Study design and data sources
We conducted a register-based cohort study using nationwide registers. Data linkages were achieved via the personal identity number (CPR number) which is assigned to all Danish residents at birth or upon receiving residency in Denmark [20]. Register keepers at Statistics Denmark carried out data collection and register linkage with all data being anonymized before the researchers gained access.
Information about parental diagnoses, birth-related variables of the child as well as both child in- and outpatient hospital contacts and emergency department (ER) contacts were obtained from The National Patient Register [21]. The diagnoses were encoded using the International Classification of Diseases, 10th Revision (ICD-10). The Danish Health Service Register for Primary Care [22] provided information on contacts to general practitioners (GPs) as well as contacts to private psychiatrists and psychologists. Information on reimbursed drug prescriptions was obtained from the Danish National Prescription Registry which uses the Anatomical Therapeutic Chemical (ATC) Classification System [23]. Data on parity and gestational age were accessed through the Danish Medical Birth Register [24] and on deaths through the Danish Death Register. Information on parental highest completed education was extracted from the Population Education Registry [25], and family income was obtained from the Income Statistics Register [26]. Date of birth, parental civil status, child gender and information on whether the parents lived both together and with the child was obtained from the Danish Civil Registration System [20]. The included Danish registers are generally considered to be high quality with complete long-term follow-up [27].
Settings, study population and follow-up
The Danish healthcare system is characterized by free access. Services at general practice (GP) and public hospitals are funded by the Danish tax system and are free of charge. Additionally, acute medical assistance is only delivered by public services. GPs serves as gatekeeper for hospitals, and contacts to the hospitals require a referral from a GP. GPs are available during regular daytime hours. In case of illness outside of normal openings hours, an out-of-hour medical service is available.
All live-born children born in Denmark from January 1, 2000 to December 31, 2016 were identified and followed during the first 12 months of their lives. Children were excluded if they died, emigrated during the first year of life, or did not live with either biological parent after birth.
Parental mental health
Parental mental health was categorized in three sub-categories (no mental health condition, minor and moderate-severe) using The National Patient Register, The Danish Health Service Register for Primary Care, and the Danish National Prescription Registry. In Denmark, minor mental health conditions are treated by GPs or psychologists in the primary healthcare sector and are thus not registered in The National Patient Register [28].
GPs serve as gatekeepers in the primary healthcare sector. These professionals refer more severe mental health conditions for assessment and treatment in the secondary healthcare sector (in- and outpatient wards at the hospitals or private psychiatrists). Diagnoses are not registered in The Danish Health Service Register for Primary Care; therefore, minor mental health conditions need to be identified in other ways. Talk therapy, psychometric tests given at GP, and contacts to psychologist are registered in The Danish Health Service Register for Primary Care and were used to identify minor mental health conditions. Furthermore, reimbursed prescriptions of antidepressant and anxiolytic medication were used to identify these conditions. Other psychotropic medication, for example antipsychotics, were not included due to broader indications such as nausea or analgesics.
Moderate to severe mental health conditions were identified as contact to a private psychiatrist or a registered psychiatric diagnosis (ICD-10 F00-99) at a psychiatric hospital. All mental health condition indicators had been measured within five years prior to the birth of the child. A reference group was defined as having no mental health-related contacts to GP, no contacts to psychologist or private psychiatrist, no reimbursed prescriptions of antidepressants or anxiolytic medication, and no registered psychiatric diagnosis within five years before the birth of the child (see Box 1).
Two groups, minor (group 1) and moderate-severe (group 2) mental health conditions, and the unexposed reference group generated a matrix of exposures for the child due to the combinations of the mother and father in the different groups.
