Article Text

Download PDFPDF
Developmental vulnerabilities in children of chronically ill parents: a population-based linked data study
  1. Megan F Bell1,2,
  2. Donna M Bayliss2,
  3. Rebecca Glauert1,
  4. Jeneva L Ohan2
  1. 1 Telethon Kids Institute, University of Western Australia, Perth, Western Australia, Australia
  2. 2 School of Psychological Science, University of Western Australia, Perth, Western Australia, Australia
  1. Correspondence to Dr Megan F Bell, Telethon Kids Institute, University of Western Australia, Perth WA 6009, Australia; megan.bell{at}telethonkids.org.au

Abstract

Background Currently, there is mixed evidence regarding the effects on children when a parent is chronically ill. Research has also primarily been conducted with adolescent samples. This study investigated developmental vulnerabilities in young children of parents with chronic illness.

Methods This study used linked administrative data. The study population included children born in Western Australia during 2003–2004 (n=19 071; mean age 5.5 years). The outcome measure was a score in the bottom 25% on any of the five developmental domains (physical, social, emotional, communicative and cognitive) of the Australian Early Development Census (2009 collection). Parental chronic illnesses were identified from hospital and cancer registry records, during the period from 1 year prior to the child’s birth and until the end of 2009.

Results Higher odds of developmental vulnerabilities in physical, social, emotional and communication domains were observed for daughters of chronically ill mothers. Sons of chronically ill mothers had increased odds of language and cognitive difficulties. Risk level increased with each additional year of exposure to maternal chronic illness. Results also indicated increased odds of developmental vulnerabilities for children of mothers experiencing multiple compared with single chronic conditions; however, results were not statistically significant (all p>0.05). No association between fathers’ chronic illness and children’s developmental outcomes was found.

Conclusions Maternal chronic illness is associated with an increased risk of poor developmental outcomes for children, particularly daughters. Healthcare services have an important role to play in linking families into appropriate family-centred services to best support the needs of chronically ill mothers.

  • record linkage
  • maternal health
  • lifecourse/childhood circumstances
  • chronic di

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Chronic physical illnesses are conditions that are complex, prolonged and difficult to treat.1 Due to a combination of medical advances and an increasing incidence of lifestyle risk factors, rates of chronic illness are increasing.1 Individuals with chronic illnesses generally experience impaired physical, social and role functioning and reduced well-being, compared with healthy counterparts.1 Awareness is also increasing that the impact of chronic illness extends beyond the patient to his or her immediate family, especially to children. A better understanding of how best to support these children may help to reduce the burden on chronically ill adults faced with the dual demands of parenting while managing their illness.

Currently, most of our knowledge of the impact of parental chronic illness relates to adolescents. In this age group, outcomes include an increased risk of poor social and behavioural functioning,2 3 academic underachievement4 5 and higher rates of internalising and externalising problems.5 6 Alternatively, investigations of these associations in samples of young children have yielded inconsistent findings: parental chronic illness has been associated with an increased risk of developmental difficulties in children,6 no increased risk7 8 and even lower risk.9 Thus, additional research is required to understand the outcomes of young children of parents with chronic illnesses. Such knowledge would help inform models of care to best support the family system of adults experiencing chronic illness.

This study uses linked population-level data to examine the association between chronic illness in parents and the physical, social, emotional, communicative and cognitive development of their young children. We hypothesise that young children of chronically ill parents will be at risk of vulnerability on multiple developmental domains and that the risk will increase with increasing duration of exposure to parent chronic illness. Outcomes of children with chronically ill parents may differ depending on the sex of the child2 3 and the parent.2 6 We therefore examine these relationships separately by parent and child sex. Furthermore, individuals who have multiple chronic conditions experience greater decreases in functioning and well-being than those who have a single chronic illness.10 We therefore identify parents with single or multiple chronic illnesses and predict that children of parents with multiple diagnoses will be most at risk.

