Maternal perinatal depression and health services utilisation in the first 2 years of life: a cohort study.

OBJECTIVES: Maternal perinatal depression is a common phenomenon, influencing infants' development. Studies have shown an inconsistent association between perinatal depression and healthcare resource utilisation. This study aimed to assess whether perinatal depression in mothers is associated with their infants' healthcare utilisation, during the first 2 years of life.
DESIGN: A cohort study based on computerised medical records.
SETTING: Nationwide primary care clinics in the second largest health maintenance organisation in Israel. PARTICIPANTS: 593 children of women with depression (the exposed group) and 2310 children of women without depression. PRIMARY AND SECONDARY OUTCOME MEASURES: Primary outcome variables included general practitioner/paediatrician (GP/Paed) visits (regular and telehealth), emergency room (ER) visits, hospital admission rates and child-development clinic visits. Secondary outcomes included antibiotic use and anaemia status. The exposure variable, perinatal depression, was based on Edinburgh Postnatal Depression Scale. A score of ≥10 was classified as depression.
RESULTS: Multivariable analysis of the number of regular visits and telehealth to the GP/Paed showed an adjusted incidence rate ratio (aIRR) of 1.08, 95% CI 1.03 to 1.13 and aIRR 0.95, 95% CI 0.82 to 1.10, respectively. Children of mothers with perinatal depression had more hospital admissions (aIRR 1.21, 95% CI 1.01 to 1.46) and more visits to child development clinics (aIRR 1.33, 95% CI 1.04 to 1.70). There was a non-significant increase in ER visits (IRR 1.26, 95% CI 0.66 to 2.42), and non-significant decrease in antibiotics prescriptions (IRR 0.95, 95% CI 0.86 to 1.05) and anaemia status (IRR 0.93, 95% CI 0.72 to 1.20).
CONCLUSION: This study shows higher health services utilisation among children of mothers with perinatal depression, including regular GP/Paed visits, hospital admission rates, and child-development clinics. © Author(s) (or their employer(s)) 2021. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

The dataset was based on a nationwide cohort, using electronic files of all patients in Maccabi Healthcare Services (MHS), the second largest health maintenance organisation in Israel.

The length of follow-up (2 years) allowed us to explore long outcomes on children of mothers with perinatal depression.

For the exposure variable, perinatal depression, we have used a validated tool, commonly used in studies of this field, the Edinburgh Postnatal Depression Scale (EPDS).

Not all women filled the EPDS in MHS, however, women who did and did not fill the EPDS were similar in most aspects, reducing the impact of selection bias.

The EPDS is a screening tool and not a diagnostic tool, leading to a non-differential misclassification bias.

Introduction

Perinatal depression is a common phenomenon affecting mothers with a reported global pooled prevalence of 17.7% and significant heterogeneity, ranging from 3% in Singapore and 38% in Chile.1 In a recenet survey conducted by the Centers for Disease Control and Prevention (CDC), about 13% of women in the United States, reported depressive symptoms after birth.2 During the perinatal period, women’s mental health may impact their functioning, parenting capability, and the health and well-being of their children.3 There are many possible influences on the infant’s development, including bonding and attachment,4 5 cognitive,6–9 emotional10 and language development,11 including IQ level.12

Parental mental health was shown to affect the healthcare resource utilisation of the child. There is an association between parental depression and a higher rate of specialty consultant visits, emergency department visits, hospital admissions and general practitioner/paediatrician (GP/Paed) visits.13 14 Timing of depression is associated with higher utilisation of primary healthcare services, with the strongest association with recent depression.15

Several studies examined the association between perinatal depression and healthcare resource utilisation. Most studies showed an association between perinatal depression and higher rates of non-routine visits (sick/emergency visits) to the GP/Paed,14–20 emergency room (ER) visits,21 hospital admission rates,19 21 specialist consultations and pharmacy claims.21 Controversy exists regarding the association between well-child visits and perinatal depression; while several studies found perinatal depression associated with decreased well-child visits,18 others did not.20 22 Anderson et al did not find any association between perinatal depression and healthcare utilisation.23 This might be explained by the substantial difference between the depressed and not depressed mothers, and because healthcare utilisation was based on self-report, which might have been subjected to measurement error.

An association was found between antidepressant use during pregnancy and higher rates of infant GP/Paed visits, specialist visits and hospital admissions.24 Screening for postpartum depression in paediatric emergency departments showed high rates of maternal depression.23 25 26 A recent study showed higher total healthcare costs in children of mothers with perinatal depression during the first 2 years of life compared with children of mothers without perinatal depression.21

This study aimed to assess whether perinatal depression in mothers is associated with their infants’ healthcare utilisation in the first 2 years of life. Variables included GP/Paed visits (regular and telehealth), ER visits, hospital admissions, child-development clinic visits and antibiotic use. Haemoglobin at the age of 12 months was also assessed to evaluate anaemia (checking haemoglobin levels at this age is recommended for all children by Israeli health policy).

