Association Between Mental Health Conditions at the Hospitalization for Birth and Postpartum Hospital Readmission

Abstract

Background:

The relationship between physical comorbidities and postpartum hospital readmission is well studied, with less research regarding the impact of mental health conditions on postpartum readmission.

Methods:

Using hospital discharge data (2016–2019) from the Hospital Cost and Utilization Project Nationwide Readmissions Database (n = 12,222,654 weighted), we evaluated the impact of mental health conditions (0, 1, 2, and ≥3), as well as five individual conditions (anxiety, depressive, bipolar, schizophrenic, and traumatic/stress-related conditions) on readmission within 42 days, 1–7 days (“early”), and 8–42 days (“late”) of hospitalization for birth.

Results:

In adjusted analyses, the rate of 42-day readmission was 2.2 times higher for individuals with ≥3 mental health conditions compared to those with none (3.38% vs. 1.56%; p < 0.001), 50% higher among individuals with 2 mental health conditions (2.33%; p < 0.001), and 40% higher among individuals with 1 mental health condition (2.17%; p < 0.001). We found increased adjusted risk of 42-day readmission for individuals with anxiety (1.98% vs. 1.59%; p < 0.001), bipolar (2.38% vs. 1.60%; p < 0.001), depressive (1.93% vs. 1.60%; p < 0.001), schizophrenic (4.00% vs. 1.61%; p < 0.001), and traumatic/stress-related conditions (2.21% vs. 1.61%; p < 0.001), relative to individuals without the respective condition. Mental health conditions had larger impacts on late (8–42 day) relative to early (1–7 day) readmission.

Conclusions:

This study found strong relationships between mental health conditions during the hospitalization for birth and readmission within 42 days. Efforts to reduce the high rates of adverse perinatal outcomes in the United States should continue to address the impact of mental health conditions during pregnancy and throughout the postpartum period.

Introduction

An estimated 1.4% of birthing individuals are readmitted within 42 days after discharge from the hospitalization for birth.¹ Such readmission is not only an indicator of an adverse maternal health event but also can cause disruptions to the maternal-infant relationship, in addition to increased costs to the family and to payers.^2,3 Given that many adverse maternal events, including readmission, are potentially preventable,^2,4,5 identifying characteristics that may increase an individual's risk of readmission may provide information to policy makers and medical professional associations for the development of guidelines and recommendations that facilitate effective prenatal and postpartum care, such as guidelines for optimal postpartum follow-up timing.⁶ At the clinical level, information about risk factors for readmission can help clinicians ensure that patients with a given risk factor have optimal health before discharge, as well as educational materials and postpartum care that may reduce the risk of readmission.⁴

It is well recognized that comorbid physical conditions, such as hypertension and diabetes, increase the risk of adverse perinatal outcomes, including readmission,^7,8 and there is growing recognition of the impact of behavioral and mental health conditions on perinatal health during delivery and throughout the postpartum period.⁹ A recent study found that the percent of individuals with at least one mental health condition during the hospitalization for birth increased more than 10-fold, from 0.6% in 2000 to 7.3% in 2018.¹⁰ Individuals with mental health conditions may be at risk for adverse perinatal outcomes in the postpartum period for a number of reasons, such as reduced self-care, increased risk of self-harm or suicidal ideation, and more frequent risky behavior, such as substance abuse.^10
–12

Previous national studies have found that pregnant individuals with mental health conditions experience increased rates of severe maternal morbidity (SMM), maternal mortality, and readmissions for postpartum psychosis,^10,11,13,14 and single-state and single-facility studies in the United States have found that mental health conditions increase the risk of readmission after hospitalization for birth.^15

–18 Using 4 years (2016–2019) of data from the Hospital Cost and Utilization Project (HCUP) Nationwide Readmissions Database (NRD), this study adds to the literature by evaluating the association between mental health conditions and hospital readmission within 42 days of discharge from the hospitalization for birth.¹⁹ Understanding the risk of readmission associated with mental health conditions is critical for identifying the full spectrum of adverse events that may be experienced by individuals with perinatal mental health conditions, which may guide resource allocation and efforts to address the perinatal mental health crisis in the United States.

Materials and Methods

Data

Data from this study came from the HCUP NRD, years 2016–2019. The data set included a nationally representative, weighted sample of ∼35 million discharges from community hospitals annually.¹⁹ The NRD includes patient demographics, hospital-level variables, and diagnosis and procedural codes using the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10).

Sample population

Of the 70,995,804 records, we identified 7,859,571 hospitalizations for birth based on diagnosis codes (O80, O82, Z370–Z374, Z3750–Z3754, Z3760–Z3764, Z3759, Z3769, Z377, and Z379), vaginal and cesarean procedural codes (10D00Z0–10D07Z8 and 10E0XZZ), and diagnosis-related group (DRG) codes (765–768, 774, and 775), but excluding abortive outcomes (O00–O08), following similar previous approaches.²⁰ We removed records with missing covariate information (0.8%) and records that must be excluded to calculate readmission (i.e., in-patient death or November or December discharge), resulting in a nationally representative obstetric analytic sample of 12,222,654 hospitalizations (6,515,742 unweighted). The HCUP NRD allows for tracking an individual across hospitalizations in a given state within a calendar year. Variables allow for calculation of the days between discharge and subsequent admission; however, exact dates are not available. As such, the analysis required removal of discharges in November and December to allow for a 42-day postdischarge lookout period.

