Matched cohort study of healthcare resource utilization and costs in young children of mothers with postpartum depression in the United States

Abstract

Objective: To assess healthcare resource utilization (HRU) and costs in children of mothers with and without postpartum depression (PPD).

Methods: Administrative claims data from the IBM Watson Health MarketScan Databases (2010–2016) were used. Women with live births (index date = delivery date) were identified and linked to their newborns. The mother–child pairs were divided into PPD and non-PPD exposure cohorts based on claims for depression, mood or adjustment disorders, or anxiety identified in the mother between 15 and 365 days after delivery. Mother–child pairs with PPD exposure were propensity score matched 1:3 to mother–child pairs without PPD exposure. Children were required to have 24 months of continuous health plan enrolment following delivery. Additional comparisons were performed between mother–child pairs with and without preterm delivery.

Results: Overall, 33,314 mother–child pairs with PPD exposure were propensity score matched to 102,364 mother–child pairs without PPD exposure. During the 24-month follow-up period, HRU across most service categories was significantly higher among children in the PPD exposure cohort than non-PPD exposure cohort. Among outpatient services, the percentages of children with a physician specialist service (68% versus 64%), early-intervention screening (40% versus 37%), and an emergency room visit (48% versus 42%) were greater in children of mothers with PPD (all p < .001). Furthermore, children of mothers with PPD incurred 12% higher total healthcare costs in the first 24 months of life compared to children of mothers without PPD ($24,572 versus $21,946; p < .001). After excluding mothers with preterm delivery, the proportion of children with ER visits, physician specialist services, and outpatient pharmacy claims was significantly higher in the PPD exposure cohort than non-PPD exposure cohort (all p < .001).

Conclusion: The results of this analysis suggest that HRU and costs over the first 24 months of life in children of mothers with PPD exceeded that of children of mothers without evidence of PPD.

Keywords:

• Administrative claims analysis
• healthcare resource utilization
• healthcare costs
• children
• mothers
• postpartum depression

JEL CLASSIFICATION CODES:

• I11
• I12

Introduction

Postpartum depression (PPD), characterized by clinically significant depressive symptoms lasting at least 2 weeks, is a mood disorder that affects women after childbirth¹. Between 8% and 20% of United States (US) women experience PPD². Onset of PPD, defined as a major depressive episode, may occur during pregnancy or in the post-partum period^1,3,4. The exact etiology of PPD remains elusive but likely results from the complex interplay of psychological, social, and biological factors^5,6. Significant risk factors for PPD include a history of mood disorders, stressful life events during pregnancy, lack of support from partner, spouse, friends, or family, low socioeconomic status, pre-term birth, and other obstetric complications^1,7.

Several pharmacological, psychological and psychosocial treatments are used for treating PPD^7–11, depending on symptom severity, level of functional impairment¹², and patient preference¹³. However, numerous barriers prevent women from receiving care¹⁰. Data on the impact of PPD on maternal and child health are well documented^8,14–17. Decreased maternal and infant bonding, as well as reduced breastfeeding initiation rates and duration, have been associated with PPD^15,18. Low birthweight, poorer child growth outcomes, delayed/impaired cognitive, emotional, and linguistic development, long-term mental health problems, and negative behavioral, social, and emotional outcomes, have been reported in infants/children with mothers having PPD, all of which can contribute to healthcare costs^{8,14,16,19,20}.

Despite associations between maternal PPD and increased infant morbidity, the few studies on the effects of mother’s PPD status on children’s healthcare resource utilization (HRU) have mixed findings^21–25. This study extends previous research on maternal PPD and HRU in children by examining real-world HRU and costs in the first 24 months of life among children with maternal PPD exposure in large, geographically diverse populations of mother–child pairs, covered by multiple insurance plan types. This study focused on the postpartum period to identify maternal PPD; mothers with PPD were identified in the 15–365 days following delivery, children were evaluated through the first 24 months of life.

