A systematic review of associations between hormone levels in hair and peripartum depression

Abstract

Peripartum depression is a global health concern, characterized by mood disturbances inclusive of pregnancy through up to 12-months postpartum. Hormones play a vital role in pregnancy maintenance, fetal development, and labor and delivery and change significantly as a function of pregnancy and childbirth. However, such life sustaining changes may have consequences related to risk for peripartum depression. To date, most studies that have examined hormones in relation to peripartum depression have focused on blood or saliva sampling approaches, though hair analysis offers unique information on trajectories of hormone concentrations over more sustained periods of time (i.e., over months). The aim of this systematic review was to provide a comprehensive review of the association between hair measures of hormones (i.e., cortisol, progesterone, estrogen, and testosterone) and depression during the peripartum period. Forty-one studies were identified for inclusion. A majority of studies reported statistically null associations. Between-person studies varied widely in reported direction and magnitude of hair hormone–depression associations, most likely attributable to a wide range of methodological approaches including timing of assessments and sample size. Studies using within-person approaches observed positive coupling of cortisol concentration and symptoms across time. Most studies focused exclusively on cortisol. We recommend future research consider both stress and reproductive hormones, prioritize within-person change in hormone levels given this is a period of dramatic change, and include contextual (e.g., social support, adverse and benevolent childhood experiences, physical and psychiatric conditions) features that may modify both changes in hormones and the association between hormone levels and depression in the peripartum period.

1. Introduction

Four million pregnant people give birth each year in the US (Martin et al., 2019), with approximately 13% experiencing a depressive episode during pregnancy and 20% experiencing depression in the postpartum period (Gavin et al., 2005). A depressive episode that occurs during pregnancy or within the first 12 months following childbirth is referred to as peripartum depression (PPD). PPD is associated with lower quality of life, disturbances in social relationships, and suicidal ideation and risk in pregnant people and new parents (Lindahl et al., 2005; Slomian et al., 2019). Meta-analytic work suggests that PPD is also associated with a range of adverse outcomes for infant offspring including higher rates of physical health concerns and impairments in cognitive and language development (Slomian et al., 2019). Several studies have also observed higher rates of internalizing and behavioral problems in children exposed to birthing parent PPD (Bagner et al., 2010; Walker et al., 2013). Moreover, birthing parent PPD symptoms are associated with differences in offspring brain development observed across childhood (Zou et al., 2019). Consequently, early identification and treatment of PPD is likely to have cascading benefits for both pregnant people and their offspring. Many of the known risk factors for PPD are the same as those identified for depression during other life stages (Dagher et al., 2021). However, PPD rates are higher than rates of major depressive disorder outside of this context (Lim et al., 2018; Underwood et al., 2016) and the dramatic neurobiological changes, including hormonal changes, that occur at this time implicate systems that are known to be associated with depression (Soares & Zitek, 2008).

Across pregnancy and the postpartum period, hormone levels—including progesterone, estrogen, testosterone, oxytocin, and cortisol—change dramatically to support pregnancy, fetal growth, and labor (Carlsen et al., 2006; Di Renzo et al., 2016; Douglas, 2010; Druckmann & Druckmann, 2005; Feldman et al., 2007; Galbally et al., 2011; Henry & Sherwin, 2012; Marceau et al., 2021; McCarthy, 2008; Murphy & Clifton, 2003; Prevost et al., 2014; Schiller et al., 2015; Schock et al., 2016; Yim et al., 2015). Progesterone, testosterone, and estradiol, the most common form of estrogen in the body during reproductive years, similarly demonstrate patterns wherein levels gradually increase across pregnancy and decrease after childbirth (Luisi et al., 2000; Marceau et al., 2021; McCarthy, 2008; Schock et al., 2016). Oxytocin levels increase during the first to third trimesters and additional oxytocin is released immediately after birth (Feldman & Bakermans-Kranenburg, 2017). In summary, hormone changes during the peripartum period support the maintenance of the pregnancy, the development of fetal organs and labor, and the production of breast milk (Galbally et al., 2011; Schock et al., 2016). Changes in hormone levels during pregnancy and the postpartum period are likely to influence brain function and behavior (Carmona et al., 2019; Servin-Barthet et al., 2023), though groups differ to the degree that they theorize these changes are “by design” (Carmona et al., 2019; Grattan & Ladyman, 2020; Servin-Barthet et al., 2023) vs. byproducts of pregnancy and childbirth (Cárdenas et al., 2020).

In the past decade, new hormone based treatments for postpartum depression have been developed, including administration of Brexanolone (Kanes et al., 2017; Lüscher & Möhler, 2019) and Zuranolone (Deligiannidis et al., 2023). These treatments are more effective than traditional first line treatments for depression (i.e., selective serotonin reuptake inhibitors; SSRIs; (Kaufman et al., 2022)). The targets for these medications were selected based on evidence that estradiol and progesterone impact several neural circuits implicated in depression, including the dorsolateral prefrontal cortex, amygdala, orbitofrontal cortex, and striatum (Albert & Newhouse, 2019; Schiller et al., 2015; Standeven et al., 2020). Prominent conceptualizations of PPD–hormone associations suggest that some individuals may be especially sensitive to hormone alterations due to life history or current psychosocial factors (Albert & Newhouse, 2019; Bloch, 2000; Nestler et al., 2002; Schiller et al., 2015; Yim et al., 2015). Schiller and colleagues (2015) suggest that hormone changes may be one potential neurobiological mechanism of peripartum depression, in a specific reproductive state, for a subset of people, alongside a host of other psychosocial risk factors.

