ABSTRACT
Objective
To determine if routine administration of antenatal corticosteroids affects the risk of infant lower respiratory tract infection and/or childhood asthma.
Design
Linked population‐based cohort analysed using a regression discontinuity design, which better controls for confounding than standard observational studies.
Setting
British Columbia, Canada.
Population
Singleton pregnancies with a maternal admission for delivery between 31 + 0 and 36 + 6 weeks' gestation from 2000 to 2016, with follow‐up to 2020.
Methods
We estimated if risks of childhood respiratory outcomes differed between pregnancies admitted just before the Canadian recommended clinical cut‐off for antenatal corticosteroid administration of 34 + 0 weeks gestation (i.e., with higher probability of exposure to antenatal corticosteroids; ‘exposed’) than those admitted just after this cut‐off (i.e., with lower probability of exposure; ‘unexposed’) using log binomial regression (infant lower respiratory infection hospitalisation) and pooled log binomial regression (asthma).
Main Outcome Measures
Infant lower respiratory tract infection hospitalisation, inpatient or outpatient asthma diagnosis at 1–18 years.
Results
In our cohort of 21 965 children, 412 (1.9%) infants were hospitalised with a lower respiratory tract infection and 2287 (10.4%) were diagnosed with asthma. Routine administration of antenatal corticosteroids was not associated with infant lower respiratory tract infection (risk ratio = 0.95 [95% CI: 0.61, 1.37], risk difference = −0.15 excess cases per 100 [95% CI: −1.30, 0.99]) or childhood asthma (rate ratio = 1.08 [95% CI: 0.88, 1.24] 5.49 excess cases per 100 by age 13 years [95% CI: −1.78, 14.39]).
Conclusions
We found no evidence that routine administration of antenatal corticosteroids affects the risk of later childhood respiratory illnesses.
Keywords: antenatal corticosteroids, asthma, lower respiratory tract infection, preterm birth, respiratory morbidity
Short abstract
Linked article: This article is commented on by Vidaeff, pp. 272–273 in this issue. To view this article visit https://doi.org/10.1111/1471‐0528.70011.
1. Introduction
Administration of antenatal corticosteroids at late preterm gestation (34–36 weeks) is controversial [1, 2, 3, 4, 5]. Although randomised trials have shown that antenatal corticosteroid administration before an imminent preterm delivery at late preterm gestation reduces the risk of respiratory morbidity by 20%–30% [6, 7], the absolute risk of respiratory morbidity is much lower in this gestational age group than at younger ages [8]. The respiratory morbidity prevented by antenatal corticosteroids at late preterm ages tends to be milder in nature, such as transient tachypnea of the newborn, and can usually be successfully treated through neonatal care [9]. As a result, it is unclear whether the neonatal respiratory benefits of routine antenatal corticosteroid administration at late preterm ages outweigh the medication's known increased risk of hypoglycemia [6] and unclear effects on longer‐term child neurodevelopment [7, 10, 11, 12, 13].
Yet, milder neonatal respiratory morbidities may have longer‐term consequences for child health. Multiple studies have linked transient tachypnea of the newborn with an increased risk of respiratory morbidity later in childhood, such as asthma and lower respiratory tract infections [14, 15, 16]. Prevention of these longer‐term conditions through a pathway of reduced neonatal respiratory morbidity would be an important argument for routine administration of antenatal corticosteroids at late preterm ages. Follow‐up studies of neonatal randomised trials of antenatal corticosteroids have found few differences in later respiratory outcomes in children who were vs. were not exposed to antenatal corticosteroids in utero, but high losses to follow‐up and small sample sizes have made it challenging to draw convincing conclusions from these studies [17, 18, 19].
Observational studies have variously reported that antenatal corticosteroids are linked with increased [20, 21] and decreased [22] risks of childhood respiratory illnesses. However, these studies are at high risk of confounding by indication because many of the clinical complications that lead to antenatal corticosteroid administration (such as hypertensive disorders of pregnancy and chorioamnionitis) are also linked with increased risks of childhood wheezing/asthma [23, 24, 25, 26]. Further, indications such as chorioamnionitis and preterm premature rupture of membranes lead to increased antibiotic use, which has been linked with increased risks of childhood respiratory morbidity [27]. As clinical risk profiles for these complications are challenging to measure accurately, standard observational approaches cannot convincingly disentangle the effects of antenatal corticosteroids on child wheezing disorders from the effects of the underlying complication that lead to the treatment [28].
The objective of this study was to determine whether routine antenatal corticosteroid administration affects the incidence of infant hospitalisation for lower respiratory tract infection or childhood asthma, using a novel design less prone to confounding than standard observational studies.
2. Methods
2.1. Study Design
We estimated the effect of routine antenatal corticosteroid administration on child respiratory health using a quasi‐experimental design known as regression discontinuity [29, 30, 31]. The intuition behind the design is that it takes advantage of the fact that clinical cut‐offs are often used to decide which individuals will get treated with a given medication, creating a situation where individuals just above a clinical cut‐off will have high rates of exposure to the medication, while individuals just below the cut‐off will not. However, as individuals immediately on either side of the cut‐off are still similar in their clinical risk profiles, comparing individuals just above vs. just below the cut‐off allows the effect of the medication on health outcomes to be examined, with groups that are comparable in most ways except for their rates of medication exposure.