Methods

Study design and data source

This was a retrospective, observational cohort study based on US administrative claims data contained in the IBM Watson Health MarketScan Commercial Claims and Encounters (Commercial) database spanning 1 January 2009 to 31 December 2016. The Commercial database contains a large convenience sample of commercially insured individuals and their dependents, covered under a variety of health plans across the USA²⁶. Included in this database are the procedure codes, diagnosis codes, and pharmacy codes associated with each fully adjudicated inpatient and outpatient medical claim, along with the outpatient pharmacy codes for each filled prescription. All study data were obtained using International Classification of Diseases, 9th and 10th Revision, Clinical Modification (ICD-9-CM and ICD-10-CM) codes, Current Procedural Terminology codes, Healthcare Common Procedure Coding System codes, and National Drug Codes. All database records are statistically de-identified and certified to be fully compliant with US patient confidentiality requirements set forth in the Health Insurance Portability and Accountability Act of 1996. Because this study used only de-identified patient records and did not involve the collection, use, or transmittal of individually identifiable data, Institutional Review Board approval to conduct this study was not necessary.

Sample selection

Women aged 15–50 years with at least one inpatient medical claim with a diagnosis code or procedure code for a delivery between 1 January 2010 and 30 September 2015 with continuous enrollment in commercial insurance plans with medical and pharmacy benefits for 12 months before (baseline) and after (follow-up) the index date (the date of delivery), were eligible for inclusion. For mothers linked to multiple children due to subsequent pregnancies (a diagnosis code or procedure code for delivery ≥6 months after the prior delivery), only the first pregnancy during the study period was included. Women with evidence of bipolar disorder, schizophrenia/schizoaffective disorder, or psychosis or any pharmacy claim for an antipsychotic medication during the 12-month baseline period, women with multifetal gestations (twins, triplets, quadruplets, etc.), and those with a family member with a delivery on the same date, were excluded.

Children with a matching family identifier and date of birth within 30 days after the index date were linked to eligible mothers and were required to have 24 months of continuous enrollment following delivery. The 30 day allowable gap between date of birth and date of delivery was selected a priori because parents can add children to their health plans in the 30 days following birth. Children linked to more than one mother or coded as any relation to an employee other than dependent were excluded.

Mother–child pairs were stratified into two cohorts based on the mother’s PPD status. Mother-child pairs where the mother met at least one of the criteria (a combination of diagnosis code and/or acquiring treatment for depression, mood or adjustment disorders or anxiety) between 15 and 365 days after the index date were included in the study (Figure 1). Patients with anxiety were required to have either a diagnosis of depression or evidence of treatment for depression in order to be included in the PPD cohort; no patients with anxiety diagnosis without evidence of depression were included in the PPD cohort. Adjustment disorders were included so that a broad range of depressive symptoms were included when identifying PPD, which like other mental health conditions, may be underreported particularly in a claims database. Other researchers have included adjustment disorders in the criteria for PPD in other studies of PPD using administrative claims^27,28 and the inclusion of adjustment disorder diagnosis codes when identifying patients with depression in claims has been validated²⁹. To avoid the inclusion of mothers experiencing depressive symptoms in the first 2 weeks of life, when the mother may be experiencing “baby blues”, patients were required to meet the PPD criteria between 15 and 365 days after the index date to be included in the PPD cohort. Women experiencing normal hormonal changes following delivery and depressed mood as a result were only included if the PPD criteria were met in the time period after the first 2 weeks following delivery. The PPD diagnosis date was defined as the earliest date where the mother had a claim meeting the PPD criteria in Figure 1. All mother–child pairs ineligible for inclusion in the PPD exposure cohort were included in the non-PPD exposure cohort. Comparisons were performed among the matched cohorts with presentation among all matched mother–infant pairs and additionally among mother–infants without preterm delivery.

Figure 1. Patient selection. Abbreviations. ECT, Electroconvulsive therapy; IP, Inpatient; OP, Outpatient; PPD, Postpartum depression; SNRIs, Serotonin and norepinephrine reuptake inhibitors; SSRIs, Selective serotonin reuptake inhibitors; TMS, Transcranial magnetic stimulation. ^aNewborns must have been linked to one mother, listed as child in relation to mother, and have had date of birth after index date and date of birth ≤1 day apart where ≥2 newborns. Additionally, mother and family member should not have had the same delivery date.

