance on standardized tests in the third year of school (aged 7–9 years). Cox proportional hazards and modified Poisson regression models were used to calculate hazard and risk ratios for outcomes in children exposed to maternal apnea compared with those not exposed.
Two hundred nine of 626188 singleton infants were exposed to maternal sleep apnea. Maternal apnea was not significantly associated with mortality (Fisher’s exact p = .48), developmental vulnerability (adjusted RR 1.29; 95% CI 0.75–2.21), special needs status (1.58; 0.61–4.07), or low numeracy test scores (1.03; 0.63–1.67) but was associated with low reading test scores (1.55; 1.08–2.23). Maternal apnea significantly increased hospitalizations in the first year of life (adjusted HR 1.81; 95% CI 1.40–2.34) and between the first and sixth birthdays (1.41; 1.14–1.75). This is partly due to admissions for suspected pediatric sleep apnea.
Maternal sleep apnea during pregnancy is associated with poorer childhood health. Its impact on developmental and cognitive outcomes warrants further investigation.
INTRODUCTION
Maternal sleep apnea confers risk for pregnancy complications which have adverse consequences for both mother and baby. These include a twofold increase in the maternal complications of gestational diabetes1–3 and pregnancy hypertension/pre-clampsia1,2,4 and increased risks for the infant outcomes of intrauterine growth restriction1/low birthweight,1,2 preterm birth,1,5,6 and neonatal intensive care admission.1,6 The chronic intermittent hypoxia and sleep fragmentation associated with sleep apnea are hypothesized to induce oxidative stress and systemic inflammation which contribute to the development of cardiometabolic complications in pregnancy and to adversely affect the uterine environment.7
The exact prevalence of sleep apnea in pregnancy is unknown,1,2 but the nuMoM2b cohort of 3100 nulliparous women recently reported clinically significant sleep-disordered breathing in 3.5% at 6–15 weeks gestation,8 increasing to 8.3% at 22–31 weeks. The severity of apnea also appears to increase with increasing gestation.8 Beyond the immediate impact on birth, it remains unknown whether infants experience long-term consequences as a result of in utero exposure to maternal sleep apnea. Thus, the aim of the current study was to examine whether exposure to maternal sleep apnea is associated with adverse health, developmental, and educational outcomes in childhood.
METHODS
We conducted a population-based retrospective data linkage study by combining state birth, hospital, and death records with information on developmental and educational achievement collected at school. We used data from New South Wales (NSW), which is the most populous state of Australia, with 7.6 million inhabitants or a third of the nation’s population and 90000 births each year. We derived the study data from the following datasets:
NSW Perinatal Data Collection, 1994–2012: These are mandated birth records of all babies born ≥20 week’s gestation and ≥400 g birthweight. Information on maternal health, pregnancy, labor, delivery, and infant characteristics is recorded by the attending midwife or medical practitioner.
NSW Admitted Patients Data Collection, July 2000–March 2014: These hospital records are a census of all discharges, transfers, and deaths from public and private hospitals and day procedure centres. Trained medical coders record diagnoses and procedures associated with each admission according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM)9 and the Australian Classification of Health Interventions.10
NSW Registry of Deaths, January 1994–March 2014: This is the mandated death registry with all deaths certified by a medical practitioner.
Australian Early Developmental Census (AEDC), 2009 and 2012: This a triennial measure of early childhood development administered nationally in the first year of school and first conducted in 2009. Teachers complete the Australian version of the Early Development Instrument (AvEDI) of over 100 items for each child to assess development in the five domains of (1) physical health and well-being, (2) social competence, (3) emotional maturity, (4) language and cognitive skills, and (5) communication skills and general knowledge. Scores for each domain are nationally standardized and students are considered “developmentally vulnerable” if they score below the 10th percentile on one or more of the five domains.11 Students with special needs due to chronic medical or physical conditions are included, but their scores are not released with the research data; hence, “special needs status” was considered a separate developmental outcome.
National Assessment Program–Literacy and Numeracy (NAPLAN), 2008–2014: All Year 3, 5, 7, and 9 students are assessed on basic skills in reading, writing, language conventions (spelling, grammar, and punctuation), and numeracy.12,13 Only results for students in government (i.e., public) schools were available in the research dataset, and due to the youth of the birth cohort, this study only considered results from the Year 3 exams.
The above datasets were linked by the NSW Centre for Health Record Linkage (http://www.cherel.org.au/) using probabilistic record linkage and ChoiceMaker software. Probabilistic linkage is used because Australia does not have a national system of unique individual identifiers. Linkage uses multiple personal identifiers to match records belonging to an individual. The process involves assigning weights to pairs of records based on the degree of matching on identifying data such as names, birth dates, and addresses. Highly weighted pairs of records are considered matches and lowly weighted pairs are considered non-matches. Clerical review is conducted of uncertain matches and the process is repeated until there are fewer than 5/1000 erroneous matches when identifying data are available.14 The accuracy of probabilistic linkage has been shown to be similar to deterministic record linkage15,16 (i.e., the process used where there is a unique individual identifier). The data were anonymized to preserve privacy and researchers were provided a linkage key to merge the relevant records for this study. Ethics approval for the data linkage and this study came from the NSW Population and Health Services Research Ethics Committee.
Study Population
The study population derived from all women who gave birth to singleton infants in NSW from January 1, 2002 to December 31, 2012. Only the first singleton birth from each woman during the study period was included in the study population.
Exposure
Sleep apnea was identified from maternal hospital records using diagnosis code G47.3 “Sleep apnea” from the International Classification of Diseases–Australian Modification (ICD-10-AM).9 This included the subcategories of central and obstructive sleep apnea (Supplementary Table S1). Women with a sleep apnea code in hospital records in the year before or during pregnancy were considered to have maternal sleep apnea, whereas women without such a code were considered not to have sleep apnea. Previous validation against medical record review has shown that conditions identified using hospital records have a specificity of over 99%.17
Outcomes and Follow-up
There were four childhood outcomes of interest: (1) mortality, (2) hospitalizations, (3) development, and (4) standardized educational test performance. For mortality and hospitalizations, all infants were followed to the sixth birthday or the end of available data (March 31, 2014), whichever came first. Thus, the length of follow-up ranged from 1.25 to 6.00 years depending on the date of birth. Mortality included any death from birth to the sixth birthday, with deaths ascertained from the birth, hospital, and death records. Hospitalizations were separated into infant and childhood periods: infant hospitalization referred to any hospitalization after the birth discharge home until the first birthday, whereas childhood hospitalization comprised any hospital admission between the first and sixth birthdays. All infants had follow-up to the first birthday. Median follow-up for hospitalizations to sixth birthday was 6.0 years (interquartile range 4.1–6.0 years).
The developmental outcomes from the AvEDI were developmental vulnerability or special needs status in the first year of school. Developmental assessment was conducted in 2009 and 2012 and therefore only available for children in our study population if they were enrolled in the first year of school during these years. Age at follow-up for these children ranged from 4 to 7 years. The educational outcomes were performance on standardized tests of reading and numeracy. Only those children enrolled in Year 3 in a government school during 2008–2014 had educational data. Age at follow-up for these children ranged from 7 to 9 years.