Outcome measures

Outcomes in this study included three separate readmission variables, including readmission within 42 days, readmission between 1 and 7 days (“early”), and readmission within 8 to 42 days (“late”).¹ Readmission analyses in other populations often assess readmission throughout 30 days, which may align with the 30-day readmission utilized in Medicare's Hospital Readmission Reduction Program. We have opted to study readmission through 42 days postpartum, as a 42 day window is a commonly utilized time frame for adverse perinatal outcomes, including readmission.^1,21 The time to readmission variable was defined as the difference between the discharge date of the index admission and the admission date of the first readmission.

Independent variables

There were two main independent variables. The first primary variable of interest was the number of mental health conditions. Categories were created to represent having zero, one, two, or three or more mental health conditions during the hospitalization for birth based on diagnosis codes for anxiety-, bipolar-, depression-, disruptive/impulse control/conduct-, eating-, obsessive-compulsive disorder-, personality-, schizophrenia-, somatic-, suicidal ideation/attempt-, and traumatic/stress-related conditions. Codes for the 11 mental health conditions were derived from a previous HCUP statistical brief.²²

Please note that substance use disorders were included as a separate covariate in this study (see Covariate Measures below). Second, we independently assessed having anxiety, bipolar, depressive, traumatic/stress, or schizophrenic conditions. The categories used for the number of mental health conditions (i.e., zero, one, two, or three or more) and the selection of the five included individual conditions were chosen based on sample size. Specifically, our analysis included 24 covariates (see Covariate Measures below). As such, we included individual conditions and the most granular number of mental health conditions such that each mental health condition and the number of mental health condition category had at least 240 (i.e., 24 covariates times the common rule of thumb of 10) readmissions.

Covariate measures

Analyses were adjusted for individual-level and hospital-level covariates, in addition to annual fixed effects. Individual-level variables included age of the birthing individual, payer, substance use, delivery type (cesarean section or vaginal), SMM, comorbid clinical conditions, quartile of median household income at the zip code level, and location of residence (large metropolitan, micropolitan, and noncore [rural]). Any substance use was defined as having a diagnosis code for alcohol-, opioid-, stimulant-, sedative-, or cannabis-related substance use disorders, as defined in a previous HCUP data brief.²²

SMM during the hospitalization for birth was defined using the CDC algorithm of 16 diagnosis-based and 5 procedure-based SMM indicators.²³ Comorbid clinical conditions included indicators from the Elixhauser comorbidity index,²⁴ but excluded conditions (congestive heart failure, pulmonary circulation disease, peripheral vascular disease, and coagulopathy) that overlap with the CDC SMM algorithm, and four conditions that overlap with mental health or substance abuse disorders (alcohol abuse, drug abuse, psychosis, and depression). The clinical comorbidity variable was defined as having zero, one or two, or three or more or more clinical comorbid conditions.

Hospital-level covariates included hospital profit status (not-for-profit, for-profit, and public), hospital teaching status/location (metropolitan nonteaching, metropolitan teaching, and nonmetropolitan), safety net hospital designation, and high cesarean section rate designation. Safety net designation was calculated annually and was based on the hospital being in the top quartile for the percent of hospitalizations for birth that were Medicaid or self-paid. High cesarean section rate designation was calculated annually and was based on the hospital being in the top quartile for the percent cesarean deliveries.

Statistical analyses

Descriptive statistics included chi square tests to test for differences in individual-level and hospital-level characteristics between hospitalizations for birth among individuals with zero, one, two, or three or more mental health conditions. Chi square tests were also used to test for differences in unadjusted readmission rates among individuals with zero, one, two, or three or more mental health conditions, as well as individuals with and without the five individual mental health conditions.

In adjusted analyses, we first conducted three separate multivariable logistic regressions to assess the impact of having one, two, or three or more mental health conditions on each of the three readmission variables (1–42 days, 1–7 days, and 8–42 days), adjusting for covariates. In addition to providing adjusted odds ratios (aOR), we calculated marginal effects, which represent average predicted probabilities of readmission for individuals with zero, one, two, or three or more mental health conditions.

We then tested for equality in the marginal effects using Wald tests to (1) assess whether the predicted probabilities of readmission associated with having one, two, or three or more mental health conditions were each independently higher than the probability of having zero mental health conditions and (2) assess whether there was a dose response in this relationship, such that having two mental health conditions had a greater effect than one mental health condition and whether three or more mental health conditions had a great effect than two mental health conditions. Marginal effects were used to facilitate comparison across logistic regression models, because comparing odds ratios across different models is not preferred.

Finally, we conducted three separate multivariable logistic regressions for the three readmission variables to assess the impact of each of the five individual mental health conditions, adjusting for covariates. We then estimated marginal effects and tested for equality in predicted probabilities of readmission among individuals with and without each of the five conditions using Wald tests.

Finally, we conducted a post hoc descriptive analysis to evaluate the most common primary diagnoses at the time of readmission. The primary diagnosis included in this analysis came from the first readmission after discharge from the hospitalization for birth.

Analyses were conducted in STATA v16. Statistical significance was assumed at p < 0.05, and the study was approved by the University of Arkansas for Medical Sciences Institutional Review Board.

Results

The analysis included 6,515,742 records, representing 12,222,654 hospitalizations for birth after weighting for the complex survey design of the HCUP NRD. Among these hospitalizations, 7.4% had mental health conditions, with 5.3% having a single mental health condition, 1.9% having two mental health conditions, and 0.2% having three or more mental health conditions (Table 1).

Table 1.