Matching algorithms

Propensity score was computed using logistic regression models, with evidence of PPD as the dependent variable and relevant baseline demographic (maternal age, geographic region, urban/rural residence, health plan type, and relationship to employee) and clinical characteristics (Deyo-Charlson comorbidity index [DCCI, categorized as 0, 1, 2, 3, and 4 or greater], type of delivery, presence of pre-term delivery, season of delivery, and total baseline non-mental health medical costs [centered and scaled]) as independent variables. Once propensity scores were computed for each mother–child pair, each PPD exposed pair was matched to 3 non-PPD exposed controls using the nearest neighbor technique. Standardized differences in matching factors between the PPD exposure and non-PPD exposure cohorts were calculated to examine the quality of the match, with standardized differences <.25 indicative of an acceptable match³⁰. Matching was done using the MatchIt package in R version 3.5.1 (R Foundation for Statistical Computing, Vienna, Austria)³¹.

Outcome measures

All-cause HRU and costs of children were measured during the 24-month post-index period and reported by maternal PPD status. Specific utilization measures assessed are described in Table 3. Early-intervention screenings included those for developmental delays including mental, behavioral, neurodevelopmental, speech, language, hearing, vision, and metabolic disorders. Healthcare costs were computed as paid amounts of adjudicated claims and included insurer payments (including coordination of benefits), health plan payments, and patient cost-sharing (copayment, deductible, and co-insurance) components. All healthcare costs were expressed in 2016 constant US dollars, adjusted using the Medical Care component of the Consumer Price Index (http://www.bls.gov/cpi/).

Other variables

Maternal demographic and clinical characteristics were measured on the index date and during the 12-month baseline period, respectively (Table 1). Children’s demographic and clinical characteristics were assessed on the index date and during the 24-month post-index period, respectively (Table 2).

Table 1. Maternal baseline characteristics: matched PPD exposure versus non-PPD exposure cohorts.

	PPD exposure cohort	Non-PPD exposure cohort	p
	n = 33,314	n = 102,364
Maternal age, years, mean (SD)	32.0 (4.6)	32.0 (4.6)	.767
Maternal age categories, n (%)
15–18	51 (0.2%)	180 (0.2%)
19–24	1,451 (4.4%)	4,376 (4.3%)
25–34	22,087 (66.3%)	68,144 (66.6%)
35–44	9,601 (28.8%)	29,319 (28.6%)
45–50	124 (0.4%)	345 (0.3%)
Geographic region, n (%)			.857
Northeast	6,891 (20.7%)	20,944 (20.5%)
North Central	8,629 (25.9%)	26,421 (25.8%)
South	10,743 (32.2%)	33,579 (32.8%)
West	6,844 (20.5%)	20,887 (20.4%)
Unknown	207 (0.6%)	533 (0.5%)
Population density, n (%)			.729
Urban	29,634 (89.0%)	90,958 (88.9%)
Rural	3,480 (10.4%)	10,887 (10.6%)
Unknown	200 (0.6%)	519 (0.5%)
Insurance plan types noted in >5%, n (%)^a			.996
PPO	20,372 (61.2%)	62,819 (61.4%)
HMO	5,148 (15.5%)	15,638 (15.3%)
POS	2,519 (7.6%)	7,969 (7.8%)
CDHP	2,238 (6.7%)	6,800 (6.6%)
Relationship to employee, n (%)			.977
Self	17,401 (52.2%)	53,852 (52.6%)
Spouse	15,724 (47.2%)	47,926 (46.8%)
Child/Other	189 (0.6%)	586 (0.6%)
DCCI, mean (SD)	0.1 (0.4)	0.1 (0.4)	.012
Any mental health disorder^b	13,048 (39.2%)	7,867 (7.7%)	<.001
Mental health disorders complicating pregnancy	927 (2.8%)	1,306 (0.4%)	<.001
Anxiety^c	5,202 (15.6%)	2,690 (2.6%)	<.001
Depression	3,920 (11.8%)	1,563 (1.5%)	<.001
Episodic mood disorder	3,348 (10.0%	2,290 (0.7%)	<.001
Major depressive disorder	3,175 (9.5%)	714 (0.7%)	<.001
Mood or adjustment disorder	3,458 (10.4%)	1,723 (1.7%)	<.001
Substance abuse^d	745 (2.2%)	2,169 (0.7%)	<.001
Suicide attempt/ideation	27 (0.1%)	44 (0.0%)	<.001
Attention deficit / hyperactivity disorder	291 (0.9%)	1,198 (0.4%)	<.001
Pain related to psychological factors	564 (1.7%)	1,627 (0.5%)	<.001
Delivery characteristics, n (%)
C-section	13,520 (40.6%)	41,324 (40.4%)	.786
Preterm delivery	2,268 (6.8%)	7,118 (7.0%)	.990
Season of delivery			.998
January–March	8,327 (25.0%)	25,795 (25.2%)
April–June	8,640 (25.9%)	26,778 (26.2%)
July–September	8,678 (26.0%)	25,965 (25.4%)
October–December	7,669 (23.0%)	23,826 (23.3%)
Total non-mental health medical costs, mean (SD)^b	$7,816 ($10,160)	$7,666 ($10,972)	<.001

