aFor each mother in the major congenital anomaly group, up to 10 mothers in the system who delivered a child without a congenital anomaly were matched on maternal age, year of infant’s birth, and parity (1, 2, and ≥3 children). There were 415 910 expected matches in the comparison group but 356 matches were not found due to extremes of maternal age or parity.
Includes 41 508 mothers with an infant with a major congenital anomaly matched (1:10) on maternal age, parity, and year of infant's birth.
Includes 4102 mothers in the multiorgan major congenital anomalies group, 37 406 in the single-organ major congenital anomalies group, and their comparisons derived from 413 742 mothers in the comparison cohort.
eAppendix. Diagnostic codes used in the study with relevant references.
eTable 1. Hazard ratios and 95% confidence intervals for mortality among women who gave birth to an infant with a major congenital anomaly (exposed) and a comparison cohort
eTable 2. Additional analyses of the hazard for mortality among women who gave birth to an infant with a major congenital anomaly (exposed) and a comparison cohort
eFigure. Conceptual framework for potential relation of covariate, exposure and outcome variables.
Cohen E, Horváth-Puhó E, Ray JG, Pedersen L, Adler N, Ording AG, Wise PH, Milstein A, Toft Sørensen H. Association Between the Birth of an Infant With Major Congenital Anomalies and Subsequent Risk of Mortality in Their Mothers. JAMA. 2016;316(23):2515-2524. doi:10.1001/jama.2016.18425
Do mothers who give birth to an infant with a major congenital anomaly have an increased risk of mortality?
In this Danish population-based cohort study of 455 250 women, mothers of infants born with a major congenital anomaly had a significantly increased mortality risk compared with women without an affected infant (absolute mortality rate difference, 0.33 per 1000 person-years; hazard ratio, 1.27). This elevated risk was noted both during the first 10 years after the child’s birth, when the mother was likely caring for a dependent child with substantial health needs, and after longer follow-up, and no single cause of death explained this association.
Having a child with a major congenital anomaly was associated with a small but significantly increased risk of mortality in the mother.
Giving birth to a child with a major birth defect is a serious life event for a woman, yet little is known about the long-term health consequences for the mother.
To assess whether birth of an infant born with a major congenital anomaly was associated with higher maternal risk of mortality.
Design, Setting, and Participants
This population-based cohort study (n = 455 250 women) used individual-level linked Danish registry data for mothers who gave birth to an infant with a major congenital anomaly (41 508) between 1979 and 2010, with follow-up until December 31, 2014. A comparison cohort (413 742) was constructed by randomly sampling, for each mother with an affected infant, up to 10 mothers matched on maternal age, parity, and year of infant’s birth.
Live birth of an infant with a major congenital anomaly as defined by the European Surveillance of Congenital Anomalies classification system.
Main Outcomes and Measures
Primary outcome was all-cause mortality. Secondary outcomes included cause-specific mortality. Hazard ratios (HRs) were adjusted for marital status, immigration status, income quartile (since 1980), educational level (since 1981), diabetes mellitus, modified Charlson comorbidity index score, hypertension, depression, history of alcohol-related disease, previous spontaneous abortion, pregnancy complications, smoking (since 1991), and body mass index (since 2004).
Mothers in both groups were a mean (SD) age of 28.9 (5.1) years at delivery. After a median (IQR) follow-up of 21 (12-28) years, there were 1275 deaths (1.60 per 1000 person-years) among 41 508 mothers of a child with a major congenital anomaly vs 10 112 deaths (1.27 per 1000 person-years) among 413 742 mothers in the comparison cohort, corresponding to an absolute mortality rate difference of 0.33 per 1000 person-years (95% CI, 0.24-0.42), an unadjusted HR of 1.27 (95% CI, 1.20-1.35), and an adjusted HR of 1.22 (95% CI, 1.15-1.29). Mothers with affected infants were more likely to die of cardiovascular disease (rate difference, 0.05 per 1000 person-years [95% CI, 0.02-0.08]; adjusted HR, 1.26 [95% CI, 1.04-1.53]), respiratory disease (rate difference, 0.02 per 1000 person-years [95% CI, 0.00-0.04]; adjusted HR, 1.45 [95% CI, 1.01-2.08]), and other natural causes (rate difference, 0.11 per 1000 person-years [95% CI, 0.07-0.15]; adjusted HR, 1.50 [95% CI, 1.27-1.76]).