Methods

Data sources

This study used anonymised administrative data merged across multiple government agencies and linked by the Western Australian (WA) Data Linkage Branch using identifiers (eg, name and address) common to the sets of records.11 Details are outlined below.

Study population

The sample included WA children born in 2003–2004 with a 2009 Australian Early Development Census (AEDC) record (n=20 653). Exclusion criteria included: (1) missing AEDC scores (n=382); (2) multiple births, with one child randomly selected for retention (excluded n=274); (3) deceased parent (n=132); (4) identified as ‘special needs’ in the AEDC dataset (n=753); or (5) diagnosis of developmental disorder (eg, autism) or cerebral palsy (n=41) in hospital records or the WA Register of Developmental Anomalies. Children identified as ‘special needs’ have a diagnosed disability (physical and intellectual), and their AEDC domain scores are not calculated. Children with a developmental disorder or cerebral palsy were likewise excluded to limit the potential for confounding developmental outcomes with pre-existing conditions. The final sample included 19 071 children (mean age 5.5 years, SD=0.29). Health and demographic information was available for all mothers and 18 876 fathers (99%).

Developmental measure

Children’s development was assessed with the AEDC, which uses the Australian version of the Early Development Instrument (AvEDI).12 The AEDC is a triennial national census, supported by the Australian Department of Education and Training, of children in the first year of compulsory schooling (in WA, the year before grade 1). Teachers complete the 104-item AvEDI for each child in their class, from which a score is calculated (from 0 to 10) for each of five developmental domains: physical, social, emotional, communication and cognition (see table 1). The method of calculating domain scores is the intellectual property of McMaster University in Canada. Domain scores are analysed at the national level and classified into percentiles. Children scoring in the bottom 10% on a domain are considered ‘developmentally vulnerable’ on that domain; those in the 11–25th percentile as ‘at risk’, and those in the 26–100th percentile as ‘on track.’ For this study, these three categories were collapsed into two (‘vulnerable/at risk’ and ‘on track’) to capture established and emerging developmental vulnerability.

Table 1

Description of domains assessed on the Australian Early Development Census

The AEDC and the Canadian Early Development Instrument,12 (the measure on which the AEDC is based) possess good convergent and concurrent validity13–17 and inter-rater and test–retest reliability.17 Both measures have also demonstrated good predictive validity for later scores on measures of academic and social abilities.14 18–20

Chronic illness measure

The list of chronic illnesses included in this study (see online supplementary materials) was adapted from multiple sources.1 21 Parental chronic illness diagnoses were assembled from the Emergency Department Data Collection (EDDC), which records presentations to public hospitals, and the Hospital Morbidity Data System (HMDS), which records public and private hospital separations. In these datasets, diagnoses are recorded using the International Classification of Diseases 10-Australian Modification (ICD-10-AM).22 Cancer diagnoses were obtained from the WA Cancer Registry, which includes information on all cancer diagnoses in WA.

Supplemental material

Parents were identified as having a chronic illness if they had at least one chronic illness diagnosis recorded in any of the three datasets during the study period. The study period started 12 months prior to the child’s birth and continued to the end of 2009 to capture the impact of chronic illness on parenting, which includes the prenatal period. Parents were categorised as ‘any diagnosis’ or ‘no diagnosis’. For supplementary analyses, parents in the ‘any diagnosis’ group were further categorised into ‘single diagnosis’ and ‘multiple diagnoses’, determined by summing all different chronic illness diagnoses recorded in the datasets. A proxy for duration of parent chronic illness was calculated as years since first record (either in the Cancer Registry or the hospital datasets) of any diagnosis (range 1–8 years).

Covariates

Child, parent and community characteristics that are associated with child developmental outcomes23–25 were included as covariates (table 2); details are below.