In the case of antenatal corticosteroids, Canadian clinical practice guidelines recommend antenatal corticosteroid administration for all pregnancies at risk of an imminent preterm birth presenting up to 33 + 6 weeks [32]. This creates a discontinuity in exposure to antenatal corticosteroids, with high rates of exposure to the medication before the clinical cut‐off and low rates after it. Because the prevalence of clinical complications that lead to antenatal corticosteroid administration is similar just before vs. just after the cut‐off, the regression discontinuity design provides a strategy to isolate the effects of antenatal corticosteroid exposure on child respiratory health from the effects of the underlying clinical complications that lead to its administration.
2.2. Study Population
Our study population was drawn from all liveborn singleton pregnancies in British Columbia (BC), Canada, with a maternal admission for delivery between 31 + 0 and 36 + 6 weeks, from April 1, 2000 to March 31, 2016. Late pregnancy terminations and pregnancies with no ultrasound‐confirmed estimate of gestational age were excluded. Obstetrical and neonatal records were obtained from the BC Perinatal Data Registry, which contains abstracted obstetrical and neonatal medical records for > 99% of deliveries in the province [33, 34]. We linked the cohort with population‐based hospital discharge data, physician billing records for outpatient and inpatient services to BC's universal health insurance program and vital statistics death records, from April 1, 2000, to March 31, 2020, providing a minimum of 4 years' follow‐up.
2.3. Study Measurements
Gestational age in days was estimated using last menstrual period confirmed or revised with early ultrasound (< 20 weeks' gestation), based on the algorithm in use during the study period [35]. Because our perinatal data registry did not begin abstracting gestational age in days by ultrasound before April 1, 2008 (it abstracted completed weeks), we calculated day‐specific estimates of gestational age for deliveries before this time using the last menstrual period estimate among women whose ultrasound‐based estimate of gestational age (in weeks) was within 1 week of their estimate based on last menstrual period (in weeks).
We calculated gestational age in days at maternal admission as the estimated gestational age at delivery minus the interval between admission and delivery. We used gestational age at the time of maternal admission for the delivery (rather than delivery) as our main exposure variable, because this best approximates the timing of decision‐making on antenatal corticosteroid administration. As not all pregnancies before 34 + 0 weeks receive antenatal corticosteroids, our estimates are analogous to an effectiveness study of the impact of clinical practice guidelines recommending routine administration of antenatal corticosteroids, rather than an efficacy study of the pharmacological effect of antenatal corticosteroids under optimal administration conditions. We compared the distribution of potential confounders among women at risk of delivery immediately before and after 34 weeks and ensured there were no abrupt changes in these factors coinciding with the change in antenatal corticosteroid administration patterns.
2.4. Childhood Respiratory Health Outcomes
We defined infant hospitalisation for respiratory tract infection as a hospital admission within 1 year of birth with an International Classification of Diseases (ICD) code for any of: acute bronchitis (ICD 9 code 466.0 or ICD 10 code J20), acute bronchiolitis (466.1 or J21), pneumonia (480–486 or J12‐J18), influenza (487, 488 or J09‐J11) or unspecified acute lower respiratory tract infection (519.8 or J22) in any position. We examined codes more specific to respiratory syncytial virus (RSV; respiratory syncytial virus bronchiolitis only [466.11 or J21.0] and acute bronchiolitis only [466.0 or J20/466.1 or J21]) as secondary outcomes only, as they are less likely to be accurate in administrative data than the broader outcome of ‘any lower respiratory tract infection’ [36, 37, 38]. Asthma was defined as a hospitalisation with a diagnostic code for asthma (ICD 9 code 493 or ICD 10 code J45) in the principal position or 2 physician visits with an asthma diagnosis, between ages 1 and 18 years [39]. We additionally examined the outcome of transient tachypnea of the newborn (ICD 9 code 770.6 or ICD 10 code P22.1) to establish the effectiveness of antenatal corticosteroids in preventing this neonatal respiratory morbidity in our cohort. A core outcome set was not used.
Infant or child deaths (n = 185) were assigned the occurrence of each adverse outcome. This was done to prevent selection bias introduced from differential losses to follow up according to antenatal corticosteroid administration status [40]. For infant lower respiratory tract infection hospitalisations, this was only done for deaths up to 1 year after birth (n = 145); for transient tachypnea of the newborn, this was done only for in‐hospital newborn deaths (n = 67).
2.5. Statistical Analysis
We implemented the regression discontinuity design [29, 30, 31] using multivariable log‐binomial regression (infant lower respiratory tract infection hospitalisation) and pooled logistic regression (asthma). Our models included a linear term for gestational age (which captures the decrease in risk of adverse outcomes with advancing gestational age), a second term to allow the gestational age slope to differ after 34 + 0 weeks and our primary coefficient of interest, an indicator of whether there was a level change or ‘jump’ in risk at the point where exposure to antenatal corticosteroids changed (i.e., the discontinuity in antenatal corticosteroid administration rates at 34 + 0 weeks) in the smoothed probabilities of risk. We used triangular weighting to upweight observations closest to the point of the discontinuity, such that the gestational days just before and just after 34 + 0 weeks were more influential in model estimates. We calculated 95% confidence intervals using bootstrap resampling. We used model marginal estimates to calculate absolute risk/rate differences. Conclusions on the presence or absence of an association were based on the clinical importance of the model point estimates and the width of associated 95% confidence intervals.
2.6. Sensitivity Analyses
As recommended, we evaluated the sensitivity of our results to different gestational age windows from the cut‐off [31]. We modified our definition of childhood asthma by limiting it to hospital‐based diagnoses and limiting it to codes > 5 years of age, given the challenges in diagnosing asthma at younger ages.
We repeated analyses after limiting our cohort to women with no prior antenatal hospitalisations before their delivery admission. This was done to assess the extent to which our results could be explained by misclassification of exposure status among pregnancies admitted after 34 + 0 weeks (classified in our design as ‘unexposed’) who had in fact been exposed to antenatal corticosteroids during an earlier admission before 34 + 0 weeks. We also repeated analyses limiting to the years post‐2008.