Conclusions and Relevance
In Denmark, having a child with a major congenital anomaly was associated with a small but statistically significantly increased mortality risk in the mother compared with women without an affected child. However, the clinical importance of this association is uncertain.
Major structural or genetic congenital anomalies affect approximately 2% to 5% of all births in the United States and Europe.1- 3 Advances in medical care have led to a continuing decrease in mortality rates for children born with congenital anomalies and a corresponding increase in the prevalence of children with complex chronic conditions.4
Mothers of children born with major congenital anomalies face serious challenges such as high financial pressures,5 as well as the burden of providing care to a child with complex needs within the home setting,6 which can impair a mother’s health. Caregiving mothers of children with severe chronic illness have been found to have impaired ability to suppress proinflammatory signals (eg, interleukin-6)7 and greater cell aging associated with high oxidative stress, low telomerase activity, and shorter telomere length.8 Parents (mostly mothers) of children with both multiple9 and specific10 chronic conditions report higher rates of chronic conditions, activity limitations, and poorer physical and mental health compared with parents of healthy children. These health issues may be related to the direct effects of caregiving demands and stress.10
This population-based cohort study assessed whether the birth of an infant with a major congenital anomaly was subsequently associated with increased risk of death of the infant’s mother. It was hypothesized that the mothers of these infants would have an increased mortality rate compared with a general population comparison cohort without exposure to a congenital anomaly. The study also aimed to compare the specific causes of death between the mothers of affected infants and the comparison cohort.
This study was conducted in Denmark (population 5.6 million), which provides its residents with universal access to health care. The study received approval from the Danish Data Protection Agency, which oversees the confidentiality of individual-level information in Danish registries, and The Hospital for Sick Children’s Research Ethics Board. Informed consent was not required for this registry-based study.
The Danish Civil Registration System (CRS) assigns a unique registration number to all Danish residents. The CRS provided demographic data and allowed complete individual-level data linkage across data sources. CRS data on migration and vital statistics are updated daily.11
We used the Medical Birth Registry to identify all women who gave birth to a live singleton infant during January 1, 1979, to December 31, 2010. The study was restricted to mothers who had a minimum of 2 years of health data before giving birth and who survived and did not emigrate at least 1 year postdelivery. The criterion of 1-year survival was instituted to exclude mortality caused by pregnancy complications, and the minimum 2-year data period allowed us to adjust for preexisting comorbidities. The Medical Birth Registry contains information on all deliveries occurring in Denmark, including the CRS registration numbers of parents and newborns, date of birth, singleton vs multiple births, gestational age, and various physical characteristics of the newborn.12 Cohorts were followed for outcomes from 1 year after the date of delivery until the first of death, emigration, or study end (December 31, 2014).
Major congenital anomalies were identified through linkage to the Danish National Patient Registry.13 The registry has tracked all Danish hospitalization data since 1977 and all outpatient and emergency department visits since 1995. It contains admission and discharge data on all hospital contacts, including up to 20 diagnoses classified according to the International Classification of Diseases, Eighth Revision (ICD-8) until 1993 and ICD-10 since 1994. Congenital anomalies were defined with the European Surveillance of Congenital Anomalies classification system.14 Major congenital anomalies diagnosed in outpatient encounters only or caused by hip dislocation or dysplasia only were excluded because of low accuracy of these diagnoses in Danish registry data.15 For mothers with multiple births of infants with major congenital anomalies, the first birth was designated as the index. Because childhood multiorgan complex chronic disease has been associated with more severe consequences compared with diseases affecting a single organ,16 major congenital anomalies were subdivided into 2 groups: major congenital anomalies affecting greater than 1 organ system (subgroup) in the European Surveillance of Congenital Anomalies classification (eg, both congenital heart disease and congenital renal anomaly), and single-organ major congenital anomalies, such as isolated congenital heart disease. All diagnostic codes used in the study are available in the eAppendix in the Supplement.
For each mother in the major congenital anomaly group, we randomly sampled up to 10 mothers in the CRS who delivered a child without a congenital anomaly, individually matched by maternal age, year of the infant’s birth, and parity (1, 2, or ≥3 children).