In addition to oxytocin, estrogen, and progesterone associations with PPD, hypothalamic-pituitary-adrenal (HPA) axis function, most often measured by cortisol output, has been implicated in depression (Nestler et al., 2002; Schiller et al., 2015; Seth et al., 2016; Yim et al., 2015). During pregnancy, the concentration of cortisol in plasma is exponentially greater than pre-pregnancy levels (Campbell et al., 1987). This high level of cortisol is believed to affect pregnant people’s responses to stress (Seth et al., 2016), given cortisol is highly involved in emotional and cognitive regulation (Jentsch et al., 2019; Stokes, 1995). However, differences in cortisol between pregnant and non-pregnant people are not always observed, and within-person approaches reveal there is a large amount of intraindividual variability over time (Marceau et al., 2021). Preliminary evidence suggests that levels of cortisone, the inactive metabolite of cortisol, should also be considered alongside cortisol to inform associations with stress because, while related, cortisone seems to have distinct associations with reports of prenatal stress (Scharlau et al., 2018).

Most research examining hormones in relation to PPD obtained estimates of hormones from blood or saliva (Orta et al., 2018). Though there are many methods of assessing hormone concentrations—including blood, urine, and saliva tissue—obtaining estimates via hair could add unique information alongside alternative methods (Hobo et al., 2020), especially in the prenatal period (Kirschbaum et al., 2009). Specifically, hair analysis provides a non-invasive method of evaluating hormone concentration and allows for assessment of hormonal changes over time without the need for multiple sample extractions (Sauvé et al., 2007). Hair grows at a rate of approximately 1 cm per month, and the steroid hormone concentrations taken from hair samples mirror average hormone levels from the corresponding preceding months (Hobo et al., 2020). Progesterone, estradiol, testosterone, and cortisol can each be extracted from hair (Stern et al., 2022; Wang et al., 2020), with preliminary studies demonstrating this may also be possible for oxytocin with developing methodology (Tabak et al., 2023). Furthermore, the collection and storage of hair samples for analysis is relatively inexpensive, hair sampling from participants does not require professional training, and collected samples are able to be stored at room temperature (Gow et al., 2010; Rodrigues et al., 2008). Proof-of-concept work suggests that hair measures of cortisol accurately reflect production up to 6 months prior and are unaffected by hair texture, hair color, or frequency of hair washes (Kirschbaum et al., 2009).

Importantly, measuring hormone levels in hair allows for the unique ability to capture hormone levels over extended periods of time. In the case of cortisol, for example, saliva samples have long been conceptualized to partly measure acute stress responses (e.g., sympathetic response to the current surrounding environment); however, contemporary literature suggests cortisol output may only be weakly associated with stress and early adversity (Malanchini et al., 2021). The extent to which cortisol is associated with depression is mixed; reviews indicate that the association may be largely dependent on smaller scale temporal fluctuations in cortisol and depression severity (Burke et al., 2005; Dedovic & Ngiam, 2015; Goldstein & Klein, 2014; Herbert, 2013). There may be utility in examining cortisol trajectories across more extended periods of time (Malanchini et al., 2020; Russel et al., 2012) as well as pregnancy-related changes that are unrelated to exposures to stress (Cárdenas et al., 2020).

Hair analysis methods for endogenous compounds were developed more recently relative to alternative measurement methods (Gow et al., 2010). The first evaluation of cortisol in human hair published in the 2000s (Sauvé et al., 2007), and multiple quantification techniques are available after hormone extraction. Interassay variability can be high. It is recommended that researchers conduct all assays from a single study in the same batch and use internal controls to ensure consistency (Russell et al., 2012). Limitations of hair analysis of hormones include the possibility of wash out with repeated water exposure, which may vary by person and lead to underestimation (Hamel et al., 2011; Staufenbiel et al., 2015). Additionally, saliva and blood extraction methods provide higher temporal precision, including diurnal fluctuations and responses to acute events. Saliva and blood measures of hormone concentrations provide estimates at the time of measurement, but hair samples provide more sustained levels and allow for charting of trajectories over the full peripartum period. For example, while testosterone and progesterone levels extracted from saliva and hair methods are moderately correlated, metrics obtained via saliva and hair are only weakly associated for cortisol assessment (Stern et al., 2022). Taken together, associations between symptoms and hormones measured via momentary (saliva, blood) and sustained (hair) sampling methods may each provide unique information. Hair analysis offers a non-invasive measure of hormone levels, is feasible within most laboratory settings with little training or storage expenses, and may be particularly well suited to examine hormonal changes across time scales relevant to the peripartum period.

Given the utility of obtaining estimates of hormones via hair analysis, the present review focused on studies that used hair sampling of hormones across the peripartum period. Notably, though not specifically focused on the perinatal period, two reviews examined associations between hair cortisol in depression (Koumantarou Malisiova et al., 2021; Psarraki et al., 2021), and concluded that the literature is inconsistent. Additionally, two reviews examined associations between depression and cortisol (and not other hormones) in pregnancy (Mlili et al., 2023; Orta et al., 2018). These reviews also concluded that the literature is inconsistent. In particular, Orta and colleagues (2018) found that most of the studies obtained cortisol estimates through blood and saliva, and Mlili and colleagues (2023) identified 9 studies which reported positive, negative, and statistically null results. As such, this systematic review aimed to build on prior research by exploring the associations between multiple hormones, measured through hair samples, and depression during the peripartum period. Hair sampling allows for unique temporal insights, enabling us to examine how hormone levels change over longer spans of time in relation to PPD risk. We hypothesized that hair hormone levels would reveal trajectories related to PPD risk. Specifically, we examined: (1) the timing, direction, and magnitude of associations between hair hormone levels and PPD, (2) whether hormone levels at one time point predicted later PPD, and (3) whether hormone trajectories better explained PPD than single time-point assessments. Additionally, we integrated findings across studies, identified gaps in the literature, and discussed limitations and future directions for this research.