3. Results
3.1. Study Population
There were 656 553 singleton live births in British Columbia between April 1, 2000, and March 31, 2016. Restricting to births with a day‐specific, ultrasound‐confirmed estimate of gestational age and a maternal delivery admission between 31 + 0 and 36 + 6 weeks left 22 298 eligible births for analysis (Figure S1). We further excluded 333 children with missing or incomplete childhood health records, producing a final sample size of 21 965.
The characteristics of pregnancies admitted for delivery before 34 + 0 weeks were similar to those admitted after 34 + 0 weeks in terms of maternal age, smoking status, pre‐pregnancy BMI, parity and neighbourhood income quintile distribution (Table 1). Pregnancies admitted for delivery before 34 + 0 weeks were more likely to have a hypertensive disorder of pregnancy (19.3 vs. 15.0%), diabetes in pregnancy (19.8 vs. 17.6%) and have had a prior antenatal hospital admission (36.4 vs. 20.7%). However, there were no abrupt differences in these characteristics at 34 + 0 weeks (a key assumption of the regression discontinuity design; Figure S2). The proportion of pregnancies exposed to antenatal corticosteroids decreased markedly after the clinical cut‐point of 34 + 0 weeks (Figure 1).
TABLE 1.
Maternal and fetal characteristics of pregnancies admitted for delivery between 31 + 0‐ and 36 + 6 weeks' gestation, British Columbia, Canada, 2000–2016.
| Characteristic | Total cohort mean (±SD) or n (%) | Maternal admission before 34 + 0 weeks mean (±SD) or n (%) | Maternal admission at or after 34 + 0 weeks mean (±SD) or n (%) |
|---|---|---|---|
| N | 21 965 | 3520 | 18 445 |
| Maternal age (years) | 31.3 ± 5.6 | 31.6 ± 5.7 | 31.3 ± 5.6 |
| Nulliparous | 10 804 (49.2) | 1814 (51.5) | 8990 (48.7) |
| Pre‐pregnancy BMI (kg/m2)a | 24.6 ± 5.6 | 24.6 ± 5.7 | 24.6 ± 5.6 |
| Smoking in pregnancy | 2408 (11.0) | 376 (10.7) | 2032 (11.0) |
| Neighbourhood income quintile | |||
| 1 | 4876 (22.2) | 766 (21.8) | 4110 (22.3) |
| 2 | 4660 (21.2) | 779 (22.1) | 3881 (21.0) |
| 3 | 4407 (20.1) | 689 (19.6) | 3718 (20.2) |
| 4 | 4213 (19.2) | 655 (18.6) | 3558 (19.3) |
| 5 | 3215 (14.6) | 520 (14.8) | 2695 (14.6) |
| Missing | 594 (2.7) | 111 (3.2) | 483 (2.6) |
| Hypertensive disorder of pregnancy | 3442 (15.7) | 681 (19.3) | 2761 (15.0) |
| Diabetes in pregnancy | 3951 (18.0) | 696 (19.8) | 3255 (17.6) |
| Iatrogenic onset of deliveryb | 9919 (45.2) | 1565 (44.5) | 8354 (45.3) |
| Prior antenatal hospital admission | 5091 (23.2) | 1280 (36.4) | 3811 (20.7) |
| Gestational age at delivery (weeks) | |||
| 31 | 484 (2.2) | 484 (13.8) | |
| 32 | 865 (3.9) | 865 (24.6) | |
| 33 | 1471 (6.7) | 1471 (41.8) | |
| 34 | 2642 (12.0) | 479 (13.6) | 2163 (11.7) |
| 35 | 4724 (21.5) | 106 (3.0) | 4618 (25.0) |
| 36 | 10 421 (47.4) | 62 (1.8) | 10 359 (56.2) |
| 37 | 1209 (5.5) | 30 (0.9) | 1179 (6.4) |
| 38 | 90 (0.4) | 10 (0.3) | 80 (0.4) |
| 39+ | 59 (0.3) | 13 (0.4) | 46 (0.2) |
| Admission‐to‐delivery interval, hours | 34.8 (±93.1) | 88.5 (±170.6) | 24.5 (±64.1) |
| Male fetus | 12 371 (56.3) | 2051 (58.3) | 10 320 (56.0) |
| Birthweight (g) | 2676.3 ± 534.2 | 2133.6 ± 513.0 | 2779.8 ± 472.0 |
Among n = 16 331 with available BMI data (2637 before 34 weeks, 13 694 after 34 weeks).
Defined as delivery following pre‐labour caesarean or induction of labour.
FIGURE 1.
Proportion of ACS administration by day of gestation at maternal admission in a cohort of 21 965 pregnancies admitted for delivery between 31 + 0‐ and 36 + 6‐weeks' gestation in British Columbia, Canada, 2000–2016.
3.2. Transient Tachypnea of the Newborn
There were 2328 (10.6%) infants with transient tachypnea of the newborn. As shown in Figure S3, there was a pronounced change in the risk of transient tachypnea of the newborn at 34 + 0 weeks, with our regression discontinuity model estimating that infants born just before 34 + 0 weeks (in whom antenatal corticosteroids were routinely administered) had a 25% [95% CI: 10% to 37%] lower risk of this outcome compared with infants born just after 34 + 0 weeks (who had minimal exposure to antenatal steroids), corresponding to an absolute risk difference of 4.6 fewer cases per 100 births [95% CI: 1.4 to 5.5 fewer cases].