Our primary outcome was time to death of the mother. A secondary outcome was cause-specific mortality of the mother based on the Danish Register of Causes of Death,17 which compiles information on causes of death from death certificates. Data were available in this registry until December 31, 2011, and categorized according to the underlying cause of death listed first on the death certificate. Cause of death was divided into death from natural (“medical”) and unnatural (“nonmedical”) causes. Deaths from natural causes were subdivided into those from cancer; cardiovascular disease (including myocardial infarction and stroke); respiratory causes; endocrine, nutritional, and metabolic causes; nervous system causes; and other natural causes. Deaths from unnatural causes were divided into those from motor vehicle crashes, suicide, and other accidents or violence.
Other variables used in the study included maternal age, marital status, and immigration status obtained from the CRS at the child’s birth. Income quartile (available starting in 1980) and level of education (available starting in 1981) were obtained at baseline from Statistics Denmark.18,19 Early cohort enrollees (1979-1980) were assigned income and education from the earliest available data. The Medical Birth Registry provided information on previous live births, stillbirths, and congenital anomalies, as well as complications during the index pregnancy, including both placental complications (preeclampsia, gestational hypertension, and placental abruption/infarction) and nonplacental complications (intrauterine hypoxia/birth asphyxia, uterine rupture, umbilical cord prolapse, vasa previa, amniotic fluid embolism, and fetal hemorrhage). Maternal medical history was ascertained from the Danish National Patient Registry and summarized with a modified Charlson comorbidity index score that excluded diabetes and alcohol-related liver disease.20 Diabetes and alcohol use have a strong confounding relation with both congenital anomalies and poor maternal health21,22 and were therefore included as separate covariates. Chronic hypertension and maternal depression were also included. Maternal smoking (data available starting in 1991) and body mass index (BMI; data available starting in 2004) were left truncated and therefore used only in secondary analyses.
Time-to-event curves for mortality were plotted with the Kaplan-Meier technique. We computed cumulative incidence estimates and plotted the cumulative incidence curves of cause-specific mortality, taking into account death by other causes as competing risks. Competing causes of death were censored for the calculation of cause-specific hazard ratios (HRs). The log-rank test was used to test the statistical significance of observed differences in survivorship between mothers who had a child with a major congenital anomaly (exposed group) to those in the comparison (referent) cohort.
We performed a survival analysis using time-to-event Cox proportional hazards regression analyses to compare the risk of each outcome between mothers in the exposed group with those in the comparison cohort, using HRs with 95% CIs. Multivariable Cox models were adjusted for covariates at baseline that were thought to be potential confounders according to biological plausibility, previous research, or clinical relevance (eFigure in the Supplement). These covariates included those related to baseline maternal demographics (marital status and immigration status), socioeconomic status (income quartile and educational level), previous maternal health (diabetes mellitus, modified Charlson comorbidity index score, chronic hypertension, history of alcohol-related disease, and depression), previous spontaneous abortion, and pregnancy complications. The proportional hazards assumption was assessed graphically with –ln[–ln(survival)] vs ln(analysis time), with no major violations observed.
Secondary analyses were planned and performed within the subgroupings of major congenital anomalies into those affecting 1 organ system or more (multiorgan) and those affecting a single organ compared with their matched comparison group, using the log-rank test. In the main analysis, stratified subgroup analyses were performed for the full cohort, comparing the major congenital anomaly group with the same referent group according to duration of follow-up (0-10 years, >10-20 years, and >20-35 years), the year of delivery (1979-1993 [ICD-8], 1994-2004 [ICD-10 with smoking data], and 2004-2010 [ICD-10 with smoking and BMI data]), infant prematurity (≤37 weeks and >37 weeks), and infant death in the first year of life. A sensitivity analysis was also performed, excluding mothers whose infants had an anomaly caused by potentially confounding exposures to teratogens or infections (teratogenic syndromes with malformations, fetal alcohol syndrome, valproate syndrome, and maternal infections resulting in malformations).
Additional post hoc analyses were performed, accounting for other life events in the observation period. The first set of analyses added to the main Cox model these time-dependent covariates: marital status, income, maternal depression, child death, and various combinations thereof. The second set of analyses stratified the main model by the number of child hospitalizations after the newborn hospitalization (0, 1-3, and ≥4), the death of the child from aged 1 year to the end of the observation period, and whether the woman in the exposed group had a subsequent pregnancy with a congenital anomaly compared with her own matches. An additional sensitivity analysis was conducted that used only inpatient data to calculate the modified Charlson comorbidity index score.