Method

2.1. Review Criteria

To identify relevant articles for this review, a systematic evaluation of the literature was performed using PRISMA guidelines on June 14, 2024. Included studies measured depression and hair hormone concentration during the peripartum period and analyzed associations between the two. The search included terms for hormones with specific types, a pregnancy or peripartum term, depression, and hair. Given that the method for extracting oxytocin from human hair is still in development (López-Arjona et al., 2021), we did not specifically query this hormone. We searched title and abstract in PubMed, PsycINFO, Web of Science, and SCOPUS databases. The specific query that was used was:

cortis* OR progesterone OR estrogen OR estradiol OR testosterone OR hormone AND pregnan* OR perinatal OR prenatal OR antenatal OR postnatal OR peripartum OR antepartum OR prepartum OR postpartum AND hair AND depress*

See the Supplement Table S1 for how this query was carried out in each database.

We used Covidence, a web-based collaboration software platform (Covidence, n.d.), to assist in the coding of articles identified across databases. Figure 1 presents the PRISMA flow chart for article review. We identified 331 articles for screening across databases; n = 138 from SCOPUS, n = 88 from Web of Science, n = 77 from PubMed, and n = 28 from PsycINFO. After accounting for duplicate studies (n = 216), 115 unique abstracts were screened for inclusion. During abstract review, studies were included for further consideration if they referenced depression and hair hormones. After abstract review, 71 full texts were retrieved for all remaining articles. References were assessed for inclusion eligibility by two independent raters. Studies were included based on study design (human empirical), hormone measurement (via hair samples), measurement timing (pregnancy or up to 12-months postpartum), and broad analytic approach (examined association between depression and at least one hormone). At this stage, 38 articles met inclusion criteria, which consisted of 34 independent studies. After completing this stage 1 search strategy, we then conducted stage 2 search strategies. Specifically, we reviewed all included references both backward (i.e., if the reference cited additional manuscripts to be included) and forward (i.e., if the reference had been cited by additional manuscripts to be included). This identified an additional 7 studies for inclusion. As such, a total of 41 studies with independent samples (comprising 45 articles) met inclusion criteria for the present review and are included on Figure 1; each unique article citation is provided in Table 1.

Figure 1.

PRISMA flow chart for article review.

Table 1.

Literature review specific study information. Please see Supplemental Material for study demographic information.