Characteristics of Hospitalizations for Birth, by Number of Mental Health Conditions

	All	No. of mental health conditions
	All	0	1	2	3+	p ^a
Unweighted N	6,515,742	6,039,983	341,807	120,451	13,501	—
Weighted N	12,222,654	11,315,966	648,707	231,922	26,059	—
Age
<20	5.3	5.3	5.2	5.4	7.9	<0.001
20–29	48.8	48.8	47.1	49.0	53.8
30–39^b	42.8	42.7	44.2	42.5	35.6
40+	3.1	3.1	3.4	3.1	2.7
Payer
Medicaid^b	41.5	41.3	43.1	46.5	60.5	<0.001
Private	53.1	53.5	50.9	46.4	27.7
Other	5.4	5.3	6.0	7.1	11.8
Any substance use
No^b	97.5	98.0	92.6	89.3	79.0	<0.001
Yes	2.5	2.0	7.4	10.7	21.0	<0.001
Delivery type
Vaginal^b	68.0	68.3	64.3	63.9	63.3	<0.001
Cesarean	32.0	31.7	35.7	36.1	36.7	<0.001
Comorbid conditions
0^b	66.4	68.1	47.0	43.5	36.5	<0.001
1–2	30.0	28.8	44.5	46.3	50.1
3+	3.6	3.1	8.6	10.2	13.5
SMM
No^b	98.5	98.6	97.8	97.8	97.2	<0.001
Yes	1.5	1.4	2.2	2.2	2.8	<0.001
Income quartile
Q1^b	27.6	27.7	25.6	25.4	31.5	<0.001
Q2	26.3	26.2	26.3	27.8	28.8
Q3	25.4	25.3	26.9	27.4	24.9
Q4	20.8	20.8	21.2	19.5	14.8
Patient rurality
Metropolitan^b	86.2	86.2	86.4	85.8	85.4	0.005
Micropolitan	8.0	8.0	8.1	8.4	9.0
Noncore	5.9	5.9	5.5	5.8	5.6
Hospital profit status
Not for profit^b	77.6	77.3	81.4	82.9	82.9	<0.001
For profit	11.4	11.7	8.5	7.1	5.3
Public	11.0	11.0	10.0	10.0	11.7
Hospital teaching status/location
Metropolitan nonteaching	20.7	21.0	17.6	16.7	14.0	<0.001
Metropolitan teaching^b	70.0	69.6	74.3	75.4	77.6
Nonmetropolitan	9.3	9.4	8.2	7.8	8.4
Safety net hospital
No^b	83.2	82.9	86.6	88.1	83.6	<0.001
Yes	16.8	17.1	13.4	11.9	16.4	<0.001
High cesarean section rate
No^b	72.5	72.0	78.4	79.3	80.1	<0.001
Yes	27.6	28.0	21.6	20.8	19.9	<0.001

p-Values from chi square tests for differences among individuals with zero, one, two, or three or more mental health conditions.

Reference category in adjusted regressions.

SMM, severe maternal morbidity.

Compared to hospitalizations without mental health conditions, hospitalizations among individuals with mental health conditions had higher rates of Medicaid coverage, cesarean section, substance use disorder, comorbid clinical conditions, and SMM (Table 1). Notably, having mental health conditions was associated with higher rates of having multiple other clinical comorbidities, with hospitalizations for births among individuals with three or more mental health conditions (13.5%), two mental health conditions (10.2%), or one mental health condition (8.6%) having higher rates of at least three clinical comorbidities compared to individuals without mental health conditions (3.1%; p < 0.001).

Individuals with three or more mental health conditions (21.0%), two mental health conditions (10.7%), or one mental health condition (7.4%) had significantly higher substance use disorders relative to individuals without mental health conditions (2.0%; p < 0.001). Individuals with mental health conditions were additionally at an increased risk of having SMM, with SMM occurring among 2.8% of individuals with three or more mental health conditions relative to 1.4% among individuals without mental health conditions (p < 0.001). Hospitalizations among individuals with mental health conditions were more likely to be at hospitals that were nonprofit or metropolitan teaching.

Table 2 provides unadjusted readmission rates stratified by mental health conditions. The left-most panel indicates the percentage of individuals who were readmitted within 42 days. We found that 4.71% of individuals with three or more mental health conditions were readmitted within 42 days compared to 1.53% of individuals without mental health conditions, 2.56% of individuals with one mental health condition, and 2.86% of individuals with 2 mental health conditions (p < 0.001).

Table 2.

Unadjusted Rates of Readmissions and Distribution of Readmissions, by Mental Health Conditions

	% Readmitted^a	% Readmitted^b		% of Readmitted individuals^c
	1–42 days	1–7 days (early)	8–42 days (late)	1–7 days	8–42 days
No. of mental health conditions
0	1.53	0.84	0.69	55.13	44.87
1	2.56	1.38	1.18	53.97	46.03
2	2.86	1.46	1.40	51.11	48.89
3+	4.71	2.04	2.67	43.40	56.60
Anxiety
No	1.57	0.86	0.71	54.81	45.19
Yes	2.59	1.43	1.16	55.10	44.90
Bipolar
No	1.60	0.88	0.72	55.01	44.99
Yes	3.72	1.69	2.03	45.40	54.60
Depression
No	1.58	0.87	0.71	55.03	44.97
Yes	2.57	1.32	1.24	51.56	48.44
Schizophrenia
No	1.61	0.88	0.72	54.93	45.07
Yes	7.54	2.82	4.71	37.47	62.53
Trauma/stress related
No	1.60	0.88	0.72	54.88	45.12
Yes	3.82	1.90	1.92	49.72	50.28

Percent of individuals readmitted within 1–42 after discharge from hospitalization for birth. Chi square tests were used to test for differences in readmission rates for each mental health condition variable. All tests were significant at p < 0.001. For this panel, n = 12,222,654 weighted and n = 6,515,742 unweighted.

