Sexual orientation-related disparities in perinatal mental health among a prospective cohort study

Abstract

Sexually minoritized women (SMW) may be at an increased risk of adverse perinatal mental health, though prior research is limited. We examined sexual orientation-related differences in perinatal mental health (i.e., stress and depression), and antidepressant utilization among those at different severities of clinically significant perinatal depressive symptoms.

Nurses’ Health Study 3 participants with prospectively assessed pregnancies (N = 6,364) received pregnancy and postpartum questionnaires. Using weighted log-binomial generalized estimating equations, we examined differences in stress (Perceived Stress Scale 4 [PSS-4]), depression (the Edinburgh Postnatal Depression Scale [EDPS] at four cut-off scores [≥7, ≥9, ≥11, ≥13]), and patterns of antidepressant utilization across five groups: completely heterosexual with no same-sex sexual partners (reference group; n = 5,178); heterosexual with same-sex sexual partners (n = 245); mostly heterosexual (n = 770); bisexual (n = 106); and lesbian (n = 47).

Compared to the completely heterosexual reference group, SMW reported increased stress during pregnancy (adjusted risk ratio [ARR]: 1.14, 95% confidence interval [1.02–1.28]). SMW reported an elevated risk of pregnancy depression at every EDPS score cutoff, with the magnitude of the disparity increasing as the score increased (ARRs: 1.09 [1.00–1.20]; 1.20 [1.05–1.36]; 1.37 [1.16–1.63]; 1.49 [1.18–1.89] for EDPS scores ≥7, ≥9, ≥11, ≥13, respectively). Disparities were highest in magnitude among the mostly heterosexual and bisexual subgroups. Utilization of postpartum antidepressants increased among the reference group with increasing symptom severity but was only associated at the highest score (≥13) among SMW subgroups.

SMW have increased risks of pregnancy stress and depression and are more likely to use perinatal antidepressants; mostly heterosexual and bisexual individuals had the highest risk of antidepressant use. Postpartum symptom severity closely corresponded to antidepressant use among completely heterosexual, but not SMW—suggesting disparities in mental health treatment. Further research priorities include determining the causes of these disparities and appropriately tailored solutions.

1. Introduction

Stress and depression are among the most common complications during pregnancy and postpartum (henceforth, the perinatal period; Woods et al., 2010; Woodya et al., 2017). Both are important risk factors for adverse pregnancy and birth outcomes such as preterm birth and low birthweight (Grote et al., 2010; Lobel et al., 2008; Szegda et al., 2014). Given that minoritized communities have disproportionately high levels of postpartum depression (Cannon and Nasrallah, 2019) and low usage of adequate postpartum mental health care (Kozhimannil et al., 2011), identifying marginalized populations at heightened risk for adverse perinatal mental health is critical to addressing perinatal health disparities in the US.

Sexually minoritized women (SMW) are those who identify as a lesbian, gay, bisexual, or another sexually minoritized identity, or have sexual or romantic attractions and/or sexual behaviors with women. Although estimates of how many SMW have given birth are outdated and imprecise due to data limitations (Goldberg et al., 2014), it has been suggested that approximately 210,000 births by SMW occur each year (Gonzales et al., 2019) or approximately 17% of pregnant people in the U.S. identify as a SMW (Everett et al., 2019). Emerging evidence points to birth and infant outcome disparities among SMW, including preterm birth and low birthweight (Barcelona et al., 2022; Downing et al., 2021; Everett et al., 2019; Leonard et al., 2022).

SMW may be at an increased risk of worse perinatal mental health relative to their heterosexual counterparts due to unique risk factors (Ross, 2005) including higher levels of poor mental health and untreated mental health needs prior to pregnancy (Steele et al., 2017) and other preconception health disparities (Limburg et al., 2020). In fact, a recent study using the Behavioral Risk Factor Surveillance System found lesbian and bisexual women had threefold greater odds of reporting frequent mental distress during pregnancy compared to heterosexual women (Gonzales et al., 2019). Additional evidence suggests high levels of perinatal anxiety and depression among SMW (Flanders et al., 2016; Goldberg and Smith, 2008; Marsland et al., 2022; Ross et al., 2007, 2012, 2017); however, findings are limited due to methodological issues, including small sample sizes, convenience recruitment methods, unvalidated measurement tools, cross-sectional data, and/or lack of a heterosexual comparison group.

SMW experiencing adverse perinatal mental health may additionally be more likely to have barriers to treatment than their heterosexual peers. SMW have increased risks of unmet mental healthcare needs during pregnancy in part due to cost-related barriers (Gonzales et al., 2019). However, even in contexts where participants had access to universal healthcare (i.e., cost was not a barrier) SMW evidenced unmet mental healthcare needs (Steele et al., 2008). Contributing factors to these unmet care needs likely include: unique interpersonal and systematic barriers among perinatal SMW seeking mental healthcare (Kirubarajan et al., 2022), gaps in the healthcare system to address SMW perinatal care needs and mental health (Carpenter, 2021; Kirubarajan et al., 2022), and pervasive invisibility, exclusion, and discrimination within perinatal healthcare interactions (Soled et al., 2023). However, to our knowledge, research has yet to explore the need for and utilization of mental health care services among SMW across pregnancy and postpartum in a large national sample.

Importantly, experiences of mental health are not homogenous across SMW subgroups. For example, bisexual women have higher odds of mood and anxiety disorders (Bostwick et al., 2010) and report more anxiety and depression compared to lesbian women (Lewis et al., 2019). Minority-specific stressors may explain these differential health disparities (Katz-Wise et al., 2017). For example, lesbian compared to bisexual women report higher levels of victimization, discrimination, and anticipated discrimination, but lower levels of identity concealment and internalized heterosexism (Puckett et al., 2016). Additionally, bisexual people are known to be excluded from sexual minority community connection (Gonzalez et al., 2021) and its buffering effect, which may contribute to their worse mental health compared to other subgroups (Doan Van et al., 2019). Consequently, understanding the unique experiences within SMW subgroups is important to detect differential experiences within this heterogeneous community.

Therefore, the goal of this study was to utilize a large, prospective cohort of US and Canadian nurses — interviewed before, during, and after pregnancy — to identify disparities by sexual orientation (broadly, and across subgroups) regarding perinatal mental health and antidepressant use among those with clinically significant perinatal depressive symptoms.