Covariates
The birth and hospital records provided information on maternal age at pregnancy, maternal country of birth (Australia/elsewhere), residential postcode, parity (primiparous/multiparous), smoking during pregnancy (any/none), mode of delivery (vaginal/caesarean) gestational age at birth (preterm < 37 weeks/term ≥37 weeks), neonatal intensive care unit/special care nursery admission (yes/no), infant sex (male/female), and birthweight. Quintile of socioeconomic disadvantage was calculated based on residential postcode on the entire population of women giving birth in NSW and applied to the current study population.18 Birthweights were grouped into those small- (<10th percentile) and large-for-gestational age (>90th percentile) using infant sex and gestational age norms.19
The developmental data provided information on age at assessment, any language other than English spoken at home, and year of assessment. Identifiers for school and teacher were used to account for clustering of the results by these variables. The educational data provided information on age at testing, non-English speaking background, parental education, parental occupation, and year of test. The highest level of parental education and parental occupation was derived by combining data for both parents. A school identifier was used to account for the clustering of test scores by school.
For the post hoc analysis, potential pediatric sleep apnea was identified by examining records of infant and childhood hospitalizations for diagnostic codes indicating sleep apnea, snoring, or other abnormality of breathing; and for procedure codes indicating tonsillectomy or adenoidectomy (Supplementary Table S2). The presence of any of these codes was considered indicative of pediatric sleep apnea or suspicion of this condition.
Statistical Analyses
Contingency tables and chi-squared tests were used to compare groups exposed and unexposed to maternal sleep apnea on maternal and birth characteristics (Table 1). Factors associated with developmental and educational assessment were similarly compared between groups (Tables 2 and 3). It is well-recognized that maternal and pregnancy characteristics are predictive of infant health, whereas sociodemographic factors such as parental education, parental occupation, and non-English speaking background (in an English-speaking country) are associated with developmental and educational outcomes. We considered these variables for inclusion in the analyses as potential confounders, depending on whether they were also significantly associated with the exposure of maternal apnea (Tables 1–3).
. | No sleep apnea N = 625979 (n, col%) . | Maternal sleep apnea N = 209 (n, col%) . | Difference between groups Test statistic, p . |
---|---|---|---|
Maternal age | χ23 = 34.84, p < .01 | ||
Missing | 56 (<0.1) | None | |
<20 years | 31907 (5.1) | 7 (3.4) | |
20–34 years | 470713 (75.2) | 127 (60.8) | |
≥35 years | 123303 (19.7) | 75 (35.9) | |
Born in Australia | 420000 (67.1) | 164 (78.5) | χ21 = 12.24, p < .01 |
Disadvantage quintile (SEIFA) | χ25 = 10.32, p = .07 | ||
Missing | 2561 (0.4) | None | |
Most disadvantaged | 116878 (18.7) | 30 (14.4) | |
Disadvantaged | 119107 (19.0) | 31 (14.8) | |
Average | 124293 (19.9) | 42 (20.1) | |
Advantaged | 128852 (20.6) | 58 (27.7) | |
Most advantaged | 134288 (21.5) | 48 (23.0) | |
Primiparous | 412051 (65.8) | 151 (72.3) | χ21 = 3.83, p = .05 |
Smoking during pregnancy | 79408 (12.7) | 27 (12.9) | χ21 = 0.01, p = .92 |
Caesarean section | 180260 (28.8) | 95 (45.5) | χ21 = 28.27, p < .01 |
Preterm birth (<37 weeks) | 38463 (6.1) | 34 (16.3) | χ21 = 37.11, p < .01 |
NICU/SCN admission | 96096 (15.4) | 67 (32.1) | χ21 = 44.86, p < .01 |
Male infant | 322994 (51.6) | 100 (47.9) | χ21 = 1.18, p = .28 |
Small-for-gestational age | 67819 (10.9) | 14 (6.7) | χ22 = 8.64, p = .01 |
Large-for-gestational age | 56968 (9.1) | 38 (18.2) | χ22 = 26.02, p < .01 |
. | No sleep apnea N = 625979 (n, col%) . | Maternal sleep apnea N = 209 (n, col%) . | Difference between groups Test statistic, p . |
---|---|---|---|
Maternal age | χ23 = 34.84, p < .01 | ||
Missing | 56 (<0.1) | None | |
<20 years | 31907 (5.1) | 7 (3.4) | |
20–34 years | 470713 (75.2) | 127 (60.8) | |
≥35 years | 123303 (19.7) | 75 (35.9) | |
Born in Australia | 420000 (67.1) | 164 (78.5) | χ21 = 12.24, p < .01 |
Disadvantage quintile (SEIFA) | χ25 = 10.32, p = .07 | ||
Missing | 2561 (0.4) | None | |
Most disadvantaged | 116878 (18.7) | 30 (14.4) | |
Disadvantaged | 119107 (19.0) | 31 (14.8) | |
Average | 124293 (19.9) | 42 (20.1) | |
Advantaged | 128852 (20.6) | 58 (27.7) | |
Most advantaged | 134288 (21.5) | 48 (23.0) | |
Primiparous | 412051 (65.8) | 151 (72.3) | χ21 = 3.83, p = .05 |
Smoking during pregnancy | 79408 (12.7) | 27 (12.9) | χ21 = 0.01, p = .92 |
Caesarean section | 180260 (28.8) | 95 (45.5) | χ21 = 28.27, p < .01 |
Preterm birth (<37 weeks) | 38463 (6.1) | 34 (16.3) | χ21 = 37.11, p < .01 |
NICU/SCN admission | 96096 (15.4) | 67 (32.1) | χ21 = 44.86, p < .01 |
Male infant | 322994 (51.6) | 100 (47.9) | χ21 = 1.18, p = .28 |
Small-for-gestational age | 67819 (10.9) | 14 (6.7) | χ22 = 8.64, p = .01 |
Large-for-gestational age | 56968 (9.1) | 38 (18.2) | χ22 = 26.02, p < .01 |
. | No sleep apnea N = 625979 (n, col%) . | Maternal sleep apnea N = 209 (n, col%) . | Difference between groups Test statistic, p . |
---|---|---|---|
Maternal age | χ23 = 34.84, p < .01 | ||
Missing | 56 (<0.1) | None | |
<20 years | 31907 (5.1) | 7 (3.4) | |
20–34 years | 470713 (75.2) | 127 (60.8) | |
≥35 years | 123303 (19.7) | 75 (35.9) | |
Born in Australia | 420000 (67.1) | 164 (78.5) | χ21 = 12.24, p < .01 |
Disadvantage quintile (SEIFA) | χ25 = 10.32, p = .07 | ||
Missing | 2561 (0.4) | None | |
Most disadvantaged | 116878 (18.7) | 30 (14.4) | |
Disadvantaged | 119107 (19.0) | 31 (14.8) | |
Average | 124293 (19.9) | 42 (20.1) | |