Methods

Study design and setting

This cohort study is based on computerised electronic medical records data of Maccabi Healthcare Services (MHS), the second-largest health maintenance organisation (HMO) in Israel, covering 2.8 million people nationwide. In a previous study, we identified all women who filled the Edinburgh Postnatal Depression Scale (EPDS) during 2015–2016 (n=27 520), estimated 70% of all women who went to mother and child clinics in MHS. Nursing staff were responsible for delivering the EPDS to women. A score of 10 or above was classified as perinatal depression.27 In the current study, we have followed the infants of mothers with or without depression according to the EPDS until they reached the age of 2 years old. All children born to women with perinatal depression (EPDS ≥10) were defined as the exposed group. Children of mothers with an EPDS score of 9 or below were considered as the unexposed group. The sample of children was chosen with a 4:1 ratio by stratified random sampling (for each child in the exposed group, four children with the same age of mother and socioeconomic status (SES) were chosen randomly).

Variables

The exposure variable, perinatal depression, was based on the EPDS as recorded for each woman in her electronic file. The EPDS is a validated, widely used scale, that was designed specifically to assess perinatal depression.28 29 The EPDS comprises 10 questions that access information about the respondent’s mood and depressive symptoms during the 7 days preceding its administration. The response to each question is scored 0–3; thus, the highest possible score is 30. In this study, a score of 10 or above was classified as depression. Women filled the EPDS during pregnancy and until 9 weeks post partum. If filled twice (antenatal and post partum), the lower score was taken to ensure a conservative estimate. Among the women who filled the questionnaire, we know the timing for 77.9%; 29.9% filled the questionnaire during pregnancy and 48% filled the questionnaire after delivery. For those who filled the questionnaire after delivery, the median number of days after delivery was 35. Women who did not fill the EPDS were excluded from this study. Women who did and did not fill the EPDS were similar in age and similar proportions lived in the periphery of the country.

Outcome variables for healthcare services utilisation in the first 2 years of life were number of visits to the GP/Paed (both regular and telehealth), visits to the ER (direct or with referral), hospital admissions, child-development clinic visits (a referral from the GP/Paed is mandatory) and antibiotic use. Anaemia status was measured at 12 months old, as it is recommended in Israel to check anaemia status for every child at this age. Anaemia was defined as a haemoglobin level of 0.105 g/L or below. We also collected maternal data: age, periphery residence index, SES, population group (Arab, orthodox Jew, other) and smoking status.

SES is a scale measuring socioeconomic status from 1 (lowest) to 10 (highest). It is defined by the Israel bureau of statistics by address. It was grouped into three levels: lower (1-4) and higher (8-10), while the middle group (5-7) was used as a reference group. Periphery residence index is also based on patients’ addresses. Sociodemographic variables exist for all women in the database, since they are inserted automatically, and are not filled by demand from the patient. Data were collected anonymously from the electronic database of MHS.

Sample size

According to MHS data, children visit their GP/Paed once every 1–2 months during the first 2 years of life (including regular visits and telehealth). The sample size was calculated to be 192 children in the exposed group and 767 in the unexposed group, assuming a difference of one visit (from 24 to 25 visits), with a significance level of 5%, 80% power and 1:4 ratio between the two groups. However, in the current study, we included all eligible children of women who filled the EPDS questionnaire, therefore the actual sample was much higher.

Statistical analysis

Descriptive statistics were calculated using mean and SD for continuous variables and percentages for categorical variables. Univariable and multivariable negative binomial models were constructed (with sample based estimation of the overdispersion parameter) to test the effect of perinatal depression and other potential predictors of healthcare resource utilisation. Akaike’s Information Criterion was used for model selection between models that were significant according to the omnibus likelihood ratio test.

Patient and public involvement

No patients involved.

Results

Descriptive statistics

Data were collected from the electronic records of MHS for 593 children of women with depression (the exposed group) and 2310 children for the unexposed group (negative EPDS, no perinatal depression) matched by maternal age and SES. The total study population consisted of 2903 children. Both groups were similar in sex distribution, mothers’ country of birth, and religious group (Arab or orthodox Jewish) (table 1). Mothers with perinatal depression were more likely to smoke during pregnancy (relative risk 1.51, 95% CI 1.19 1.94).