	Study (Last Name, Year)	Sample size	Peripartum timing of hair hormones (hair length assayed)	Hormones assayed from hair	Depression timing	Depression measure	Results
Hormone: Broader sampling of hormones in addition to cortisol
Study design: Longitudinal assessments—pregnancy
1	(Robertson et al., 2023)	34 pregnant people; 34 non-pregnant people	First trimester (12 weeks, M=12.47, SD=1.21), second trimester (26 weeks, M=26.16, SD=1.41), and third trimester (38 weeks, M=37.62, SD=1.17); 3cm	Cortisol, DHEA, testosterone	First trimester (12 weeks, M=12.47, SD=1.21), second trimester (26 weeks, M=26.16, SD=1.41), and third trimester (38 weeks, M=37.62, SD=1.17)	BDI	Within-person: Positive associations of hair cortisol and cumulative psychological distress only in pregnant people (statistically null for depression specifically). Between-person: Negative association between cortisol and depression symptoms in pregnant people; positive association between cortisol and depression only in non-pregnant people; positive association between testosterone and depression only in non-pregnant people. Statistically null associations between cortisol-to-DHEA ratio and depression in both samples; Statistically null associations between cortisol-to-testosterone ratio and depression in both samples.
Study design: Longitudinal—spanning pregnancy to postpartum
2	(Jahangard et al., 2019)	98	12 weeks postpartum (additional timing information not provided; 6 cm)	Cortisol, cortisone, progesterone, DHEA, testosterone	12 weeks postpartum (additional timing information not provided)	BDI and EPDS	Between-person: Hair steroid levels (cortisol, cortisone, and progesterone) significantly lower in people with depression both before and after delivery with measurement extracted postpartum; Large effect sizes for cortisol and cortisone; Small effect size for progesterone.
Hormone: Cortisol and/or cortisone only
Study design: Longitudinal assessments—pregnancy and postpartum
3	(King et al., 2022)	85	Pregnancy (range=12-37 weeks gestation), one month (range=3-8 weeks) and six months (range=5-8 months) postpartum (5 cm)	Cortisol	Pregnancy (range=12-37 weeks gestation), one month (range=3-8 weeks), and six months (5-8 months) postpartum	CES-D	Within-person: Fluctuations in psychosocial adversity were positively coupled with cortisol; no significant within-person or between-person association between depression and cortisol.
4	(Mustonen et al., 2019)	595	Second trimester (24 weeks gestation, range=21-28 weeks) and/or post-delivery (range=1-3 days after birth) (5 cm)	Cortisol	Gestational weeks 14, 24, and 34	EPDS	Within-person: The trajectory characterized by consistently high levels of depression had higher cortisol postpartum than other groups; small effect size. Between-person: No significant associations between cortisol and depression symptoms at any timepoint.
5	(Feng et al., 2024)	237	First trimester, second trimester, third trimester, three and six months postpartum (3 cm)	Cortisol, cortisone	Preconception, first trimester, second trimester, third trimester, three and six months postpartum	BDI and EPDS	Between-person: No association between corticosteroids across pregnancy with depression.
6	(Kishan et al., 2023)	31	Third trimester (>32 weeks of pregnancy) and postpartum (6-8 weeks after delivery	Cortisol	Third trimester (>32 weeks of pregnancy) and postpartum (6-8 weeks after delivery)	EPDS	Between-person: Prenatal cortisol levels were significantly positively associated with depression; medium effect size.
7	(Romero-Gonzalez et al., 2018)	80	First trimester (M=11.48 weeks gestation; SD=3.72), second trimester (M=24.19 weeks of gestation; SD=3.75), third trimester (M=34.49 weeks of gestation; SD=2.46), and postpartum (M=15.79 days postpartum; SD=9.78; 3 cm)	Cortisol	First trimester (M=11.48 weeks gestation; SD=3.72), second trimester (M=24.19 weeks of gestation; SD=3.75), third trimester (M=34.49 weeks of gestation; SD=2.46), and postpartum (M=15.79 days postpartum; SD=9.78)	SCL-90-R	Between-person: No significant associations between cortisol and depression at any time point small effect sizes.
8	(Orta et al., 2018, 2019)	97	Pregnancy (M=13.1 weeks gestational age, SD=3.9; 3cm) and full-term delivery (M=39.0 weeks gestational age, SD=1.0; 6-9 cm)	Cortisol	Pregnancy (M=13.1 weeks gestational age, SD=3.9)	PHQ-9	Between-person: No significant differences in cortisol collected at enrollment or delivery and depression symptoms at enrollment, assessed either continuously or by dichotomizing groups (i.e., < 5 symptoms, 5-9 symptoms, 10-14 symptoms, or ≥15 symptoms); small effect sizes.
Study design: Longitudinal assessments—pregnancy
9	(Caparros-Gonzalez et al., 2017)	44	First trimester (M=12.36 weeks gestation, SD=3.60; 3 cm), second trimester (M=25.32 weeks gestation, SD=3.24; 3 cm), and third trimester (M=34.94 weeks gestation, SD=3.34; 3 cm)	Cortisol	First trimester (M=12.36 weeks gestation, SD=3.60), at second trimester (M=25.32 weeks gestation, SD=3.24), at third trimester (M=34.94 weeks gestation, SD=3.34); EPDS postpartum (M = 15.79 days after birth, SD = 9.78)	SCL-90-R, the Spanish version of the EPDS	Between-person: Cortisol at first and third trimesters each significantly positively associated with depression symptoms postpartum; Postpartum depression group had higher cumulative hair cortisol levels across all three trimesters.
10	(Hoffman et al., 2016)	90	16 weeks gestation (range=16-18 weeks; 3 cm); 28 weeks gestation (range=28-30 weeks; 3 cm), 40 weeks gestation (range=38-42 weeks; 3 cm)	Cortisol	16, 22, 28, 34, 40 weeks gestation (additional timing information not provided)	CES-D	Between-person: Cortisol collected at 16 weeks gestation positively associated with depression at 40 weeks gestation. Cortisol collected at 28 weeks gestation positively associated with depression at 16 and 28 weeks gestation. Cortisol collected at 40 weeks gestation was positively associated with depression at 16 and 28 weeks gestation; small to medium effect sizes.
11	(Juvinao-Quintero et al., 2023)	2,581	First trimester (1-12 weeks gestation), second trimester (13-24 weeks gestation), and third trimester (25-36 weeks gestation)	Cortisol	First trimester (1-12 weeks gestation), second trimester (13-24 weeks gestation), and third trimester (25-36 weeks gestation)	PHQ-9	Between-person: No significant association between cortisol collected at first, second, and third trimester with depression at first, second, and third trimester. Those with depressive symptoms at first trimester had 0.11 lower HCC in the first trimester than those without any or with mild depressive symptoms at first trimester.
12	(Scharlau et al., 2018)	62	Second trimester (range=24-26 weeks gestation; 1 cm) and third trimester (range=34-36 weeks gestation; 1 cm)	Cortisol, cortisone	Second trimester (range=24-26 weeks gestation) and third trimester (range=34-36 weeks gestation)	PHQ-9	Between-person: Depression was significantly negatively associated with cortisone collected second trimester; medium effect size.
13	(Wiley et al., 2023)	80	First trimester (range=8-16 weeks gestation; 3 cm) and 30 weeks gestation (3 cm)	Cortisol	First trimester (range=8-16 weeks gestation), 30 weeks gestation, and 12 months postpartum	BDI	Between-person: No significant associations between depression and cortisol at any time point.
14	(Romero-Gonzalez et al., 2020)	762 (non-pregnant, n = 171; pregnant and recruited at 1st trimester, n = 124; 2nd trimester, n = 200; 3rd trimester, n = 190; or longitudinal participation across trimesters, n = 77)	Varying pregnancy periods: first trimester (M=10.93 weeks gestation, SD=3.97; 3 cm), second trimester (M=24.69 weeks gestation; SD=3.38; 3 cm), and/or third trimester (M=34.18 weeks gestation, SD=3.23; 3 cm)	Cortisol	Varying pregnancy periods: first trimester (M=10.93 weeks gestation, SD=3.97), second trimester (M=24.69 weeks gestation, SD=3.38), and/or third trimester (M=34.18 weeks gestation, SD=3.23)	SCL-90	Between-person: Cortisol was not significantly associated with depression at any timepoint.
15	(Wilczyńska et al., 2024)	38	Before intervention (21.82 ± 4.30 week of gestation) and 8 weeks after intervention (1 cm)	Cortisol	Before intervention (21.82 ± 4.30 week of gestation) and 8 weeks after intervention	BDI-II	Between-person: No significant association between depression symptoms at time 1 or time 2 with cortisol collected at time 1 or time 2 in either the high-intensity interval intervention group nor the education intervention group; small effect sizes.
Study design: Longitudinal assessments—postpartum
16	(Galbally et al., 2019, 2022, 2023)	Galbally, 2019 n=241; Galbally, 2022 n=198; Galbally, 2023 n=190	Post-delivery (additional timing information not provided; 3 cm) and 12 months postpartum (additional timing information not provided; 3 cm)	Cortisol	SCID-IV before 20 weeks of pregnancy and EPDS before 20 weeks of pregnancy, at third trimester, delivery, six months postpartum, and 12 months postpartum	SCID-IV and EPDS	Between-person: No significant difference in maternal HCC across pregnancy and at 12 months postpartum between women with and without a diagnosis of depression (Galbally, 2019). No significant association between average HCC during pregnancy and depressive symptoms at six months postpartum (Galbally, 2022). No significant association between depressive symptoms at 20 weeks of pregnancy and average maternal antenatal HCC (Galbally, 2023).