Box 1. Criteria for exposure. Consensus definition of groups of mental conditions.
|
|
Exposure
|
Specification
At least one criterion fulfilled
|
Further criteria
All criteria fulfilled
|
Healthcare sector and registry
|
Group 1
Minor mental health conditions
|
Medication - At least 2 prescriptions of:
- antidepressant medicine (ATC N06AB, N06AX)
- anxiolytic (benzodiazepines: ATC N03AE, N05BA, N05CD, N05CF)
Services at general practice
- At least two ‘talk therapy’
- At least two psychometric tests
Other services
- At least one contact to private psychologist
|
No contacts to psychiatric hospital and no psychiatric hospital diagnoses
No records of contact to private psychiatrist
|
Primary healthcare sector
The Danish Health Service Register for Primary Care
Danish National Prescription Registry
|
Group 2
Moderate to severe mental health conditions
|
- Any registered psychiatric diagnosis (ICD-10 F00-99) at psychiatric hospital
- Mental health conditions treated at private psychiatrists (including child and adolescent psychiatrists)
|
|
Secondary healthcare sector
The National Patient Register
The Danish Health Service Register for Primary Care
|
Reference group
No mental health condition
|
None of above
- No registrations of psychiatric diagnoses and no mental health condition-related contacts to GP, psychologist or private psychiatrist and
- No prescriptions of antidepressants or anxiolytic drugs
|
|
|
Exposure was measured in a period of 5 years before the birth of the child
|
|
Outcomes
All healthcare service contacts made within the first year of a child’s life were identified. Vaccinations and routine childcare visits were not included. Contact to GP daytime or out-of-hour service was defined as either consultation, telephone contact, or physical visit. Due to changes in the delivery of out-of-hours service in the Capital Region of Denmark in 2014, children from this region were excluded from the out-of-hour service analyses from 2014-2016. An inpatient contact was defined as any admission to hospital, and an outpatient hospital contact as any registered contact to an outpatient unit. Due to changes in registration of neonatal outpatient contacts during the study period, only outpatient contacts after the neonatal period (the first 28 days of life) were included. Any registered contact to the emergency department was defined as an ER contact.
Covariates
Covariates, that could potentially confound the association between parental mental health conditions and child use of healthcare services were identified by reviewing the literature. A directed acyclic graph [29, 30] was constructed to evaluate the identified covariates (see supplements). Except for family income, which was extracted from the calendar year before the child’s birth, all covariates were extracted at the time of birth.
Based on the International Standard Classification of Education (ISCED 2011) [31], parental education was grouped according to highest completed education and divided into three groups: 1) Early childhood education (primary and lower secondary education (ISCED levels 0-2)); 2) General upper secondary education ( high school programs, vocational upper secondary education, vocational training and education (ISCED 3-4)); and 3) Short-, medium-length or long-length higher education( first-, second- or third-cycle programs, tertiary education, bachelor or equivalent, Master’s or equivalent, Doctoral, PhD programs or the equivalent (5-8)).
Family income was defined as the equivalated disposable income for the family, which took into account the number of children and adults in the household. We adjusted for family income, parental age at time of birth of the child and calendar year using restricted cubic splines with three knots [32].
Family type was classified as either living with two biological or adoptive parents or not.
Small-for-gestational-age (SGA) was defined as a birthweight for gestational age under the 2.3-percentile and large-for-gestational-age (LGA) as a birthweight for gestational age over the 97.7-percentile based on national guidelines [33, 34]. Parity was divided into two groups according to whether the child was the mother’s first child or not.
Statistics
Baseline information was analyzed and reported as percentages based on maternal and paternal, respectively, mental health condition. In the following analyses, the maternal and paternal groups were combined and a total of eight exposure groups and one reference group were generated.
The median number of events for each outcome variable was calculated, and total number and percentages of children in each exposure group with an outcome above the median were calculated. Both crude and adjusted Poisson regression analyses [35] were applied to the number of outcome events for each child during follow-up to estimate the incidence rate ratio (IRR) of experiencing the outcome.
Siblings appeared in the cohort and should have been treated in the analyses as dependent. A mixed effects model, however, was not possible to apply to such a large dataset. Instead, we made sensitivity analyses, which only included first-born children of mothers.
Non-fatal birth defects, prematurity and other significant conditions in early life might necessitate a higher number of healthcare contacts during the first year of life. To examine whether such conditions affected the estimates, a sensitivity analysis excluding children admitted for at least one week during first month of life was performed.
Data was analyzed using Stata SE 15.1.