3.3. Infant Hospitalisation With Lower Respiratory Tract Infection
There were 412 (1.9%) infants hospitalised with a lower respiratory tract infection, including 330 (1.5%) with RSV bronchiolitis. The majority of these hospitalisations occurred during the first 6 months after birth, with approximately 20% occurring in the second month after birth (Figure S4a). Figure S5a shows descriptive patterns of infant hospitalisation with a lower respiratory tract infection by week of maternal admission for delivery. Although risks decreased with advancing gestational age at maternal admission, there was no evidence of an abrupt change in risk at the gestational age associated with a change of antenatal corticosteroid administration practices (34 weeks). When formalised in the regression discontinuity model, we similarly found no difference in risks among infants admitted just before vs. just after 34 + 0 weeks (estimate risk ratio for births with a maternal admission just before 34 + 0 of 1.0 [95% CI: 0.6 to 1.4], Table 2). There was likewise no difference in risks of our secondary outcomes of RSV bronchiolitis infant hospitalisations (risk ratio 1.03 [95% CI: 0.63 to 1.70]; risk difference 0.06 excess cases per 100 births [95% CI: −0.73 to 1.05]) or bronchiolitis infant hospitalisations (risk ratio 0.94 [95% CI: 0.60 to 1.50]; risk difference − 0.16 excess cases per 100 births [95% CI: −1.22 to 1.17].
TABLE 2.
Routine antenatal corticosteroid administration and risk of childhood respiratory illnesses in a cohort of 21 965 pregnancies admitted for delivery between 31 + 0‐ and 36 + 6‐weeks' gestation in British Columbia, Canada, 2000–2016.
| Outcome | Estimated risk of respiratory illness in infants born just before (vs. just after) a maternal delivery admission at 34 + 0 weeks' gestation | |
|---|---|---|
| Risk or rate ratio [95% CI] | Excess cases per 100 births a | |
| Infant hospitalisation for lower respiratory tract infection | 0.95 (0.61, 1.37) | −0.15 (−1.30, 0.99) |
| Asthma diagnosis | 1.08 (0.88, 1.24) | 5.49 (−1.78, 14.39) |
Absolute difference in cumulative risk by age 13 for asthma outcomes.
3.4. Childhood Asthma
Our cohort had 2287 (10%) children diagnosed with asthma, of whom more than half were first diagnosed before age 3 (Figure S4b) and most based on physician billing records (Table S2). Risk of asthma decreased with advancing gestational age at maternal admission (Figure S5b), but we saw no evidence of an abrupt change in risk at 34 weeks. Our regression discontinuity model estimated a risk ratio of 1.08 for children with a maternal admission just before vs. just after 34 + 0 weeks (95% CI: 0.88 to 1.24, Table 2).
3.5. Sensitivity Analyses
Conclusions were unchanged when restricting analyses to the subgroup of women with no prior antenatal hospitalisations, limiting to admissions post‐2008, modifying our definition of asthma or when the gestational age window included in the models was narrowed for infant lower respiratory tract infection hospitalisation models (Table S1). When gestational age windows were narrowed for the outcome of childhood asthma, estimates suggested an increased risk of asthma associated with admission just before 34 + 0 weeks, which were not compatible with a null effect for the 7‐day window (Table S1). However, post hoc analyses suggested this was likely due to the influence of an outlier at 33 + 1 weeks' (seen in Figure 2) that caused the model to estimate a rising rate of asthma with advancing gestational age during the period before 34 + 0 weeks (Figure S6), which is biologically unlikely.
FIGURE 2.
Crude day‐specific risks (open circles) for (a) lower respiratory tract infection infant hospitalisation and (b) childhood asthma, with regression discontinuity model estimates (solid line) and 95% confidence intervals (shaded areas) superimposed in a cohort of 21 965 pregnancies admitted for delivery between 31 + 0‐ and 36 + 6‐weeks' gestation in British Columbia, Canada, 2000–2016.
4. Discussion
4.1. Summary
In this large, population‐based study, we found no evidence that routine administration of antenatal corticosteroids was linked with asthma or infant hospitalisation with lower respiratory tract infection. This suggests that the respiratory benefits of routine antenatal corticosteroid administration may be limited to the neonatal period and provides some evidence that the previously observed association between mild respiratory morbidity at birth and later childhood respiratory morbidity [14, 15, 16] may not be causal.
4.2. Comparison With the Literature
Follow‐up studies of the original randomised trials of antenatal corticosteroids on neonatal respiratory morbidity have examined participants' longer‐term respiratory health [17, 18, 19, 41]. Although these follow‐up studies generally suggest that antenatal corticosteroids do not reduce the risk of longer‐term respiratory morbidity, the sample sizes of most of the follow‐up studies have been small (n < 500), limiting their power to detect differences between groups. Further, high losses to follow‐up make it challenging to rule out selection bias. For example, follow‐up of the original Auckland Steroid Trial participants found no differences in spirometry‐assessed lung function at age 30 or self‐reported asthma at age 50 [17, 42]. However, the cohorts of n = 534 and n = 424, respectively, represented only 56% and 44% of the original surviving trial participants. A follow‐up of the ASTECS‐2 trial of antenatal corticosteroids prior to elective term caesarean delivery found no differences in parent‐reported wheezing or asthma at ages 8–15 years [19]. The study found weak evidence that there was a higher risk of childhood hospitalisation for respiratory infection among infants who were not exposed to antenatal corticosteroids (11% controls vs. 7% betamethasone, risk ratio = 1.5; 95% CI: 0.8–2.9). The sample size of n = 407 represented only 47% of eligible participants, with participation rates differing by treatment arm and neonatal respiratory morbidity status. In a follow‐up of 447 (69% of survivors) from the 1976 National Heart, Lung and Blood Institute Collaborative trial, parents with infants in the control arm reported more respiratory complications (including wheezing or signs of acute respiratory tract infection) than those of infants exposed to antenatal corticosteroids (3.5% vs. 0.5%, p = 0.02) at age 18 months, but these differences were no longer present at the 36‐month visit [18]. Our findings complement this randomised trial evidence by demonstrating little evidence of an effect of routine administration of antenatal corticosteroids on later childhood respiratory illness in a large, population‐based cohort with minimal losses to follow‐up and a quasi‐experimental design that supports causal inference.