17	(Lang et al., 2021)	240	1-6 days (M=2.75 days, SD=1.55) post-delivery, 12 weeks (M=86.36 days, SD=8.40) postpartum (3 cm)	Cortisol, cortisone	12 weeks (M=86.36 days, SD=8.40) postpartum (3 cm)	DSM-V	Between-person: No evidence of group (PPD, non-depressed, adjustment disorder) differences in cortisol collected at 1-6 days post-delivery or 12 weeks postpartum; significant difference in cortisone 1-6 days post-delivery and cortisol/cortisone ratio collected at 16 days post-delivery between groups such that cortisone and cortisol/cortisone ratio was significantly higher in non-depressed than in the adjustment disorder group; Small effect sizes.
18	(Liu et al., 2016)	41	9 months (39 weeks) and 12 months (52 weeks) postpartum; 3 cm)	Cortisol	9 (39 weeks) and 12 months (52 weeks) postpartum	BDI	Between-person: No significant association between depression and cortisol; small effect sizes.
19	(Louis-Jacques et al., 2024)	96	1-2 months postpartum, 3-4 months postpartum, and 5-6 months postpartum	Cortisol	1-2 months postpartum, 3-4 months postpartum, and 5-6 months postpartum	EPDS	Between-person: Cortisol collected at 1-2 months postpartum significantly negatively associated with depression at 1-2 months postpartum; small effect size; results were no longer significant after FDR correction.
20	(Braig et al., 2016, 2023)	Braig, 2016 n=768; Braig, 2023 n=596	Delivery (0 months; Median=1.5 days, range=0-3 days; 3 cm), 6 months and 12 months postpartum (3 cm)	Cortisol	Delivery (0 months; Median=1.5 days, range=0-3 days; 3 cm), 6 months (3 cm) and 12 months postpartum	HADS	Between-person: Cortisol did not differ significantly across the quartiles of depression symptoms (Braig, 2016). No significant association between depression and cortisol; small effect size (Braig, 2023).
21	(Stickel et al., 2021)	196	1-6 days (M=2.39 days, SD=1.44) post-delivery (3 cm) and 12 weeks (M=87.67 days post-delivery, SD=3.18) postpartum (3 cm)	Cortisol, cortisone	1-6 days (M=2.39 days, SD=1.44), 3 weeks, 6 weeks, 9 weeks, 12 weeks postpartum; DSM-5 clinical interview at 12 weeks (M=87.67 days post-delivery, SD=3.18) postpartum	EPDS	Between-person: No significant association between cortisol collected at 1-6 days post-delivery or 12 weeks post-delivery and depression symptoms 1-6 days, 3 weeks, 6 weeks, 9 weeks, 12 week post-delivery. Decreases in cortisol from 3rd trimester to 12 weeks postpartum was only observed in the non-depressed and postpartum adjustment disorder groups at 12 weeks postpartum and not the postpartum depression group; Small effect sizes for decreases in cortisol.
Study design: Cross-sectional—pregnancy
22	(Bowers et al., 2018)	30	Mid-pregnancy (gestational weeks not provided; 3 cm)	Cortisol	Mid-pregnancy (gestational weeks not provided)	BSI-18	Between-person: Cortisol collected mid-pregnancy positively significantly correlated with depression only for participants endorsing two or more adverse childhood experiences; large effect size.
23	(Dobernecker et al., 2023)	149	During pregnancy (11-38 weeks gestation; 2 cm)	Cortisol	Pregnancy (11-38 weeks gestation)	HSCL-25	Between-person: Cortisol was significantly negatively associated with depression.
24	(Freedman et al., 2021)	181	16 weeks gestation (range=15-17 weeks; 2.5 cm)	Cortisol, cortisone	16 weeks gestation (range=15-17 weeks) and at one month postpartum (Newborn age adjusting for gestational age M=28.0 days, SD=9.7, range=14-59 days)	CES-D	Cortisol significantly positively associated with greater depression at 16 weeks only in female-fetus pregnancies.
25	(Hunter et al., 2021)	23	Third trimester (28-33 weeks gestation (2.5 cm)	Cortisol, Cortisone, Cortisol/cortisone ratio	Third trimester (28-33 weeks gestation (2.5 cm)	CES-D	Between-person: Cortisol, cortisone, and cortisol/cortisone ratio were significantly negatively associated with depressive symptoms.
26	(Khoury et al., 2020)	51	Third trimester (M=34.90 weeks gestation, SD=3.38; 4 cm)	Cortisol	Third trimester (M=34.90 weeks gestation, SD=3.38), and at four months postpartum (M=4.47 months, SD=0.73)	EPDS	Between-person: No significant association between cortisol and depression measured at either third trimester or at four months postpartum; small effect sizes.
27	(Penner et al., 2023)	51	Third trimester (4cm)	Cortisol	Third trimester	EPDS	Between-person: No significant association between cortisol and depression; small effect size.
28	(Steudte-Schmiedgen et al., 2023)	212	Third trimester: Four weeks (M=2.96 weeks, SD=1.17, range=0-6 weeks) before the anticipated birth date (2 cm)	Cortisol, cortisone, cortisone/cortisone ratio	four weeks (M=2.96 weeks, SD=1.17, range=0-6 weeks) before the anticipated birth date	EPDS	Between-person: No significant associations between depression and cortisol, cortisone, or cortisol/cortisone ratio.
29	(Wikenius et al., 2016)	181	Second trimester (M=24.8 weeks gestation, SD=3.9; 1 cm)	Cortisol	Second trimester (M=24.8 weeks gestation, SD=3.9)	EPDS	Between-person: No significant association between depression and cortisol.
Study design: Cross-sectional—postpartum
30	(Bergunde et al., 2024)	269	8 weeks postpartum (M=8.3 weeks, SD=1.4; 2 cm)	Cortisol, cortisone, cortisol/cortisone ratio	8 weeks postpartum (M=8.3 weeks, SD=1.4)	EPDS	Between-person: Hair cortisol/cortisone ratio was significantly positively associated with depressive symptoms; small effect size.
31	(Broeks et al., 2023)	89	Postpartum (14.3 weeks postpartum; range=11-19) weeks; 3 cm)	Cortisol	16, 24, and 36 weeks gestation and postpartum (14.3 weeks postpartum; range=11-19)	EPDS	Between-person: No significant difference in postpartum cortisol for participants with or without a depressive disorder at postpartum; No significant association between depression symptoms during pregnancy (mean of symptoms at 16, 24, and 36 weeks gestation) with postpartum cortisol.
32	(Bryson et al., 2021)	546	At child ages one, two, and three year postpartum (additional timing information not provided; 3 cm)	Cortisol	At child ages one, two, and three year postpartum	DASS short-form	Between-person: Dichotomized high depression symptom severity at one year postpartum was associated with higher cortisol one year postpartum; Associations were not observed between any continuous depression score and cortisol at one year postpartum.
33	(Jaramillo et al., 2023)	235	Postpartum (M=8.36 weeks post-delivery)	Cortisol, cortisone, cortisol/cortisone ratio	Pregnancy (M=27.14 weeks gestation) and postpartum (M=8.45 weeks post-delivery)	EPDS	Between-person: Depression significantly positively associated with cortisol/cortisone ratio; medium effect size.
34	(Juncker et al., 2022)	96	<10 days post-delivery (additional timing information not provided; 3 cm)	Cortisol, cortisone	Days 10 and 24 post-delivery	EPDS	Between-person: No significant association between cortisol and depression measured at days 10 or 24 post-delivery.
35	(Kajanoja et al., 2020)	130	1-3 days post-delivery (additional timing information not provided; 5 cm)	Cortisol	Third trimester (34 weeks gestation)	EPDS	Between-person: No significant association between depression measured at third trimester and cortisol collected 1-3 days post-delivery.
36	(Kramer et al., 2009)	117	Postpartum (9 cm)	Cortisol	24-26 weeks gestation	CES-D	Between-person: No significant association between depressive symptoms during pregnancy and cortisol collected postpartum.
37	(Pintye et al., 2023)	153	6 weeks postpartum (3 cm)	Cortisol	Pregnancy (Median=20, IQR=16-26)	PHQ-2	Between-person: No significant association between depression during pregnancy and postpartum cortisol.
38	(Schreier et al., 2016)	180	Post-delivery (3-9 cm)	Cortisol	29 weeks gestation	EPDS	Between-person: No significant association between depression at 29 weeks gestation and cortisol collected post-delivery; small effect size.
39	(Tarullo et al., 2017)	121	6 months postpartum (M=6.67, SD=0.43; 3 cm)	Cortisol	6 months postpartum (M=6.67, SD=0.43)	CES-D	Between-person: No significant association between cortisol and depression; small effect size.
40	(van der Voorn et al., 2018)	172	First day postpartum (1 cm)	Cortisol, cortisone	First/second trimester, third trimester, and first day postpartum	HDS	Between-person: No significant association between depression at first/second trimester, third trimester, or first day postpartum and cortisol or cortisone collected first day postpartum.
41	(Zhang et al., 2019)	349	2 days, one-month, and two-month postpartum (10 mg)	Cortisol	4 different time points before delivery; 2 days, one-month, and two-month postpartum	HAMD	Between-person: Cortisol collected 2 days postpartum (pre-intervention) was significantly higher in the depressed group compared with the control group.

