Mortality Among Parents of Children With Major Congenital Anomalies

This study took place in Denmark, which provides universal publicly funded health coverage for all residents. Unique identification numbers assigned to all residents allow accurate linkage of information across all national registries.¹⁶  To define our cohorts of mothers and fathers, we used the Danish Medical Birth Registry¹⁷  and Civil Registry System.¹⁸  We identified mothers and fathers with their first infant born during our study period (1986–2015), although they may have had children before the beginning of the study period. Infants with unknown maternal information were not included. The study was approved by the Research Ethics Board of The Hospital for Sick Children (1000053060).

Our primary exposure was the birth of a child with a major anomaly. We used the Danish National Patient Registry¹⁹  to identify infants diagnosed with a major anomaly after their birth. This registry has been used to track all hospitalizations in Denmark since 1977 and all outpatient hospital clinic visits since 1995 by using diagnosis codes from the International Classification of Diseases, Eighth Revision through 1993 and the International Classification of Diseases, 10th Revision since 1994. We defined congenital anomalies according to the European Surveillance of Congenital Anomalies Classification System using inpatient and both primary and secondary discharge diagnoses, with minor adaptations developed for use of this system with Danish data.²⁰  We excluded mothers (n = 7963) and fathers (n = 7688) for whom the first child born in the study period had a minor anomaly, such as mild skeletal anomalies (eg, rib hypoplasia) or mild craniofacial anomalies (eg, plagiocephaly). We further subdivided major anomalies into 2 groups: single-organ anomalies and anomalies affecting >1 organ system. All diagnostic codes used in the study are listed in Supplemental Tables 6–13.

The primary outcome was time-to-death, and the secondary outcome was cause-specific mortality. Date of death was ascertained by using the Danish Civil Registration System.¹⁸  Cause-specific mortality was ascertained by using the Danish Register of Causes of Death,²¹  which compiles cause of death from death certificates. Cause-specific mortality was defined as deaths from medical causes (natural causes) or deaths from injury (unnatural causes) by using specified underlying cause of death. Medical causes were classified as follows: cardiovascular disease, including myocardial infarction, cerebrovascular injury, and venous thromboembolism; cancer; respiratory diseases; endocrine, nutritional, and metabolic causes; nervous system causes; and other medical deaths. Deaths from injury were defined as deaths from motor vehicle collisions, suicide, other unintentional injuries or violence, and other injuries.

Maternal pregnancy characteristics were collected from the Danish Medical Birth Registry and Danish National Patient Registry. These included maternal age at delivery, parity, pregnancy-related complications, and known exposure to teratogenic medications (ascertained from the Danish National Prescription Registry²² ). Health status for both parents at baseline was ascertained from the Danish National Patient Registry and the Danish Psychiatric Central Research Registry²³  to capture health conditions that may be associated both with the birth of a child with a congenital anomaly and with premature mortality. Health status variables collected included conditions in the Charlson Comorbidity Index (modified to remove codes included in the definition of hypertension),²⁴  history of hypertension, and mental illness and/or substance abuse. We included hypertension as a separate covariate because of reported associations between gestational hypertension and congenital heart anomalies in offspring.^25,26  Maternal substance use in pregnancy, particularly alcohol, is associated with multiple different congenital anomalies,²⁷  and paternal alcohol use has also been reported to be associated with some congenital anomalies.²⁸  Socioeconomic characteristics and income quartile were obtained from the Income Statistics Register,²⁹  immigration status from the Civil Registration System, and education from the Population Education Register.³⁰ 