All analyses were conducted with SAS version 9.4. Testing was 2-sided and P<.05 was considered significant.
Among 1 924 497 eligible mothers during the study period, 42 731 (2.2%) gave birth to an infant with a major congenital anomaly. Almost all of these, 41 508, were the mother’s first delivery of an infant with a major anomaly. These mothers composed the exposed study cohort (Figure 1). Among infants with a major congenital anomaly, 4102 (9.9%) had multiorgan congenital anomalies and 37 406 (90.1%) had single-organ congenital anomalies. The comparison cohort was composed of 413 742 mothers.
The 2 cohorts of mothers were a mean (SD) age of 28.9 (5.1) years at delivery. No large differences were noted between the 2 groups in matching variables (year of birth, age at delivery, or parity), demographics (marital status, immigration status, education, and income), pregnancy history (history of spontaneous abortions and stillbirths), index pregnancy complications, and maternal medical history (Table 1). Among the infants, low birth weight was more common in those with major congenital anomalies (14.5%) than in the comparison cohort (4.2%), as was prematurity (13.5% vs 4.5%), 5-minute Apgar score less than 7 at 5 minutes (6.1% vs 1.8%), and child mortality before the end of the observation period (3.1% vs 0.5%).
After a median follow-up of 21 years (interquartile range, 12-28 years), 3.1% (1275) of mothers of infants born with major congenital anomalies died during the follow-up period (mortality rate, 1.60 per 1000 person-years [95% CI, 1.51-1.69]) at a mean age of 49.2 years (SD, 9.6 years), whereas 2.4% (10 112) of those in the comparison cohort died at a mean age of 49.5 years (SD, 9.4 years) (mortality rate, 1.27 per 1000 person-years [95% CI, 1.24-1.29]). The corresponding absolute mortality rate difference was 0.33 per 1000 person-years (95% CI, 0.24-0.42), the unadjusted HR was 1.27 (95% CI, 1.20-1.35), and the adjusted HR was 1.22 (95% CI, 1.15-1.29) (Figure 2).
Analyses stratified by time after birth, by whether the infant had single-organ or multiorgan congenital anomalies, by prematurity, and by infant death are presented in Table 2. The adjusted HR for mortality was 1.31 (95% CI, 1.08-1.59) among mothers whose infants had multiorgan congenital anomalies and 1.21 (95% CI, 1.14-1.29) for those whose infants had single-organ congenital anomalies (Figure 3). The mortality difference between the groups was observed in all periods, starting with the first 10 years of follow-up (adjusted HR, 1.31 [95% CI, 1.13-1.53]), as well as after greater than 10 to 20 and greater than 20 to 35 years of follow-up. The increased mortality risk was observed in all 3 birth year periods, but not significantly so in 2004-2010. Mortality risk was increased among mothers of term infants with major congenital anomalies (adjusted HR, 1.18 [95% CI, 1.10-1.26]). The adjusted HR of mortality among mothers of preterm infants with major congenital anomalies (<37 weeks) was 1.10 (95% CI, 0.95-1.28), and among mothers whose infants died in the first year of life, it was 1.19 (95% CI, 0.30-4.75). Exclusion of major congenital anomalies caused by maternal teratogens and infections did not change the findings for the primary outcome.
Table 3 compares the groups by cause of mortality. After adjusting for covariates, mothers of infants born with major congenital anomalies were more likely to die from all natural causes (rate difference, 0.25 per 1000 person-years [95% CI, 0.16-0.34]; adjusted HR, 1.21 [95% CI, 1.12-1.30]). They were more likely to die of cardiovascular diseases (rate difference, 0.05 per 1000 person-years [95% CI, 0.02-0.08]; adjusted HR, 1.26 [95% CI, 1.04-1.53]), including myocardial infarction (rate difference, 0.02 per 1000 person-years [95% CI, 0.01-0.04]; adjusted HR, 1.97 [95% CI, 1.28-3.02]), and were also more likely to die of respiratory diseases (rate difference, 0.02 per 1000 person-years [95% CI, 0.00-0.04]; adjusted HR, 1.45 [95% CI, 1.01-2.08]) and other natural causes (rate difference, 0.11 per 1000 person-years [95% CI, 0.07-0.15]; adjusted HR, 1.50 [95% CI, 1.27-1.76]). Cancer deaths were marginally increased (rate difference, 0.06 per 1000 person-years [95% CI, 0.00-0.12]; adjusted HR, 1.11 [95% CI, 1.00-1.22]). Mothers of infants born with major congenital anomalies were no more likely to die from unnatural (nonmedical) causes (rate difference, 0.03 per 1000 person-years [95% CI, 0.00-0.06]; adjusted HR, 1.12 [95% CI, 0.92-1.36]).
