Depressive Symptoms in Parents of Children With Chronic Health Conditions

A Meta-analysis

Martin Pinquart, PHD


J Pediatr Psychol. 2019;44(2):139-149. 

In This Article


The 460 studies provided data on 58,290 parents of children with chronic conditions. About 78.4% of the parents were mothers, 81.3% were married, and the parents had a mean age of 37.78 years (SD = 3.64). The children had a mean age of 8.88 years (SD = 3.14) and 46.7% of them were girls. The children most often had cancer (N = 10,987), asthma (N = 6,763), cystic fibrosis (N = 6,132), epilepsy (N = 4,411), diabetes (N = 4,459; predominantly type I diabetes), heart disease (N = 3,772), and cerebral palsy (N = 2,526). Their health condition had been diagnosed, on average, 3.82 years before the assessment of parental depression (SD = 2.99). Only 21 studies provided longitudinal data. The included studies have been published between 1959 and 2018, with 51.2% of the studies being published after 2007. Parental depression was assessed with versions of the Beck Depression Inventory (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961; 151 studies), the Center for Epidemiological Studies Depression Scale (Radloff, 1977; 72 studies), the Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983; 60 studies), versions of the Symptom Checklist by Derogatis (e.g., SCL-90R; Derogatis, 1977; 58 studies), structured clinical interviews (14 studies), and related instruments (105 studies, see Supplementary Appendix 2).

Overall Elevations of Depressive Symptoms and Rates of Clinical Depression

With regard to the first research question, we found, on average, small to moderate elevations of depressive symptoms among families with a child with a chronic condition when compared against parents of healthy children or test norms (g = .46 SD units; Table I). This effect size indicated that 5.0% of the interindividual variability of depressive symptoms could be explained by having a child with a chronic condition. The effect size was heterogeneous, thus indicating the need to search for moderator variables.

Twelve studies reported prevalence rates of depressive disorders based on structured clinical interviews (targeting parents of children with asthma, severe burns, cancer, cleft lip and palate, chronic pain, cystic fibrosis, epilepsy, inflammatory bowel disease, and liver disease). These studies provide a weighted mean prevalence rate of depressive disorders of 20.9%. For comparing rates of clinical depression in these parents with prevalence rates in the general population, we weighted the global prevalence rates of mood disorders in men (4.0%) and women (7.3%; Steel et al., 2014) with the percentages of mothers and fathers of children with a chronic health condition who had been assessed for clinical depression. This comparison indicates that the prevalence rate of clinical depression in these parents was 2.98 times higher than the rate in the general population.

Moderator Effects of Study Characteristics

Condition-specific analyses were computed if at least four effect sizes were available for an individual condition. With regard to the second research question, we found that depressive symptoms varied between chronic conditions (Table II). Moderate elevations were found in parents of children with neuromuscular disorders (e.g., Duchenne muscular dystrophy; g = .75), cancer (g = .63), cerebral palsy (g = .60), inherited hematological diseases, such as thalassemia major and sickle cell disease (g= .60), chronic pain (e.g., migraine; g = .51), and spina bifida (g = .50). Small elevations of depressive symptoms were found in in parents of children with asthma (g = .46), kidney/liver/renal diseases (g = .44), epilepsy (g = .43), cystic fibrosis (g = .43), metabolic diseases (such as phenylketonuria, g = .41), gastroenterological diseases (g = .40), cardiovascular diseases (g = .35), sensory impairment (g = .35), obesity (g = .33), skin diseases (e.g., eczema, g = .32), and diabetes (g = .23). Furthermore, there were very small, yet statistically significant elevations of depressive symptoms if the child had arthritis/rheumatism (g = .19). In contrast, no significant elevations of depressive symptoms were found in parents of children with food allergies, cleft lip and palate, or HIV infection/AIDS. The largest effect size indicates that 12.3% of the interindividual variability in depressive symptoms could be explained by having a child with a neuromuscular disorder.

The nonoverlap of the 95% CIs indicates that elevations of parental depressive symptoms were greater in the cases of cancer than in the cases of arthritis, cleft lip and palate, diabetes, and cardiovascular diseases. Effect sizes were heterogeneous in 3 of 21 chronic conditions (cancer, cleft lip and palate, and cardiovascular diseases). Additional analyses showed that heterogeneity of effect sizes in parents of children with cancer, cardiovascular diseases, and cleft lip and palate could be explained by differences in duration of the chronic condition: comparisons of parents of children with duration of the chronic condition above versus below the median indicate significantly stronger elevations of depressive symptoms in the case of shorter duration, cancer: g = .58 versus g = .36, Q(1) = 4.40, p < .05; cleft lip: g = .42 versus g = −.41, Q(1) = 4.41, p < .05. There was a similar trend among the parents of children with cardiovascular diseases, g = .44 versus g = .22, Q(1) = 2.88, p < .10, although the difference was only marginally significant. Because longer duration of cancer is often associated with the successful completion of therapy, we also compared levels of depression in parents of completers and parents of children who were still in active treatment. Depression levels varied between both groups, Q(1) = 5.05, p < .02, with parents of completers reporting lower levels of depressive symptoms (g = .41, based on 36 samples) than parents of children who were in active treatment (g = .66, based on 76 samples). All effect sizes in the subgroups with shorter versus longer duration of the condition and active treatment versus completion were homogeneous.

When testing the moderator effect of the duration of the chronic condition, we excluded studies on progressive conditions, such as HIV infection/AIDS, cystic fibrosis, and muscular dystrophy. In the remaining studies, we found that a longer duration of the condition was associated with smaller elevations of parental depressive symptoms (Table II). We did not have enough studies on moderator effect of the duration of progressive conditions, as the duration was often not reported in these studies.

Between-group differences in parental depressive symptoms did not vary by the child's age. However, there were greater between-group differences if the sample included larger percentages of mothers. In contrast, between-group differences did not vary by marital status (Table II).

The meta-analysis also identified a moderator effect of county of residence (Table I). Between-group differences in depressive symptoms were greater if the families lived in an economically less developed country (g = .76) rather than in a developed country, such as the United States or in Western Europe (g = .37). In contrast to studies from developed countries, effect sizes varied across studies in the less developed countries.

The sum measure of study quality did not moderate the size of the observed between-group differences. Follow-up analyses (not shown) also did not find moderator effects of the individual criteria of study quality. Effects sizes of published and unpublished studies did also not differ significantly. Egger's test did not indicate funnel plot asymmetry, t(723) = .28, and the trim-and-fill analysis did not find evidence for a possible overestimation of the effect size because of publication bias.