Although several observational studies have examined the association between antenatal corticosteroid use and childhood respiratory health [20, 21, 22], with conflicting findings, these findings should be interpreted with considerable caution because of their high potential for confounding. Given that antenatal corticosteroids are given in the real‐world setting to pregnancies with a systematically higher risk profile than those who are unexposed to steroids (by virtue of having a clinical condition suggesting an imminent preterm birth), it is unlikely that standard observational approaches that simply control for observed confounders can fully disentangle the causal effect of antenatal steroids from other differences in risk profiles between groups [43].
4.3. Strengths and Limitations
Our use of the regression discontinuity design, which exploits the abrupt change in the probability of exposure to antenatal corticosteroid administration at 34 + 0 weeks, leveraged the strengths of population‐based data (large sample sizes, minimal losses to follow‐up), while reducing the potential for confounding arising from differences in risk profiles of individuals who do vs. do not receive steroids as part of their clinical care. The effect of antenatal corticosteroids on transient tachypnea of the newborn estimated by our regression discontinuity design (a risk reduction of 25% [95% CI: 10%–37%]) aligned very closely with the estimated effect obtained from a meta‐analysis of randomised trials at late preterm ages (a 23% reduction [95% CI: 8%–44%] [44]). This supports the validity of the design to examine longer‐term respiratory outcomes that are more challenging to study in a randomised trial.
A sizeable number of pregnancies were excluded from our cohort due to a lack of day‐specific ultrasound‐confirmed estimate of gestational age. As reported in a previous study from this cohort, a majority of these exclusions were due to administrative reasons (between the years 2000–2008, our provincial perinatal registry only abstracted completed weeks at the time of the dating scan, not days); we thus have no reason to believe this introduced meaningful bias. Our prior comparison of individuals who did vs. did not have a day‐specific, ultrasound estimate of gestational age found that excluded pregnancies were modestly younger (1 year), more likely to be parous (44 vs. 47%) and more likely to smoke (13 vs. 8%), but there were no clinically important differences in pre‐pregnancy body mass index, maternal comorbidities, birthweight, preterm birth or neonatal health outcomes [45].
Our perinatal data registry does not contain information on the dose or timing of antenatal corticosteroid administration. Canadian guidelines recommend two 12 mg doses of IM betamethasone 24 h apart or four 6 mg doses of IM dexamethasone 12 h apart. Repeat courses are not recommended [32, 46]. However, as the need for corticosteroids is typically established shortly after admission, the date of maternal admission to hospital is a reasonable proxy for gestational age of administration. Further, our data show a clear discontinuity at 34 weeks (Figure 1). Pregnancies admitted for delivery after 34 + 0 weeks may have still been exposed to antenatal corticosteroids during a prior antenatal hospitalisation before 34 weeks. However, only 21% of maternal admissions between 34 + 0 and 36 + 6 weeks had a prior hospitalisation and our sensitivity analyses limited to pregnancies with no prior hospitalisations produced similar results as our primary analyses.
Diagnostic codes in administrative health databases are imperfect measures of disease status, and it is possible that the accuracy of the codes differs by gestational age. However, validation studies have shown that the diagnostic codes for asthma and lower respiratory tract infections generally have good accuracy [47, 48] (e.g., sensitivity > 90%, specificity > 80% for childhood asthma in one Canadian study [47]), and we have no reason to believe this error would be differential at the point of the discontinuity at 34 + 0 weeks.
An important consideration of the regression discontinuity design is that it estimates the effect of routine antenatal steroid administration on our outcomes at the point of the discontinuity: 34 + 0 weeks. Thus, the generalisability of our findings is weaker at gestational ages further from 34 + 0 weeks, and it is possible that routine administration of antenatal corticosteroids does have an effect on child respiratory morbidity at later or earlier gestational ages. However, as 34 + 0 weeks is the gestational age in the late preterm period with the highest baseline rates of neonatal respiratory morbidity, it seems likely that this would be the late preterm gestational age at which any population‐level protective effects of routine antenatal corticosteroids resulting from prevention of mild neonatal respiratory morbidity would most likely be seen. Our rate ratio for asthma (RR = 1.08) suggested a null effect, but because of the high prevalence of this condition, cannot rule out a meaningful difference in absolute terms at the population level.
4.4. Conclusions
We found no evidence that routine administration of antenatal corticosteroids is linked with altered rates of later childhood respiratory morbidity. These findings are valuable for clinical practice guideline authors seeking to establish recommendations for use of this medication at late preterm ages, when the balance of harms and benefits is less clear.
Author Contributions
J.A.H., S.H., E.C.S., J.Y.T., M.A.S., J.L. and M.S. conceived the research question, J.A.H. acquired approvals and access to the study data, J.A.H. and M.C.C. conducted data management and statistical analyses, all authors reviewed results and contributed to the interpretation of findings, J.A.H. wrote the first draft of the manuscript, S.H., E.C.S., J.Y.T., M.A.S., J.L., P.S., M.S. and M.C.C. critically revised the manuscript for scientific content.