Advantaged | 128852 (20.6) | 58 (27.7) | |
Most advantaged | 134288 (21.5) | 48 (23.0) | |
Primiparous | 412051 (65.8) | 151 (72.3) | χ21 = 3.83, p = .05 |
Smoking during pregnancy | 79408 (12.7) | 27 (12.9) | χ21 = 0.01, p = .92 |
Caesarean section | 180260 (28.8) | 95 (45.5) | χ21 = 28.27, p < .01 |
Preterm birth (<37 weeks) | 38463 (6.1) | 34 (16.3) | χ21 = 37.11, p < .01 |
NICU/SCN admission | 96096 (15.4) | 67 (32.1) | χ21 = 44.86, p < .01 |
Male infant | 322994 (51.6) | 100 (47.9) | χ21 = 1.18, p = .28 |
Small-for-gestational age | 67819 (10.9) | 14 (6.7) | χ22 = 8.64, p = .01 |
Large-for-gestational age | 56968 (9.1) | 38 (18.2) | χ22 = 26.02, p < .01 |
. | No sleep apnea N = 625979 (n, col%) . | Maternal sleep apnea N = 209 (n, col%) . | Difference between groups Test statistic, p . |
---|---|---|---|
Maternal age | χ23 = 34.84, p < .01 | ||
Missing | 56 (<0.1) | None | |
<20 years | 31907 (5.1) | 7 (3.4) | |
20–34 years | 470713 (75.2) | 127 (60.8) | |
≥35 years | 123303 (19.7) | 75 (35.9) | |
Born in Australia | 420000 (67.1) | 164 (78.5) | χ21 = 12.24, p < .01 |
Disadvantage quintile (SEIFA) | χ25 = 10.32, p = .07 | ||
Missing | 2561 (0.4) | None | |
Most disadvantaged | 116878 (18.7) | 30 (14.4) | |
Disadvantaged | 119107 (19.0) | 31 (14.8) | |
Average | 124293 (19.9) | 42 (20.1) | |
Advantaged | 128852 (20.6) | 58 (27.7) | |
Most advantaged | 134288 (21.5) | 48 (23.0) | |
Primiparous | 412051 (65.8) | 151 (72.3) | χ21 = 3.83, p = .05 |
Smoking during pregnancy | 79408 (12.7) | 27 (12.9) | χ21 = 0.01, p = .92 |
Caesarean section | 180260 (28.8) | 95 (45.5) | χ21 = 28.27, p < .01 |
Preterm birth (<37 weeks) | 38463 (6.1) | 34 (16.3) | χ21 = 37.11, p < .01 |
NICU/SCN admission | 96096 (15.4) | 67 (32.1) | χ21 = 44.86, p < .01 |
Male infant | 322994 (51.6) | 100 (47.9) | χ21 = 1.18, p = .28 |
Small-for-gestational age | 67819 (10.9) | 14 (6.7) | χ22 = 8.64, p = .01 |
Large-for-gestational age | 56968 (9.1) | 38 (18.2) | χ22 = 26.02, p < .01 |
. | No apnea N = 114,986 n (col%) . | Maternal sleep apnea N = 39 n (col%) . | Difference between groups Test statistic, p . |
---|---|---|---|
Child gender | χ21 = 0.08, p = .78 | ||
Male | 59323 (51.6) | 21 (53.9) | |
Female | 55663 (48.4) | 18 (46.2) | |
Socioeconomic disadvantage at birth (quintile) | χ25 = 6.42, p = .58 | ||
Missing | 87 (<0.1) | None | |
Most disadvantaged | 21547 (18.7) | X (10) | |
2 | 23065 (20.1) | X (15) | |
3 | 22774 (19.8) | X (28) | |
4 | 23647 (20.6) | X (21) | |
Least disadvantaged | 23866 (20.8) | X (26) | |
Age at assessment (years) | χ21 = 0.18, p = .67 | ||
≤4–5 | 36083 (31.4) | 11 (28.2) | |
6–≥7 | 78904 (68.6) | 28 (71.8) | |
Language other than English at home | 24575 (21.4) | 6 (15.4) | χ21 = 0.83, p = .36 |
Year of assessment | χ21 = 3.71, p = .05 | ||
2009 | 67573 (58.8) | 17 (43.6) | |
2012 | 47413 (41.2) | 22 (56.4) |
. | No apnea N = 114,986 n (col%) . | Maternal sleep apnea N = 39 n (col%) . | Difference between groups Test statistic, p . |
---|---|---|---|
Child gender | χ21 = 0.08, p = .78 | ||
Male | 59323 (51.6) | 21 (53.9) | |
Female | 55663 (48.4) | 18 (46.2) | |
Socioeconomic disadvantage at birth (quintile) | χ25 = 6.42, p = .58 | ||
Missing | 87 (<0.1) | None | |
Most disadvantaged | 21547 (18.7) | X (10) | |
2 | 23065 (20.1) | X (15) | |
3 | 22774 (19.8) | X (28) | |
4 | 23647 (20.6) | X (21) | |
Least disadvantaged | 23866 (20.8) | X (26) | |
Age at assessment (years) | χ21 = 0.18, p = .67 | ||
≤4–5 | 36083 (31.4) | 11 (28.2) | |
6–≥7 | 78904 (68.6) | 28 (71.8) | |
Language other than English at home | 24575 (21.4) | 6 (15.4) | χ21 = 0.83, p = .36 |
Year of assessment | χ21 = 3.71, p = .05 | ||
2009 | 67573 (58.8) | 17 (43.6) | |
2012 | 47413 (41.2) | 22 (56.4) |
For the purposes of statistical disclosure control, X indicates censored cell due to n < 5 and adjacent cells which can be used to identify cells with small n. Rounded percentages for that variable are also reported.
. | No apnea N = 114,986 n (col%) . | Maternal sleep apnea N = 39 n (col%) . | Difference between groups Test statistic, p . |
---|---|---|---|
Child gender | χ21 = 0.08, p = .78 | ||
Male | 59323 (51.6) | 21 (53.9) | |
Female | 55663 (48.4) | 18 (46.2) | |
Socioeconomic disadvantage at birth (quintile) | χ25 = 6.42, p = .58 | ||
Missing | 87 (<0.1) | None | |
Most disadvantaged | 21547 (18.7) | X (10) | |
2 | 23065 (20.1) | X (15) | |
3 | 22774 (19.8) | X (28) | |
4 | 23647 (20.6) | X (21) | |
Least disadvantaged | 23866 (20.8) | X (26) | |
Age at assessment (years) | χ21 = 0.18, p = .67 | ||
≤4–5 | 36083 (31.4) | 11 (28.2) | |
6–≥7 | 78904 (68.6) | 28 (71.8) | |
Language other than English at home | 24575 (21.4) | 6 (15.4) | χ21 = 0.83, p = .36 |
Year of assessment | χ21 = 3.71, p = .05 | ||
2009 | 67573 (58.8) | 17 (43.6) | |
2012 | 47413 (41.2) | 22 (56.4) |
. | No apnea N = 114,986 n (col%) . | Maternal sleep apnea N = 39 n (col%) . | Difference between groups Test statistic, p . |
---|---|---|---|
Child gender | χ21 = 0.08, p = .78 | ||
Male | 59323 (51.6) | 21 (53.9) | |
Female | 55663 (48.4) | 18 (46.2) | |
Socioeconomic disadvantage at birth (quintile) | χ25 = 6.42, p = .58 | ||
Missing | 87 (<0.1) | None | |
Most disadvantaged | 21547 (18.7) | X (10) | |
2 | 23065 (20.1) | X (15) | |
3 | 22774 (19.8) | X (28) | |
4 | 23647 (20.6) | X (21) | |
Least disadvantaged | 23866 (20.8) | X (26) | |
Age at assessment (years) | χ21 = 0.18, p = .67 | ||
≤4–5 | 36083 (31.4) | 11 (28.2) | |
6–≥7 | 78904 (68.6) | 28 (71.8) | |
Language other than English at home | 24575 (21.4) | 6 (15.4) | χ21 = 0.83, p = .36 |
Year of assessment | χ21 = 3.71, p = .05 | ||
2009 | 67573 (58.8) | 17 (43.6) | |
2012 | 47413 (41.2) | 22 (56.4) |
For the purposes of statistical disclosure control, X indicates censored cell due to n < 5 and adjacent cells which can be used to identify cells with small n. Rounded percentages for that variable are also reported.