Percent of individuals readmitted within 1–7 days and 8–42 days after discharge from the hospitalization for birth. Chi square tests were used to test for differences in readmission rates for each mental health condition variable. All tests were significant at p < 0.001. For this panel, n = 12,222,654 weighted and n = 6,515,742 unweighted.

Percent of readmitted individuals who were readmitted within 1–7 days or within 8–42 days. Chi square tests were used to test for differences in the percent of readmitted individuals who were readmitted within each time frame. All tests were significant at p < 0.001, with the exception of the test for having anxiety versus not having anxiety. For this panel, n = 197,433 weighted and n = 105,414 unweighted.

Individuals with each of the selected mental health conditions had higher rates of 42-day readmission relative to individuals without the given mental health condition, with 2.59% of individuals with anxiety conditions (p < 0.001), 3.72% of individuals with bipolar conditions (p < 0.001), 2.57% of individuals with depressive conditions (p < 0.001), 7.54% of individuals with schizophrenic conditions (p < 0.001), and 3.82% of individuals with traumatic or stress-related conditions (p < 0.001) having readmission within 42 days. The middle panel of Table 2 provides the readmission rates between 1 and 7 (“early”) and between 8 and 42 (“late”) days after discharge. Individuals with one, two, or three or more or more mental health conditions, as well as those with individual mental health conditions had higher readmission rates relative to their counterparts without mental health conditions in both the early and late time periods (all p < 0.001).

The right-most panel of Table 2 indicates the distribution of readmission timing among individuals who were readmitted. A higher percentage of readmissions were in the late time period among individuals with more mental health conditions (e.g., 3 or more mental health conditions vs. 0), as well as among individuals with bipolar, schizophrenic, and traumatic or stress-related conditions. A higher percentage of readmissions were in the early stage among individuals with depressive conditions.

Table 3 provides adjusted outcomes associated with risk of readmission among individuals with zero, one, two, or three or more mental health conditions. There was an increased risk of readmission for all three readmission variables (1–42 days, 1–7 days, and 8–42 days) associated with having one, two, or three or more mental health conditions. For example, individuals with three or more mental health conditions had 2.2 times the risk of readmission within 42 days (3.38% vs. 1.56%; p < 0.001), 1.8 times the risk of readmission within 7 days (1.53% vs. 0.86%), and 2.6 times the risk of readmission between 8 and 42 days (1.81% vs. 0.70%; p < 0.001).

Table 3.

Adjusted Associations Between Number of Mental Health Conditions and Readmission

	Logistic regression			Marginal effects from logistic regression
	OR	95% CI	p	ME^a	95% CI	p-(category vs. zero)^a	p-(dose response)^a
Readmission within 1–42 days
No. of mental health conditions
0	Ref.	Ref.	Ref.	1.56	1.53–1.57	Ref.	Ref.
1	1.41	1.37–1.45	<0.001	2.17	2.12–2.23	<0.001	<0.001
2	1.52	1.46–1.58	<0.001	2.33	2.24–2.42	<0.001	0.001
3+	2.23	2.02–2.46	<0.001	3.38	3.07–3.70	<0.001	<0.001
Readmission within 1–7 days (“early”)
No. of mental health conditions
0	Ref.	Ref.	Ref.	0.86	0.84–0.87	Ref.	Ref.
1	1.36	1.31–1.41	<0.001	1.16	1.12–1.20	<0.001	<0.001
2	1.39	1.32–1.47	<0.001	1.19	1.13–1.25	<0.001	0.418
3+	1.80	1.58–2.06	<0.001	1.53	1.33–1.73	<0.001	0.001
Readmission within 8–42 days (“late”)
No. of mental health conditions
0	Ref.	Ref.	Ref.	0.70	0.69–0.71	Ref.	Ref.
1	1.45	1.41–1.51	<0.001	1.01	0.98–1.04	<0.001	<0.001
2	1.65	1.56–1.74	<0.001	1.14	1.08–1.20	<0.001	<0.001
3+	2.64	2.34–2.99	<0.001	1.81	1.60–2.03	<0.001	<0.001

Logistic regressions conducted separately for three separate readmission outcomes, including readmission within 1–42 days, readmission within 1–7 days, and readmission within 8–42 days. Models were adjusted for year, age, payer, substance use, delivery type, number of comorbid conditions, SMM, income quartile of the patient's zip code, patient rurality, hospital profit status, hospital teaching status and location, safety net hospital designation, and high cesarean section rate designation.

The number of mental health condition variable was created using diagnosis codes for 11 mental health conditions: anxiety, bipolar, depressive, disruptive/impulse control/conduct, eating, obsessive-compulsive, personality, schizophrenia, somatic symptoms, suicidal ideation/attempt, and traumatic/stress conditions.

Marginal effects represent average predicted probabilities based on the adjusted logistic regression. The first p-value (next to right-most column) represents the p-value associated with a test for equivalence of the average predicted probability among individuals with 1, 2, or 3+ mental health conditions relative to the average predicted probability among those with 0 mental health conditions. The second p-value (right-most column) represents the p-value associated with a test for equivalence of the average predicted probability among individuals with 1 versus 0, 2 versus 1, and 3 versus 2 mental health conditions to assess a dose–response relationship.

95% CI, 95% confidence interval; OR, odds ratios.