The first set of additional analyses, which considered marital status, income, maternal depression, and child death as time-dependent covariates individually or in combination, resulted in mild attenuation of the estimate of the effect size (eTable 1 in Supplement). In the second set of analyses, the adjusted HRs ranged from 1.02 (95% CI, 0.89-1.17) among children with no subsequent hospitalizations to 1.71 (95% CI, 1.56-1.88) among those with greater than or equal to 4 hospitalizations (eTable 2 in Supplement). Adjusted HRs were 0.76 (95% CI, 0.49-1.18) for death of the child during the observation period and 1.15 (95% CI, 0.79-1.68) among women who had a subsequent child with a congenital anomaly. HR estimates did not change with exclusion of outpatient Charlson comorbidity index score ascertainment.
The birth of a child with major congenital anomalies was associated with a small increased risk of death of mothers. This elevated risk was noted both during the first 10 years after the child’s birth, when the mother was likely caring for a dependent child with substantial health needs, and after longer follow-up. No single cause of death explained this association.
Previous population-level longitudinal studies reporting increased mortality among caregiving mothers have focused solely on bereaved parents.23- 25 Caregiver health studies of nonbereaved mothers caring for chronically ill children have described small cohorts with poor self-reported health outcomes9,10 or biomarkers of advanced cellular aging.8 High rates of chronic stress are reported in these caregivers.26 Animal and human studies of allostatic load have linked chronic stress exposure with health-damaging behaviors and adverse physiologic changes that increase disease risk.27 Deaths from diseases that were most closely linked to chronic stress exposure and to allostatic load, most notably cardiovascular diseases,28,29 were more prevalent among the mothers of children with major congenital anomalies. Cancer-related deaths were only slightly elevated in the exposed group, which is consistent with previous studies that have suggested a null or weak relation of chronic stress and cancer.30,31 Furthermore, high risk was observed in mothers who gave birth to infants with more severe anomalies (multiorgan congenital anomalies), as well as those whose children had 4 or more subsequent hospitalizations. Such children likely pose greater caregiving challenges. The relatively modest effect sizes in subgroups defined by infant risk factors (eg, prematurity and infant death by 1 year) could be due to the independent effect of these infant characteristics on mortality in mothers or a lack of precision in outcome estimates because of small sample size.
An alternative explanation is that the association of offspring congenital anomalies and mortality risk of the mother is due to other unmeasured factors, including unknown genetic factors or behavioral factors such as poor diet, sedentary behavior, smoking, or alcohol consumption that may affect risk of both major congenital anomalies32 and premature adult mortality.33,34 Adjustments were made for as many relevant measurable maternal factors as possible, but some (eg, smoking, BMI) were available for only part of the study period and others (eg, nutritional intake history, hereditary factors) were unavailable.
Studies of mortality among persons caring for dependent adults (mainly the elderly) have reported mixed findings. One study found increased mortality among spouses who reported caregiving strain,35 whereas others have reported reduced mortality among informal family caregivers.36- 38 The latter reports may stem from the so-called healthy caregiver effect, whereby healthy people are more likely to assume caregiving roles or may obtain health benefits from caregiving roles.39 Differences in the findings of studies of caregivers for adults vs children may reflect differences in the age of caregivers, the nature of the care, or the extent of choice in taking on a caregiving role. There are differences as well in the dominant role assumed by a single individual (the mother) in early childhood caregiving situations in Denmark and elsewhere, the duration for which caregiving is necessary, or the lack of preparation for the role.