. | No apnea N = 172296 n (col%) . | Maternal sleep apnea N = 49 n (col%) . | Difference between groups Statistic, p . |
---|---|---|---|
Child gender | χ21 = 0.89, p = .35 | ||
Male | 88967 (51.6) | 22 (44.9) | |
Female | 83329 (48.4) | 27 (55.1) | |
Age at test (years) | χ21 = 12.29, p < .01 | ||
≤7 | 11181 (6.5) | X (18) | |
8 | 140344 (81.5) | X (76) | |
≥9 | 20771 (12.1) | X (6) | |
Non-English speaking background | 46808 (27.2) | 6 (12.2) | χ21 = 5.51, p = .02 |
Parental education | χ23 = 2.10, p = .55 | ||
Unknown and ≤Year 12 | 47404 (27.5) | 9 (18.4) | |
Certificate I–IV | 49961 (29.0) | 16 (32.7) | |
Diploma | 23609 (13.7) | 8 (16.3) | |
Bachelor degree or above | 51322 (29.8) | 16 (32.7) | |
Parental occupation | χ24 = 9.60, p = .05 | ||
Unknown or not in paid work | 38439 (22.3) | X (18) | |
Hospitality and laborers | 25866 (15.0) | X (8) | |
Trades, office workers, sales and service staff | 34900 (20.3) | X (12) | |
Managers and associate professionals | 37266 (21.6) | X (25) | |
Senior management and professionals | 35825 (20.8) | X (37) | |
Year of test | χ26 = 6.90, p = .33 | ||
2008 | X (<0.1) | None | |
2009 | X (<0.1) | None | |
2010 | 15099 (8.8) | X (6) | |
2011 | 46824 (27.2) | X (14) | |
2012 | 42977 (24.9) | X (25) | |
2013 | 35450 (20.6) | X (27) | |
2014 | 31940 (18.5) | X (29) |
. | No apnea N = 172296 n (col%) . | Maternal sleep apnea N = 49 n (col%) . | Difference between groups Statistic, p . |
---|---|---|---|
Child gender | χ21 = 0.89, p = .35 | ||
Male | 88967 (51.6) | 22 (44.9) | |
Female | 83329 (48.4) | 27 (55.1) | |
Age at test (years) | χ21 = 12.29, p < .01 | ||
≤7 | 11181 (6.5) | X (18) | |
8 | 140344 (81.5) | X (76) | |
≥9 | 20771 (12.1) | X (6) | |
Non-English speaking background | 46808 (27.2) | 6 (12.2) | χ21 = 5.51, p = .02 |
Parental education | χ23 = 2.10, p = .55 | ||
Unknown and ≤Year 12 | 47404 (27.5) | 9 (18.4) | |
Certificate I–IV | 49961 (29.0) | 16 (32.7) | |
Diploma | 23609 (13.7) | 8 (16.3) | |
Bachelor degree or above | 51322 (29.8) | 16 (32.7) | |
Parental occupation | χ24 = 9.60, p = .05 | ||
Unknown or not in paid work | 38439 (22.3) | X (18) | |
Hospitality and laborers | 25866 (15.0) | X (8) | |
Trades, office workers, sales and service staff | 34900 (20.3) | X (12) | |
Managers and associate professionals | 37266 (21.6) | X (25) | |
Senior management and professionals | 35825 (20.8) | X (37) | |
Year of test | χ26 = 6.90, p = .33 | ||
2008 | X (<0.1) | None | |
2009 | X (<0.1) | None | |
2010 | 15099 (8.8) | X (6) | |
2011 | 46824 (27.2) | X (14) | |
2012 | 42977 (24.9) | X (25) | |
2013 | 35450 (20.6) | X (27) | |
2014 | 31940 (18.5) | X (29) |
For the purposes of statistical disclosure control, X indicates censored cell due to n < 5 or adjacent cells which can be used to identify cells with small n. Rounded percentages for that variable are also reported.
. | No apnea N = 172296 n (col%) . | Maternal sleep apnea N = 49 n (col%) . | Difference between groups Statistic, p . |
---|---|---|---|
Child gender | χ21 = 0.89, p = .35 | ||
Male | 88967 (51.6) | 22 (44.9) | |
Female | 83329 (48.4) | 27 (55.1) | |
Age at test (years) | χ21 = 12.29, p < .01 | ||
≤7 | 11181 (6.5) | X (18) | |
8 | 140344 (81.5) | X (76) | |
≥9 | 20771 (12.1) | X (6) | |
Non-English speaking background | 46808 (27.2) | 6 (12.2) | χ21 = 5.51, p = .02 |
Parental education | χ23 = 2.10, p = .55 | ||
Unknown and ≤Year 12 | 47404 (27.5) | 9 (18.4) | |
Certificate I–IV | 49961 (29.0) | 16 (32.7) | |
Diploma | 23609 (13.7) | 8 (16.3) | |
Bachelor degree or above | 51322 (29.8) | 16 (32.7) | |
Parental occupation | χ24 = 9.60, p = .05 | ||
Unknown or not in paid work | 38439 (22.3) | X (18) | |
Hospitality and laborers | 25866 (15.0) | X (8) | |
Trades, office workers, sales and service staff | 34900 (20.3) | X (12) | |
Managers and associate professionals | 37266 (21.6) | X (25) | |
Senior management and professionals | 35825 (20.8) | X (37) | |
Year of test | χ26 = 6.90, p = .33 | ||
2008 | X (<0.1) | None | |
2009 | X (<0.1) | None | |
2010 | 15099 (8.8) | X (6) | |
2011 | 46824 (27.2) | X (14) | |
2012 | 42977 (24.9) | X (25) | |
2013 | 35450 (20.6) | X (27) | |
2014 | 31940 (18.5) | X (29) |
. | No apnea N = 172296 n (col%) . | Maternal sleep apnea N = 49 n (col%) . | Difference between groups Statistic, p . |
---|---|---|---|
Child gender | χ21 = 0.89, p = .35 | ||
Male | 88967 (51.6) | 22 (44.9) | |
Female | 83329 (48.4) | 27 (55.1) | |
Age at test (years) | χ21 = 12.29, p < .01 | ||
≤7 | 11181 (6.5) | X (18) | |
8 | 140344 (81.5) | X (76) | |
≥9 | 20771 (12.1) | X (6) | |
Non-English speaking background | 46808 (27.2) | 6 (12.2) | χ21 = 5.51, p = .02 |
Parental education | χ23 = 2.10, p = .55 | ||
Unknown and ≤Year 12 | 47404 (27.5) | 9 (18.4) | |
Certificate I–IV | 49961 (29.0) | 16 (32.7) | |
Diploma | 23609 (13.7) | 8 (16.3) | |
Bachelor degree or above | 51322 (29.8) | 16 (32.7) | |
Parental occupation | χ24 = 9.60, p = .05 | ||
Unknown or not in paid work | 38439 (22.3) | X (18) | |
Hospitality and laborers | 25866 (15.0) | X (8) | |
Trades, office workers, sales and service staff | 34900 (20.3) | X (12) | |
Managers and associate professionals | 37266 (21.6) | X (25) | |
Senior management and professionals | 35825 (20.8) | X (37) | |
Year of test | χ26 = 6.90, p = .33 | ||
2008 | X (<0.1) | None | |
2009 | X (<0.1) | None | |
2010 | 15099 (8.8) | X (6) | |
2011 | 46824 (27.2) | X (14) | |
2012 | 42977 (24.9) | X (25) | |
2013 | 35450 (20.6) | X (27) | |
2014 | 31940 (18.5) | X (29) |
For the purposes of statistical disclosure control, X indicates censored cell due to n < 5 or adjacent cells which can be used to identify cells with small n. Rounded percentages for that variable are also reported.