Table 3 additionally highlights that for readmission within 42 days and readmission within 8–42 days, there was a dose–response relationship, such that individuals with a greater number of mental health conditions were associated with increased risk of readmission (1–42 days) and late readmission (8–42 days). For example, individuals with one mental health condition had increased risk of 42-day readmission relative to individuals with no mental health condition (2.17% vs. 1.56%; p < 0.001), and individuals with two mental health conditions had increased risk of 42-day readmission relative to individuals with one mental health condition (2.33% vs. 2.17%; p = 0.001).

Table 4 provides adjusted outcomes associated with the risk of readmission among individuals with mental health conditions. Individuals with anxiety (aOR: 1.25; p < 0.001), bipolar (aOR: 1.50; p < 0.001), depressive (aOR: 1.21; p < 0.001), schizophrenic (aOR: 2.57; p < 0.001), or traumatic or stress-related (aOR: 1.38; p < 0.001) conditions had increased risk of 42-day readmission relative to individuals without the respective mental health condition. Individuals with these five conditions additionally had an increased risk for 1–7 day and 8–42 day readmissions.

Table 4.

Adjusted Associations of Mental Health Conditions and Readmission

	Logistic regression			Marginal effects from logistic regression
	OR	95% CI	p	ME ^a	95% CI	p-(category vs. zero)^b
Readmission within 42 days
Anxiety conditions
No	Ref.	Ref.	Ref.	1.59	1.57–1.61	Ref.
Yes	1.25	1.21–1.28	<0.001	1.98	1.92–2.03	<0.001
Bipolar conditions
No	Ref.	Ref.	Ref.	1.60	1.58–1.63	Ref.
Yes	1.50	1.43–1.58	<0.001	2.38	2.26–2.51	<0.001
Depressive conditions
No	Ref.	Ref.	Ref.	1.60	1.58–1.62	Ref.
Yes	1.21	1.17–1.25	<0.001	1.93	1.86–1.99	<0.001
Schizophrenic conditions
No	Ref.	Ref.	Ref.	1.61	1.59–1.63	Ref.
Yes	2.57	2.32–2.84	<0.001	4.00	3.62–4.39	<0.001
Traumatic/stress conditions
No	Ref.	Ref.	Ref.	1.61	1.59–1.63	Ref.
Yes	1.38	1.29–1.48	<0.001	2.21	2.06–2.36	<0.001
Readmission within 1–7 days
Anxiety conditions
No	Ref.	Ref.	Ref.	0.87	0.86–0.89	Ref.
Yes	1.28	1.23–1.33	<0.001	1.11	1.07–1.15	<0.001
Bipolar conditions
No	Ref.	Ref.	Ref.	0.88	0.87–0.90	Ref.
Yes	1.33	1.23–1.43	<0.001	1.17	1.08–1.25	<0.001
Depression conditions
No	Ref.	Ref.	Ref.	0.88	0.87–0.90	Ref.
Yes	1.12	1.07–1.17	<0.001	0.98	0.94–1.03	<0.001
Schizophrenic conditions
No	Ref.	Ref.	Ref.	0.88	0.87–0.90	Ref.
Yes	1.90	1.64–2.20	<0.001	1.66	1.42–1.90	<0.001
Traumatic/stress conditions
No	Ref.	Ref.	Ref.	0.88	0.87–0.90	Ref.
Yes	1.35	1.23–1.48	<0.001	1.19	1.08–1.30	<0.001
Readmission within 8–42 days
Anxiety conditions
No	Ref.	Ref.	Ref.	0.72	0.71–0.73	Ref.
Yes	1.20	1.15–1.26	<0.001	0.87	0.83–0.90	<0.001
Bipolar conditions
No	Ref.	Ref.	Ref.	0.72	0.71–0.73	Ref.
Yes	1.67	1.55–1.79	<0.001	1.20	1.12–1.28	<0.001
Depressive conditions
No	Ref.	Ref.	Ref.	0.72	0.71–0.73	Ref.
Yes	1.32	1.26–1.38	<0.001	0.94	0.90–0.99	<0.001
Schizophrenic conditions
No	Ref.	Ref.	Ref.	0.73	0.72–0.74	Ref.
Yes	3.05	2.70–3.45	<0.001	2.17	1.91–2.43	<0.001
Traumatic/stress conditions
No	Ref.	Ref.	Ref.	0.73	0.72–0.74	Ref.
Yes	1.39	1.27–1.53	<0.001	1.01	0.92–1.10	<0.001

Marginal effects represent average predicted probabilities based on the adjusted logistic regression.

p Values represent the p-value associated with a test for equivalence of the average predicted probability among individuals with the given mental health condition relative to those without the mental health condition.

To better understand the relationship between mental health conditions and readmission, we evaluated the primary diagnosis at readmission among individuals who were readmitted within 42 days. Individuals with zero, one, or two mental health conditions, as well as individuals with anxiety, depressive, or traumatic/stress-related conditions shared the same two most common primary diagnoses at readmission (O14.15: pre-eclampsia complicating the puerperium and O86.12: endometritis following delivery). The most common primary diagnosis for individuals with three or more mental health conditions, as well as for individuals with bipolar conditions or with schizophrenic conditions was the code for mental disorders complicating the puerperium (O99.345). This diagnosis (O99.345) was present for 8.4% of individuals with three or more mental health conditions, 6.7% of individuals with bipolar conditions, and 15.5% of individuals with schizophrenic conditions.

Discussion

This study evaluated the association between mental health conditions identified in the discharge records for hospitalizations for birth and hospital readmission within 42 days using 4 years of data (2016–2019) from the HCUP NRD. Given the increased risk of readmission among individuals with mental health conditions, the findings from this study have important implications regarding the need for postpartum support for individuals with perinatal mental health conditions.