2. Methods

2.1. Sample

We examined pregnancies reported in the Pregnancy Health Sub-study within the Nurses’ Health Study 3 (NHS3), a prospective cohort study of adult nurses and nursing students born on or after January 1, 1965, and who live in the US (>95%) or Canada. Enrollment began in 2010 and is ongoing. Participants complete biennial online questionnaires (“modules”) and are regularly queried about pregnancy status. Participants who report a current pregnancy are invited to participate in the Pregnancy Health Sub-study; they receive a pregnancy questionnaire at ~20–25 weeks’ gestation and a postpartum questionnaire at ~5 weeks’ postpartum. Participants are eligible to take the Pregnancy Health Sub-study for each new pregnancy while enrolled in the NHS3.

Eligible participants reported at least one pregnancy between 2010 and February 2023, responded to both Pregnancy Health Sub-study questionnaires with non-missing information for at least one perinatal mental health outcome, and reported their sexual orientation. All study variables were from the Pregnancy Health Sub-study except for demographic variables (i.e., sexual orientation, race, ethnicity, relationship status, education level, employment status, and lifetime gravidity at baseline) which were extracted from NHS3 modules. For parsimony, we refer to our sample as women, although gender identity had not yet been systematically measured for the majority of the sample.

2.2. Outcomes

We examined three, self-reported mental health outcomes in the perinatal period: perceived stress (measured only during pregnancy), depression symptoms, and antidepressant use (both measured separately during pregnancy and postpartum).

Perceived stress was measured using the Perceived Stress Scale 4 (PSS-4). The PSS-4 is a common, self-report screening tool used to assess perceptions of situations occurring over the last month as stressful (Cohen et al., 1983) and is a useful and valid tool to measure stress during pregnancy (e.g., acceptable internal consistency α = 0.79, concurrent validity with the mental health component of the Short-Form-12 (r = −0.62; p < 0.001); Karam et al., 2012). Four items were scored using a five-point Likert scale with scores ranging 0–16. We examined perceived stress continuously (raw scores) and dichotomously using a cut-off score of ≥6 to indicate “high perceived stress” and <6 to indicate “low perceived stress,” consistent with prior studies (Kehler et al., 2022).

Depression symptoms were measured using the Edinburgh Postnatal Depression Scale (EPDS). The EDPS is a self-report screening tool that assesses perinatal depression symptomology over the last seven days and has proven an effective screening tool for major and minor depression among numerous validation studies (α = 0.87; Cox et al., 1987; Cox, 2019). Ten items were scored on a four-point Likert scale with scores ranging 0–30. We examined depressive symptoms dichotomously using multiple cut-off scores, including ≥7 (maximizing sensitivity), ≥9, ≥11, and ≥13 (maximizing specificity). Each cut-off score indicated “clinically significant depression symptoms” for its corresponding score; scores <7 indicated “not clinically significant depression symptoms.” We used multiple cut-off scores due to lack of validation of the EPDS among SMW and heterogeneity in the literature regarding clinically significant cut-off scores (Gibson et al., 2009; Levis et al., 2020; McBride et al., 2014). Moreover, evidence suggests cut-off scores may be trimester-specific (Kozinszky and Dudas, 2015), lower for minor depression (Chaudron et al., 2010) and to detect major depression among marginalized groups (Tandon et al., 2012). Stress and depression are strongly related, but each can independently predict adverse mental health symptoms (Cohen et al., 1983) and the PSS-4 has convergent validity with the EDPS (r = 0.76; p < 0.001; Karam et al., 2012).

Finally, we examined antidepressant medication use. Respondents were asked, “Regarding this pregnancy, did you take antidepressants (e.g., Prozac, Paxil, Zoloft, Celexa, etc.)?” with the option to select all that apply across many time periods. We examined antidepressant use at two time points: pregnancy (comprised of 1st, 2nd, 3rd-trimester, and during labor) and postpartum (i.e., “since delivery”).

2.3. Exposure

Sexual orientation was assessed among NHS3 participants at two time points (the 5th and 10th modules). For each pregnancy, we used the sexual orientation measured closest to the date of the corresponding pregnancy questionnaire. For those who had not yet taken the 10th module or were missing sexual orientation data at that point, we used data from the 5th module.

We assessed sexual orientation using items that measured current sexual identity and lifetime same-sex sexual partners. We measured sexual identity using an adapted item from the Minnesota Adolescent Health Survey (Remafedi et al., 1992) that asked, “Which of the following best described your feelings?” with the following response options: completely heterosexual (attracted to persons of the opposite sex), mostly heterosexual, bisexual (equally attracted to men and women), mostly homosexual, completely homosexual (gay/lesbian, attracted to persons of the same sex), and not sure. We assessed same-sex sexual partners using an item that asked, “During your LIFETIME, have you EVER had sexual contact with a FEMALE?” We assigned participants to the completely heterosexual reference group if they identified as completely heterosexual and did not report same-sex partners (hereafter, completely heterosexual). Next, we classified sexually minoritized participants into one of the following subgroups: identified as completely heterosexual while also reporting same-sex partners (hereafter, heterosexual with same-sex partners), mostly heterosexual, bisexual, mostly or completely homosexual (hereafter, combined into a lesbian subgroup) consistent with prior NHS3 research (Charlton et al., 2013, 2020). We considered disparities both by aggregating SMW, as well as across five subgroups (inclusive of the completely heterosexual reference group).

2.4. Control variables

We controlled for mental health history at the pregnancy level using two pregnancy questionnaire items. Pre-pregnancy adverse mental health is likely a mediator, rather than a confounder, of any associations between sexual orientation and perinatal depression. Consequently, we present models with and without adjustment for mental health history to determine whether sexual orientation differences in perinatal mental health persist above and beyond those that operate through known differences in pre-pregnancy mental health. We assessed pre-pregnancy depression symptoms using the item, “Before this pregnancy, was there ever a period of time when you were feeling depressed or down or when you lost interest in pleasurable activities most of the day, nearly every day, for at least 2 weeks?” (yes/no). We additionally assessed pre-pregnancy antidepressant use with the item, “Before this pregnancy, did a health professional ever prescribe a medication for you for depression?” (yes/no). As sexual orientation is a fundamental aspect of a person’s identity, many control variables that are commonly adjusted for in studies of adverse perinatal outcomes may be mediators in the present analysis. We risk introducing bias if we adjust for mediators, and in this analysis, specifically, adjustment of mediators may remove the effect of heterosexism and discrimination in community and healthcare settings that may contribute to adverse outcomes among diverse SMW.

2.5. Other study variables

We examined person- and pregnancy-level variables to characterize the sample and inform imputation of missing data; none were included in any statistical models. Person-level variables — assessed in the baseline NHS3 module — included race, ethnicity, relationship status, number of previous pregnancies, employment status, and education level. Pregnancy-level variables included pregnancy intention, age at pregnancy, time to conception among planned pregnancies, pregnancy outcomes, pregnancy complications, and gestation type. Postpartum depression diagnosis was also collected; we chose to model antidepressant use versus diagnosis since use was modeled at both perinatal time points and captures the prevalence of treated depression compared to only new cases (i.e., incidence).