For mortality, the rate of death was reported and Fisher’s exact test was used to determine whether there was a significant difference between groups. Due to the small number of events in the maternal sleep apnea group, no further analysis was undertaken.
For infant and childhood hospitalization, Cox proportional hazards regression was used to determine hazard ratios and associated 95% confidence intervals for those with maternal sleep apnea compared with those without maternal apnea. For infant hospitalizations, children were censored at death, their first birthday or the last day for which hospital data were available (March 31, 2014), whichever came first. For childhood hospitalizations, children were censored at death, their sixth birthday or the last day of data availability (March 31, 2014), whichever came first. We also carried out competing risks models for the two hospitalization outcomes which did not censor children at death, but the results of these did not differ from the main models described above (Supplementary Table S3).
The crude models for hospitalization included only maternal apnea as a predictor. The adjusted model controlled for maternal age, country of birth, quintile of socioeconomic disadvantage, and parity. We conducted a post hoc analysis investigating whether pediatric sleep apnea contributed to hospitalizations. Chi-squared tests were used to examine whether the rate of pediatric apnea differed between the groups exposed and unexposed to maternal sleep apnea. The Cox regression models for infant and childhood hospitalizations were also re-run after excluding those children with admissions related to pediatric sleep apnea.
For the developmental and educational outcomes, modified Poisson regression models20 were used to calculate the relative risks and associated 95% confidence intervals for children whose mothers had sleep apnea during pregnancy compared with those who did not. Generalized estimating equations were used to account for clustering with exchangeable matrices used to estimate the intraclass correlations. Correlation of developmental outcomes between children in the same school or rated by the same teacher was controlled for by including a school identifier and a teacher identifier as cluster variables. Correlation of educational outcomes was taken into account by controlling for clustering by school in all models for reading and numeracy. All developmental models included year of assessment, but models for educational achievement did not as year of testing did not differ between maternal sleep apnea groups. The adjusted model for the developmental outcomes accounted for maternal age, maternal country of birth, quintile of socioeconomic disadvantage, and parity. The adjusted model for the educational outcomes controlled for maternal age, country of birth, socioeconomic disadvantage, parity, age at test, non-English speaking background, and parental occupation. We also stratified the developmental and educational models by the sex of the child to explore potential effect modification of sleep apnea on known sex differences in development (Supplementary Table S4).
For all outcomes, we conducted a simple test of mediation by including preterm birth in the adjusted models and compared the results of these models with those without preterm birth (Supplementary Table S5). Similarly, we conducted additional analyses for the developmental and educational outcomes in which we included history of infant and childhood hospitalization as potential mediators of the relationship between maternal apnea and childhood development (Supplementary Table S6) and stratified by the sex of the child (Supplementary Table S7). Since these results do not much differ from the main results, we have presented only the above models for simplicity.
All analyses were conducted using SAS 9.3 (SAS Institute, NC). Two-sided p-value of <.05 was considered statistically significant and p-values between 0.05 and 0.10 were interpreted as a tendency toward a difference between groups. There was no formal adjustment for multiple testing.
RESULTS
There were 626188 women with singleton babies in the study population and 209 (0.03%) had sleep apnea in the year before or during pregnancy. Table 1 shows the pregnancy characteristics by maternal sleep apnea. Women with sleep apnea were older, more likely Australian-born, and tended to be less disadvantaged than women without sleep apnea. There was a tendency for more women with sleep apnea to be primiparous. There was no statistical difference between groups on smoking during pregnancy. Infants whose mothers had sleep apnea were more likely to be preterm. More infants exposed to maternal sleep apnea had a NICU/SCN admission and more were large for gestational age.
Table 2 shows the characteristics of the 115025 children with developmental data. The maternal sleep apnea groups did not differ significantly on speaking a language other than English at home or age at developmental assessment, but there appeared to be proportionally more children in the maternal sleep apnea group assessed in 2012 than for the no-apnea group.
Table 3 describes the 172345 children with educational data and indicates that more of those exposed to maternal sleep apnea sat the tests at a younger age (7 years or younger) and fewer were from a non-English speaking background. Parental education and occupation did not differ significantly between groups, although those with maternal apnea group tended to have higher rates of employment, including more management and professional roles. Year of testing did not differ significantly between groups.
Table 4 shows the relative frequency of the outcomes in the children exposed to maternal sleep apnea and those unexposed, whereas Table 5 shows the results of crude and adjusted models for each of the outcomes.
Outcome . | Denominator N . | No apnea n (col%) . | Maternal sleep apnea n (col%) . | Difference between groups Test statistic, p . |
---|---|---|---|---|
Any mortality up to 6 years of age | 626188 | 6474 (1.03) | X (1.44) | Fisher’s exact, p = .48 |
Any hospitalization before first birthday | 626188 | 102683 (16.4) | 58 (27.8) | χ21 = 19.6, p < .01 |
Excluding infants admitted for potential sleep apnea | 623438 | 99939 (16.0) | 52 (26.0) | χ21 = 13.8, p < .01 |
Any hospitalization between first and sixth birthdays | 620101 | 184401 (29.8) | 81 (39.3) | χ21 = 9.03, p < .01 |
Excluding children admitted for potential sleep apnea | 586470 | 150798 (25.7) | 53 (29.8) | χ21 = 1.53, p = .22 |
Developmentally vulnerable | 115025* | 21915 (19.1) | 9 (23.1) | χ21 = 2.48, p = .29 |
Special needs status | 115025* | 5329 (4.6) | X (7.7) | χ21 = 0.82, p = .36 |
Low reading test score | 172345* | 33574 (19.5) | 11 (22.5) | χ21 = 0.27, p = .60 |
Low numeracy test score | 172345* | 35799 (20.8) | 8 (16.3) | χ21 = 0.59, p = .44 |
Outcome . | Denominator N . | No apnea n (col%) . | Maternal sleep apnea n (col%) . | Difference between groups Test statistic, p . |
---|---|---|---|---|
Any mortality up to 6 years of age | 626188 | 6474 (1.03) | X (1.44) | Fisher’s exact, p = .48 |
Any hospitalization before first birthday | 626188 | 102683 (16.4) | 58 (27.8) | χ21 = 19.6, p < .01 |
Excluding infants admitted for potential sleep apnea | 623438 | 99939 (16.0) | 52 (26.0) | χ21 = 13.8, p < .01 |
Any hospitalization between first and sixth birthdays | 620101 | 184401 (29.8) | 81 (39.3) | χ21 = 9.03, p < .01 |
Excluding children admitted for potential sleep apnea | 586470 | 150798 (25.7) | 53 (29.8) | χ21 = 1.53, p = .22 |
Developmentally vulnerable | 115025* | 21915 (19.1) | 9 (23.1) | χ21 = 2.48, p = .29 |
Special needs status | 115025* | 5329 (4.6) | X (7.7) | χ21 = 0.82, p = .36 |
Low reading test score | 172345* | 33574 (19.5) | 11 (22.5) | χ21 = 0.27, p = .60 |
Low numeracy test score | 172345* | 35799 (20.8) | 8 (16.3) | χ21 = 0.59, p = .44 |
*Number of children with linked developmental or educational data. X indicates cell size < 5 and is censored due to potentially identifiable individuals.