In this study, we found that individuals with three or more mental health conditions had over twice the adjusted rates of readmission compared to individuals without mental health conditions (3.38% vs. 1.56%), and individuals with one or two mental health conditions had ∼40% and 50% increased risk of readmission, respectively. When assessing early readmission within 1–7 days and late readmission within 8–42 days, we found that mental health conditions had a stronger impact on late readmission. In fact, individuals with three or more mental health conditions had 2.6 times the rate of late readmission relative to those without mental health conditions.

We analyzed five individual mental health conditions in this study, including anxiety, bipolar, depressive, schizophrenic, and traumatic/stress-related conditions. We found increased risk of readmission associated with all five mental health conditions; however, we found a larger risk of readmission-associated schizophrenic conditions, followed by bipolar conditions and traumatic/stress-related conditions. A previous evaluation of data from an academic medical center similarly found varying risks of readmission among individuals with different mental health conditions, such as increased readmission risk among individuals with psychiatric conditions, but not among individuals with depression/anxiety.¹⁶

In our study, the unadjusted rates of readmission within 42 days among individuals with depression or anxiety were both 2.6%, which align with the previous study that found a 30-day readmission rate of 2.5% among individuals with depression or anxiety.¹⁶ A separate recent study evaluated risk factors for readmission for postpartum psychosis specifically and found that ∼0.6 per 1,000 individuals had readmission within 60 days after discharge from hospitalization for birth, with significantly higher risk of readmission for psychosis among individuals with mental health conditions or substance abuse conditions during the hospitalization for birth.¹⁴ Similar to our study, which found that individuals with schizophrenia-related conditions at the time of the hospitalization for birth had the highest risk for readmission, this previous study¹⁴ found that individuals with schizophrenia-related conditions had the greatest risk for readmission for psychosis.

When assessing the primary diagnosis at the time of readmission, we found that mental health-related primary diagnoses were common among individuals with bipolar or schizophrenic conditions as well as for those with three or more mental health conditions. Among individuals with zero, one, or two mental health conditions, as well as for those with anxiety, depressive, or traumatic/stress-related conditions, the most common primary readmission diagnoses were pre-eclampsia and endometritis. Understanding the reason for readmission among different unique populations may be important for developing educational materials and follow-up guidelines to prevent readmission. For example, we found that the impact of multiple mental health conditions (3+) had a relatively larger impact on late readmission (8–42 days). This, in combination with the more frequent mental health-specific readmission diagnoses among this population, suggests the need for follow-up care that specifically emphasizes mental health screening and treatment throughout the postpartum period.

In this study, we used the number of mental health conditions, as well as individual mental health conditions, and did not specifically assess severity. A previous national study evaluated the impact of substance use on readmission adjusted for the severity of mental illness, and the study classified mental illness into severity categories, including none, mild, moderate, and severe.²⁵ This important metric was based on ICD-9 codes and used data through 2014; future efforts to develop a similar measure with ICD-10 codes would be a valuable tool in evaluating the impact of mental health severity on maternal health.

It is well known that medical comorbidities during hospitalization for birth, such as diabetes or hypertension, increase the risk of subsequent hospitalization^26
–28; we found an increased risk of readmission associated with mental health conditions, which adds to facility- and state-level analyses^15

–18; of the impact of mental health conditions on readmission, as well as national evaluations of substance use disorders²⁹; and of readmission for psychosis.¹⁴ Individuals with mental health conditions may be at risk for adverse perinatal outcomes for several reasons, such as increased risk of self-harm and substance abuse, reduced social support, fears of stigma in the clinical setting, and reduced adherence to self-care behaviors,¹¹ such as adequate high-quality sleep. This study found higher rates of SMM, clinical conditions, and substance use among individuals with mental health conditions, as well as the increased risk of readmission associated with mental health conditions independent of such clinical conditions.

Given the nature of this study, we are unable to suggest a causal relationship between mental health conditions, other clinical conditions, and SMM at the time of birth; however, the association between these conditions is supported by previous studies^11,13,30 and suggests the need for a multifaceted approach, including efforts at the health policy level and facility levels. At facility level, hospitals should strongly consider establishing standards for patient and family education regarding the importance of follow-up, as well as potential warning signs of adverse maternal events. Hospitals should additionally make efforts to ensure timely follow-up after discharge from the hospitalization for birth. The American College of Obstetricians and Gynecologists (ACOG) has updated its recommendations for postpartum follow-up to recommend that individuals have contact with their provider within 3 weeks postpartum.³¹

Within these guidelines, ACOG highlights that individuals with chronic conditions, including mood disorders, should be provided education regarding the importance of follow-up care and that there should be efforts to facilitate scheduling follow-up care before discharge from hospitalization for birth. We add to these suggestions to highlight the critical importance of considering mental health conditions in efforts to identify individuals in need for postpartum coordination of care. In addition, given the increased risk of readmission associated with mental health conditions during the late readmission time period (8–42 days), as well as the frequency of mental health-related primary diagnoses among individuals who were readmitted, ACOG's recommendation to have a comprehensive postpartum visit within 12 weeks after birth, including assessments of mood and emotional well-being, is relevant to our findings.³¹

This study had several limitations. While representative of the United States, the HCUP NRD is constructed using hospital discharge data and does not include information on a number of important factors, such as health care utilization in the outpatient setting, granular geographic information, race/ethnicity, or the ability to identify many disadvantaged populations. Given the higher rates of perinatal mental health conditions and adverse perinatal outcomes among many racial/ethnic groups, including Black, Hispanic, Native American/Native Alaskan, Pacific Islander, and Asian individuals,^9,32,33 it is critical for future national studies to evaluate racial/ethnic inequities in mental health-related postpartum readmission. In this study, mental health conditions were identified using diagnosis codes present at the hospitalization for birth, and we were unable to differentiate between conditions that may have been present before pregnancy or newly during pregnancy.

