Heavy prenatal alcohol exposure and healthcare use during childhood and adolescence: a Danish nationwide cohort study 1997–2022

Heavy prenatal alcohol exposure remains a key public health concern due to its potential adverse effects on fetal development leading to an increased risk of life-long physical, cognitive, and behavioural impairments [1]. Prenatal alcohol exposure can also lead to fetal alcohol spectrum disorders (FASD). The global prevalence of any alcohol use during pregnancy has been estimated at 9.8% [2], and the prevalence of FASD at 19.8 per 1000 children in the European Region [3]. Prevalences might be overestimates due to secular trends of lower levels of alcohol drinking in pregnancy [4], nonetheless they represent best available estimates and pregnant women often tend to conceal their alcohol use, and FASD is underdiagnosed. While heavy prenatal alcohol exposure is associated with increased child morbidity and mortality [5, 6], there are few studies on healthcare use among children with heavy prenatal alcohol exposure and existing studies primarily focus on selected populations. Some studies have found FASD associated with higher healthcare use [7, 8], while other studies have described barriers to health care for children with FASD, including lack of knowledge, complex needs, misinterpretation of the children’s behaviour, and caregiver incapacities [9, 10]. Children affected by prenatal alcohol exposure constitute a vulnerable group with both potential life-long impairments caused by alcohol and a high risk of co-occurring childhood adversities such as parental mental disorders, out of home care placement, neglect, and abuse [5, 11]. Patterns of healthcare use can highlight attention points for health care providers and inform policy efforts to promote health equity. This nationwide cohort study aimed to assess healthcare utilisation among children with heavy prenatal alcohol exposure compared to all other Danish children, using national registry data. To achieve a nuanced understanding of healthcare-seeking behaviour and system-level responsiveness, we aimed to: (1) quantify use across eight distinct healthcare services, including somatic and psychiatric hospitals, general practice (GP), and participation in the universal GP-based Danish child health programme; (2) describe age- and sex-specific rates in utilisation; (3) describe hospital admission characteristics, including length of stay, among alcohol-exposed children. These insights may help identify service gaps and guide targeted improvements in care for this vulnerable population.

In total, 1,457,962 live-born children were followed from 1997 to 2022 with a mean total follow-up age of 12.3 years, after exclusion of 59,381 children (figure S1). Of these, 5898 children (0.4%) were identified as exposed. The subpopulation for the outcome of the preventive child health programme consisted of 1,124,326 (4496 [0.4%] exposed) children. Table S5 shows the total number of healthcare contacts and person-time for healthcare outcomes within sex and age groups.

For calculating start of pregnancy in regard to exposure time, 2.1% had missing information on gestational age and 0.25% of all hospital contacts had missing end-of-contacts. For the covariates of maternal education and ethnicity, no information was registered for 1.15% and 0.003% respectively.

Compared to reference children, children with heavy prenatal alcohol exposure had higher mortality (72 [1.2%] vs. 8175 [0.6%], p < 0.001) while fewer emigrated during the study period (142 [2.4%] vs. 68,802 [4.7%], p < 0.001).

Mothers of exposed children were more often adolescent mothers and had lower educational levels, fewer were immigrants or descendants of immigrants. More exposed children were preterm and first-borns compared to reference children, and fewer were multiples. Compared to reference children, more children were prenatally exposed to other substance use (27.1% vs. 0.4%) and had mothers with mental disorders (35.8% vs. 4.3%) (Table 1).

Of the eight healthcare services assessed, Fig. 1A presents age- and sex-specific rates for six healthcare contacts. Rates were higher among children with heavy prenatal alcohol exposure compared to reference children for acute hospital contacts, planned outpatient hospital contacts, planned hospital admissions, psychiatric hospital contacts, GP consultations, and GP additional services. At different ages, boys and girls exhibited varying healthcare utilisation patterns. Sex-specific utilisation patterns among exposed children overall resembled that of reference children. Longer length of stay for planned hospital admissions and neonatal admissions were observed for exposed children compared to reference children (Fig. 1B).

For hospital outcomes, more exposed children than reference children had at least one contact, e.g. 23.2% vs. 12.5% had at least one neonatal admission (table S6). For GP outcomes, 99% of all children had at least one contact.

For all hospital and GP outcomes and within all age groups, adjusted incidence rate ratio estimates suggested relative increases in the mean number of contacts for exposed children compared to reference children, with only one estimate for planned hospital admissions in ages 12–18 not being statistically significant. Estimates adjusted for year, sex, and age of child suggested relative increases between 19 and 101% in somatic hospital use, e.g. for ages 0–5 years, 6–11 years, and 12–18 years, respectively: IRR for exposed children compared to reference children for acute hospital contacts were 1.24 [95% CI 1.20–1.28], 1.19 [1.14–1.25], and 1.34 [1.26–1.43]; for planned outpatient contacts 2.01 [1.92–2.10], 1.29 [1.21–1.37], and 1.20 [1.12–1.28]; and for neonatal admissions 1.82 [1.73–1.90]). Likewise relative increases between 12 and 36% for GP use, and two to three times as many psychiatric hospital contacts for exposed children compared to reference children (aIRR 3.55 [2.98–4.24] for ages 0–5, 2.68 [2.41–2.98] for ages 6–11, and 2.19 [1.95–2.46] for ages 12–18 (Fig. 2). While estimates additionally adjusted for maternal age, ethnicity, and education were lower, patterns of higher healthcare use among exposed persisted.

