Health Outcomes 50 Years After Preterm Birth in Participants of a Trial of Antenatal Betamethasone

Preterm birth (before 37 weeks’ gestation) affects 1 in 10 births internationally, resulting in significant neonatal and childhood morbidity and mortality.^1,2 Preterm birth also has implications for adult health through the persistence of childhood disability and the potential for early-life adverse events to influence later health.^3–5 

Preterm birth has been associated with higher rates of cardiovascular risk factors and disease such as hypertension, diabetes mellitus, stroke, and ischemic heart disease.^6–9 The associations with higher blood pressure and insulin resistance are well established, but the effect on cardiovascular disease is much less consistent, with some studies finding clearly increased risk and others finding higher risk only in specific subgroups.^8–13 A population-based study of 2.4 million people in Finland and Norway also found higher odds of asthma and chronic obstructive pulmonary disease (COPD) in preterm-born adults compared with those born at term, with stronger effects in those born very or extremely preterm.¹⁴ 

Although poorer neurodevelopmental outcomes after preterm birth are observed during childhood, cognitive and education outcomes during adulthood have been less consistent, with some studies finding little difference or better outcomes after preterm birth and others showing worse outcomes.^15–19 Those born preterm also have more internalizing and avoidant behavior and less externalizing behavior than those born at term.²⁰ A population-based cohort study of 3.3 million people in Sweden observed higher rates of psychotic or bipolar disorder, autism, attention-deficit hyperactivity disorder, and suicide attempts in those born preterm, with higher risk at lower gestational ages.²¹ 

However, most studies of the impact of preterm birth on adult outcomes have assessed effects in early adulthood, and older cohorts studied have been born before the era of modern intensive care and antenatal corticosteroids.^6–8,13 

We therefore investigated a range of health outcomes in later life in a prospective cohort from the first randomized trial of antenatal corticosteroids for preterm birth, comparing those born preterm with those born at term.^22,23 

The Auckland Steroid Trial (AST) and the follow-up study protocol have been described elsewhere.^22–24 Briefly, between December 1969 and February 1974, a double-blind, randomized, placebo-controlled trial of antenatal corticosteroids for women at risk of preterm birth was performed at the National Women’s Hospital, Auckland, New Zealand. Surviving offspring were assessed for cardiovascular risk factors, respiratory, and general health when aged 30 years.^{12,18,25–28} 

Eligible participants included all surviving offspring born to women who participated in the AST, whether randomized to antenatal placebo or betamethasone. Outcomes for participants who died prior to follow-up were assessed based on mortality and administrative data from the NZ Ministry of Health.

We invited participants to complete a questionnaire and to consent to access administrative datasets. The questionnaire was based on the New Zealand Health Survey 2019/2020, the International Primary Care Airways Guidelines questionnaire, the International Physical Activity Questionnaire and the Washington Group Short Set on Functioning.^29–32 Administrative datasets included those managed by the NZ Ministry of Health (National Minimum Dataset [hospital admissions since 1988], National Non-Admitted Patients Collection [emergency department and outpatient visits since 2006], Mortality Collection [national mortality records since 1970], Pharmaceutical Data [national community dispensing records of pharmaceuticals funded by the NZ government since 1992]), TestSafe (laboratory tests data since 2010 for Auckland and Northland regions; includes 36% of NZ’s population and NZ’s largest city, where participants were born), NZ Qualifications Authority (national qualifications data since 1991), Ministry of Justice (national convictions data), and the Accident Compensation Corporation (accident claims data since 1974).

Ethics approval was obtained from the Northern A Health and Disability Ethics Committee (20/NTA/37). Written informed consent was obtained from each participant. A waiver of consent for accessing administrative health records was approved for potential participants who had died. This study is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology guideline (Supplemental Appendix 1).