Abbreviations. CDHP, Consumer-directed health plan; DCCI, Deyo-Charlson comorbidity index; EPO, Exclusive provider organization; HDHP, High-deductible health plan; HMO, Health maintenance organization; p, Statistical significance; POS, Point of service; PPD, Postpartum depression; PPO, Preferred provider organization; SD, Standard deviation.

^a Other insurance plan types noted among ≤5% of PPD exposure and non-PPD exposure cohorts were comprehensive, EPO, POS with capitation, HDHP and missing/unknown entries.

^b Maternal behavioral health disorders and non-mental health medical costs were measured during the 12-month baseline period prior to index date.

^c Anxiety disorder diagnoses included anxiety state, unspecified; panic disorder without agoraphobia; generalized anxiety disorder; and other anxiety states.

^d Substance abuse excludes tobacco and antidepressant use and/or abuse.

Table 2. Demographic and clinical characteristics of children over 24-month follow-up period: matched PPD exposure versus non-PPD exposure cohorts.

	PPD exposure cohort	Non-PPD exposure cohort	p	Absolute difference
	n = 33,314	n = 102,364
Male, n (%)	17,193 (51.6%)	52,285 (51.1%)	.092	0.5%
Age, days, mean (SD)^a	134.7 (98.5)	–
Age, n (%)^a
<1 month	4,670 (14.0%)	–
1–3 months	10,043 (30.2%)	–
3–6 months	8,105 (24.3%)	–
6–12 months	10,496 (31.5%)	–
Clinical characteristics noted in >5%, n (%)^b
Low birthweight^c	2,157 (6.5%)	6,506 (6.4%)	1.000	0.1%
Major and minor congenital anomalies	9,066 (27.2%)	25,995 (25.4%)	<.001	1.8%
Feeding problems	12,172 (36.5%)	34,259 (33.5%)	<.001	3.1%
Fussiness/excessive crying	4,336 (13.0%)	9,322 (9.1%)	<.001	3.9%
Vomiting	5,793 (17.4%)	14,702 (14.4%)	<.001	3.0%
Special screening for developmental handicaps	1,927 (5.8%)	5,807 (5.7%)	1.000	0.1%
Vaccinations to the child, n (%)	32,134 (96.5%)	98,634 (96.4%)	1.000	0.1%

Abbrevaitions. p, Statistical significance; PPD, Postpartum depression; SD, Standard deviation.

^a Age is measured on the maternal PPD diagnosis date (earliest date where mothers meet PPD criteria).

^b Other clinical characteristics noted in ≤5% of children in either groups were abnormal weight gain, delayed milestones, physical retardation, child physical abuse, and child abandonment or neglect.

^c Low birth weight was measured on index date.