Future studies should further evaluate potential differences in readmission risk based on timing of mental health diagnosis (e.g., antenatal) and consider the influence of antenatal services (e.g., group prenatal care) on the increased readmission risk associated with mental health condition. In addition, if milder severities of mental health diagnoses were not recorded or if mental health conditions are more likely to be noted for those with other clinical conditions, the increased risk of readmission may be overestimated.

The strengths of this study include using 4 years of data from the largest readmission database in the United States and adopting previously utilized measures for the primary metrics in the study, including (1) a 42-day²¹ postreadmission window, (2) defining mental health conditions based on a previous HCUP-published statistical brief,²² and (3) following the CDC definition for SMM.²³ Use of such standardized metrics allows for comparison across studies and populations.

Conclusions

This study found strong associations between mental health conditions and risk of readmission after the hospitalization for birth, highlighting the need for continued efforts toward addressing the perinatal mental health crisis in the United States. Additional research is needed to better understand mechanisms that may prevent postpartum readmission among birthing individuals with mental health conditions.

Footnotes

Authors' Contributions

C.B.: conceptualization, data curation and analysis, writing—original draft, and writing—review and editing; S.K.: conceptualization, writing—original draft, and writing—review and editing; K.S.: conceptualization, writing—original draft, and writing—review and editing; J.M.: conceptualization and writing—review and editing; B.A.: conceptualization, writing—original draft, and writing—review and editing

Disclaimer

The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the funders. Funders had no role in the design, analysis, or writing of this article.

Author Disclosure Statement

C.B. reports her position as a senior research fellow with the Institute for Medicaid Innovation. All other authors reported no conflicts of interest.

Funding Information

Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health (NIH) (5P20GM109096), as well as the National Institute on Minority Health and Health Disparities of the NIH (K01MD018072). The project described was also supported by University of Arkansas for Medical Sciences Translational Research Institute funding (KL2 TR003108) and (UL1 TR003107) through the National Center for Advancing Translational Sciences and the National Institute of Nursing Research (1R21NR020677-01) of the NIH.

References

Matas

, Mitchell

, Sharma

, et al. Severe maternal morbidity at delivery and postpartum readmission in the United States. Paediatr Perinat Epidemiol, 2021; 35(6):627–634; doi: https://doi.org/10.1111/ppe.12762

Decker

, Thibault

, Davis

, et al. A nationwide analysis of maternal morbidity and acute postpartum readmissions in women with epilepsy. Epilepsy Behav, 2021; 117:107874; doi: https://doi.org/10.1016/j.yebeh.2021.107874

Wen

, Krenitsky

, Clapp

, et al. Fragmentation of postpartum readmissions in the United States. Obstet Gynecol, 2020; 223(2):252.e1–e252.e14; doi: https://doi.org/10.1016/j.ajog.2020.01.022

Trost

, Beauregard

, Smoots

, et al. Preventing pregnancy-related mental health deaths: insights from 14 US Maternal Mortality Review Committees, 2008–2017. Health Aff, 2021; 40(10):1551–1559; doi: https://doi.org/10.1377/hlthaff.2021.00615

Mogos

, Salemi

, Spooner

, et al. Hypertensive disorders of pregnancy and postpartum readmission in the United States: national surveillance of the revolving door. J Hypertens, 2018; 36(3):608–618; doi: 10.1097/HJH.0000000000001594

Interrante

, Admon

, Stuebe

, et al. After childbirth: better data can help align postpartum needs with a new standard of care. Womens Health Issues, 2022; 32(3):208–212; doi: https://doi.org/10.1016/j.whi.2021.12.001

Mourad

, Wen

, Friedman

, et al. Postpartum readmissions among women with diabetes. Obstet Gynecol, 2020; 135(1):80–89; doi: https://doi.org/10.1097%2FAOG.0000000000003551

Clapp

, Little

, Zheng

, et al. A multi-state analysis of postpartum readmissions in the United States. Obstet Gynecol, 2016; 215(1):113.e1–e113.e10; doi: https://doi.org/10.1016/j.ajog.2016.01.174

Moore

, McLemore

, Glenn

, et al. Policy opportunities to improve prevention, diagnosis, and treatment of perinatal mental health conditions. Health Aff, 2021; 40(10):1534–1542; doi: https://doi.org/10.1377/hlthaff.2021.00779

10.

Logue

, Wen

, Monk

, et al. Trends in and complications associated with mental health condition diagnoses during delivery hospitalizations. Obstet Gynecol, 2022; 226(3):e405.e1–e405.e16; doi: https://doi.org/10.1016/j.ajog.2021.09.021

11.

Howard

, Khalifeh

. Perinatal mental health: A review of progress and challenges. World Psychiatry, 2020; 19(3):313–327; doi: https://doi.org/10.1002/wps.20769

12.

Slomian

, Honvo

, Emonts

, et al. Consequences of maternal postpartum depression: A systematic review of maternal and infant outcomes. Womens Health, 2019; 15:1745506519844044; doi: 10.1177/1745506519844044

13.