Table 1

Demographic and clinical characteristics of children and mothers according to maternal, perinatal depression, defined by the Edinburgh Postpartum Depression Scale

Characteristics	With perinatal depression n=593	Without perinatal depression n=2310
	N (%)	N (%)
Children
Sex
Male	293 (49.4)	1237 (53.5)
Female	300 (50.6)	1073 (46.5)
Mothers
Age
Mean±SD	33.4±0.24	32.8±0.12
Born in Israel
Yes	505 (85.2)	1994 (86.3)
No	88 (14.8)	316 (13.7)
Smoking status
Yes	73 (12.3)	173 (7.5)
No	520 (87.7)	2137 (92.5)
Children and mothers
Living in a peripheral region
Yes	38 (6.4)	206 (8.9)
No	555 (93.6)	2104 (91.1)
Population group
Arab	49 (8.3)	169 (7.3)
Orthodox Jewish	30 (5.1)	126 (5.5)
Others	514 (86.7)	2015 (87.2)
Socioeconomic status
1–4	124 (20.9)	501 (21.7)
5–7	325 (54.8)	1227 (53.1)
8–10	144 (24.3)	583 (25.2)

Univariable analysis

In a univariable analysis, children of women with perinatal depression had more regular visits to their GP/Paed (incidence rate ratio (IRR) 1.07, 95% CI 1.02 to 1.13, p value 0.005) but fewer telehealth visits (IRR 0.96, 95% CI 0.83 to 1.12, p value 0.653). They had more ER visits (IRR 1.26, 95% CI 0.66 to 2.42, p value 0.482), hospital admissions (IRR 1.19, 95% CI 1.00 to 1.43, p value 0.053) and visits to child development clinics (IRR 1.31, 95% CI 1.03 to 1.68, p value 0.03), but receive less antibiotics prescriptions (IRR 0.95, 95% CI 0.86 to 1.05, p value 0.313) and have less anaemia (IRR 0.93, 95% CI 0.72 to 1.20, p value 0.572) (table 2).

Table 2

Primary outcomes—children’s healthcare service utilisation during the first 2 years of life, univariable analysis

Characteristics	With perinatal depression n=593	Without perinatal depression n=2310	Incidence rate ratio (95% CI)	P value
Number of regular visits to general practitioner/paediatrician mean±SD	28.74±0.60	26.8±0.31	1.07 (1.02 to 1.13)	0.005
Number of telehealth visits to the GP/Paed mean±SD	3.78±0.26	3.91±0.12	0.96 (0.83 to 1.12)	0.653
Emergency room visits (ever)	12 (2%)	37 (1.6%)	1.26 (0.66 to 2.42)	0.482
Hospital admissions (ever)	154 (26%)	502 (21.7%)	1.19 (1.00 to 1.43)	0.053
Visits to child development clinics (ever)	85 (14.3%)	252 (10.9%)	1.31 (1.03 to 1.68)	0.03
Number of prescriptions for antibiotics mean±SD	2.49±0.11	2.61±0.06	0.95 (0.86 to 1.05)	0.313

GP/Paed, general practitioner/ paediatrician.

Multivariable analysis

Multivariable analysis showed that children of mothers with perinatal depression have more regular visits to the GP/Paed (adjusted IRR (aIRR) 1.08, 95% CI 1.03 to 1.13, p value 0.002) (figure 1). Arab or orthodox Jewish children were less likely to go to their GP/Paed (aIRR 0.86, 95% CI 0.79 to 0.94, p value 0.001 and aIRR 0.76, 95% CI 0.69 to 0.84, p value <0.001, respectively). Males and children of mothers who were born in Israel were more likely to go to their GP/Paed (aIRR 1.08, 95% CI 1.04 to 1.12, p value <0.001 and aIRR 1.06, 95% CI 1.00 to 1.12 p value 0.039, respectively). In contrast, no difference was demonstrated for telehealth visits for children born to mothers with perinatal depression (aIRR 0.95, 95% CI 0.82 to 1.10, p value 0.491) (figure 2). Children of mothers who live in the periphery or are Arab or orthodox Jewish were less likely to use this method of communication with their GP/Paed (aIRR 0.75, 95% CI 0.60 to 0.93, p value 0.009; aIRR 0.21, 95% CI 0.16 to 0.29, p value <0.001; aIRR 0.18, 95% CI 0.12 to 0.25, p value <0.001respectively).

Figure 1

Regular visits to the general practitioner/paediatrician, multivariable analysis.

Figure 2

Telehealth visits to the general practitioner/paediatrician, multivariate analysis.

Three factors were found significant in hospital admissions; children of mothers with perinatal depression were more likely to be admitted to the hospital (aIRR 1.21, 95% CI 1.01 to 1.46, p value 0.034). Lower rates of admissions were recorded for females (aIRR 0.82, 95% CI 0.70 to 0.96, p value 0.014) and children of older mothers (aIRR 0.98, 95% CI 0.97 to 0.99, p value 0.026).

For child development clinic visits, two significant factors were found; children of mothers with perinatal depression and males had more of these visits, aIRR 1.33, 95% CI 1.04 to 1.70, p value 0.024 and aIRR 1.30, 95% CI 1.05 to 1.62, p value 0.016, respectively.