We used time-to-event Cox proportional hazards regression models to compare the risk of each outcome among 4 categories of parents: mothers with no child with a major congenital anomaly (reference group), mothers with a child with a major anomaly, fathers without a child with a major anomaly, and fathers with a child with a major anomaly. We calculated hazard ratios (HRs) and 95% confidence intervals (CIs) for both unadjusted and adjusted models. In the analyses of cause-specific mortality, we calculated cause-specific hazards, and competing causes of death were censored when calculating the cause-specific HRs. All estimates were adjusted for the following covariates measured at the time of birth: parents’ age (years), child’s year of birth (in 3 categories: 1986–1995, 1996–2005, and 2006–2015), parity, modified Charlson Comorbidity Index score, hypertension, previous mental health diagnosis (including substance and alcohol abuse), marital status, income quartile, education, and immigration status. The proportional hazards assumption was assessed graphically by plotting log (–log [survival function]) versus log of time. Time-to-event curves for overall mortality were plotted with the Kaplan-Meier technique. We computed cumulative incidence estimates and plotted cumulative incidence curves of cause-specific mortality, taking into account death by other causes as competing risks.

Stratified analyses were planned a priori and performed by (1) grouping major congenital anomalies into those affecting multiorgan versus single-organ anomalies, (2) duration of follow-up (0–10, >10–20, and >20–31 years), (3) infant gestational age (<37, ≥37 weeks), (4) child death during the study period, and (5) number of child hospitalizations in the first year of life (0, 1–3, 4+). We conducted 3 sensitivity analyses: (1) excluding mothers and fathers with antenatal exposure to teratogens or prescription medications that may have caused anomalies; (2) restricting the analysis to mothers’ firstborn child (excluding those who had given birth before 1986); and (3) excluding mothers and fathers with a Charlson Comorbidity Index score >1. Analyses were conducted by using SAS software version 9.4 (SAS Institute, Inc, Cary, NC).

Our final cohort consisted of 965 310 mothers (of whom 20 952 [2.2%] had a child with a major anomaly) and 951 022 fathers (of whom 20 655 [2.2%] had a child with a major anomaly) (Fig 1). Among infants with a major anomaly, 8.9% had multiorgan anomalies. There were no large differences between parents with and without an affected infant regarding sociodemographic characteristics, including age at delivery (Table 1). However, more mothers of infants with anomalies had a history of diabetes (1% vs 0.4%) and Charlson Comorbidity Index scores ≥1 (3.5% vs 2.5%). Both mothers and fathers of affected infants were more likely to have mental illness and/or alcohol or substance abuse than mothers and fathers of unaffected infants (mothers: 3.9% vs 3.0%; fathers 3.1% vs 2.7%). These potential confounders were included in multivariable models. Infants born with a major anomaly were more likely to be boys (61% vs 52%), to have a low birth weight (15% vs 4.7%), and to be born prematurely (14.5% vs 5.2%). They were also more likely to die during the follow-up period (3% vs 0.3%).

After a median follow-up of 17.9 years (interquartile range: 9.5 to 25.5 years), mothers of affected infants had an increased overall mortality (mortality rate = 1.19 per 1000 person-years [95% CI: 1.08 to 1.31]) compared with the reference group of mothers without an affected infant (mortality rate = 0.90 per 1000 person-years [95% CI: 0.88 to 0.91]). This corresponded to an adjusted hazard ratio (aHR) of 1.20 (95% CI: 1.09 to 1.32). Fathers of unaffected infants also had an increased overall mortality (mortality rate = 1.96 per 1000 person-years [95% CI: 1.94 to 1.99]). The aHR was 1.62 (95% CI: 1.59 to 1.66). Fathers of affected infants had the highest overall mortality (2.36 per 1000 person-years [95% CI: 2.20 to 2.52]). The aHR was 1.76 (95% CI: 1.64 to 1.88) (Table 2, Fig 2). This increased mortality rate indicates that survival was shorter for mothers of affected infants and fathers of unaffected and affected infants and that death occurred at an earlier age.