Note. “Significance” denotes statistical significance. All information concerning timing of assessments and effect sizes are reported when available for extraction from the indicated manuscripts. BDI = Beck Depression Inventory; BSI-18 = Brief Symptom Index-18; CESD = Center for Epidemiological Studies of Depression Scale; DASS = Depression, Anxiety, Stress Scales; EPDS = Edinburgh Postnatal Depression Scale; HADS = Hospital Anxiety and Depression Scale; HDS = Hospital Depression Subscale; HSCL-25 = Hopkins Symptom Checklist; Inventory to Diagnose Depression (IDD); PHQ-9 = Patient Health Questionnaire; SCL-90-R = Symptom Check-list-90-Revised; SCID-IV = Structured Clinical Interview for DSM-IV.

2. Results

Summary

See Table 1 for specific study information and Figure 2 for a summary of findings. Studies included in this systematic review employed a range of methods in examining hair hormones in relation to depression in the peripartum period. Below, we present results according to hormones assayed, timing of assessments, and statistical approaches across all studies identified.

Figure 2.

Visual diagram of included study findings

Study design considerations

Approximately half of included studies included measured hair hormones only once (n = 21; 51%), usually measuring 3 cm of hair to assess the past 3 months of hormone concentrations, and ranging in time of assessment from “mid pregnancy” to 12-months postpartum. Of the 21 studies with only 1 point of assessment, 12 were collected postpartum and 9 were collected during pregnancy. The remaining studies varied in the number of assessment timepoints, with 8 studies (20%) using 2 timepoints, 10 studies (24%) using 3 timepoints, 1 study (2%) using 4 timepoints, and 1 study (2%) using 5 timepoints; in the majority of these studies, the data collected across timepoints were then collapsed into summary scores. The timing of depression assessment largely mirrored timing of hormone assessments, though ranged in assessment methods. All studies used self-report questionnaires to assess depressive symptoms (e.g., Beck Depression Inventory, Edinburgh Postnatal Depression Scale, Patient Health Questionnaire). Only 1 study also used a diagnostic interview to assess depression in addition to self-report measures (Galbally et al., 2019).