For the number of prescriptions for antibiotics, perinatal depression was not found significant in the multivariate analysis. Males and Jewish orthodox background were less likely to get prescriptions (aIRR 0.83, 95% CI 0.77 to 0.89, p value <0.001 and aIRR 0.71, 95% CI 0.59 to 0.85, p value <0.001, respectively). Children of mothers born in Israel were more likely to get prescriptions (aIRR 1.12, 95% CI 1.01 to 1.26, p value 0.05). Children from lower SES groups were more likely to get prescriptions, and children from higher SES groups were less likely to get prescriptions (aIRR 1.18, 95% CI 1.06 to 1.31, p value 0.004 and aIRR 0.83, 95% CI 0.75 to 0.91, p value <0.001 respectively).

Discussion

Main findings

This study suggests that perinatal depression affects children’s healthcare utilisation from birth to 2 years old. These children more frequently visit their GP/Paed’s and child development clinics and have higher hospital admission rates. There was also a non-significant increase in ER visits. However, there was no significant difference in the rates of telehealth visits or the number of antibiotic prescriptions issued.

Strengths

The strengths of the study include the dataset which is based on a nationwide cohort, using electronic files of all patients in MHS, the second largest HMO in Isarel. All HMOs in Israel have fully computerised healthcare systems, and therefore sociodemogrphic data were available for all participants in the study. The length of follow-up allowed us to explore long outcomes on children of mothers with perinatal depression. For the exposure variable, perinatal depression, we have used a validated tool, commonly used in studies of this field.

Weaknesses

The sample was based on the filled EPDS questionnaires in the electronic database of MHS. Not all women filled the EPDS. However, women who did and did not fill the EPDS were similar in most aspects27 such as age (33.2±5.63 vs 32.4±6.17, respectively) or women who live in the periphery (8% vs 7.5%, respectively). This similarity reduces the impact of selection bias. Second, the EPDS is a screening tool and not a diagnostic tool, leading to a non-differential misclassification bias. Third as this study is based on electronic data mining, information bias may exist, due to missing data such as private visits, outside the HMO. However, this is not likely, as in Israel, GP/Paed visits are fully reimbursed.

Interpretation

In this study, we report an increase of 8% in the number of regular visits to the GP/Paed by children of mothers with perinatal depression. Many studies have differentiated between sick/emergency visits and well-child visits. While the first is increased for children of mothers with perinatal depression in most studies the second is more controversial. In this study, we could not differentiate between the two visit types. In addition to frontal visits, we also examined telehealth visits. To our knowledge, this is the first study that examined this type of visits in the context of health services utilisation. We report no significant difference for telehealth visits of children born to mothers with or without perinatal depression.

We did not have data about parity and the place of the child in relation to his or her siblings. This might be an important variable that could confound the findings.

Mandl et al suggested the excessive use of healthcare services may be a sign of help-seeking behaviour of the mother or a sign of anxiety to the well-being of her child.17 It has been suggested that some women might find it easier to seek help for their child rather than for themselves.16 On the other hand, mothers with perinatal depression may feel unconfident about their children’s health issues, thus bringing them more quickly to see their GP/Paed.15 Non-specific baby-related complaints may reflect maternal distress (crying, irritability, feeding or weight problems).19 Another explanation could suggest that mothers of infants with more health problems may be more prone to depression.16 However, in this study, we identified children who were born following maternal, perinatal depression. While children of mothers with perinatal depression significantly visit their GP/Paed more often, telehealth visits are less used by these mothers. This may also reflect the more soothing effect of a regular visit vs the distant and technical telehealth visit.

Higher rates of visits to child-development clinics may result from several factors. Children of mothers with perinatal depression show higher rates of developmental problems, including cognitive, emotional and language development.30 Furthermore, it may be related to the more worried mother as a reflection of her depression.

Minorities in Israel, such as Arabs and orthodox Jewish populations, were shown to visit the GP/Paed less frequently, in both regular visits and telehealth. Orthodox Jewish children also received fewer antibiotic prescriptions during the first 2 years of life. As there are no restrictions to consumption for these population groups, these findings may be explained in several aspects. We hypothesise that as these populations tend to have larger families, with more children than the average Israeli family,31 32 they may be more experienced with raising children and dealing with health problems. These populations also have a high level of community support that may help when minor health problems arise. However, these assumptions should be investigated in future studies.

Conclusion

This study shows higher health services utilisation among children of mothers with perinatal depression, including GP/Paed visits, hospital admissions and child-development clinics. This emphasises the need to increase awareness of perinatal depression in mothers, thus reducing the burden of their children’s health service utilisation. Further studies are needed to examine if increased awareness and early diagnosis of mothers’ perinatal depression reduce the use of their children’s health services.