Outcome . | Denominator N . | No apnea n (col%) . | Maternal sleep apnea n (col%) . | Difference between groups Test statistic, p . |
---|---|---|---|---|
Any mortality up to 6 years of age | 626188 | 6474 (1.03) | X (1.44) | Fisher’s exact, p = .48 |
Any hospitalization before first birthday | 626188 | 102683 (16.4) | 58 (27.8) | χ21 = 19.6, p < .01 |
Excluding infants admitted for potential sleep apnea | 623438 | 99939 (16.0) | 52 (26.0) | χ21 = 13.8, p < .01 |
Any hospitalization between first and sixth birthdays | 620101 | 184401 (29.8) | 81 (39.3) | χ21 = 9.03, p < .01 |
Excluding children admitted for potential sleep apnea | 586470 | 150798 (25.7) | 53 (29.8) | χ21 = 1.53, p = .22 |
Developmentally vulnerable | 115025* | 21915 (19.1) | 9 (23.1) | χ21 = 2.48, p = .29 |
Special needs status | 115025* | 5329 (4.6) | X (7.7) | χ21 = 0.82, p = .36 |
Low reading test score | 172345* | 33574 (19.5) | 11 (22.5) | χ21 = 0.27, p = .60 |
Low numeracy test score | 172345* | 35799 (20.8) | 8 (16.3) | χ21 = 0.59, p = .44 |
Outcome . | Denominator N . | No apnea n (col%) . | Maternal sleep apnea n (col%) . | Difference between groups Test statistic, p . |
---|---|---|---|---|
Any mortality up to 6 years of age | 626188 | 6474 (1.03) | X (1.44) | Fisher’s exact, p = .48 |
Any hospitalization before first birthday | 626188 | 102683 (16.4) | 58 (27.8) | χ21 = 19.6, p < .01 |
Excluding infants admitted for potential sleep apnea | 623438 | 99939 (16.0) | 52 (26.0) | χ21 = 13.8, p < .01 |
Any hospitalization between first and sixth birthdays | 620101 | 184401 (29.8) | 81 (39.3) | χ21 = 9.03, p < .01 |
Excluding children admitted for potential sleep apnea | 586470 | 150798 (25.7) | 53 (29.8) | χ21 = 1.53, p = .22 |
Developmentally vulnerable | 115025* | 21915 (19.1) | 9 (23.1) | χ21 = 2.48, p = .29 |
Special needs status | 115025* | 5329 (4.6) | X (7.7) | χ21 = 0.82, p = .36 |
Low reading test score | 172345* | 33574 (19.5) | 11 (22.5) | χ21 = 0.27, p = .60 |
Low numeracy test score | 172345* | 35799 (20.8) | 8 (16.3) | χ21 = 0.59, p = .44 |
*Number of children with linked developmental or educational data. X indicates cell size < 5 and is censored due to potentially identifiable individuals.
Outcome . | Crude HR (95% CI) . | p . | HR adjusted for maternal factorsa (95% CI) . | p . |
---|---|---|---|---|
Any mortality up to 6 years of age | 1.39 (0.45–4.32) | .57 | — | — |
Any hospitalization before first birthday | 1.84 (1.42–2.38) | <.01 | 1.81 (1.40–2.34) | <.01 |
Excluding infants admitted for potential sleep apnea | 1.74 (1.33–2.28) | <.01 | 1.71 (1.31–2.24) | <.01 |
Any hospitalization between first and sixth birthdays | 1.43 (1.15–1.78) | <.01 | 1.41 (1.14–1.75) | <.01 |
Excluding children admitted for potential sleep apnea | 1.26 (0.99–1.60) | .06 | 1.25 (0.98–1.59) | .07 |
Outcome | Crude RR (95% CI) | RR adjusted for maternal factorsa and differences between groupsb,c (95% CI) | ||
Developmentally vulnerable | 1.20 (0.72–2.03) | .48 | 1.29 (0.75–2.21)b | .35 |
Special needs status | 1.63 (0.64–4.16) | .30 | 1.58 (0.61–4.07)b | .34 |
Low reading test score | 1.33 (0.96–1.85) | .09 | 1.55 (1.08–2.23)c | .02 |
Low numeracy test score | 0.93 (0.60–1.42) | .72 | 1.03 (0.63–1.67)c | .92 |
Outcome . | Crude HR (95% CI) . | p . | HR adjusted for maternal factorsa (95% CI) . | p . |
---|---|---|---|---|
Any mortality up to 6 years of age | 1.39 (0.45–4.32) | .57 | — | — |
Any hospitalization before first birthday | 1.84 (1.42–2.38) | <.01 | 1.81 (1.40–2.34) | <.01 |
Excluding infants admitted for potential sleep apnea | 1.74 (1.33–2.28) | <.01 | 1.71 (1.31–2.24) | <.01 |
Any hospitalization between first and sixth birthdays | 1.43 (1.15–1.78) | <.01 | 1.41 (1.14–1.75) | <.01 |
Excluding children admitted for potential sleep apnea | 1.26 (0.99–1.60) | .06 | 1.25 (0.98–1.59) | .07 |
Outcome | Crude RR (95% CI) | RR adjusted for maternal factorsa and differences between groupsb,c (95% CI) | ||
Developmentally vulnerable | 1.20 (0.72–2.03) | .48 | 1.29 (0.75–2.21)b | .35 |
Special needs status | 1.63 (0.64–4.16) | .30 | 1.58 (0.61–4.07)b | .34 |
Low reading test score | 1.33 (0.96–1.85) | .09 | 1.55 (1.08–2.23)c | .02 |
Low numeracy test score | 0.93 (0.60–1.42) | .72 | 1.03 (0.63–1.67)c | .92 |
All models compared outcomes for infants exposed to maternal sleep apnea relative to those not exposed (referent). Bold font indicates p < .05.
aMaternal factors were maternal age, country of birth, socioeconomic disadvantage, and parity.
bAll adjusted models for developmental outcomes included year of assessment as a covariate.
cAll adjusted models for educational outcomes included age at test, non-English speaking background, and parental occupation.