Table 2

Sociodemographic characteristics of the study cohort

Child characteristics

Child sex and Aboriginality were determined from the Midwives Notification System, provided by the WA Health Department, and Birth Registrations, provided by the Department of the Attorney General. These datasets contain information on birth and pregnancy details for all live births in WA, as well as demographic information on the child, mother and father (if reported by the mother). Child age was calculated in months from birth to the end of 2009. Children who spoke English as a second language (ESL) were identified from the AEDC dataset.

Aboriginality was included as a proxy variable for a range of sociopolitical and historical factors that Aboriginal children and their families may experience and that may impact on developmental and health outcomes.26 27 This variable is therefore intended as a proxy for social, and not racial, determinants of health and developmental outcomes.

Previous research has shown that children who experience chronic illness are at risk of developmental vulnerability.28 Child chronic illness diagnoses (corresponding to the list used by Bell and colleagues28) were therefore identified to account for intergenerational health disadvantage. Child diagnoses were identified from the same datasets as parent diagnoses, from birth until the end of 2009.

Parent characteristics

Parental age and mother’s marital status (at the time of the cohort member’s birth) were obtained from the Midwives Notification System and Birth Registrations.

As chronic illness is associated with poor mental health,29 30 and comorbid parent physical and mental illness is associated with poorer child outcomes,9 29 a dichotomous variable indicating diagnosis/no diagnosis of mental illness for mothers and fathers (separately) was also included as a covariate. Parent mental illness diagnoses over the entire study period (identified as records with a ICD-10-AM ‘F’ code) were assembled from the HMDS and EDDC in addition to public outpatient data from the Mental Health Information System.

Community characteristics

Community-level covariates were obtained from the AEDC dataset, which identifies the child’s community of residence at the time of AEDC completion. Each community is assigned a value for remoteness and socioeconomic status, determined by the Australian Standard Geographical Classification (ASGC)31 and Index of Relative Socioeconomic Disadvantage (IRSD),32 respectively. The ASGC classifies an area as metropolitan, inner/outer regional or remote/very remote, based on access to goods, services and opportunities for social interaction. Due to smaller cohort numbers outside the metropolitan areas, these five categories were collapsed into three (metropolitan, regional and remote). The IRSD is derived from census information that reflects area-level disadvantage (eg, low income, low educational attainment, high unemployment and unskilled occupations). Communities were given a score ranging from 1 (most disadvantaged) to 5 (least disadvantaged).

Analysis

Logistic regression models were fitted with maximum likelihood estimation using SAS V.9.3 for Windows.33 Analyses estimated the odds of children being classified as vulnerable/at risk on each of the five AEDC domains as an outcome of parent chronic illness. Covariates were transformed into binary or continuous categorical variables with the category representing lower risk coded as the reference group. Unadjusted and adjusted ORs and 95% CIs were estimated for each AEDC domain. Adjusted models controlled for the covariates listed in table 2, and parental chronic illness information was entered simultaneously. Analyses were run separately by child sex. Holm’s p value correction was applied to account for multiple hypothesis testing.34 35

Missing data

Inspection of frequencies showed that children who were Aboriginal, ESL, born to teenage mothers and living in remote locations or disadvantaged communities were more likely to have missing data for parent demographic variables. Missingness was therefore included as an extra category in regression models to limit potential bias. Adjusted models were also run including only cases with complete information (n=18 261). The ORs from the complete case analysis were up to 0.05 points larger than the ORs from the missing indicator analysis. However, CIs were slightly wider. Therefore, for completeness of the sample and robustness of estimates, the results presented below are from analyses using the missing indicator method.

Results

Prevalence

Table 3 presents frequencies of parental chronic illness diagnoses. During the study period, 1399 mothers (7.34% of the total cohort) were diagnosed with a chronic illness, of whom 95 mothers (6.79%; 0.50% of total cohort) had multiple diagnoses (range 1–4 diagnoses). Cancer was the most common diagnosis (47.89%), followed by respiratory (18.37%) and cardiovascular (12.22%) diseases. The average duration of exposure to maternal chronic illness was 2.90 years (SD=2.08).