Stratified analyses are presented in Table 3. Mothers of infants with multiorgan anomalies had an increased mortality risk compared with those with a single-organ anomaly (for multiorgan anomalies, aHR = 1.69 [95% CI: 1.27 to 2.24] and for single-organ anomalies, aHR = 1.16 [95% CI: 1.05 to 1.28]). This finding was not observed in fathers (aHRs = 1.49 [95% CI: 1.15 to 1.92] for multiorgan anomalies and 1.78 [95% CI: 1.66 to 1.91] for single-organ anomalies). Stratification by duration of follow-up revealed that highest mortality, among both mothers and fathers of an affected infant occurred during the first 10 years of follow-up (aHRs for exposed mothers = 1.27 [95% CI: 1.04 to 1.56], for exposed fathers = 1.92 [95% CI: 1.66 to 2.23]). Mothers and fathers whose infants died during the first year of life had a mortality risk similar to the whole cohort, regardless of exposure status. Stratification by gestational age revealed that mortality risk was slightly increased for both mothers and fathers of infants with major anomalies born at 37 weeks or later (aHR for exposed mothers and fathers = 1.14 [95% CI: 1.03 to 1.27] and 1.76 [95% CI: 1.63 to 1.90], respectively), compared with parents of infants born earlier than 37 weeks’ gestational age (aHR for exposed mothers and fathers = 1.07 [95% CI: 0.85 to 1.34] and 1.42 [95% CI: 1.18 to 1.71], respectively). Mothers, but not fathers, of affected infants had the highest mortality when their infants had ≥4 hospitalizations during the first year of life (aHRs for exposed mothers of infants with ≥4 hospitalizations = 1.79 [95% CI: 1.55 to 2.06]). Sensitivity analyses restricted to mothers’ first child, excluding parents with teratogen exposure, and excluding parents with a comorbidity included in the Charlson Comorbidity Index did not alter the main findings (Supplemental Table 4).

Analyses addressing specific causes of death are presented in Fig 3 and Supplemental Table 5. Mothers of infants with a major anomaly had an elevated mortality risk from all medical causes (aHR = 1.20 [95% CI: 1.09 to 1.33]), as did fathers of unaffected infants (aHR = 1.35 [95% CI: 1.32 to 1.38]), and fathers of affected infants (aHR = 1.51 [95% CI 1.40 to 1.64]), compared with mothers of unaffected infants. The most common cause of death was cancer, accounting for 17 961 of 32 205 deaths (37.9%). Cardiovascular diseases accounted for 7634 deaths (16.1%). Whereas mortality risk was similar across all exposure groups for cancer, it was increased for cardiovascular disease among mothers of affected infants (aHR = 1.41 [95% CI: 1.10 to 1.83]), fathers of unaffected infants (aHR = 2.29 [95% CI: 2.16 to 2.42]), and, in particular, fathers of affected infants (aHR = 2.69 [95% CI: 2.29 to 3.15]). Increased mortality risk associated with endocrine or metabolic disease and other unspecified medical causes death was also observed among mothers of affected infants. Whereas fathers had a greater risk of mortality than mothers from endocrine and nervous system conditions, as well as injury, the risk was similar among fathers with and without an affected infant. Mothers of affected infants had an increased risk of death from other unintentional injuries or violence and other injury; this was not the case with fathers.

In this study of mortality risk among parents of infants born with a major congenital anomaly, we found an increased risk of death for both mothers and fathers, largely due to medical causes, particularly cardiovascular disease. An increased parental mortality risk was seen during the first 10 years of the child’s life, which then mildly attenuated over time. The mortality was relatively early in life, which suggests an increased risk of premature mortality in parents of infants with anomalies. The risk was highest among mothers of more severely affected infants. Overall, the pattern of effects across various causes of death was similar between mothers and fathers and did not change in sensitivity analyses.

To our knowledge, this is the first study to report increased mortality risk among fathers of infants with congenital anomalies. In previous population-level cohort studies using Danish registry data by our research team, researchers have reported increased mortality⁷  and increased cardiovascular⁸  and psychiatric morbidity³¹  risk among mothers of affected infants but have not addressed paternal outcomes.