Outcome . | Crude HR (95% CI) . | p . | HR adjusted for maternal factorsa (95% CI) . | p . |
---|---|---|---|---|
Any mortality up to 6 years of age | 1.39 (0.45–4.32) | .57 | — | — |
Any hospitalization before first birthday | 1.84 (1.42–2.38) | <.01 | 1.81 (1.40–2.34) | <.01 |
Excluding infants admitted for potential sleep apnea | 1.74 (1.33–2.28) | <.01 | 1.71 (1.31–2.24) | <.01 |
Any hospitalization between first and sixth birthdays | 1.43 (1.15–1.78) | <.01 | 1.41 (1.14–1.75) | <.01 |
Excluding children admitted for potential sleep apnea | 1.26 (0.99–1.60) | .06 | 1.25 (0.98–1.59) | .07 |
Outcome | Crude RR (95% CI) | RR adjusted for maternal factorsa and differences between groupsb,c (95% CI) | ||
Developmentally vulnerable | 1.20 (0.72–2.03) | .48 | 1.29 (0.75–2.21)b | .35 |
Special needs status | 1.63 (0.64–4.16) | .30 | 1.58 (0.61–4.07)b | .34 |
Low reading test score | 1.33 (0.96–1.85) | .09 | 1.55 (1.08–2.23)c | .02 |
Low numeracy test score | 0.93 (0.60–1.42) | .72 | 1.03 (0.63–1.67)c | .92 |
Outcome . | Crude HR (95% CI) . | p . | HR adjusted for maternal factorsa (95% CI) . | p . |
---|---|---|---|---|
Any mortality up to 6 years of age | 1.39 (0.45–4.32) | .57 | — | — |
Any hospitalization before first birthday | 1.84 (1.42–2.38) | <.01 | 1.81 (1.40–2.34) | <.01 |
Excluding infants admitted for potential sleep apnea | 1.74 (1.33–2.28) | <.01 | 1.71 (1.31–2.24) | <.01 |
Any hospitalization between first and sixth birthdays | 1.43 (1.15–1.78) | <.01 | 1.41 (1.14–1.75) | <.01 |
Excluding children admitted for potential sleep apnea | 1.26 (0.99–1.60) | .06 | 1.25 (0.98–1.59) | .07 |
Outcome | Crude RR (95% CI) | RR adjusted for maternal factorsa and differences between groupsb,c (95% CI) | ||
Developmentally vulnerable | 1.20 (0.72–2.03) | .48 | 1.29 (0.75–2.21)b | .35 |
Special needs status | 1.63 (0.64–4.16) | .30 | 1.58 (0.61–4.07)b | .34 |
Low reading test score | 1.33 (0.96–1.85) | .09 | 1.55 (1.08–2.23)c | .02 |
Low numeracy test score | 0.93 (0.60–1.42) | .72 | 1.03 (0.63–1.67)c | .92 |
All models compared outcomes for infants exposed to maternal sleep apnea relative to those not exposed (referent). Bold font indicates p < .05.
aMaternal factors were maternal age, country of birth, socioeconomic disadvantage, and parity.
bAll adjusted models for developmental outcomes included year of assessment as a covariate.
cAll adjusted models for educational outcomes included age at test, non-English speaking background, and parental occupation.
Mortality
The rate of any death up to age 6 did not differ between groups (Table 4). No further analysis was undertaken due to the small numbers in the maternal sleep apnea group.
Hospitalizations
Infants of mothers exposed to maternal sleep apnea were significantly more likely to be admitted in hospital in the first year of life and to be admitted in hospital between first and sixth birthdays (Table 4). The associations remained after controlling for maternal factors that might be associated with healthcare seeking or access to health services (Table 5). Comparison of these results with models that also include preterm birth shows a very small reduction in the association, suggesting a limited role for mediation through preterm birth (Supplementary Table S4).
Pediatric Sleep Apnea as a Reason for Hospitalization (Post hoc Analysis)
Infants exposed to maternal sleep apnea were more likely to be admitted in hospital for potential pediatric sleep apnea. Of all infants hospitalized in the first year of life, 10.3% (n = 6) of those exposed to maternal sleep apnea were admitted for potential pediatric apnea, compared with 2.7% of the unexposed (X2(1) = 13.1; p < .0001). Of all children hospitalized between the first and sixth birthdays, 34.6% (n = 28) exposed to maternal sleep apnea during pregnancy was admitted for pediatric sleep apnea compared with 18.2% whose mothers did not have sleep apnea during pregnancy (X2(1) = 14.5; p < .0001).
After excluding infants admitted for pediatric sleep apnea, exposure to maternal sleep apnea remained significantly associated with hospitalizations in the first year of life. Excluding children admitted for apnea between the first and sixth birthdays attenuated the association between maternal apnea and childhood hospitalizations such that no significantly increased hazard was seen (Table 5).
Developmental Outcomes
Developmental vulnerability (RR 1.20; 95% CI 0.72–2.03) and special needs status (1.63; 0.64–4.16) did not differ in children exposed to maternal sleep apnea compared with those not exposed (Table 5). Adjustment for year of assessment and maternal factors did not appreciably change the results (Table 5).
Educational Outcomes
There was no significant difference between the children of mothers with sleep apnea compared with those without apnea on test scores for reading (RR 1.33; 95% CI 0.96–1.85) and numeracy (0.93; 0.60–1.42) (Table 5). Adjustment for maternal and social factors resulted in a significantly increased risk for low reading scores for children exposed to maternal sleep apnea (1.55; 1.08–2.23) but did not appreciably change the result for numeracy (1.03; 0.63–1.67). Comparison with models which included preterm birth, and separately, infant and childhood hospitalizations, showed that the associations were not considerably attenuated, suggesting that there was limited mediation of the observed associations by preterm birth or history of hospitalizations (Supplementary Tables S4 and S5).
DISCUSSION
This is the first population-based study to examine long-term childhood outcomes associated with exposure to maternal sleep apnea in pregnancy. The findings suggest that exposure to maternal sleep apnea confers risk for infant and childhood hospitalizations, although it does not appear to consistently increase the risk of poor development or educational achievement.
The risk of infant and childhood hospitalizations associated with maternal sleep apnea is partly attributable to higher rates of admissions for potential pediatric sleep apnea. For hospitalizations between ages one and six, we found no significant association between maternal apnea and childhood hospitalizations after excluding those children admitted for potential pediatric apnea. This is consistent with the view that genetics are responsible for about 40% of the predisposition to sleep apnea,21 and similarities in craniofacial structure, body fat distribution, and neural control of the airways between mother and child are likely to be responsible for the correlation seen between maternal and childhood apnea. A previous study of 74 women found that four times as many women who had sleep-disordered breathing identified in the second trimester subsequently reported snoring in their infants compared with women free from apnea (41.7% vs. 7.5%).22 Increased vigilance, awareness, and healthcare seeking for pediatric sleep apnea after a personal experience of sleep apnea by the mother may also be responsible for the higher rates of pediatric apnea admissions in this group, although childhood admissions overall do not appear to be driven by more healthcare access in those women with sleep apnea in the present study.