Statistical analysis

All study variables were analyzed descriptively for both the study cohorts. Data were stratified by the mother’s PPD status (presence/absence). Categorical variables were summarized as count and percentages. Continuous variables were reported as means and standard deviations (SD). Statistical comparisons between the matched cohorts were evaluated using paired t-tests, Wilcoxon sign rank tests, or McNemar tests as appropriate. The p-values were adjusted for multiple comparisons using the Holm Bonferroni method³². An overall α < .05 (after adjustment) was considered, a priori, statistically significant. A sensitivity analysis excluding preterm deliveries from the matched cohorts was conducted to test the robustness of the findings from the main analysis. Descriptive analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC).

Results

Study sample

A total of 365,897 patients met the study eligibility criteria and were appropriately linked to a child. After propensity score matching, 33,314 PPD exposed and 102,364 non-PPD exposed control mother–child pairs were included (Figure 1).

Demographic and clinical characteristics

Maternal

After matching, mothers in PPD exposure and non-PPD exposure cohorts had similar demographic and clinical characteristics. The mean age among cases and controls was 32.0 years (SD = 4.6 years) and mean DCCI was similar (0.1 [SD = 0.4]) across both the cohorts. The prevalence of a mental health diagnosis during the baseline period was significantly higher in the matched PPD exposure cohort than the matched non-PPD exposure cohort (39.2% vs. 7.7%, p < .001). The most common mental health diagnoses in the PPD exposed mothers during the 12-month baseline period, including the full length of pregnancy, were anxiety (15.6%), depression (11.8%), major depressive disorder (9.5%) and mood or adjustment disorder (10.4%). In the matched non-PPD exposure cohort, the most common diagnoses were anxiety (2.6%), depression (1.5%), and mood/adjustment disorder (1.7%). All mental health disorders examined were significantly more prevalent in the PPD exposure cohort than the non-PPD exposure cohort (all p < .001) (Table 1).

Children

Within the matched PPD exposure cohort, over half of the children were male and 44% of children were 0–3 months of age when the mother was first diagnosed with PPD: 14.0% were <1 month, 30.2% were 1–3 months, 24.3% were 3–6 months, and 31.5% were 6–12 months of age (Table 2). The three clinical outcomes with the greatest difference between the PPD versus non-PPD exposure cohorts were fussiness/excessive crying (13.0% vs. 9.1%; p < .001), feeding problems (36.5% vs. 33.5%; p < .001), and vomiting (17.4% vs. 14.4%; p < .001). Major and minor congenital anomalies were also significantly higher among children of mothers with PPD than in children of mothers without PPD (27.2% vs. 25.4%; p < .001).

HRU and costs among children

There were significant differences in most utilization measures between the two cohorts. Specifically, more children of mothers with PPD had ER visits (48.4% vs. 41.8%; p < .001), outpatient physician specialist services (67.9% vs. 64.0%; p < .001), early-intervention screening (39.9% vs. 37.1%; p < .001) and outpatient pharmacy claims (92.2% vs. 89.0%; p < .001) compared to children of mothers without PPD. Children of mothers with PPD had a significantly higher mean number of ER visits, all-cause physician office visits, and pharmacy claims than children of mothers without PPD. Similar proportions of children in PPD exposure and non-PPD exposure cohorts had any well-child office visit over the 24-month follow-up period (PPD: 99.6%, non-PPD: 99.5%). Similarly, vaccination rates in the follow-up period were similar among children of mothers with PPD (96.5%) and children of mothers without PPD (96.4%, p = 1.00; Table 3). Though the proportion of children with an outpatient office visit was similar across both the cohorts, children of mothers with PPD had significantly more visits (10.2 vs. 9.2; p < .001; Table 3). The most common prescriptions across both PPD exposure and non-PPD exposure cohorts were antibiotics (79.9% and 74.4% respectively), anti-infectants (61.2% and 56.1%), and sympathomimetic medications (30.0% and 27.5%) (Figure 2).

Figure 2. Most commonly prescribed medications among children over 24-month follow-up period: matched PPD exposure versus non-PPD exposure cohorts. *p < 0.001 for PPD exposure versus non-PPD exposure cohorts. Abbreviations. PPD, Postpartum depression. Sympathomimetic drugs included acetaminophen, albuterol, aspirin, ephedrine, epinephrine, and ibuprofen.