Outcomes were assessed using the questionnaire, administrative datasets, or both. In cases of inconsistency between datasets or in which only a single administrative data source was available, all available data were reviewed by a panel of 5 clinicians unaware of treatment allocation or preterm birth status to adjudicate the outcome. The primary outcome was a composite of cardiovascular events or risk factors, defined as a history of a major adverse cardiovascular event (MACE; admission for myocardial infarction, coronary revascularization, peripheral vascular disease, stroke, heart failure, or cardiovascular death) or the presence of at least 1 cardiovascular risk factor (diabetes mellitus, prediabetes, treated hypertension, treated dyslipidemia). Key secondary outcomes included the components of the primary outcome and a range of respiratory, mental health, educational, and other health outcomes (Supplemental Appendix 2).

The statistical analysis plan was prepared after recruitment and before analysis (Appendix 3, Statistical Analysis Plan). All analyses were performed using SAS version 9.4 (SAS Institute). Denominators are the number of participants who consented to follow-up with data available for the outcome unless otherwise specified. Participants missing data for outcomes were removed from analyses of that outcome, and missing data were not imputed.

Binary outcomes were compared between those born preterm and at term using generalized linear modeling (GLM, binomial distribution, log link function) to estimate relative risk and 95% CIs. Continuous outcomes were compared by GLM (normal distribution, identity link function) and reported as mean differences (95% CIs). Analyses were undertaken unadjusted and adjusted for multiple pregnancies and randomized treatment allocation (betamethasone or placebo).

For the time-to-event analysis, a Cox proportional hazards model was used, reported as hazards ratios (95% CIs), after verifying the assumption of proportionality following visualization of Kaplan-Meier curves.

Significance testing used type 3 fixed effects from the GLM model and joint testing for the Cox proportional hazards model. Outcomes were tested at the 5% significance level. No adjustment to the significance level was made for secondary outcomes or subgroup and sensitivity analyses, as the study prioritized identifying important effects of preterm birth for hypothesis generation at the risk of a type 1 error with multiple comparisons. Exploratory subgroup and sensitivity analyses were performed only for the primary and key secondary outcomes and are listed in the statistical analysis plan.

In the AST, 1121 live infants were born to 1025 women, and 964 survived to age 1 year (Figure 1). Between January 2020 and May 2022, 424 adult survivors were recruited to this follow-up study (46.2% of those presumed alive), of whom 405 (95.5%) completed a questionnaire and consented to data linkage, 9 (2.1%) only completed a questionnaire, and 10 (2.4%) only consented to data linkage. For 46 paticipants who died between infancy (1 year) and follow-up, only linked data were available.

Participants had similar baseline characteristics to those eligible who did not participate, although participants were more likely to be female (47.7% vs 40.8%) and born preterm (69.4% vs 59.8%) (Table 1). A higher proportion of the preterm birth group were a result of planned preterm birth (69/326 [21.2%] vs 10/144 [6.9%]), were from a multiple pregnancy (52/326 [16.0%] vs 7/144 [4.9%]), and had never smoked (170/293 [58%] vs 52/118 [44.1%]) (Table 2).

The primary outcome composite of cardiovascular events or risk factors occurred in 112/325 (34.5%) of those born preterm and 43/144 (29.9%) of those born at term, with an adjusted relative risk (aRR) of 1.14 (95% CI, 0.85–1.54; P = .37).

The cardiovascular risk factor components of the primary outcome (diabetes mellitus, prediabetes, treated hypertension, and treated dyslipidemia) were not significantly different between groups (Table 3). However, the proportion with at least 1 admission for a MACE was lower in those born preterm (9/326 [2.8%]) than those born at term (10/144 [6.9%]; aRR 0.33, 95% CI, 0.14–0.79; P = .013). Age at first admission for MACE was also older in the preterm group compared with those born at term (adjusted hazard ratio [aHR] 0.31, 95% CI, 0.12–0.79; P = .014).

A self-reported diagnosis of high blood pressure and the composite of treated hypertension or self-reported diagnosis of high blood pressure were both more common in those born preterm (aRR 1.74, 95% CI, 1.16–2.61; P = .007 and aRR 1.63, 95% CI, 1.12–2.36; P = .010, respectively), but there was no difference between groups for dyslipidemia (self-report, laboratory diagnosis of hypercholesterolemia and low-density lipoprotein cholesterol) (Table 3).