While statistical approaches varied, a large number of included studies included the analysis of hormone–symptom associations as an ancillary aim to the primary analyses. Study designs that included repeated assessment designs, which leveraged the increased reliability and statistical power of multiple assessments over time, were more likely to observe an observation between hair hormones in hair and depression. This may also be in part reflective of the specific aims of these studies to examine hormone–depression associations.

Other hormones assayed form hair

Only 2 studies assayed hormones other than cortisol or cortisone. Robertson and colleagues (2023) examined cortisol, DHEA, and testosterone at 3 timepoints across the prenatal period in pregnant people (n=34) and across the same timespan in non-pregnant people (n=34). The research team used both between-person and within-person analytic strategies to examine the associations between hormones and depression over time. No statistically significant within-person hormone and depression associations were found. However, they did observe that cortisol fluctuations were positively associated with cumulative psychological distress in pregnant participants. Further, in between-person analyses, they observed a negative association between cortisol and depression in pregnant participants, but a positive association in non-pregnant participants. Testosterone was negatively associated with depression in non-pregnant participants only. As the sole study that has measured multiple hormones via hair analysis and used both within- and between-person approaches, this study stands out as uniquely situated to inform current knowledge regarding associations between hormone levels in hair and peripartum depression. Jahangard and colleagues (2019) also examined multiple hormones (i.e., cortisol, progesterone, and testosterone). They observed that cortisol and progesterone, but not testosterone, were statistically significantly lower in pregnancy and postpartum periods for people with depression compared to people without depression.

Within- versus between-person statistical approaches

The majority of the articles (n = 38; 93%) used a solely between-person approach. Notably, however, (1) Mustonen et al. (2019) uniquely examined symptom trajectories in relation to postpartum cortisol, (2) King et al. (2022) examined within-person variability in cortisol across the peripartum, and (3) Robertson et al. (2023) examined hair hormone trajectories across pregnancy.

Associations between hair cortisol and depression

Given the limited representation of estrogen, progesterone, and testosterone across studies, the remainder of this systematic review will, by necessity, focus on hair cortisol concentrations. First, the majority of studies included in the present review (n = 25; 61%) demonstrated associations that were not statistically significant (were “statistically null”) between hair cortisol and depressive symptoms. Half of these investigations were cross-sectional (n = 13) and all utilized between-person statistical approaches, with the exception of King and colleagues (2022) who found that within-person fluctuations in hair cortisol related to psychosocial adversity, but not depressive symptoms.

Of the studies that described statistically significant associations between hair cortisol and PPD (n = 16; 39%), 5 (12%) reported a negative association and 10 (24%) reported a positive association. Timing of assessment was an important factor in understanding these results, as studies that examined cross-sectional associations in the second trimester (e.g., Scharlau et al., 2018) or in the postnatal period (e.g., Jahangard et al., 2019) tended to find that lower levels of hair cortisol was related to higher concurrent depression symptoms. In contrast, studies that considered multiple measures across time tended to find bidirectional associations between hair cortisol and PPD, such that higher hair cortisol levels across time were related to higher levels of later PPD and higher levels of PPD across time were related to later hair cortisol. These analytic strategies ranged from cumulative hair cortisol measures across pregnancy in prospective association with postpartum symptoms (e.g., Caparros-Gonzalez et al., 2017), to hair cortisol values positively associated with depression symptoms across trimesters in pregnancy (e.g., Hoffman et al., 2016), as well as consistently high depression symptoms across pregnancy positively associated with postpartum hair cortisol (Mustonen et al., 2019).

Several studies demonstrated a statistically significant association for only some people or in some cases (i.e., depending on life history, characteristics of the fetus, in more severe cases, or depending on the timing of assessment). For example, Bowers and colleagues (2018) found that higher hair cortisol was associated with more depressive symptoms only for pregnant people who reported at least two adverse childhood experiences. Additionally, Freedman and colleagues (2021) found that second trimester hair cortisol was positively associated with depression symptoms only for pregnant people carrying female fetuses. Bryson and colleagues (2021) found that hair cortisol was positively associated with depressive symptoms, only when the depression measure was dichotomized such that the top 15% of depression scores were associated with higher hair cortisol. Timing of assessment was also an important component to identifying statistically significant versus null associations, as demonstrated by Mustonen and colleagues (2019). In this study, the researchers assessed both cross-sectional associations and trajectories of depression symptoms with hair cortisol at delivery and found an association only when trajectories of symptoms were considered (Mustonen et al., 2019). Relatedly, one study found that difficulty identifying feelings (i.e., alexithymia) was associated with higher hair cortisol in the days following birth (Kajanoja et al., 2020). King and colleagues (2022) found that within-person fluctuation of psychosocial adversity experiences specifically were positively associated with hair cortisol measures across pregnancy and postpartum.