Table 3. All-cause healthcare resource utilization among children over 24-month follow-up period: matched PPD exposure versus non-PPD exposure cohorts.

	PPD exposure cohort	Non-PPD exposure cohort	p	Absolute difference
	n = 33,314	n = 102,364
Inpatient admissions
Any inpatient admission, n (%)	24,698 (74.1%)	76,063 (74.3%)	1.000	−0.2%
Number of admissions, mean (SD)	0.9 (0.7)	0.8 (0.7)	.004	0.02
Length of stay, mean (SD)	2.8 (6.9)	2.6 (6.4)	.431	0.13
Any physician specialist service, n (%)	10,458 (31.4%)	32,055 (31.3%)	1.000	0.1%
Outpatient services
Any outpatient service, n (%)	33,282 (99.9%)	102,244 (99.9%)	1.000	0.0%
Any physician specialist service, n (%)	22,603 (67.9%)	65,472 (64.0%)	<.001	3.9%
Emergency room visits
Any emergency room visit, n (%)	16,109 (48.4%)	42,822 (41.8%)	<.001	6.5%
Number of visits, mean (SD)	1.0 (1.5)	0.8 (1.4)	<.001	0.19
Physician office visits
Any physician office visit, n (%)	33,225 (99.7%)	102,083 (99.7%)	1.000	0.0%
Number of visits, mean (SD)	20.8 (10.2)	19.0 (9.2)	<.001	1.78
Any well-child office visit, n (%)	33,163 (99.6%)	101,843 (99.5%)	1.000	0.1%
Number of visits, mean (SD)	8.2 (1.9)	8.2 (2.0)	<.001	−0.06
Any early-intervention screening, n (%)^a	13,283 (39.9%)	37,924 (37.1%)	<.001	2.8%
Other outpatient services
Any other outpatient service, n (%)	33,238 (99.8%)	102,033 (99.7%)	1.000	0.10%
Number of services, mean (SD)	16.0 (18.5)	14.5 (14.9)	<.001	1.5
Outpatient pharmacy
Any outpatient pharmacy claim, n (%)	30,701 (92.2%)	91,094 (89.0%)	<.001	3.20%
Number of claims, mean (SD)	9.9 (10.8)	8.3 (9.6)	<.001	1.6

Abbreviations. p, Statistical significance; PPD, Postpartum depression; SD, Standard deviation.

Mean number of visits for each healthcare service component (e.g. inpatient, emergency room) are calculated among patients with at least one encounter of that type.

^a Early-intervention screenings included those for developmental delays including mental, behavioral, neurodevelopmental, speech, language, hearing, vision, and metabolic disorders.

During the follow-up period, children of mothers with PPD incurred significantly higher healthcare costs. Total healthcare costs among children of mothers with PPD were 12.0% higher than those of mothers without PPD (Mean [SD]: $24,572 [$105,872] and $21,946 [$99,911], respectively; p < .05; Figure 3).

Figure 3. All-cause healthcare costs among children over 24-month follow-up period: matched PPD exposure versus non-PPD exposure cohorts. *p < 0.05, **p < 0.001 for PPD exposure versus non-PPD exposure cohorts. Abbreviations. PPD, Postpartum depression. ^aMean costs for each healthcare service component (e.g. inpatient, emergency room) were calculated among patients with at least one encounter of that type.

Sensitivity analysis

Sensitivity analysis excluding mothers with pre-term delivery and baseline mental health diagnoses were conducted (results not shown). The proportion of children with ER visits, physician specialist services, and outpatient pharmacy claims remained significantly higher in the PPD exposure cohort than non-PPD exposure cohort after excluding mothers with preterm delivery and baseline mental health diagnosis (all p < .001). After excluding pre-term deliveries, the mean healthcare costs were higher in the PPD exposure cohort compared with the non-PPD exposure cohort ($18,103 vs. $16,327) but the difference was not statistically significant. When excluding mothers with a pre-existing mental health diagnosis, the mean healthcare costs were significantly higher for the PPD exposure cohort than the non-PPD exposure cohort ($25,639 vs. $21,691; p < .001).