2.6. Analytic strategy

The unit of analysis was the pregnancy. We first examined associations between sexual orientation and mental health to understand disparities in study outcomes. Next, we examined perinatal antidepressant use as a function of EDPS scores — within strata delineated by sexual orientation — to elucidate potential heterogeneity in antidepressant use by sexual orientation at different symptom severity levels and disparities in unmet mental healthcare needs. All models used log-binomial generalized estimating equations clustered at the participant level (to account for non-independence introduced by repeated pregnancies) and weighted using informative cluster sizes (i.e., people who experience adverse perinatal mental health may be less likely to have subsequent pregnancies). We chose to use weighting for cluster sizes (i.e., number of observed pregnancies) rather than control for pregnancy order as a fixed effect. This is because pregnancy order is not strictly a confounder, but instead a potential consequence of the outcome, and informative weighting is the recommended statistical technique to reduce bias in these circumstances (Chaurasia et al., 2018). In addition, given the method of survey administration, it is not reasonable to assume that every participant had complete enumeration of their lifetime pregnancies. All analyses were performed in R 4.2.0 (R Core Team, 2013).

2.7. Missing

Between 2–5% of pregnancies were missing data on study outcomes, and 2% were missing data on control variables. We imputed missing data using “Amelia” in R to account for clustering (Zhang, 2016). We trained the imputed data sets using outcome, exposure, and control variables, except for pregnancy outcome and intention due to positivity violations and collinearity, respectively; we re-ran the imputation on the subset of participants eligible for each sensitivity analysis that used these two variables. We imputed 10 datasets and combined estimates from all analyses using Rubin’s Rules (Rubin, 1987).

2.8. Sensitivity analyses

We performed sensitivity analyses restricted to first observed pregnancies, live births, and intended pregnancies because adverse perinatal mental health may influence the decision to have subsequent pregnancies. Additionally, pregnancy loss (Quenby et al., 2021) and pregnancy intention (Abajobir et al., 2016) influence perinatal mental health and may vary by sexual orientation. We performed a final sensitivity analysis examining medication utilization only among those who endorsed clinically meaningful depression symptoms (EDPS ≥7).

3. Results

Among the 50,411 NHS3 participants, 7,637 reported having one or more pregnancies and were enrolled in the Pregnancy Health Sub-study (N = 11,044 unique pregnancies). Among these, 1,435 were ineligible due to missing sexual orientation data, 1,433 were ineligible because they did not complete both questionnaires, and six were missing all mental health data. Among the 4,763 eligible participants, there were 6,364 unique pregnancies. Among the eligible participants, 931 unique participants (20%) identified as SMW at some point in the study, which represented 18% of the unique pregnancies.

Table 1 shows pregnancy-level study variables stratified by subgroup and a combined SMW group. Supplemental Figs. 1–3 show histograms of the distributions of perceived stress and depression symptoms by sexual orientation. Some notable features include the sample is majority non-Hispanic White (>92%), employed (>79%), and received less than a college education (>58%). Pregnancies among the lesbian subgroup made up only 1% of total pregnancies, were the least likely to endorse a prior pregnancy compared to any other subgroup (0.30 mean lifetime gravida status vs 0.53 among pregnancies to completely heterosexual participants), and had the highest proportion of multiple gestations (13% vs 2%), pregnancy complications (55% vs 44%), pregnancies at an older age (≥35 years old; 58% vs 24%), and intended pregnancies (100% vs 97%) which occurred through actively trying to conceive (100% vs 79%) and resulted in live births (100% vs 98%).

Table 1.

Distribution of demographics and pregnancy-level characteristics among Nurses’ Health Study 3 participants in the Pregnancy Health Sub-study; N = 4,763 unique participants with N = 6,364 pregnancies^a.