Table 3

Prevalence of different chronic illness diagnoses

There were 1297 fathers (6.80% of total cohort) with a chronic illness diagnosis during the study period, of whom 123 (9.48%; 0.65% of total cohort) had multiple diagnoses (range 1–4 diagnoses). For fathers, musculoskeletal disorders were the most common (38.55%), followed by cancer (25.37%) and cardiovascular disease (21.43%). The average duration of exposure to paternal chronic illness was 3.26 years (SD=2.12).

There were 92 children (0.48% of total cohort) who had two parents diagnosed with a chronic illness during the study period.

Regression analyses

Table 4 shows the unadjusted and adjusted ORs for children with a mother or father with chronic illness, compared with children whose parents did not have a chronic illness diagnosis recorded in the available data. In the unadjusted models, maternal chronic illness was associated with increased odds of daughters being classified as vulnerable/at risk on all AEDC domains and increased odds of sons being classified as vulnerable/at risk on and the language and cognitive domain. After adjusting for sociodemographic factors, odds were attenuated but remained statistically significant for daughters of mothers with a chronic illness for developmental vulnerability on the physical (29% increase), social (40% increase) and communication (47% increase) domains. ORs for the emotional and cognitive domains were elevated but not significant. Results for sons of mothers with chronic illness were also attenuated after adjustment, with significantly increased odds (37%) of developmental vulnerability for cognitive skills only.

Table 4

Unadjusted and fully adjusted* odds of children being classified as developmentally vulnerable/at risk on the Australian Early Development Census as an outcome of parental chronic Illness diagnosis

Paternal chronic illness was not associated with significantly increased odds of developmental vulnerability for sons or daughters on any AEDC domain in either the unadjusted or fully adjusted models (table 4). Accordingly, no further analysis of paternal chronic illness was conducted.

Duration of exposure

Table 5 displays the fully adjusted ORs estimating the impact of duration of exposure to maternal chronic illness on developmental vulnerability. For daughters of mothers with chronic illness, each additional year of exposure was associated with a significant increase in odds of developmental vulnerability on the physical (7% increase), social (6% increase), emotional (5% increase), communication (9% increase) and cognitive (5% increase) domains. For sons of chronically ill mothers, only the OR for the language and cognitive skills domain was significant, indicating a 7% increase in odds of developmental vulnerability with each additional year of exposure.

Table 5

Fully adjusted* odds of children being classified as developmentally vulnerable/at risk on the Australian Early Development Census as an outcome of duration of exposure (in years) to mother’s chronic illness

Supplementary analyses

Supplementary analyses investigated whether multimorbidity in mothers was associated with poorer outcomes for children. Table 6 shows the fully adjusted models.

Table 6

Fully adjusted* odds of children being classified as developmentally vulnerable/at risk on the Australian Early Development Census as an outcome of mother’s single or multiple chronic illness diagnosis

Compared with their peers, daughters of mothers with a single or multiple chronic illness diagnosis/diagnoses had increased odds of developmental vulnerability on all domains. Daughters of mothers experiencing multimorbidity (relative to no chronic illness) were particularly at risk of poorer social and physical development, with an increase in odds of 121% and 177%, respectively. There was also a large, but non-significant, increase in odds of communicative vulnerability (109%) for these girls. For sons, a single maternal chronic illness diagnosis was associated with increased odds of developmental vulnerability on the language and cognitive skills domain (39% increase) only. Results for sons of mothers with multiple diagnoses were non-significant but compatible with increased odds of developmental vulnerability on all domains.

Compared with daughters of mothers with a single chronic illness diagnosis, daughters of mothers experiencing multimorbidity had significantly increased odds of poor physical development (131% increase). All other ORs comparing children of mothers with multiple chronic illness diagnoses to children of mothers with a single diagnosis were non-significant but compatible with maternal multimorbidity being associated with increased odds of developmental vulnerability on all AEDC domains for girls, and for four of five domains for boys, with language and cognitive skills being the exception.