One putative explanation for our findings is the chronic stress of caregiving. Cardiovascular and endocrine or metabolic diseases are thought to be the causes of death most related to chronic stress.^12,32  The effects of chronic stress may occur through different mechanisms. Chronic activation of the hypothalamic-pituitary-adrenocortical axis and sympathetic-adrenal-medullary system can lead to a proinflammatory state, contributing to cardiovascular disease.^12,33  Models of chronic stress also indicate effects on health behaviors, such as smoking, diet, and exercise, which may lead to poorer cardiometabolic outcomes.³⁴  These health behaviors, as well as chronic stress, are associated with inhibition of telomerase activity, leading to shorter telomere length and possible cardiovascular aging.^35,36  It is notable that the mortality observed was younger than typical deaths from chronic conditions such as cardiovascular disease,³⁷  indicating a need for further investigation to elucidate the mechanisms of these deaths. It would also be meaningful to continue to follow parents to establish if increased mortality continues into later adulthood when a greater burden of chronic disease would be expected.

An alternative explanation could be that infants and parents share a common genetic predisposition, environmental exposure, or health behaviors that lead to both the anomaly and parental illness.³⁸  For mothers, it is possible that other intrapartum factors that were not fully captured in our multivariable models were related to both the anomaly and later cardiovascular morbidity, although we were able to adjust for maternal morbidities such as diabetes.³⁹  Our finding revealing similar effects for mothers and fathers in analyses adjusting for multiple factors, including age, supports the hypothesis that the experience of caregiving plays a meaningful role in leading to mortality risk for both parents.

The finding of similar overall effect sizes for mothers and fathers is consistent with the limited evidence suggesting that gender-based differences in caregiver health are often smaller than expected. Some gender differences were observed. Consistent with life-expectancy data across the world, fathers had an increased mortality risk compared with mothers, regardless of exposure status.⁴⁰  Mothers, but not fathers, had increased mortality risk with more severe manifestations in the infant. Mothers had an increased mortality risk associated with multiorgan anomalies, and mothers were noted to have the highest risk among those with medically fragile infants (≥4 hospitalizations during the first year of life). Gender differences were also observed in specific injuries as causes of death. Although our study did not measure each parents’ participation in caregiving, there may be some gender-based differences.⁴¹  Because of social norms, mothers may share a higher load of caregiving⁴¹  and be more likely to report distress in the setting of their child’s illness.⁴²  Additionally, mothers undergo the stresses of giving birth and shared risk factors particular to pregnancy.⁴³  Further research is needed to better understand the ways that both mothers and fathers experience the stresses of caregiving, and how these stresses affect morbidity.

Our study has several strengths. The exposure (birth of an infant with a major anomaly) occurred at a discrete time before the outcome and was defined by using standard diagnostic codes. We used population-level registries, which minimized selection bias and provided comprehensive exposure and outcome data, with studies revealing high validity for prospectively collected health measures in Danish registry databases.^16,44,45  This design afforded a large enough sample size and long follow-up time, with complete follow-up, to generate stable estimates.

Several limitations also must be considered. There may be limitations in data quality related to validity and completeness of administrative databases and linking between databases. A small number of infants could not be linked with fathers, and a small number of mothers could not be linked to infants. Also, we could not verify paternity. Although rates of nonpaternity vary, in a large review of international studies, a median rate of 3.7% was found.⁴⁶  The exposure was ascertained at birth and in the first year of life. Infants with anomalies diagnosed later would not have been captured and we were not able to draw inferences related to life circumstances that occur after the newborn period.

Imprecise estimates may have resulted from misclassification of deaths from death certificates or small sample sizes in some subgroup analyses. We also lacked data on certain variables that could be confounders, such as diet and smoking. It is also possible that parents and infants had shared genetic conditions that were not accounted for in this analysis. In addition, the study took place in a single country with universal health care and a robust support system for parents, potentially limiting generalizability. Further research is needed to determine if similar patterns exist in other populations.

Fathers and mothers of a child born with a major congenital anomaly have an increased risk of mortality, compared with unexposed parents, primarily due to cardiovascular disease. Our findings support including fathers in programs aimed at protecting parental caregivers’ health.

aHR

adjusted hazard ratio

CI

confidence interval

HR

hazard ratio