	Completely heterosexual	All sexual minority subgroupsb	Heterosexual, same-sex partners	Mostly heterosexual	Bisexual	Lesbian
	(82%, n = 5,178)	(18%, n = 1,168)	(4%, n = 245)	(12%, n = 77O)	(2%, n = 1O6)	(1%, n = 47)
Race, n (%)
White	4,887 (94%)	1,105 (95%)	235 (96%)	728 (95%)	97 (92%)	45 (96%)
Black	51 (1%)	12 (1%)	5 (2%)	4 (1%)	2 (2%)	1 (2%)
Asian	109 (2%)	11 (1%)	0 (0%)	11 (1%)	0 (0%)	0 (0%)
Indigenous American	66 (1%)	27 (2%)	1 (0%)	20 (3%)	5 (5%)	1 (2%)
Middle Eastern	6 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Another race	33 (1%)	4 (0%)	1 (0%)	3 (0%)	0 (0%)	0 (0%)
Missing	26 (1%)	9 (1%)	3 (1%)	4 (1%)	2 (2%)	0 (0%)
Ethnicity, n (%)
Hispanic	183 (3%)	55 (5%)	13 (5%)	35 (5%)	7 (7%)	0 (0%)
Non-Hispanic	4,972 (96%)	1,109 (95%)	213 (94%)	732 (95%)	99 (93%)	47 (100%)
Missing	23 (0%)	4 (0%)	1 (0%)	3 (0%)	0 (0%)	0 (0%)
Relationship statusc, n (%)
Married/Partnered	2,880 (56%)	593 (51%)	121 (49%)	402 (52%)	41 (39%)	29 (62%)
Single	1,812 (35%)	443 (38%)	108 (44%)	271 (35%)	49 (46%)	15 (32%)
Missing	486 (9%)	132 (11%)	16 (7%)	97 (13%)	16 (15%)	3 (6%)
Educational leveld, n (%)
< College degree	3,222 (62%)	716 (61%)	157 (64%)	458 (59%)	68 (64%)	33 (70%)
≥ College degree	1,554 (20%)	326 (28%)	71 (29%)	221 (29%)	23 (22%)	11 (23%)
Missing	402 (8%)	126 (11%)	17 (7%)	91 (12%)	15 (14%)	3 (6%)
Employment statuse, n (%)
Not working	222 (4%)	86 (7%)	16 (7%)	62 (8%)	5 (5%)	3 (6%)
Working part-time	3,440 (66%)	751 (64%)	164 (67%)	485 (63%)	70 (66%)	32 (68%)
Working full-time	1,011 (20%)	192 (16%)	41 (17%)	130 (17%)	14 (13%)	7 (15%)
Missing	505 (10%)	139 (12%)	24 (10%)	93 (12%)	17 (16%)	5 (11%)
Lifetime gravidity at baseline, mean (SD)	0.53 (1.03)	0.56 (1.03)	0.50 (0.90)	0.60 (1.09)	0.61 (1.02)	0.30 (0.67)
Age at pregnancy, n (%)
≤ 24 or younger	35 (1%)	2 (0%)	0 (0%)	1 (0%)	1 (1%)	0 (0%)
25 to 34	3,236 (62%)	613 (52%)	115 (47%)	432 (56%)	50 (47%)	16 (34%)
35 to 39	1,027 (20%)	302 (26%)	78 (32%)	167 (22%)	34 (32%)	23 (49%)
≥40	201 (4%)	54 (5%)	9 (4%)	34 (4%)	7 (7%)	4 (9%)
Missing	679 (13%)	197 (17%)	43 (18%)	136 (18%)	14 (13%)	4 (9%)
Pregnancy intention, n (%)
Actively trying to become pregnant	4,086 (79%)	933 (80%)	195 (80%)	608 (79%)	83 (78%)	47 (100%)
Not actively trying; glad to become pregnant	925 (18%)	198 (17%)	43 (18%)	136 (18%)	19 (18%)	0 (0%)
Wanted to be pregnant, but not now	124 (2%)	22 (2%)	5 (2%)	15 (2%)	2 (2%)	0 (0%)
Did not want to be pregnant now or in future	37 (1%)	9 (1%)	1 (0%)	7 (1%)	1 (1%)	0 (0%)
Missing	6 (0%)	6 (1%)	1 (0%)	4 (1%)	1 (1%)	0 (0%)
Time to conception, n (%)
Actively trying for ≤ 1 year	3,443 (66%)	766 (66%)	159 (65%)	509 (66%)	65 (61%)	33 (70%)
Actively trying for >1 year	624 (12%)	145 (12%)	33 (13%)	93 (12%)	13 (12%)	6 (13%)
Not actively trying	1,086 (21%)	229 (20%)	49 (20%)	158 (21%)	22 (21%)	0 (0%)
Missing	25 (0%)	28 (2%)	4 (2%)	10 (1%)	6 (6%)	8 (17%)
Gestational type, n (%)
Singleton	4,819 (93%)	1,077 (92%)	227 (93%)	713 (93%)	98 (92%)	39 (83%)
Multiples	112 (2%)	35 (3%)	9 (4%)	17 (2%)	3 (3%)	6 (13%)
Missing	247 (5%)	56 (5%)	9 (4%)	40 (5%)	5 (5%)	2 (4%)
Pregnancy outcome, n (%)
Live birth	5,049 (98%)	1,139 (98%)	239 (98%)	751 (98%)	102 (96%)	47 (100%)
Another outcomef	41 (1%)	10 (1%)	1 (0%)	6 (1%)	3 (3%)	0 (0%)
Missing	88 (2%)	19 (2%)	5 (2%)	13 (2%)	1 (1%)	0 (0%)
Pregnancy complicationsg, n (%)	2,289 (44%)	573 (49%)	117 (48%)	381 (49%)	49 (46%)	26 (55%)
Missing	418 (8%)	98 (8%)	22 (9%)	64 (8%)	8 (8%)	4 (9%)
Postpartum depression diagnosis, n (%)	203 (4%)	81 (7%)	13 (5%)	56 (7%)	10 (9%)	2 (4%)
Missing	200 (4%)	44 (4%)	7 (3%)	30 (4%)	5 (5%)	2 (4%)
Mental health
Antidepressant use prior to pregnancy, n (%)	1,334 (26%)	511 (44%)	104 (42%)	335 (44%)	52 (49%)	20 (43%)
Missing	94 (2%)	22 (2%)	8 (3%)	14 (2%)	0 (0%)	0 (0%)
Depression symptoms prior to pregnancy, n (%)	1,745 (34%)	656 (56%)	107 (44%)	449 (58%)	72 (68%)	28 (60%)
Missing	111 (2%)	28 (2%)	10 (4%)	18 (2%)	0 (0%)	0 (0%)
Pregnancy antidepressant use, n (%)	413 (8%)	187 (16%)	30 (12%)	121 (16%)	30 (28%)	6 (13%)
Missing	Missing cannot be determined
Postpartum antidepressant use, n (%)	469 (9%)	175 (15%)	34 (14%)	112 (15%)	25 (24%)	4 (9%)
Missing	Missing cannot be determined
Pregnancy depression (EPDSh), mean (SD)	5.69 (3.28)	6.56 (3.95)	6.28 (3.61)	6.53 (3.92)	7.55 (4.58)	6.24 (4.30)
Score ≥7, n (%)	1,478 (29%)	417 (36%)	83 (34%)	271 (35%)	48 (45%)	15 (32%)
Score ≥9, n (%)	862 (17%)	278 (24%)	59 (24%)	178 (23%)	29 (27%)	12 (26%)
Score ≥ 11, n (%)	458 (9%)	181 (15%)	35 (14%)	112 (15%)	26 (25%)	8 (17%)
Score ≥13, n (%)	232 (4%)	101 (9%)	17 (7%)	63 (8%)	16 (15%)	5 (11%)
Missing	183 (4%)	47 (4%)	12 (5%)	31 (4%)	3 (3%)	1 (2%)
Perceived stress (PSS-4i), mean (SD)	3.36 (2.66)	3.92 (2.84)	3.70 (2.65)	3.93 (2.85)	4.60 (3.02)	3.45 (2.97)
Score ≥6, n (%)	1,024 (20%)	310 (27%)	57 (23%)	211 (27%)	33 (31%)	9 (19%)
Missing	106 (2%)	22 (2%)	8 (3%)	14 (2%)	0 (0%)	0 (0%)
Postpartum depression (EPDSg), mean (SD)	5.66 (3.21)	6.42 (3.65)	6.29 (3.50)	6.47 (3.69)	6.82 (3.93)	5.30 (2.78)
Score ≥ 7, n (%)	1,449 (28%)	424 (36%)	92 (38%)	280 (36%)	42 (40%)	10 (21%)
Score ≥ 9, n (%)	821 (16%)	255 (22%)	51 (21%)	171 (22%)	27 (25%)	6 (13%)
Score ≥ 11, n (%)	438 (8%)	139 (12%)	23 (9%)	97 (13%)	16 (15%)	3 (6%)
Score ≥ 13, n (%)	225 (4%)	80 (7%)	15 (6%)	52 (7%)	12 (11%)	1 (2%)
Missing	237 (5%)	53 (5%)	8 (3%)	36 (5%)	5 (5%)	4 (9%)

^aFor this table, descriptive statistics are shown using complete case sample before imputation, so percentages might sum to <100% due to missing.