Discussion

This retrospective analysis assessed real-world HRU and costs in children of mothers with PPD compared to children of mothers without PPD. Over a 24-month follow-up period, children of mothers with PPD incurred 12.0% higher healthcare costs compared to children of mothers without PPD. We also found a higher proportion of children with PPD exposure had ER visits, outpatient physician specialist services, early-intervention screening, and outpatient pharmacy claims. Baseline demographic and non-mental clinical characteristics were well-balanced across both PPD and non-PPD exposure cohorts, while the prevalence of adverse clinical morbidities such as major and minor congenital anomalies, feeding problems, fussiness/excessive crying, and vomiting over the 24 month follow-up period was significantly higher among children of mothers with PPD than those of mothers without PPD (all p < .001). However, no differences were observed in the proportion of children who attended well-child outpatient office visits or received vaccinations, indicating higher utilization of non-routine care. Furthermore, the higher healthcare costs observed in children of mothers with PPD in the first 24 months of life suggest the presence of maternal PPD is associated with significantly higher HRU and cost in early childhood.

To our knowledge, this is the first major claims-based analysis to focus on differences in HRU and cost outcomes from children of mothers with PPD to those of mothers without PPD in the first 24 months of life. Several studies have shown negative implications of maternal PPD status on infant/children’s health^{8,15–17,33,34}. However, the impact of maternal PPD status on the children’s HRU and cost during early childhood has been inadequately investigated. Findings of this study may be even more robust when investigated in the non-commercially insured population, given the previously reported associations between low socio-economic status (SES) and a higher risk of PPD and greater impact on HRU and cost^35,36.

Previous US-based studies have reported increased ER visits, hospitalizations, and sick visits during the postpartum period among infants and children of mothers with depression^21,37,38. A study by Minkovitz et al.³⁷ found increased odds of ER visits among infants of depressed mothers. In a study by Sills et al.³⁸, the rate of ER visits was higher among children of parents with depression than those without a diagnosis of depression. Similarly, infants of mothers with depression diagnosed during pregnancy and the postpartum period were more likely to have ≥6 sick/ER visits than infants of mothers with no depression²¹. The current study also observed higher rates of ER visits among children of mothers with PPD cohort relative to children of mothers without PPD. Thus, higher ER utilization is a robust finding, consistently observed across studies differing in design, population, and time period.

Regardless of maternal PPD status, most children received a vaccination or a well-child visit over the first 24 months and the number of well-child visits received was comparable. This is in contrast to observations of a prospective cohort study that found that maternal depressive symptoms at early infancy significantly decreased well-child visits³⁷. The same study showed decreased compliance with vaccination schedules in infants of mothers with depressive symptoms³⁷. Though compliance with recommended vaccination schedules was not assessed as an outcome in the current study, the overall rate of vaccination did not differ significantly in children of mothers with PPD. However, our study included only privately insured patients, whereas 41% of patients were covered by public insurance in the previous study³⁷. Child vaccination rates are also lower among publicly insured, low-SES populations^39,40; differences in outcomes may be attributable to differences in the study population. The study by Minkovitz et al.³⁴ was also conducted prior to the passage of the Affordable Care Act in the USA, requiring health plan coverage of preventive care and vaccinations, which may further explain the differences observed in the current study. Further research is necessary to determine if vaccination compliance in children is similar based on maternal PPD status.

The higher frequency of all-cause outpatient office visits among children of mothers with PPD in conjunction with the similar prevalence of well-child visits in both cohorts in the current study, suggests children of mothers with PPD may have more frequent non-routine office visits. This finding is consistent with a prior prospective study which reported an association between maternal depressive symptoms and high frequency (≥3 visits) of infant non-routine visits to a pediatrician⁴¹. During the 24-month follow-up period, children of mothers with PPD had significantly higher HRU across most service categories compared to those without PPD. Among outpatient services, the percentage of children with a physician specialist service, early-intervention screening, or an ER visit were 3–7% greater in children of mothers with PPD than those of mothers without PPD (p < .001).