Body size did not differ markedly between preterm and term-born participants, with no significant difference between groups in height, weight, body mass index (BMI), and proportion with overweight or obesity (Table 3).

There was no significant difference between groups for chronic kidney disease (Table 4).

Death from any cause from randomization to 50-year follow-up was more than twice as common in the preterm group than those born at term (172/767 [22.4%] vs 35/354 [9.9%]; aRR 2.29, 95% CI, 1.63–3.23; P < .0001). However, deaths after 28 days and after 1 year were similar between groups.

Respiratory outcomes were similar between groups, although the proportion with pharamceuticals dispensed for asthma or COPD was higher in the preterm group (88/288 [30.6%] vs 28/113 [24.8%]; aRR 1.22, 95% CI, 0.84–1.76; P = .30) (Table 4).

Diagnosis or treatment of a mental health disorder was less common in those born preterm (115/301 [38.2%]) compared with those born at term (65/123 [52.9%]; aRR 0.73, 95% CI, 0.59–0.92; P = .007) (Table 5). Depression was also less common in the preterm group (86/300 [28.7%] vs 48/123 [39.0%]; aRR 0.74, 95% CI, 0.56–0.99; P = .044), but bipolar affective disorder, anxiety disorder, inpatient admissions for mental health disorders, and the dispensing of pharmaceuticals were not different between groups. Schizophrenia was uncommon (3/301 [1.0%] in the preterm group vs 0/123 in the term group).

Self-reported general health was similar between groups, as was the proportion with fair or poor general health. Overall functional difficulties were not different between groups, but hearing difficulty was less common in the preterm group (4/295 [1.4%] vs 7/120 [5.8%]; aRR 0.20, 95% CI, 0.05–0.71; P = .014) (Table 6).

Educational outcomes and proportion with a criminal conviction were not significantly different between preterm and term groups, although there was a higher proportion in paid employment in the preterm group (P = .023) (Table 6).

Other health outcomes were not significantly different between groups (Table 6).

In subgroup analyses, there was some evidence that preterm birth may be more likely to be associated with overweight or obesity in male but not female individuals (P = .013 for interaction, Supplemental Table 1, Appendix 4). Preterm birth may also be less likely to be associated with asthma or COPD in participants born small for gestational age vs those born appropriate or large for gestational age (P = .038).

Sensitivity analysis adjusting for socioeconomic status (SES) based on the NZDep2018 as a potential confounder increased the magnitude of the exposure effect for the primary outcome and shifted the CI to exclude no effect (aRR 1.44, 95% CI, 1.04–1.99; P = .028; Supplemental Table 2, Appendix 4).³³ Including BMI in the adjusted model to assess its effect as a mediator abolished the differences between groups (aRR 0.97, 95% CI, 0.68–1.38; P = .87). Other sensitivity analyses did not markedly affect the results of the primary outcome (Supplemental Table 3, Appendix 4).

The effect estimate for treated hypertension was also increased by adjustment for SES, but other adjustments had minimal effect on the results. In contrast, for the diagnosis or treatment of a mental health disorder, adjustment for SES appeared to attenuate the effect of preterm birth (aRR 0.82, 95% CI, 0.64–1.04; P = .10), whereas adjustment for BMI and birthweight had minimal effect (Supplemental Table 2, Appendix 4). Sensitivity analysis for the remaining key secondary outcomes did not noticeably affect the results (Supplemental Table 2, Appendix 4)

We compared outcomes of those aged 50 years between those born preterm and those born at term in a prospective cohort study of offspring of participants in the first randomized trial of antenatal corticosteroids for women at risk of preterm birth.