Ethics Statement
This study was approved by the University of British Columbia/British Columbia Children's & Women's Hospital Research Ethics Board (H18‐00620) on March 15, 2018.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Data S1.
Acknowledgements
We thank Tim Choi from Population Data BC, Vancouver, Canada for his invaluable assistance in data access.
Funding: This work was supported by Canadian Institutes of Health Research by a Canadian Institutes of Health Research Project Grant to J.A.H., E.C.S. and S.H. J.A.H. holds a Canada Research Chair in Perinatal Population Health. E.C.S. holds a William Dawson Scholar award at McGill University. M.S. holds a Chercheur‐Boursier award from the Fonds de recherche du Québec–Santé. J.L. holds a Michael Smith Foundation for Health Research Health Professional Investigator Award. P.M.S. holds a Fonds de Recherche du Québec–Santé Doctoral Award.
Linked article: This article is commented on by Vidaeff, pp. 272–273 in this issue. To view this article visit https://doi.org/10.1111/1471‐0528.70011.
Data Availability Statement
Access to data provided by the Data Stewards is subject to approval but can be requested for research projects through the Data Stewards or their designated service providers. The following data sets were used in this study: BC Perinatal Data Registry, Discharge Abstract Database, Medical Services Plan Billing Records, Consolidation File and Vital Statistics Deaths. Further information regarding these data sets can be found by visiting the PopData project webpage at https://my.popdata.bc.ca/project_listings/18‐124/collection_approval_dates. All inferences, opinions and conclusions drawn in this publication are those of the author(s) and do not reflect the opinions or policies of the Data Steward(s). Statistical code to replicate our analyses is available at https://osf.io/xu9n.
References
- 1. Nowik C. M., Davies G. A., and Smith G. N., “We Should Proceed With Caution When It Comes to Antenatal Corticosteroids After 34 Weeks,” Journal of Obstetrics and Gynaecology Canada 39, no. 1 (2017): 49–51 In eng, 10.1016/j.jogc.2016.05.011. [DOI] [PubMed] [Google Scholar]
- 2. Smith G. C., Rowitch D., and Mol B. W., “The Role of Prenatal Steroids at 34‐36 Weeks of Gestation,” Archives of Disease in Childhood, Fetal and Neonatal Edition 102, no. 4 (2017): F284–F285. [DOI] [PubMed] [Google Scholar]
- 3. Kamath‐Rayne B. D., Rozance P. J., Goldenberg R. L., and Jobe A. H., “Antenatal Corticosteroids Beyond 34 Weeks Gestation: What Do We Do Now?,” American Journal of Obstetrics and Gynecology 215, no. 4 (2016): 423–430. [DOI] [PubMed] [Google Scholar]
- 4. Gyamfi‐Bannerman C., Saade G., and Blackwell S. C., “Concerns Regarding the Use of Antenatal Betamethasone in Late Preterm Gestation: Is the Evidence of Potential Harm Being Overstated?,” BJOG : An International Journal of Obstetrics and Gynaecology 125, no. 8 (2018): 923–924. [DOI] [PubMed] [Google Scholar]
- 5. Vidaeff A. C. and Asztalos E., “Late Preterm Corticosteroids Exposure and Neurodevelopmental Outcomes,” JAMA 331, no. 19 (2024): 1626–1627. [DOI] [PubMed] [Google Scholar]
- 6. Gyamfi‐Bannerman C., Thom E. A., Blackwell S. C., et al., “Antenatal Betamethasone for Women at Risk for Late Preterm Delivery,” New England Journal of Medicine 374, no. 14 (2016): 1311–1320. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. McGoldrick E., Stewart F., Parker R., and Dalziel S. R., “Antenatal Corticosteroids for Accelerating Fetal Lung Maturation for Women at Risk of Preterm Birth,” Cochrane Database of Systematic Reviews 12, no. 12 (2020): CD004454. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Jobe A. H. and Goldenberg R. L., “Antenatal Corticosteroids: An Assessment of Anticipated Benefits and Potential Risks,” American Journal of Obstetrics and Gynecology 219, no. 1 (2018): 62–74. [DOI] [PubMed] [Google Scholar]
- 9. Asztalos E. V., Murphy K. E., and Matthews S. G., “A Growing Dilemma: Antenatal Corticosteroids and Long‐Term Consequences,” American Journal of Perinatology 39, no. 6 (2022): 592–600. [DOI] [PubMed] [Google Scholar]
- 10. Gyamfi‐Bannerman C., Clifton R. G., Tita A. T. N., et al., “Neurodevelopmental Outcomes After Late Preterm Antenatal Corticosteroids: The ALPS Follow‐Up Study,” JAMA 331, no. 19 (2024): 1629–1637. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Sotiriadis A., Tsiami A., Papatheodorou S., Baschat A. A., Sarafidis K., and Makrydimas G., “Neurodevelopmental Outcome After a Single Course of Antenatal Steroids in Children Born Preterm: A Systematic Review and Meta‐Analysis,” Obstetrics and Gynecology 125, no. 6 (2015): 1385–1396. [DOI] [PubMed] [Google Scholar]
- 12. Stock S. J., Thomson A. J., Papworth S., and Royal College of O, Gynaecologists , “Antenatal Corticosteroids to Reduce Neonatal Morbidity and Mortality: Green‐Top Guideline No. 74,” BJOG: An International Journal of Obstetrics & Gynaecology 129, no. 8 (2022): e35–e60. [DOI] [PubMed] [Google Scholar]
- 13. Sultana S., Vogel J. P., and Oladapo O. T., “The Efficacy of Antenatal Corticosteroids to Improve Preterm Newborn Outcomes in Low‐Resource Countries: Are We There Yet?,” BJOG 130, no. Suppl 3 (2023): 84–91. [DOI] [PubMed] [Google Scholar]