Brown

, Adams

, George

, et al. Mental health conditions increase severe maternal morbidity by 50 percent and cost $102 million yearly in the United States. Health Aff, 2021; 40(10):1575–1584; doi: https://doi.org/10.1377/hlthaff.2021.00759

14.

Albers

, Wen

, Monk

, et al. Postpartum psychosis during delivery hospitalizations and postpartum readmissions, 2016–2019. AJOG MFM, 2023; 5(5):100905; doi: https://doi.org/10.1016/j.ajogmf.2023.100905

15.

Girsen

, Leonard

, Butwick

, et al. Early postpartum readmissions: Identifying risk factors at birth hospitalization. AJOG Global Rep, 2022; 2(4):100094; doi: https://doi.org/10.1016/j.xagr.2022.100094

16.

Kumar

, Rao

, O'Rourke

, et al. Relationship between depression and/or anxiety and hospital readmission among women after childbirth. J Obstet Gynecol Neonatal Nurs, 2019; 48(5):552–562; doi: https://doi.org/10.1016/j.jogn.2019.07.001

17.

Jiang

, Alexander

, Christopher

, et al. Relationship of depression to increased risk of mortality and rehospitalization in patients with congestive heart failure. Arch Intern Med, 2001; 161(15):1849–1856; doi: 10.1001/archinte.161.15.1849

18.

Glazer

, Harrell

, Balbierz

, et al. Postpartum hospital readmissions and emergency department visits among high-risk, Medicaid-insured women in New York City. J Womens Health, 2022; 31(9):1305–1313; doi: https://doi.org/10.1089/jwh.2021.0338

19.

Healthcare Cost and Utilization Project (HCUP). Overview of the Nationwide Readmissions Database (NRD). Rockville, MD: Healthcare Cost and Utilization Project; 2021. Available from: https://www.hcup-us.ahrq.gov/nrdoverview.jsp [Last accessed: October 28, 2022 ].

20.

Kuklina

, Whiteman

, Hillis

, et al. An enhanced method for identifying obstetric deliveries: implications for estimating maternal morbidity. Matern Child Health J, 2008; 12(4):469–477; doi: https://doi.org/10.1007/s10995-007-0256-6

21.

Hoyert

. Maternal mortality rates in the United States, 2020. NCHS Health E-Stats; 2022; doi: https://dx-doi-org-s.web.bisu.edu.cn/10.15620/cdc:113967

22.

Owens

, Fingar

, McDermott

, et al. Inpatient stays involving mental and substance use disorders, 2016. HCUP Statistical Brief #249. Agency for Healthcare Research and Quality, Rockville, MD; 2019. Available from: www.hcup-us.ahrq.gov/reports/statbriefs/sb249-Mental-Substance-Use-Disorder-Hospital-Stays-2016.pdf [Last accessed: October 28, 2022 ].

23.

Centers for Disease Control and Prevention. Severe maternal morbidity in the United States; 2021. Available from: https://www.cdc.gov/reproductivehealth/MaternalInfantHealth/SevereMaternalMorbidity.html [Last accessed: October 28, 2022 ].

24.

Elixhauser Comorbidity Software for ICD-10-CM (beta version) Healthcare Cost and Utilization Project (HCUP). Agency for Healthcare Research and Quality, Rockville, MD; October 29, 2021. Available from: www.hcup-us.ahrq.gov/toolssoftware/comorbidityicd10/comorbidity_icd10.jsp [Last accessed: October 28, 2022 ].

25.

Salemi

, Raza

, Modak

, et al. The association between use of opiates, cocaine, and amphetamines during pregnancy and maternal postpartum readmission in the United States: A retrospective analysis of the Nationwide Readmissions Database. Drug Alcohol Depend, 2020; 210:107963.

26.

Jarvie

, Metz

, Davis

, et al. Short-term risk of cardiovascular readmission following a hypertensive disorder of pregnancy. Heart, 2018; 104(14):1187–1194; doi: https://doi.org/10.1016/j.drugalcdep.2020.107963

27.

Aseltine

Jr. , Yan

, Fleischman

, et al. Racial and ethnic disparities in hospital readmissions after delivery. Obstet Gynecol, 2015; 126(5):1040–1047; doi: 10.1097/AOG.0000000000001090

28.

Miller

, Wen

, Elkind

, et al. Infection during delivery hospitalization and risk of readmission for postpartum stroke. Stroke, 2019; 50(10):2685–2691.

29.

Wen

, Batista

, Wright

, et al. Postpartum readmissions among women with opioid use disorder. AJOG MFM, 2019; 1(1):89–98; doi: https://doi.org/10.1016/j.ajogmf.2019.02.004

30.

Leonard

, Kennedy

, Carmichael

, et al. An expanded obstetric comorbidity scoring system for predicting severe maternal morbidity. Obstet Gynecol, 2020; 136(3):440–449; doi: https://doi.org/10.1097%2FAOG.0000000000004022

31.

American College of Obstetricians and Gynecologists. Optimizing postpartum care. ACOG Committee Opinion No. 736. Obstet Gynecol, 2018; 131:e140–e150; doi: 10.1097/AOG.0000000000002633

32.

Kozhimannil

, Interrante

, Tofte

, et al. Severe maternal morbidity and mortality among indigenous women in the United States. Obstet Gynecol, 2020; 135(2):294–300; doi: https://doi.org/10.1097%2FAOG.0000000000003647

33.

Howell

. Reducing disparities in severe maternal morbidity and mortality. Clin Obstet Gynecol, 2018; 61(2):387–399; doi: https://doi.org/10.1097%2FGRF.0000000000000349