The composite of cardiovascular events and risk factors was not statistically significantly different between preterm and term groups. However, the exposure effect estimates for some cardiovascular risk factors and MACE were in opposite directions. Cardiovascular risk factors other than diabetes had exposure effect estimates suggestive of an increase with preterm birth, whereas admissions for MACE were less common after preterm birth. A similar discrepancy between cardiovascular events and risk factors was reported in the Helsinki birth cohort study, a population-based cohort study of 19 000 people born between 1922–1944.^10,13 These seemingly paradoxical findings suggest additional cardiovascular disease determinants other than the risk factors measured. These could include differences in health behaviors. We found a higher proportion of the preterm group had never smoked, consistent with previous reports that preterm birth is associated with less risk-taking behavior.²⁰ Those aware of being born preterm also may be more likely to seek preventive treatments, potentially resulting in a reduced risk of cardiovascular disease but a greater prevalence of risk factors if defined by a treatment such as treated dyslipidemia or treated hypertension. However, there was a relatively small number of cardiovascular events in this cohort, increasing the risk of a chance finding.

Our findings contrast with a large population-based cohort study in Sweden, birth years 1973 to 1994, in which the risk of ischemic heart disease and stroke was higher in those born preterm.^8,9 Changes in neonatal care may affect the comparability of cohorts. For example, older cohorts such as the Helsinki birth cohort predate antenatal corticosteroids and neonatal intensive care, while the more recent Swedish cohort occurred during the era of antenatal corticosteroid use, although actual exposure to antenatal corticosteroid was not reported.^8,13 Interestingly, in both our study and the Helsinki birth cohort, smoking rates were markedly lower in those born preterm than those born at term, although these were similar across gestations in the Swedish cohort.^7,8,10 Furthermore, our study and the Helsinki birth cohort both used term birth at 37 and higher weeks’ gestation as the reference group, whereas the Swedish cohort used 39–41 weeks, thus excluding from the reference group those born at early term gestations (37–38 weeks) who were also reported to have an increased risk of ischemic heart disease. This offers a further potential explanation for differences in findings for cardiovascular events.

Hypertension is the cardiovascular risk factor with the most well-established relationship with preterm birth.^6,11,34 Our finding of a higher prevalence of self-reported high blood pressure with and without treated hypertension in the preterm group is consistent with these reports and also with previous findings from this cohort of higher systolic blood pressure at 30 years compared with those born at term.¹² Interestingly, we found the proportion with chronic kidney disease, a potential mediator of elevated blood pressure, appeared greater in those born preterm, although not statistically significant, with an exposure effect estimate similar to that reported in a recent meta-analysis.³⁵ 

The risk of diabetes was similar between groups. In the same cohort, follow-up at 30 years identified possible insulin resistance but no difference in blood glucose concentration during oral glucose tolerance tests in those born preterm, similar to findings from another study using intravenous glucose tolerance tests.^10,12 A large Swedish cohort study found an aHR for type 2 diabetes of 1.85 (95% CI [1.24–2.77]) for very preterm birth compared with term birth but no differences for moderate to late preterm birth (aHR 1.07, 95% CI [0.82–1.40]).⁷ It is possible that the follow-up of our cohort was not long enough for differences in insulin resistance to manifest as differences in type 2 diabetes or that our cohort had too few participants born at earlier gestational ages to see an effect as marked as that in the Swedish cohort.⁷ It is also possible that the earlier differences in insulin resistance were a chance finding or not large enough to be clinically important.

We found respiratory outcomes were similar between the preterm and term groups. This is important, since there have been concerns about progressive deterioration in lung function in preterm survivors in the era of antenatal corticosteroid use.³⁶ Our findings differ from a recent large population study and from previous reports indicating poorer respiratory function in adults born preterm.^14,37,38 However, the increase in asthma reported after moderate and late preterm birth compared with term birth was relatively small (odds ratio 1.24–1.42 for moderate and 1.17–1.28 for late preterm).¹⁴ The number of participants with COPD in our study was small, so we had limited power to detect a small effect.