We also investigated potential diagnosis-specific effects by running models on the three most prevalent maternal diagnoses (cancer, musculoskeletal disorders and respiratory disorders). These models excluded mothers with multimorbidity. Overall, results across the three diagnostic groups were consistent with the ‘single diagnosis’ group, with many ORs elevated but non-significant. In particular, results were consistent with girls experiencing social and communicative difficulties and boys experiencing cognitive difficulties. Chronic maternal respiratory illness was significantly associated with cognitive difficulties in both boys and girls (see online supplementary tables).

Discussion

This study examined the developmental outcomes of young children of parents with chronic illness. Compared with children of well parents, daughters of chronically ill mothers were at risk of physical, social and communication difficulties and sons were at increased risk of cognitive difficulties, with the risk increasing with each additional year of exposure to mother’s chronic illness. We did not find an association between paternal chronic illness and poor developmental outcomes in children. Our findings concur with another study that found that physical illness in mothers, but not fathers, was associated with greater emotional and behavioural problems in children.36 However, there is some evidence that paternal ill health is associated with children’s maladjustment,37 suggesting more research is needed on these relationships.

Related literature may offer explanations for the lack of association between paternal chronic illness and developmental vulnerabilities for children in this study. For instance, if mothers are the primary caregivers, maternal chronic illness may place an added burden on the family, contributing to children’s developmental vulnerability. This is supported by findings that maternal chronic illness impacts negatively on both the child and the spouse, with illness demands leading to feelings of depression and marital and parenting difficulties in the father.38 Limited resources in both parents may therefore negatively impact on children’s development. Alternatively, fathers may be less involved in caregiving, particularly when ill, so reduced exposure to paternal illness may mean a lesser impact on children’s development.36 Future research should examine such features of the family context that may mediate the relationship between parental chronic illness and children’s outcomes.

We also found differences in outcomes for sons and daughters of chronically ill mothers. Daughters appeared to be most at risk, consistent with studies on older children that have found a greater incidence of social, emotional, physical and cognitive difficulties in adolescent girls, compared with boys.3 6 Others have proposed various explanations for such findings, including girls taking on additional family responsibilities or being more emotionally involved in their families than boys.39 40 However, given the young age of children in our sample, it is unlikely that these factors could explain our findings. Future research using qualitative measures of the reactions of young children to maternal chronic illness may provide insights into the mechanisms of risk transmission found in this study.

Our results also suggested that children whose mothers had multiple chronic illnesses were at increased risk of developmental vulnerability compared with those whose mothers had a single diagnosis. However, due to the limited number of mothers in our sample identified with multimorbidity, most of these results were not statistically significant. Although multimorbidity is more common in the elderly, a substantial minority of adults of child-rearing age have multiple chronic conditions.41 Individuals suffering multiple chronic conditions generally experience increased healthcare costs and decreased quality of life,10 but there is limited knowledge of how this impacts on children. Our results suggest that further research on these relationships would be beneficial.

Understanding how maternal chronic illness can impact on children’s early development is important for developing interventions to minimise negative effects and invites speculation based on previous findings. For example, poor maternal health during the prenatal period can lead to preterm delivery or intrauterine growth retardation,42 both of which are associated with poorer developmental outcomes.43 Post-birth, chronic illness may limit mothers physically, financially, emotionally or cognitively, meaning that they have fewer resources and are less available to care for their child.37 In addition, parents’ emotional responses to their illness can predict children’s outcomes more strongly than the severity or duration of the illness,29 indicating the importance of appropriate family support services when a mother is chronically unwell. We accounted for parental mental illness in this study but did not specifically examine the association of co-occurring parental chronic physical and mental illness on children’s outcomes. This is an important area of future research on the impact of parental chronic illness on young children.