- 14. Alhassen Z., Vali P., Guglani L., Lakshminrusimha S., and Ryan R. M., “Recent Advances in Pathophysiology and Management of Transient Tachypnea of Newborn,” Journal of Perinatology 41, no. 1 (2021): 6–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Heinonen S., Suvari L., Gissler M., Pitkanen O., Andersson S., and Helve O., “Transient Tachypnea of the Newborn Is Associated With an Increased Risk of Hospitalization due to Respiratory Syncytial Virus Bronchiolitis,” Pediatric Infectious Disease Journal 38, no. 4 (2019): 419–421. [DOI] [PubMed] [Google Scholar]
- 16. Smith G. C., Wood A. M., White I. R., Pell J. P., Cameron A. D., and Dobbie R., “Neonatal Respiratory Morbidity at Term and the Risk of Childhood Asthma,” Archives of Disease in Childhood 89, no. 10 (2004): 956–960. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Dalziel S. R., Rea H. H., Walker N. K., et al., “Long Term Effects of Antenatal Betamethasone on Lung Function: 30 Year Follow Up of a Randomised Controlled Trial,” Thorax 61, no. 8 (2006): 678–683. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Effects of Antenatal Dexamethasone Administration in the Infant: Long‐Term Follow‐Up,” Journal of Pediatrics 104, no. 2 (1984): 259–267. [DOI] [PubMed] [Google Scholar]
- 19. Stutchfield P. R., Whitaker R., Gliddon A. E., Hobson L., Kotecha S., and Doull I. J., “Behavioural, Educational and Respiratory Outcomes of Antenatal Betamethasone for Term Caesarean Section (ASTECS Trial),” Archives of Disease in Childhood. Fetal and Neonatal Edition 98, no. 3 (2013): F195–F200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Nixon P. A., Washburn L. K., and O'Shea T. M., “Antenatal Steroid Exposure and Pulmonary Outcomes in Adolescents Born With Very Low Birth Weight,” Journal of Perinatology 33, no. 10 (2013): 806–810. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Pole J. D., Mustard C. A., To T., Beyene J., and Allen A. C., “Antenatal Steroid Therapy for Fetal Lung Maturation: Is There an Association With Childhood Asthma?,” Journal of Asthma 46, no. 1 (2009): 47–52. [DOI] [PubMed] [Google Scholar]
- 22. Hung Y. L., Hsieh W. S., Chou H. C., Yang Y. H., Chen C. Y., and Tsao P. N., “Antenatal Steroid Treatment Reduces Childhood Asthma Risk in Very Low Birth Weight Infants Without Bronchopulmonary Dysplasia,” Journal of Perinatal Medicine 38, no. 1 (2010): 95–102. [DOI] [PubMed] [Google Scholar]
- 23. Byberg K. K., Lundholm C., Brew B. K., Rejno G., and Almqvist C., “Pre‐Eclampsia and Risk of Early‐Childhood Asthma: A Register Study With Sibling Comparison and an Exploration of Intermediate Variables,” International Journal of Epidemiology 51, no. 3 (2022): 749–758. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Conlan N., Maher G. M., Al Khalaf S. Y., McCarthy F. P., and Khashan A. S., “Association Between Hypertensive Disorders of Pregnancy and the Risk of Asthma, Eczema and Allergies in Offspring: A Systematic Review and Meta‐Analysis,” Clinical and Experimental Allergy 51, no. 1 (2021): 29–38. [DOI] [PubMed] [Google Scholar]
- 25. Getahun D., Strickland D., Zeiger R. S., et al., “Effect of Chorioamnionitis on Early Childhood Asthma,” Archives of Pediatrics & Adolescent Medicine 164, no. 2 (2010): 187–192. [DOI] [PubMed] [Google Scholar]
- 26. Kumar R., Yu Y., Story R. E., et al., “Prematurity, Chorioamnionitis, and the Development of Recurrent Wheezing: A Prospective Birth Cohort Study,” Journal of Allergy and Clinical Immunology 121, no. 4 (2008): 878, e6–884. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Zhao D., Su H., Cheng J., et al., “Prenatal Antibiotic Use and Risk of Childhood Wheeze/Asthma: A Meta‐Analysis,” Pediatric Allergy and Immunology 26, no. 8 (2015): 756–764. [DOI] [PubMed] [Google Scholar]
- 28. Sendor R. and Stürmer T., “Core Concepts in Pharmacoepidemiology: Confounding by Indication and the Role of Active Comparators,” Pharmacoepidemiology and Drug Safety 31, no. 3 (2022): 261–269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Bor J., Moscoe E., Mutevedzi P., Newell M. L., and Barnighausen T., “Regression Discontinuity Designs in Epidemiology: Causal Inference Without Randomized Trials,” Epidemiology 25, no. 5 (2014): 729–737. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. O'Keeffe A. G., Geneletti S., Baio G., Sharples L. D., Nazareth I., and Petersen I., “Regression Discontinuity Designs: An Approach to the Evaluation of Treatment Efficacy in Primary Care Using Observational Data,” BMJ 349 (2014): g5293. [DOI] [PubMed] [Google Scholar]
- 31. Smith L. M., Levesque L. E., Kaufman J. S., and Strumpf E. C., “Strategies for Evaluating the Assumptions of the Regression Discontinuity Design: A Case Study Using a Human Papillomavirus Vaccination Programme,” International Journal of Epidemiology 46, no. 3 (2017): 939–949. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Society of Obstetricians and Gynaecologists of Canada Expert Opinion , “Antenatal Corticosteroid Therapy for Fetal Lung Maturation,” Journal of Obstetrics and Gynaecology Canada 25, no. 1 (2003): 45–48. [DOI] [PubMed] [Google Scholar]
- 33. Frosst G., Hutcheon J., Joseph K. S., Kinniburgh B., Johnson C., and Lee L., “Validating the British Columbia Perinatal Data Registry: A Chart Re‐Abstraction Study,” BMC Pregnancy and Childbirth 15 (2015): 123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Perinatal Services BC , “British Columbia Data Registry. In: Population Data BC, ed. Vancouver, BC,” (2018), http://www.perinatalservicesbc.ca/health‐professionals/data‐surveillance/perinatal‐data‐registry.