3. Conclusions

The aims of the present review were to systematically examine the research to date investigating hair hormone levels and depression across the peripartum period in order to integrate findings across studies, identify current gaps in the literature, and offer future directions. Our review suggests that a majority of the research in this area focuses exclusively on cortisol and few studies have considered trajectories of change across time. The most consistent finding was statistically null associations between hair cortisol and depressive symptoms (n = 25; 61%). In contrast, studies that used within-person analyses demonstrated some evidence that hormone trajectories across the peripartum period are associated with psychosocial functioning (i.e., depression, stress, distress), though associations with depression specifically are not consistent across these studies. Timing of assessment may be particularly important to observing a between-person hair cortisol–PPD association, as well as the direction of this association. Investigations that considered multiple timepoints and within-person trajectories revealed more evidence of positive associations between hair cortisol and PPD or psychosocial adversity. This may be, in part, because within-person approaches were able to leverage increased statistical power due to removing between-person noise when we are interested in within-person associations. Finally, some studies explored potential moderators, finding that the pattern of associations differed based on pregnant people’s life histories, characteristics of the fetus, or study methodology.

Ultimately, findings support that the cumulative evidence to date is mixed regarding associations between cortisol and PPD. Previous systematic reviews regarding the cortisol–depression association both within (Mlili et al., 2023; Orta et al., 2018) and outside of (Burke et al., 2005; Goldstein & Klein, 2014; Herbert, 2013; Knorr et al., 2010; Koumantarou Malisiova et al., 2021; Psarraki et al., 2021) the perinatal period have discussed similar results. Koumantarou Malisiova and colleagues (2021) highlight that depression and stress may have opposing effects on cortisol, wherein studies examining depression specifically seem to find heightened cortisol concentrations and studies examining stress seem to find blunted cortisol concentrations. In the perinatal period, Orta and colleagues (2018) suggest heightened cortisol during recovery periods following acute stressors may be the most related to depression, though Mlili and colleagues (2023) suggest relatively lower cortisol concentrations in pregnancy may be the most related to depression. There is evidence synthesized in previous reviews of distinct alterations in waking cortisol, diurnal changes, stress response, and potential for bidirectional effects between cortisol alterations and depression symptoms (Burke et al., 2005; Goldstein & Klein, 2014; Herbert, 2013). A limitation of hair samples of hormones is the inability to differentiate briefer fluctuations in dynamics (e.g., across short windows of time [e.g., in response to acute stressors]). Given the importance of assessment timing in the literature outside of peripartum, lower temporal resolution may contribute to why we are seeing mixed patterns here. While hair hormone measures may eventually provide complementary information to the present research on briefer measures, the limited research to date suggests we do not know, yet, how trajectories across the peripartum months relate to depression.

Altogether, our systematic review revealed considerable gaps in the literature—particularly the predominant focus on cortisol and the inconsistency in the association between hair cortisol levels and depression during the peripartum period, which may be due to the reliance on cross-sectional data and between-person analyses. As such, we need (1) incorporation of multiple hormones in consideration of how hormones across the peripartum period may be related to PPD, especially estrogen, progesterone, and testosterone (2) explicit focus on timing of assessment, including investigations of within-person change in hair hormone concentrations and PPD, and (3) consideration of pregnant person context and life history to further inform these questions in future research. To the first point, there are strong conceptual reasons to investigate multiple hormones in relation to peripartum mood, especially estrogen and progesterone (Albert & Newhouse, 2019; Schiller et al., 2015; Standeven et al., 2020), though a majority of the research to date has focused exclusively on cortisol (and cortisone).

Secondly, a review of hair hormone levels across pregnancy demonstrated that there is considerable methodological variability across studies to date when assessing mean levels and change in levels across pregnancy (Marceau et al., 2020). These varying methodological approaches to basic research have implications for applied research. In the present review, we observed that the hormone and depression relation in the peripartum period is dynamic in that each of these levels of functioning fluctuate over time and their association with each other likely fluctuates as well. Endocrine measurement requires high levels of precision to isolate signal from noise. Within-person designs accommodate this need with integrated repeated measures; however, the studies reviewed primarily examined between-person differences at single time points, which do not capture hormonal fluctuations potentially critical to understanding depression risk in this sensitive period. It will be imperative for future work to examine the degree to which hair hormones and depression fluctuate—in general and as an individual difference across people—as well as how this relates to overall functioning and health. The recent publication by Robertson and colleagues (2023) demonstrates this point eloquently in finding within-person associations between cortisol and psychological distress in pregnant people, echoing what King and colleagues (2022) also found.

Thirdly, the present review also identified factors that may drive individual differences in cortisol or symptoms over time, namely pregnant person life history and sex of the fetus. These characteristics are likely only examples of what may be many influences on pregnant person functioning and hormone changes across the perinatal period. A priori power analyses including the consideration of moderating variables are suggested for specific analytic plans as part of the study design stage. Included studies ranged in size from N = 23 to N = 5,337, and sample size and statistical power should be considered in interpreting the quality of the studies composing the literature to date.

Altogether, longitudinal studies charting individual differences in trajectories of hormones across the perinatal period would identify how hormones contribute to risk for PPD and—critically—potentially also identify for whom. The answers to these two questions could provide the basis for more personalized treatment approaches, including identification of what mechanisms to intervene on and who would benefit from the intervention. Successful intervention in the peripartum period is beneficial for both pregnant people and their prospective offspring, and early identification and intervention with pregnant people may be especially important to altering cascading effects. Personalized medicine approaches in the peripartum period targeting hormone changes may be imperative to supporting individuals and their young offspring, but more specificity to the mechanisms of hormone–PPD associations is needed. Preliminary evidence for the efficacy of novel hormone-based treatments for postpartum depression are promising, and yet more research is needed to understand the basic process of how hormones contribute to risk for PPD, the mechanisms targeted with the hormone treatments, and whether and when to incorporate these treatments into personalized medicine.

Highlights.

• Peripartum hormone levels may be a contributor to risk for depressive symptoms.

• Across 41 studies, a majority reported statistically null associations.

• Mixed evidence for an association between hair cortisol and peripartum depression.

• Studies varied widely in sampling approaches to examine these associations.

• We recommend within-person studies on multiple hormones to advance the field.