Although childhood hospitalizations associated with maternal sleep apnea seem mostly due to admissions linked to pediatric sleep apnea, infant hospitalizations remained elevated after the exclusion of infants admitted for sleep-disordered breathing. The reasons for this are unclear. Maternal sleep apnea is thought to affect pregnancy and fetal and infant health through the systemic inflammation and oxidative stress associated with the chronic intermittent hypoxia produced by apnoeic episodes.23,24 Experimental studies in animal models have shown that chronic exposure to intermittent hypoxia leads to intrauterine growth restriction and smaller offspring.25 We did not find an association between maternal sleep apnea during pregnancy and small-for-gestational age infants in the current cohort,6 suggesting that growth restriction is not responsible for the increase in observed hospitalizations. Control for socioeconomic factors known to be associated with healthcare-seeking and healthcare access in women with sleep apnea did not significantly reduce the association between maternal apnea and hospitalizations in the first year of life. As a simple test of mediation, we examined the impact of including preterm birth in our models as preterm birth increases risk for hospitalizations in infancy and is also elevated in maternal sleep apnea.1,2 However, this only slightly attenuated the relationship with a 60% increased hazard remaining for infant hospitalizations, suggesting that maternal sleep apnea confers risk for infant health independently of preterm birth.
It is possible that an adverse impact of prenatal exposure to maternal sleep apnea may not be readily apparent at birth but may nonetheless contribute to vulnerability to disease and the risk of hospitalizations, especially in infancy when the immune system is still in development. A recent study found that telomere lengths are shorter in term infants of women deemed high-risk for sleep apnea compared with those whose mothers were considered low-risk.26 Short telomere length is a marker of cellular aging which in early life is thought to reflect susceptibility to later disease.27 We lacked the numbers in the present study to examine whether certain conditions such as infections were particularly increased in infants exposed to maternal apnea, and future studies will need to test whether this is the case.
We found that maternal apnea was not consistently linked to adverse developmental or educational outcomes. Risk for low reading scores was significantly elevated in those exposed to maternal sleep apnea although we can think of no reason why there might be a differential impact of maternal apnea on reading compared with numeracy or general development. We cannot discount the possibility of chance findings, especially with the number of statistical comparisons made in this study; however, neither can we discount a small detrimental effect of maternal sleep apnea since the risk estimates for three of four outcomes are nonsignificantly elevated. Consideration of preterm birth and a history of hospitalizations did not considerably change the risk estimates, suggesting a limited role for these mediating factors (Supplementary Tables S5 and S6).
We also considered a differential impact of maternal sleep apnea on male and female children, given known sex differences in development. Post hoc analyses stratifying the results by the sex of the child showed that maternal sleep apnea was associated with an increased risk of being developmentally vulnerable in the first year of school for boys, but not for girls (Supplementary Table S7). Stratification of the other developmental and educational models did not indicate a statistically significant impact of maternal sleep apnea that depended on the sex of the child. For low reading scores which were significantly increased with maternal sleep apnea, the sex-stratified results show a similar risk increase in both girls and boys (of roughly 40%) (Supplementary Table S7). These results are however based on small numbers of children with the outcomes; they should be interpreted with caution and the findings will require replication in future studies.
One previous study, of 74 women with uncomplicated pregnancies and their term infants, has reported no association between maternal sleep-disordered breathing (apnea-hypopnea index > 5) and infant neuro-motor development in the first six months of life or general development at 1 year of age.22 However, that study did find poorer parent-reported social development in infants exposed to maternal sleep-disordered breathing, possibly due to increased infant sleep-disordered breathing in this group.22 Taken together with the findings of the present study, the long-term impact of maternal sleep apnea appears specific to child health, with any impact on neurocognitive development likely mediated by the increased risk for sleep-disordered breathing in the children rather than as a direct consequence of prenatal exposure to maternal apnea.
Strengths of the present study include its population coverage and the generalizability of results to the pregnant population at large. We were also able to examine long-term outcomes associated with maternal sleep apnea beyond the first year of life and well into the school ages or middle childhood. The main limitation of the current study is the ascertainment of maternal sleep apnea from hospital records. Although the hospital records have high specificity for medical conditions, meaning that the maternal apnea group contains cases of clinically significant apnea, we assumed that women without a diagnosis code in hospital records did not have apnea. Women could have had undiagnosed apnea or apnea that went unrecorded because they were not admitted in hospital prior to or during pregnancy. Further, sleep apnea identified this way is likely to have been treated. We did not have information on the severity of the apnea or any information on treatment. Treatment would reduce the disparity between the infants exposed to maternal apnea and those who were not. Both of these factors are likely to have biased the results conservatively toward null findings.
We did not have data on pre-pregnancy body mass index (BMI) and therefore cannot rule out residual confounding due to maternal obesity. Obesity is known to predispose to maternal sleep apnea, and there is emerging research on the negative impact of maternal obesity and childhood health.28 Obesity may have an independent impact on childhood health, it may affect childhood health through predisposing to incident maternal sleep apnea, and it could be associated with childhood health through the shared parent–child socioeconomic environment. We have controlled for socioeconomic variables in our modeling, which should reduce the confounding impact of obesity through this last pathway. Future studies need to explore the independent contributions of maternal obesity and maternal sleep apnea by including data on pre-pregnancy BMI or by studying the subgroup of nonobese women with sleep apnea.
Loss to follow-up may have affected the results for the developmental and educational outcomes. Since we made use of population datasets, relocation interstate or overseas would be the main reason for such losses; however, as relocation is unlikely to be systematically associated with maternal sleep apnea, the results are unlikely to be biased. Lastly, the post hoc analyses and developmental and educational outcomes were based on small numbers of children exposed to maternal sleep apnea, so the results need confirmation by future studies. Primary cohort studies involving the clinical measurement of maternal apnea, its severity, and the level of treatment adherence will be useful for extending the current findings and elucidating the mechanisms underlying the link between maternal apnea and infant health.
To conclude, prenatal exposure to maternal sleep apnea seems to confer risk for infant and childhood hospitalizations. This appears partly driven by increased risk for pediatric sleep apnea; the reasons for increased infant hospitalizations are unknown. Exposure to maternal apnea does not significantly hinder children’s development and this is reassuring for women with sleep apnea in pregnancy.
SUPPLEMENTARY MATERIAL
Supplementary material is available at SLEEP online.
FUNDING
This work was supported by an Australian National Health and Medical Research Council (NHMRC) Centre for Research Excellence grant (1001066) and a New South Wales Ministry of Health Population Health and Health Services Research Support Program Grant. YSB was supported by the 2016 Rob Pierce Grant-in-Aid from the Australasian Sleep Association. CLR is supported by an NHMRC Senior Research Fellowship (1021025). JBF was supported by an Australian Research Council Future Fellowship (FT120100069). The funding sources had no involvement in the study design; collection, analysis, and interpretation of the data; or the decision to submit this paper for publication.
DISCLOSURE STATEMENT
None declared.
ACKNOWLEDGMENTS
This study uses unit record data from population health data collections, the Australian Early Development Census (AEDC) and the National Assessment Program—Literacy and Numeracy (NAPLAN). We thank the Ministry of Health, the NSW Department of Education and Communities and the Australian Government Department of Education for provision of population data, and the NSW Centre for Health Record Linkage for record linkage. The findings and views reported in this article are those of the authors and should not be attributed to any of these departments.
REFERENCES
Comments