- 35. Society of Obstetricians and Gynaecologists of Canada Expert Opinion , “Determination of Gestational Age by Ultrasound,” Journal of Obstetrics and Gynaecology Canada 36, no. 2 (2014): 171–181. [DOI] [PubMed] [Google Scholar]
- 36. Stockman L. J., Curns A. T., Anderson L. J., and Fischer‐Langley G., “Respiratory Syncytial Virus‐Associated Hospitalizations Among Infants and Young Children in the United States, 1997‐2006,” Pediatric Infectious Disease Journal 31, no. 1 (2012): 5–9. [DOI] [PubMed] [Google Scholar]
- 37. Turi K. N., Wu P., Escobar G. J., et al., “Prevalence of Infant Bronchiolitis‐Coded Healthcare Encounters Attributable to RSV,” Health Sci Rep 1, no. 12 (2018): e91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Flaherman V., Li S., Ragins A., Masaquel A., Kipnis P., and Escobar G. J., “Respiratory Syncytial Virus Testing During Bronchiolitis Episodes of Care in an Integrated Health Care Delivery System: A Retrospective Cohort Study,” Clinical Therapeutics 32, no. 13 (2010): 2220–2229. [DOI] [PubMed] [Google Scholar]
- 39. Ducharme F. M., Dell S. D., Radhakrishnan D., et al., “Diagnosis and Management of Asthma in Preschoolers: A Canadian Thoracic Society and Canadian Paediatric Society Position Paper,” Paediatrics & Child Health 20, no. 7 (2015): 353–361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Rouse D. J., Hirtz D. G., Thom E., et al., “A Randomized, Controlled Trial of Magnesium Sulfate for the Prevention of Cerebral Palsy,” New England Journal of Medicine 359, no. 9 (2008): 895–905. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Dessens A. B., Haas H. S., and Koppe J. G., “Twenty‐Year Follow‐Up of Antenatal Corticosteroid Treatment,” Pediatrics 105, no. 6 (2000): E77. [DOI] [PubMed] [Google Scholar]
- 42. Walters A. G. B., Gamble G. D., Crowther C. A., et al., “Cardiovascular Outcomes 50 Years After Antenatal Exposure to Betamethasone: Follow‐Up of a Randomised Double‐Blind, Placebo‐Controlled Trial,” PLoS Medicine 21, no. 4 (2024): e1004378. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Hutcheon J. A. and Liauw J., “Counterpoint: The Value of Benchmarking in Observational Studies of the Longer Term Safety of Antenatal Corticosteroids,” Paediatric and Perinatal Epidemiology 37, no. 1 (2023): 15–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Saccone G. and Berghella V., “Antenatal Corticosteroids for Maturity of Term or Near Term Fetuses: Systematic Review and Meta‐Analysis of Randomized Controlled Trials,” BMJ (Clinical Research Ed.) 355 (2016): i5044. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Hutcheon J. A., Harper S., Liauw J., Skoll M. A., Srour M., and Strumpf E. C., “Antenatal Corticosteroid Administration and Early School Age Child Development: A Regression Discontinuity Study in British Columbia, Canada,” PLoS Medicine 17, no. 12 (2020): e1003435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Skoll A., Boutin A., Bujold E., et al., “364‐Antenatal Corticosteroid Therapy for Improving Neonatal Outcomes,” Journal of Obstetrics and Gynaecology Canada JOGC 40, no. 9 (2018): 1219–1239. [DOI] [PubMed] [Google Scholar]
- 47. To T., Dell S., Dick P. T., et al., “Case Verification of Children With Asthma in Ontario,” Pediatric Allergy and Immunology 17, no. 1 (2006): 69–76. [DOI] [PubMed] [Google Scholar]
- 48. Pisesky A., Benchimol E. I., Wong C. A., et al., “Incidence of Hospitalization for Respiratory Syncytial Virus Infection Amongst Children in Ontario, Canada: A Population‐Based Study Using Validated Health Administrative Data,” PLoS One 11, no. 3 (2016): e0150416. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1.
Data Availability Statement
Access to data provided by the Data Stewards is subject to approval but can be requested for research projects through the Data Stewards or their designated service providers. The following data sets were used in this study: BC Perinatal Data Registry, Discharge Abstract Database, Medical Services Plan Billing Records, Consolidation File and Vital Statistics Deaths. Further information regarding these data sets can be found by visiting the PopData project webpage at https://my.popdata.bc.ca/project_listings/18‐124/collection_approval_dates. All inferences, opinions and conclusions drawn in this publication are those of the author(s) and do not reflect the opinions or policies of the Data Steward(s). Statistical code to replicate our analyses is available at https://osf.io/xu9n.