^bSexual minority includes the following subgroups: heterosexual with same-sex partners, mostly heterosexual, bisexual, and lesbian women.

^cRelationship status was dichotomized into married (i.e., endorsed being married or living with a partner) and single (i.e., endorsed being never married, divorced, separated, or widowed).

^dEducation status was binarized into less than a college degree (i.e., participants who endorsed some high school, high school graduate or the equivalence, trade/vocational school certificate/degree, some college) and college degree of higher (i.e., participants who endorsed an associate’s degree, bachelor’s degree, master’s degree, or doctoral degree).

^eEmployment is only inclusive of hours worked as a nurse.

^fAnother outcome includes: stillbirth, miscarriage, induced abortion, tubal or ectopic pregnancy, and molar pregnancy.

^gPregnancy complications include gestational diabetes, gestational hypertension, pre-eclampsia/toxemia, bacterial vaginosis, urinary tract infection, other genitourinary infection, mastitis, preterm labor, premature rupture of membranes, bleeding during pregnancy, or neonatal intensive care stay.

^hThe Edinburgh Postnatal Depression Scale (EDPS).

ⁱThe Perceived Stress Scale 4 (PSS-4).

Pregnancies to bisexual participants had the highest proportion of single participants (46% unmarried or not living with a partner vs 35% among completely heterosexual participants), and exhibited the highest levels (mean scores and case percentages) of perceived stress and depression compared to any sexual orientation subgroup. Importantly, pregnancies to bisexual and lesbian participants had nearly double the proportion of pre-pregnancy antidepressant use (49% and 43%, respectively, vs 26%) and pre-pregnancy depression symptoms (68% and 60%, vs 34%). However, pregnancies to mostly heterosexual and bisexual participants had double the proportion of antidepressant use during pregnancy (16% and 28% vs 8%) and postpartum (15% and 24%, vs 9%).

Model results for associations between sexual orientation, stress, depression, and antidepressant use are shown in Table 2, with and without adjustment for prior adverse mental health history and antidepressant use. Associations between sexual orientation and continuously measured scores for stress and depression are shown in Supplemental Table 1; findings are consistent with those shown in the main text.

Table 2.

Risk of adverse perinatal mental health across sexual orientation groups among Pregnancy Health Sub-study participants,^a N = 4,763 unique participants with N = 6,364 pregnancies.

Mental health during pregnancy		Completely heterosexual	All sexual minority subgroupsc	Heterosexual with same-sex partners	Mostly heterosexual	Bisexual	Lesbian
		(82%, n = 5,178)	(18%, n = 1,168)	(4%, n = 245)	(12%, n = 770)	(2%, n = 106)	(1%, n = 47)
			Risk Ratio (RR) (95% CI)	RR (95% CI)	RR (95% CI)	RR (95% CI)	RR (95% CI)
Perceived Stress Score (PSS) ≥6	Unadjusted Adjustedb	1.00 (ref)	1.35 (1.20, 1.52) 1.14 (1.02, 1.28)	1.20 (0.93, 1.54) 1.07 (0.84, 1.37)	1.40 (1.23, 1.61) 1.17 (1.02, 1.34)	1.49 (1.10, 2.03) 1.21 (0.89, 1.63)	0.96 (0.53, 1.73) 0.82 (0.46, 1.46)
Edinburgh Postnatal Depression Scale (EDPS), score ≥7	Unadjusted Adjustedb	1.00 (ref)	1.25 (1.14, 1.38) 1.09 (1.00, 1.20)	1.15 (0.95, 1.40) 1.06 (0.88, 1.28)	1.26 (1.13, 1.40) 1.09 (0.97, 1.21)	1.54 (1.24, 1.92) 1.28 (1.03, 1.58)	1.06 (0.65, 1.71) 0.93 (0.59, 1.48)
EPDS, score ≥9	Unadjusted Adjustedb	1.00 (ref)	1.44 (1.27, 1.63) 1.20 (1.05, 1.36)	1.42 (1.11, 1.83) 1.27 (1.00, 1.61)	1.42 (1.23, 1.65) 1.17 (1.01, 1.35)	1.61 (1.16, 2.24) 1.26 (0.92, 1.73)	1.38 (0.79, 2.43) 1.17 (0.69, 2.01)
EPDS, score ≥11	Unadjusted Adjustedb	1.00 (ref)	1.76 (1.49, 2.09) 1.37 (1.16, 1.63)	1.59 (1.11, 2.27) 1.36 (0.97, 1.91)	1.67 (1.36, 2.05) 1.28 (1.04, 1.57)	2.80 (1.95, 4.02) 1.99 (1.40, 2.83)	1.84 (0.91, 3.73) 1.47 (0.75, 2.86)
EPDS, score ≥13	Unadjusted Adjustedb	1.00 (ref)	2.03 (1.60, 2.57) 1.49 (1.18, 1.89)	1.69 (1.01, 2.84) 1.38 (0.84, 2.26)	1.92 (1.44, 2.54) 1.39 (1.05, 1.84)	3.32 (2.00, 5.52) 2.26 (1.39, 3.67)	2.63 (1.15, 6.03) 1.95 (0.87, 4.34)
Pregnancy antidepressant use	Unadjusted Adjustedb	1.00 (ref)	1.95 (1.63, 2.33) 1.21 (1.03, 1.43)	1.50 (1.02, 2.20) 0.99 (0.70, 1.41)	1.92 (1.55, 2.37) 1.18 (0.97, 1.43)	3.40 (2.34, 4.94) 1.87 (1.36, 2.57)	1.51 (0.63, 3.59) 1.02 (0.48, 2.18)
Mental health postpartum
EPDS, score ≥7	Unadjusted Adjustedb	1.00 (ref)	1.29 (1.18, 1.42) 1.12 (1.02, 1.23)	1.35 (1.13, 1.62) 1.22 (1.02, 1.45)	1.28 (1.15, 1.43) 1.10 (0.99, 1.23)	1.44 (1.12, 1.84) 1.19 (0.94, 1.52)	0.88 (0.54, 1.45) 0.76 (0.47, 1.24)
EPDS, score ≥9	Unadjusted Adjustedb	1.00 (ref)	1.38 (1.20, 1.57) 1.13 (0.99, 1.29)	1.33 (1.02, 1.73) 1.18 (0.91, 1.53)	1.39 (1.19, 1.62) 1.12 (0.96, 1.31)	1.63 (1.16, 2.29) 1.24 (0.89, 1.74)	0.86 (0.41, 1.83) 0.73 (0.35, 1.51)
EPDS, score ≥11	Unadjusted Adjustedb	1.00 (ref)	1.41 (1.17, 1.70) 1.11 (0.92, 1.35)	1.24 (0.83, 1.84) 1.06 (0.72, 1.56)	1.45 (1.17, 1.79) 1.13 (0.91, 1.40)	1.73 (1.06, 2.83) 1.29 (0.80, 2.08)	0.90 (0.31, 2.67) 0.72 (0.25, 2.13)
EPDS, score ≥13	Unadjusted Adjustedb	1.00 (ref)	1.58 (1.22, 2.05) 1.16 (0.90, 1.51)	1.58 (0.95, 2.61) 1.26 (0.76, 2.08)	1.51 (1.11, 2.04) 1.09 (0.80, 1.48)	2.62 (1.46, 4.69) 1.77 (1.00, 3.14)	0.55 (0.08, 3.98) 0.41 (0.06, 2.96)
Postpartum antidepressant use	Unadjusted Adjustedb	1.00 (ref)	1.64 (1.37, 1.96) 1.05 (0.89, 1.24)	1.50 (1.04, 2.17) 1.00 (0.71, 1.41)	1.61 (1.30, 1.99) 1.03 (0.85, 1.25)	2.53 (1.69, 3.80) 1.47 (1.03, 2.11)	0.80 (0.25, 2.62) 0.56 (0.19, 1.67)