Exposed children had lower participation in the child health programme compared to reference children: 23.1% vs. 40.5% fully participated in the seven exams (table S7), and a median of five vs. six exams was observed (Fig. 3B). When adjusted for sex and birth year of child, exposed children were 31% less likely to have participated in above five out of seven exams (adjusted risk ratio 0.69, 95% CI 0.67–0.72) (Fig. 3A).

Supplementary analyses showed the higher use of health care to be evident for both children solely exposed to heavy prenatal alcohol exposure and for children with both heavy prenatal alcohol exposure and respectively other prenatal substance exposure or maternal mental disorders, when compared to children with no exposure of heavy prenatal alcohol exposure and respectively no other prenatal substance exposure or no maternal mental disorders (figure S2 and S3). Likewise, the lower participation in the child health programme was evident for both children with solely exposure and double exposures (figure S4). Higher healthcare use was found consistently for both boys and girls. Tests for interaction with sex were insignificant for all healthcare outcomes except neonatal admissions (figure S5).

Results were robust in regard to choice of correlation structure for the GEE-models: Sensitivity analysis with an independent correlation structure did not overall affect results, however results for planned hospital admissions age 12–18 became significant (figure S6). Main findings remained consistent in all sensitivity analyses with restriction to: singleton first-borns where estimates were slightly lower (figure S7, S10), extended data availability (figure S8, S10), and maternal exposure diagnoses (figure S9, S10).

Regarding extended data availability: Before incorporating a data source late in the study, the proportion of exposed children declined over time, but remained consistent afterward (Table 1).

Quantifying patterns of healthcare use among children with heavy prenatal alcohol exposure, we found higher use of hospitals and GP, and lower participation in the child health programme compared to reference children.

Findings of higher healthcare use are in line with previous studies on children with FASD [7, 8, 26]. Findings of lower participation in the child health programme remained consistent when restricting the analysis of participation to children born full-term, indicating that missed participation was not explained by the higher incidence of exposed children born preterm (figure S10).

Healthcare utilisation depends on both need and care-seeking behaviour. This study found that less than one in four exposed children (23.1%) attended all seven preventive child health programme exams, indicating that their higher use of hospital and GP services likely reflects greater health needs or delayed care seeking rather than a general healthcare-seeking prone behaviour. The lower uptake of the free-of-charge preventive child health programme suggests that, despite free access, non-financial barriers affect this vulnerable group. Therefore, higher healthcare use does not necessarily indicate optimal care. In settings with financial barriers, even greater disparities in preventive service use and health outcomes might be expected. With the cohort consisting of all live-born children in Denmark, this study also provides patterns of healthcare use for the general population of Danish children.

In accordance with literature describing accumulations of risk factors among children with prenatal alcohol exposure, we found exposed children more likely to have other prenatal substance use exposure (27.1% vs. 0.4%) and maternal mental disorders (35.8% vs. 4.3%). Supplementary analyses showed that both as single risk factors and as double exposures, heavy prenatal alcohol exposure, other prenatal substance use exposure, and maternal mental disorders were associated with patterns of higher healthcare use. This is in line with studies that have shown higher healthcare use among children with different risk profiles such as maternal mental- and alcohol use disorders [27], maternal alcohol and substance use disorders [28, 29], and among children with a high level of childhood adversities [30]. We also found that 16.4% of exposed children had mothers younger than 21 years of age at time of birth compared to 2.4% of reference children, which underlines the importance of preconception care.

Nationwide registry data enabled this large-scale study, supplementing existing studies by including different healthcare services and ensured no susceptibility to selective inclusion or recall bias. Danish registers are generally believed to be of high validity and completeness. FASD is overlooked, and since assessment of prenatal alcohol exposure relies on self-reports prone to social desirability bias, underreporting is likely. To improve identification and reduce misclassification, we assessed heavy prenatal alcohol exposure based on both maternal and child diagnoses, an approach previously used in registry-based studies [5] and inspired by work of O’Leary and colleagues [24]. While misclassification remains possible, it would likely bias results towards the null and thus reflects a conservative, best-informed estimate in a Danish context.

The exposure definition holds no information on timing, frequency, or quantity of alcohol intake, and represents only the most heavy use which is a small fraction of any alcohol use during pregnancy. Data on privately funded or anonymous enrollments in alcohol or substance use treatment clinics are incomplete. Data on alcohol clinic enrollments covers the years from 2006 and forwards, however sensitivity analysis estimates aligned with main estimates. We did not include information on maternal smoking in this study, which is however expected to co-occur with prenatal alcohol exposure. We chose to estimate results with adjustment for two sets of covariates: one adjusted for age, sex and year, and one additionally adjusted for sociodemographic factors. Several additional covariates, including sociodemographic characteristics, are likely part of a broader risk context associated with prenatal alcohol exposure. To reflect differences in healthcare use associated with heavy prenatal alcohol exposure, careful consideration of model adjustments is essential to avoid bias from overadjustment. Therefore, we consider the lightly adjusted models more appropriate for capturing overall patterns and also believe these most relevant from a clinical and public health perspective. Independently of the chosen adjustment, patterns of higher healthcare use among exposed children persisted.

This study shows that despite a comprehensive nationwide free preventive child health programme, the programme which requires self-referral might not be well-suited for caregivers in vulnerable families and therefore fail to capture the possibly most vulnerable children and families. Overall, these findings highlight the long-term individual consequences and societal burden of heavy prenatal alcohol exposure and indicate a need for adequate support to ensure health equity and child well-being for a vulnerable group of children.