This study used a population-based cohort design to avoid selection bias, leveraging the opportunity provided by the Danish CRS to link data among registries at the individual level with virtually complete follow-up for the study population of death and emigration.11 The exposure (birth of a child with a major congenital anomaly) occurred at a discrete time and was identified with a defined set of diagnostic codes. The overall rate of major congenital anomalies closely matches rates submitted by Denmark to the European Surveillance of Congenital Anomalies from 1980-2012.2 Exposed mothers were matched to a comparison group on parity, maternal age, and year of delivery, avoiding key sources of potential confounding. The study also used many important covariates through routinely collected data, thereby avoiding potential measurement bias. Although the outcome of death is relatively uncommon in women of childbearing age, the study population was large enough and the outcome period was sufficiently long to provide stable estimates.
Several limitations must be considered when study findings are interpreted. The study population benefited from universal free health care, with relatively generous family assistance and extensive support for families of children with chronic medical conditions.40 Findings may differ in other countries. However, given that medical and social support should be protective of poor maternal health outcomes, even stronger effects might be expected in other settings. Study findings cannot be generalized to other groups of children with chronic conditions that are not due to major congenital anomalies. For cause of death, potential misclassification and incomplete data can lead to imprecise estimates.17 Some important covariates were left truncated and so could not be used in the full analysis (eg, smoking and BMI). Others required extrapolation from data collected after the exposure for a small subset of the cohort (eg, income and education were only available starting in 1980 and 1981, respectively). The results were consistent on stratifying the analysis by the era in which the child was born, although not significantly in 2004-2010. This is likely due to the small sample size and short time to accrue deaths, but also could be due to other factors such as controlling for BMI in this period. Some subgroup analyses had small sample sizes and hence few outcome events (eg, prematurity and infant death) and imprecise estimates of effect size. Covariates in the main model were analyzed as fixed-in-time variables in Cox modeling. This analysis could not account for all potential confounders and mediators between a woman’s having a child with a major congenital anomaly and her risk of dying later in life. Although additional analyses were conducted to account for other life events in the observation period, the potential dynamic effects of variables that may mediate stress or other factors such as the various health trajectories and caregiving needs of children born with major congenital anomalies were not considered. Danish registry data also do not contain information on other potentially important mediators such as the degree of caregiving provided by individual mothers, or care provided by other individuals or in settings outside the home.
In Denmark, having a child with a major congenital anomaly was associated with a small but statistically significantly increased mortality risk in the mother compared with women without an affected child. However, the clinical importance of this association is uncertain.
Corresponding Author: Eyal Cohen, MD, MSc, Department of Pediatrics, The Hospital for Sick Children, University of Toronto, 555 University Ave, Toronto, ON, M5G1X8 Canada (email@example.com).
Author Contributions: Drs Cohen and Horváth-Puhó had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Cohen, Horváth-Puhó, Pedersen, Adler, Milstein, Toft Sørensen.
Acquisition, analysis, or interpretation of data: Cohen, Horváth-Puhó, Ray, Pedersen, Gulbech Ording, Wise, Milstein, Toft Sørensen.
Drafting of the manuscript: Cohen.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Horváth-Puhó, Pedersen, Toft Sørensen.
Obtained funding: Cohen, Milstein, Toft Sørensen.
Administrative, technical, or material support: Cohen, Pedersen, Milstein, Toft Sørensen.
Supervision: Pedersen, Toft Sørensen.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: Dr Cohen was supported by the Commonwealth Fund, a private independent foundation based in New York City, and the Canadian Foundation for Health Care Improvement as a 2015/2016 Harkness Fellow, and by the Canadian Institutes of Health Research (FDN-143315). Drs Sørensen and Pedersen were supported by the Program for Clinical Research Infrastructure (PROCRIN) established by the Lundbeck Foundation and the Norvo Nordisk Foundation. Dr Ray holds a Canadian Institutes for Health Research Chair in Reproductive and Child Health Services and Policy Research, cofunded by the SickKids Foundation.
Role of the Funder/Sponsor: The funders played no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The views herein are those of the authors and not those of the funders, their directors, officers, or staff.
Additional Contributions: We thank Donald Redelmeier, MD, MSHSR, from the Institute for Clinical Evaluative Sciences and Katherine Nelson, MD, from The Hospital for Sick Children (both in Toronto, Canada) for reading and commenting on earlier versions of the manuscript. No one received financial compensation for his or her contributions.