^aEach row denotes the estimates from a single log-binomial generalized estimating equations regression model specifying mental health (row headers) as the dependent variable and sexual orientation (column headers) as a categorical variable.

^bAdjusted estimates included covariate adjustment for mental health history and prior antidepressant use. The estimates shown are generated from a total of 22 separate regression models fitted to the data.

^cSexual minority includes the following subgroups: heterosexual with same-sex partners, mostly heterosexual, bisexual, and lesbian women.

Across all unadjusted models, we observed an elevated risk of adverse mental health and increased antidepressant use among SMW compared to completely heterosexual respondents, though these attenuated in adjusted models. Compared to pregnancies among completely heterosexual women, pregnancies to SMW were associated with increased stress during pregnancy (adjusted risk ratio [ARR] for PSS-4: 1.14, 95% confidence interval [CI] [1.02–1.28]). These increases were driven by elevated risks among the mostly heterosexual subgroup (ARR: 1.17 [1.02–1.34]) and non-significant higher magnitudes of risk among the bisexual subgroup (ARR: 1.21 [0.89–1.63]). No meaningful differences in stress were observed for the subgroups of heterosexual with same-sex partners or lesbian participants.

Regarding depression during pregnancy, SMW were at elevated risk of depression at the pregnancy level for every level of EDPS score cutoff, with the magnitude of the disparity increasing as the cutoff became more stringent (ARRs: 1.09 [1.00–1.20]; 1.20 [1.05–1.36]; 1.37 [1.16–1.63]; 1.49 [1.18–1.89] for EDPS scores ≥7, ≥9, ≥11, ≥13, respectively). These increases were driven by elevated risks among the mostly heterosexual and bisexual subgroups, with some evidence for elevated risks among the heterosexual with same-sex partner subgroup at certain cut-offs (e.g., ARR: 1.27 [1.00–1.61] for score ≥9, although all scores crossed the null). For the lesbian subgroup in all adjusted models, the relative risk of depression included the null at every EDPS score. We observed similar patterns for antidepressant use during pregnancy (ARRs for all SMW: 1.21 [1.03–1.43], mostly heterosexual participants: 1.18 [0.97–1.43], and bisexual participants: 1.87 [1.36–2.57]). No meaningful differences were observed for the lesbian subgroup.

Regarding postpartum depression, pregnancies among SMW evidenced higher magnitudes of risk in adjusted models but rarely reached statistical significance (ARRs: 1.12 [1.02–1.23]; 1.13 [0.99–1.29]; 1.11 [0.92–1.35]; 1.16 [0.90–1.51] for EDPS scores ≥7, ≥9, ≥11, ≥13, respectively). We did not observe the same pattern of increasing magnitudes of disparity as depression severity increased that we observed with pregnancy depression, and, in fact, observed the opposite pattern among the lesbian subgroup. Among subgroups at the lowest cutoffs, we observed elevated risks among the heterosexual with same-sex partners subgroup (ARR: 1.22 [1.02–1.45]) and marginal, non-significant increases among the mostly heterosexual and bisexual subgroups (ARRs 1.10 [0.99–1.23]; 1.19 [0.94–1.52], respectively). We observed similar trends for postpartum antidepressant use, with increased use among the bisexual subgroup only (ARR: 1.47, 95% CI [1.03–2.11]) and no other meaningful differences.

Fig. 1 shows associations between perinatal antidepressant use and depression severity, stratified by sexually minoritized status (numeric estimates provided in Supplemental Table 2). We were unable to examine differences within subgroups due to small cells. Among the completely heterosexual reference group, use of antidepressants during pregnancy and postpartum increased as the severity of depression increased (e.g., ARR comparing EDPS score 11–13 to <7 during pregnancy: 1.54 [1.13–2.09]; ARR comparing score ≥13 to <7: 1.72 [1.33–2.21]). Among SMW, although the magnitudes of risk were similar to the heterosexual comparison group across symptom severity during pregnancy, during postpartum the estimates typically included the null; the exception was for pregnancies among SMW with a score of ≥13, which was associated with increased antidepressant use (relative to heterosexual sample with scores <7) during postpartum (ARR: 1.46 [1.05, 2.04)]. At all other EDPS severity scores during postpartum, SMW had no greater use of antidepressants than those with scores <7. Tests of interaction between EDPS score categories and sexual orientation showed no statistical heterogeneity by strata (p > 0.05).

Fig. 1.

Likelihood of antidepressant use at different severities of depression symptoms, during pregnancy and postpartum, across sexual orientation groups among Pregnancy Health Sub-study participants, N = 4,763 participants with N = 6,364 pregnancies

All models adjusted for mental health history and prior antidepressant use; Sexual minority women include the following subgroups: heterosexual with same-sex partners, mostly heterosexual, bisexual, and lesbian women.