Family-centred care for parental chronic illness has long been recommended in clinical practice.44 45 However, this approach is not routinely implemented,45 and there remain barriers to implementing family-centred care, such as pre-existing family problems, limited awareness among professionals and a lack of standardised intervention guidelines.44 45 Yet, there is promising evidence for the effectiveness of such interventions in improving children’s coping skills and decreasing family emotional distress.46 47 Healthcare professionals are in a unique position to link families into appropriate services: young children generally have more exposure to healthcare than education services, and healthcare services can readily identify chronically ill patients who are parents. Increased recognition of the importance of family-centred interventions for chronically ill parents, as well as efforts to address the barriers to implementing them, are both essential to achieve improved support for at-risk families.

One limitation of this study was the potential for selection bias through use of hospital data to identify chronic illness. Hospitals are disproportionately accessed by disadvantaged groups due to issues of availability and affordability of primary healthcare services.48 Individuals attending hospital for chronic illnesses are also more likely to have more severe and/or poorly managed symptoms, and these features of parental chronic illness may affect children’s development differently. We were unable to evaluate severity of illness from the information available in the data, but this should be investigated in future research. Furthermore, our measure of duration of exposure to chronic illness was only a proxy, as it could not be ascertained if first presentation to hospital equated to onset of disease. Investigations of the impact of duration of exposure using specific measures would be of interest.

It is also possible that we underestimated rates of chronic illness. Prevalence estimates indicate that cardiovascular and endocrine/metabolic diseases and musculoskeletal disorders are the most common chronic illnesses seen in primary care,49 whereas the most common diagnoses in our sample were cancer, respiratory diseases and musculoskeletal disorders. Use of the Cancer Registry enabled identification of all parental cancer diagnoses; however, other diagnoses are likely under-identified due to reliance on hospital data. Lastly, it is possible that children in our comparison group have parents with a chronic illness that was not recorded in the available data. Accordingly, we expect our results are more conservative, relative to analyses using exclusive samples of children with and without chronically ill parents.

In conclusion, our findings indicate that maternal, but not paternal, chronic illness is associated with an increased risk of adverse outcomes for children. Specifically, young girls with chronically ill mothers are at risk of poorer social, physical and communication development, and boys are at risk of cognitive difficulties. Maternal health status should be considered when supporting children’s early development, and treatment of chronic illness should consider the broader impacts on the patient’s family. Future research into potential mediating and moderating factors at the child, family and system level will provide important information for development of services to support at-risk families.

What is already known on this subject

  • Chronic illness in parents has been associated with poor social, emotional, psychological and academic outcomes in adolescents. Very little is known about the outcomes of young children with chronically ill parents who may be differently impacted.

What this study adds

  • Children of chronically ill mothers, but not fathers, are at risk of poor social, emotional, cognitive, physical and communicative development. Girls are particularly likely to experience developmental difficulties. There is a significant opportunity for health services to better support families affected by maternal chronic illness, with the goal of improving outcomes for both mothers and children.

Acknowledgments

This article does not necessarily reflect the views of the government departments involved in this research. Thank you to the people of Western Australia for the use of their administrative data. Thank you also to the WA Data Linkage Branch and all data custodians for supporting the project, providing the data and the ongoing review of project outputs. We acknowledge the excellent partnership between the custodians of the WA Department of Health and the Commonwealth Department of Education.

References

Footnotes

  • Contributors All authors devised the idea and designed the study. MFB conducted data cleaning and analysis and prepared the manuscript with input from DB, RG and JLO. All authors approve the final manuscript as submitted.

  • Funding All phases of this study were supported by the Australian Research Council (grant number LP100200507).

  • Competing interests None declared.

  • Ethics approval Ethics approval for this study was granted by the Department of Health Human Research Ethics Committee, the University of Western Australian Human Research Ethics Committee and the Western Australian Aboriginal Health Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Patient consent for publication Not required.