The mean total healthcare costs were lower in both the matched infants of mothers with PPD and infants of mothers without PPD cohorts after excluding preterm deliveries. Infants of mothers with PPD had higher total healthcare costs than those without PPD, but the difference was no longer significant after excluding preterm deliveries ($18,103 vs. $16,327, p = .077), suggesting preterm birth is a driver of all-cause healthcare costs and supporting the previously identified relationship between preterm delivery and risk of PPD^42–44. It is also possible that comorbid conditions in the preterm infant leading to increased HRU and costs could be a contributing factor to the mother’s depressive symptoms, instead of the mother’s mental health contributing to the infant’s health. However, the proportion of children with ER visits, outpatient physician specialist services, and outpatient pharmacy claims remained significantly higher in children in the PPD exposure cohort. Taken together, these results suggest that though the differences in healthcare costs in the PPD and non-PPD cohort may be driven by preterm deliveries, differences in healthcare resource use persist with higher resource use in the PPD exposure cohort.

Several limitations inherent to claims-based analysis apply to this study. The analysis was conducted on the commercially-insured population; thus, the results of this analysis may not be generalizable to PPD patients with other/no insurance. The population studied included people with private insurance and children with 24 months continuous enrollment, likely representing persons of higher SES. As patients were identified through administrative claims data, there is potential misclassification of PPD status, covariates, or study outcomes due to coding limitations. Mental health conditions are often undiagnosed, so some women in the non-PPD cohort may have been inappropriately classified. Thus, these study results may underestimate the magnitude of difference in healthcare utilization and costs between children of mothers with and without PPD. Furthermore, the MarketScan Commercial database relies on administrative claims data; therefore, the data may be subject to misclassification resulting from diagnostic coding errors. Although multiple confounding factors were adjusted using propensity score matching, some parameters, unquantifiable using claims data, may influence the healthcare costs and utilization differences between the cohorts. Children’s outcomes were assessed overall; the temporal relationship between PPD diagnosis and child outcomes was not described. This relationship could have included an inverse association between health complications in the child leading to depression in the mothers.

Conclusion

In conclusion, this study demonstrates that the presence of maternal PPD was associated with higher HRU and economic burden among children of mothers with PPD. Future studies are warranted to establish optimal treatment strategies for maternal PPD to reduce maternal morbidity as well as the resource and cost burden among children.

Transparency

Declaration of funding

This study was funded by Sage Therapeutics, Inc., Cambridge, MA, USA.

Declaration of conflict of interests

Ming-Yi Huang and Adi Eldar-Lissai are employed by Sage Therapeutics, Inc. Elizabeth Packnett, Nicole Zimmerman, and Meghan Moynihan are employed by IBM Watson Health as consultants and received funding from Sage Therapeutics to conduct this study. Tiffany Moore Simas was a paid consultant of Sage Therapeutics. The peer reviewers on this manuscript have received an honorarium from JME for their review work. In addition, a reviewer on this manuscript has disclosed that they have worked for NICE Scientific Advice on pharmacological interventions for postpartum depression. The reviewers have no other relevant financial relationships or otherwise to disclose.

Author contributions

Adi Eldar-Lissai, Elizabeth R. Packnett, and Ming-Yi Huang were involved in the conception of this study. Elizabeth R. Packnett, Nicole M. Zimmerman, and Meghan Moynihan were involved in the analysis of the data. All authors were involved in the design of the study, the interpretation of the data, and the drafting or revision of the manuscript. All authors have read and approved the final version of this submitted manuscript, and all authors agree to be accountable for all aspects of the work.

Previous presentations

This research was presented in part at the 2019 American College of Obstetrics and Gynecology Annual Meeting conference in Memphis, TN.

Acknowledgements

Medical writing services were provided by Santosh Tiwari, Ph.D. of IBM Watson Health with editorial support from Elizabeth Hoit Marchlewicz, Ph.D. of IBM Watson Health. These services were paid for by Sage Therapeutics, Inc.