Observations from sensitivity analyses restricting to first observed pregnancies (Supplementary Table 3), live births (Supplemental Table 4), and intended pregnancies (Supplemental Table 5) did not meaningfully differ from results shown in main analyses — neither did sensitivity analyses examining medication utilization among only those who met clinically meaningful depression criteria (Supplemental Table 6). While we observed increased magnitudes of risks of severe depression (pregnancy EDPS score ≥13) among the lesbian subgroup in the analysis restricted to first observed pregnancies (ARR: 2.44 [1.09–5.48]) and intended pregnancies (ARR: 2.17 [0.97–4.87]), there were very few observations that met clinical criteria so we cannot rule out spurious findings.

4. Discussion

This study examined sexual orientation-related differences in perinatal stress, depression, and antidepressant use in a cohort of nurses living in the US or Canada. We found perinatal SMW are more likely to experience stress, depression, and antidepressant use compared to their completely heterosexual counterparts. We also found SMW’s use of antidepressant medications was differentially associated with symptom severity during postpartum, namely only once symptoms became severe (≥13). Pregnancies to mostly heterosexual and bisexual participants exhibited the highest disparities in perceived stress, depression symptoms (during pregnancy among bisexual participants only), and antidepressant use; SMW experience differential risks for adverse perinatal mental health and should be examined as distinct populations.

Our findings overall support prior evidence from small convenience samples of SMW. Raw depression scores were generally consistent with prior scores; a study of lesbian and bisexual women reported a mean postpartum EDPS score of 7.4 (Yager et al., 2010), just one point higher than our combined SMW group score of 6.4. Our sample had a similar percentage of SMW endorsing clinically significant pregnancy depression symptoms compared to a Canadian sample using a ≥13 cut-off (9% vs 6%), although a lower percentage of postpartum symptoms (7% vs 17%; Ross et al., 2007). This is notable given the structural support provided to postpartum parents in Canada (e.g., 12–18 months of paid parental leave) compared to parents in the US. A final study compared EDPS scores between perinatal SMW recruited through consecutive (e. g., every eligible participant is selected) and convenience methods and found higher scores among the consecutive sample (7.1 vs 4.7 in pregnancy and 5.4 vs 8.9 postpartum; Flanders et al., 2016), which our combined SMW group scores fell between (6.6 during pregnancy and 6.4 postpartum).

Our findings underscore the need to consider SMW subgroups with more granularity, given their heterogeneity of risks. Higher levels of depression among bisexual women were consistent with prior studies of bisexual and male-partnered SMW. Mean EDPS scores were similar to scores in the literature of lesbian and female-partnered SMW during pregnancy (mean score of 6.2 among our study vs 5.6–6.4 in the previous literature) and postpartum (5.3 vs 5.2), and were lower compared to bisexual and male-partnered SMW during pregnancy again in our sample and the literature (7.6 vs 7.3–8.8) as well as during postpartum (6.8 vs 6.8; Goldberg et al., 2020; Ross et al., 2007, 2017). Similar trends of higher scores among bisexual compared to lesbian women were also reported in a recent online convenience sample of SMW from the US and Australia (Marsland et al., 2022); however, sample scores were notably higher among both groups (12.8 among bisexual women and 9.0 among lesbian women).

No study to our knowledge has looked at perinatal mental health among heterosexual women with same-sex partners or mostly heterosexual women. This is important as the mostly heterosexual subgroup had notably elevated risks of pregnancy stress and depression symptoms secondary to bisexual women, and both groups drove most outcomes among unadjusted scores. Without precise and nuanced measurement of sexual orientation, some SMW may be grouped into a heterosexual reference group; our findings suggest this would obfuscate mental health disparities. Results also indicate these are additional communities at risk of poor perinatal mental health and warrant investigation as a distinct disparity group. Furthermore, it is unclear if sexual identity compared to sexual partnership may differentially impact perinatal mental health; our data were unable to examine these questions.

We are the first study to our knowledge to look at multiple EDPS cut-off scores among SMW. Our findings suggest evidence of sexual orientation-related disparities at every cut-off score, indicating disparities persist across all levels of symptom severity. However, we also found that, during pregnancy, the disparity increased in magnitude with increasing symptoms severity (i.e., specificity) suggesting that SMW are disproportionately experiencing more severe symptoms during this time and may need more screening and treatment.

Additionally, we found that antidepressant use during postpartum increased with symptom severity among the reference group of completely heterosexual participants, as expected, but was only significantly associated with increased antidepressant use at the most severe symptom severity levels ≥13 among SMW. We interpret these findings to imply that perinatal SMW may be prescribed antidepressants differentially compared to their completely heterosexual peers, specifically only once the level of symptom severity becomes severe. This finding speaks to an underlying concern regarding barriers to perinatal mental health treatment among SMW. Barriers likely exist at the individual-level (e.g., fear of having parental rights taken away with a mental illness diagnosis) and system-level; health professionals have reported insufficient preparation to provide perinatal care to SMW (Johansson et al., 2020; Wingo et al., 2018), which likely extends to perinatal mental health care as well.

However, the magnitude of sexual orientation-related perinatal mental health disparities attenuated somewhat after pregnancy. Elevated risks of depression among pregnant SMW, rather than during postpartum, may be explained by the frequency of receiving prenatal vs postpartum care within a heteronormative healthcare system, which has been well documented (Permezel et al., 2023). Yet, as this trend was driven by the mostly heterosexual and bisexual subgroups, it is also possible that the reinforced heteronormative model of pregnancy and parenting in healthcare and society causes increased stress among those who felt invisible and “straightened” by their pregnancy experience (Manley et al., 2018; Soled et al., 2023), on top of the invisibility, erasure, and exclusion experienced from the greater sexual minority community during non-perinatal times (Ross et al., 2018). Conversely, a recent study using medical records of SMW from a Chicago medical center found mental health worsened across the perinatal period, with 138% increased odds of screening positive for postpartum depression using a 9-item Patient Health Questionnaire (Lapping-Carr et al., 2023). It is possible that sample differences drove these opposing trends, namely increased racially minoritized and single participants in the Chicago sample. Future research with large and diverse samples of SMW should investigate known factors that may mediate depression among these groups (e.g., pregnancy intention) as well as minority-specific factors (e.g., intersectional discrimination).

Furthermore, the magnitudes of effect were more elevated prior to adjustment of a history of antidepressant use or poor mental health, highlighting the importance of mental health differences prior to the perinatal period. In line with prior work showing a higher endorsement of lifetime depression among pregnant lesbian and bisexual women compared to heterosexual women (Gonzales et al., 2019), SMW participants had higher levels of pre-pregnancy depression and antidepressant use. Prior work has also found higher use of mental health services among SMW in the preconception period compared to pregnancy and found those who conceived via sexual intercourse with men to be more likely to report unmet needs for mental health services compared to those who conceived through a sperm bank or known donor (Steele et al., 2008).