Diagnosis or treatment of a mental health disorder was less common in those born preterm. Although many, but not all, studies have identified an increased risk of mental health disorders in those born preterm, there are many potential confounders.³⁹ Adjustments for current SES markedly attenuated the difference between groups for diagnosis or treatment of a mental health disorder. Current SES may be acting as a confounder, as a proxy for parental SES, but due to using current SES mediation cannot be ruled out. This was an exploratory sensitivity analyses, and the findings should be interpreted with caution. Psychosis is of particular interest given previous findings of higher rates of psychosis and psychotropic use in preterm-born adults.²¹ In our study, all 3 cases of schizophrenia were in the preterm group, but the small numbers make it difficult to draw conclusions.

The highest level of educational achievement was similar between adults born preterm and at term in the present study. This is consistent with an earlier follow-up of this cohort showing no differences between groups in Weschler Abbreviated Scale of Intelligence testing and educational outcomes from those aged 31 years.¹⁸ The effects of being born preterm on educational outcomes is controversial, with studies suggesting that preterm birth has negative impacts on educational measures in childhood but less consistent effects in adulthood.^17,40–43 In our study, most preterm-born participants were moderately preterm, whereas findings of poorer educational outcomes are more consistent in studies of those born very or extremely preterm.^17,40,41 

The strengths of this study include that it is the longest follow-up of a cohort born in the time of antenatal corticosteroids for preterm birth and neonatal intensive care. We also combined self-reported questionnaire data and information from administrative datasets to improve outcome ascertainment, as self-reported data alone is known to underestimate chronic diseases such as hypertension, dyslipidemia, and diabetes mellitus.^44,45 

The main limitation of this study was the relatively small sample size, limiting the power to detect small differences between groups and limiting the validity of further subgroup analyses. The low follow-up rate (46% of those presumed to be alive) also increases the risk of attrition bias. However, most baseline characteristics were similar between those assessed and those lost to follow-up, and there were few baseline differences between the preterm and term groups.

Missing data for individuals who did not answer specific questions or either did not consent or did not have data available in specific administrative datasets could introduce bias if this information is not missing at random. Specifically, data on BMI were missing for 16% of participants alive at follow-up, and laboratory data were missing for 28%.

A further limitation was that we did not undertake in-person assessments of blood pressure and blood tests due to geographical dispersal of the cohort over 50 years, so the outcomes of hypertension, dyslipidemia, and diabetes may be underestimated.

The gestational ages in the preterm group (median 34.1 weeks, IQR 32.3–35.1) meant that most participants were born moderately or late preterm, and relatively few were born very or extremely preterm. This may limit the generalizability of the findings to preterm birth at earlier gestations. However, the data remain highly relevant, as most preterm infants are born at moderate or late gestations, with global estimates indicating approximately 85% of preterm births occur at a gestational age of 32 or more weeks.¹ Generalizability may also be reduced by the exclusion of women in preterm labor who delivered immediately on arrival, as these women would not have been eligible for the AST.

Preterm-born adults aged 50 years were at a higher risk for high blood pressure than those born at term and had a similar risk of diabetes, prediabetes, and dyslipidemia; however, their risk of cardiovascular events was lower. Worsened cardiovascular outcomes may not be a universal finding for those born preterm, particularly those with moderate preterm birth.

We thank the mothers and their offspring who participated in the original trial and this follow-up study, without whom this work would not have been possible, and acknowledge the seminal contributions of the original trialists, the late Ross Howie and Mont Liggins, who generously made available the original data and documentation from the AST. We also thank Trisha Meagher-Lundberg for study coordination and Coila Bevan for participant tracing.

aHR

adjusted hazard ratio

aMD

adjusted mean difference

aRR

adjusted relative risk

AST

Auckland Steroid Trial

BMI

body mass index

COPD

chronic obstructive pulmonary disease

GLM

generalized linear modeling

HR

hazard ratio

LDL

low-density lipoprotein

MACE

major adverse cardiovascular event

MD

mean difference

RR

relative risk

SES

socioeconomic status.