Yet, compared to lifetime mental health, preconception mental health literature among SMW has been mixed. Emotional challenges have been consistently reported among qualitative reports of SMW trying to conceive (Rogers, 2020; Topper and Bauermeister, 2022); yet quantitative data of lesbian and bisexual women trying to conceive showed no elevated anxiety and depression symptoms (Borneskog et al., 2013; Yager et al., 2010). The discrepancy between qualitative and quantitative results could be explained by reduced disclosure in questionnaires due to fear that their access to parenthood will be revoked if labeled mentally unfit. It is also possible that these mental health instruments may not be sensitive to the unique stressors SMW experience and need to be revised to capture minority-specific stressors. Additionally, this could suggest mental health resources and support prior to pregnancy, especially among bisexual women and women partnered with men, could be critical to connecting SMW to care and the primary prevention of adverse perinatal mental health, as well as its sequelae of adverse birth and infant outcomes.

4.1. Limitations

This study is the first to our knowledge to document perinatal mental health outcomes by sexual identity groups in a national sample and overall has a small amount of missingness among study variables; yet, there are several limitations. The sample is comprised of nurses with a high level of health and healthcare system literacy; this may influence access to mental health treatment as well as knowledge of perinatal mental health risks, reducing the severity of mental health symptoms in this sample compared to a population sample (i.e., the estimates may underestimate the true population disparity; Tambling et al., 2021). Only participants who were pregnant were invited to participate in the study, excluding those who contributed to a pregnancy without being the gestational carrier such as through reciprocal IVF. Additionally, participants whose pregnancies were not carried to 20–25 weeks’ gestation were not surveyed (i.e., outcomes that ended in a miscarriage or elective abortion), which implies our estimates may be more conservative than if these pregnancies were included. Prior to 2022, participants provided the questionnaire designed for males would not receive female-specific questions (e.g., pregnancy), thus experiences among transgender men are likely excluded. However, limited gender identity data were collected at enrollment so participants’ gender could not be accurately assessed to elucidate variation across gender identity. Additionally, although the NHS3 has a large sample of racial and ethnic minority participants, this analytic sample did not. Prior research has revealed perinatal and infant health disparities are most pronounced among SMW of color (Everett et al., 2021) so future research around individuals who are multiply marginalized — such as by sexual orientation and race/ethnicity — is particularly needed.

We examined only stress, depression, and antidepressant use, but SMW experience high levels of post-traumatic stress disorder (PTSD; Weiss et al., 2015) and adverse childhood experiences (Bochicchio et al., 2024), which both may contribute to higher levels of adverse perinatal mental health. In fact, one study found high levels of fear of childbirth among sexually minoritized childbearing people which also highly co-occurred with anxiety and depression symptoms (Hallström et al., 2022). Measuring multiple dimensions of perinatal mental health including PTSD, obsessive-compulsive disorder, and psychosis is needed moving forward. It is also important to explore mental health symptomology further given the variance in sociocultural expressions of mental distress (e.g., somatic versus affective symptoms among Chinese Americans; Zhu, 2018) and stigma around mental health (e.g., stigma around depression in Black populations; Campbell and Mowbray, 2016) that may result in differential measurement error of certain psychological experiences among SMW with multiple marginalized identities.

We could not adjudicate between null results and being underpowered, especially among the lesbian subgroup and when adjusting for a mental health history and prior antidepressant use. Small sample sizes further restricted us from examining differences across additional demographic variables such as race, ethnicity, and education, which we originally planned to adjust for in the analyses. We also could not differentiate between “no” and “missingness” (i.e., non-response) among antidepressant use; however, the low percent of missingness (e.g., 2–5% for pregnancy stress and perinatal depression) among other variables reassures us that misclassification due to missingness is likely low for this item. Participants who endorsed same-sex reproduction were excluded from survey questions regarding infertility diagnoses so we could not accurately report this variable. Additionally, pregnancy intention was measured at the time of pregnancy and may not have represented preconception intention.

The postpartum timepoint is proximal to the date of birth (~5 weeks postpartum) and may not have fully captured postpartum depression, worsening of symptoms, and antidepressant use across the entire postpartum period; depression symptoms can increase across the first 12 months postpartum (Rosander et al., 2021) and linger up to three years postpartum (Putnick et al., 2020). We could not ascertain the exact timeframe (e.g., weeks’ gestation) questionnaires were completed which limits the time-specific implications of participants’ mental health symptoms. It is unlikely, however, that these measurement limitations would be differential by sexual orientation, thus we are confident that any biases introduced by these errors would be towards, rather than away, from the null.

4.2. Implications

To understand the full picture of SMW’s perinatal mental health, future research must control for those engaged in treatment (e.g., medication, therapy), evaluate the spectrum of mental health disorders, appraise baseline mental health levels prior to the event of interest, and elucidate mental health trajectories from preconception through late postpartum. Results should additionally be replicated in samples that represent the general population, have larger numbers of SMW, and include individuals with diverse gender identities. Accurate and repeated data collection on sexual orientation and gender identity is critical in population databases that assess reproductive and perinatal health outcomes such as in the CDC’s Pregnancy Risk Assessment Monitoring System. Future research needs to examine if and how sexual orientation constructs (e.g., sexual identity versus partnership) contribute to adverse perinatal outcomes (including poor mental health) and barriers to accessing care during the perinatal period. Perinatal mental health measurement tools need to be validated among SMW to determine if stress and depression are being assessed in culturally competent ways as well as determine the most appropriate cut-off scores for clinically significant symptoms particularly given the high levels of poor mental health SMW experience at baseline. Additionally, culturally competent training is also needed among primary care, perinatal, and mental health care professionals. Furthermore, clinical perinatal screening guidelines should be responsive to the higher risk of mental distress in SMW, for example starting screening during the preconception period and follow-up screening through late postpartum.

5. Conclusions

This study is the largest study to our knowledge to examine sexual orientation-related differences in perinatal stress, depression, and antidepressant use, and the only study to include nuanced sexual orientation groups. We showed SMW have elevated risks for pregnancy stress and perinatal depression and antidepressant use compared to their completely heterosexual counterparts, that these risks are not equally shared among all SMW, and that their antidepressant use does not mirror symptom severity. This study adds to the building evidence base that SMW experience perinatal mental health disparities and calls for further research and clinical interventions to understand and reduce this disparity during a vulnerable time in the lifespan.

Abbreviations:

SMW

Sexually minoritized women

NHS3

Nurses’ Health Study 3

PSS-4

Perceived Stress Scale 4

EPDS

Edinburgh Postnatal Depression Scale