Comparing Childhood, Chronic, and Recent Adversity on Internalizing Symptoms and Hypothalamic-Pituitary-Adrenal Axis Functioning

Abstract

Objective:

To examine how childhood, chronic, and recent adversity associate with internalizing symptoms and hypothalamic–pituitary–adrenal (HPA) axis activity among women in a low-resourced context.

Methods:

Data come from the Bachpan Study (2014-2022), a longitudinal cohort of mothers in rural Pakistan. Childhood adversity was assessed using retrospective reports of adverse childhood experiences. Chronic adult adversity was defined as adversities experienced across two or more prior study waves (baseline, 1-, 2-, and 3-years), and recent adversity was assessed at the 6-year wave. Outcomes, measured at the 6-year wave, included depressive symptoms (Patient Health Questionnaire–9), anxiety symptoms (Generalized Anxiety Disorder–7), and hair cortisol as an indicator of HPA axis activity. Generalized linear models examined associations for each adversity separately and simultaneously.

Results:

In our sample of 900 mothers, we find that childhood, chronic, and recent adversity were each associated with higher depressive and anxiety symptoms, particularly at the highest levels of exposure. Chronic adult adversity and recent adversity showed dose–response relationships with internalizing symptoms. When examined together, recent adversity was the most robust predictor of depressive and anxiety symptoms. Meanwhile, associations between adversity and cortisol were more modest, but suggest evidence of blunted cortisol levels at the highest levels of childhood and recent adversity exposure.

Conclusions:

Findings highlight the importance of considering the differential effects of childhood, chronic, and recent adversity for health. While early intervention remains critical, addressing ongoing stressors may yield meaningful health benefits. Even if adversities are not eliminated, reducing them can have a significant impact on health.

Adversity exposure (e.g., poverty, violence) is associated with future poor health outcomes, including elevated internalizing symptoms and dysregulation of stress-responsive biological systems (Jones et al., 2018). Although these associations are well-established, less is known about how adversity exposure across the life course shape these psychological and physiological outcomes. For outcomes in adulthood, a life course perspective that distinguishes differential impacts between childhood, chronic, or recent adversity may be particularly important for informing prevention and intervention strategies. Internalizing symptoms like depressive and anxiety symptoms and hypothalamic–pituitary–adrenal (HPA) axis activity represent indicators of how adversity-related stress may shape mental health and physiological regulation (Koss & Gunnar, 2018). Investigating these associations in women in low-resourced contexts, who face a heightened prevalence of adversity, is critical for understanding how these relationships operate across settings and for informing global health efforts.

To understand why childhood, chronic, and recent adversity may differentially relate to health, it is important to examine both psychological and physiological outcomes. Depression and anxiety are two of the most common mental health disorders and are leading contributors to global disease burden (World Health Organization, 2022). These internalizing symptoms have been consistently associated with exposure to adversities such as adverse childhood experiences and intimate partner violence (IPV), reflecting the psychological toll of these experiences on mental health (LeMasters et al., 2021). In addition to psychological outcomes, adversity has been associated with alterations in stress-responsive physiological systems, such as the HPA axis (Danese & McEwen, 2012). The HPA axis regulates physiological responses to stress through hormones, such as cortisol, with short-term activation of this system supporting responses to acute stressors (McEwen, 1998). However, frequent or sustained activation of the HPA axis associated with adversity exposure can lead to either heightened or blunted HPA axis activity, reflecting increased stress sensitivity or a downregulation of stress response over time (Fries et al., 2005). Consistent with this complex system, prior studies on adversity and HPA axis functioning have found a range of patterns, including evidence for both elevated and blunted activity, as well as null findings (Ikram et al., 2022; Niu et al., 2025; Oresta et al., 2021). For example, research has found that acute adversity may result in increased HPA activity, while chronic adversity can lead to downregulation of the HPA axis, which results in hypocortisolism over time (Koss & Gunnar, 2018). Dysregulation of the HPA axis is a form of biological embedding of adversity, with implications for metabolism, immune functioning, and long-term mental health risk (Adam et al., 2017).

Although prior research has established that adversity is associated with internalizing symptoms and HPA axis dysregulation, it remains unclear whether these associations depend on when adversity occurs. A life course perspective provides a framework for understanding how childhood, chronic, and recent adversity exposure, which are each distinct constructs of adversity, may shape health (Ben-Shlomo & Kuh, 2002). Within this framework, three complementary models have been proposed to explain how adversity exposure at different periods and durations relates to health: sensitive period, accumulation of risk, and recency (Morrissey & Kinderman, 2020). The sensitive period model posits that exposure to adversity at critical developmental periods, such as childhood, has lasting effects on adult health, above and beyond later stressors (Jones et al., 2018). This model suggests that programs delivered during or before a critical developmental period would be the most effective. In contrast, the accumulation of risk model emphasizes that the additive effects of persistent adversity over time, rather than the specific timing of exposure, is what is important for later life health (Lindström et al., 2014). Evidence for the accumulation model would indicate that programs that reduce prolonged exposure would be particularly important. The recency model hypothesizes proximal or recent adverse experiences are the most strongly linked to health, suggesting the need for programs that address current stressors (Dunn et al., 2018). These three models are not mutually exclusive, as childhood, chronic, and recent adversity may each be differentially associated with health outcomes (Morrissey & Kinderman, 2020).

Empirical research examining the relationship between adversity and internalizing symptoms and HPA axis activity provides support for all three frameworks. For example, research has demonstrated that persistent childhood maltreatment is associated with both increased internalizing and externalizing childhood symptoms, as well as with comorbid psychopathology in adulthood (Russotti et al., 2021). A longitudinal study in Sweden found that both childhood adversity and adversity accumulated over time were associated with higher allostatic load, a marker of stress-related biological risk (Gustafsson et al., 2012). Similarly, economic stress in both childhood and adulthood has been linked to poor psychological health (Lindström et al., 2014). Beyond finding support for multiple models, some research suggests that adversity occurring at particular periods or durations may be more strongly associated with health outcomes. In a U.S. birth cohort followed to age 37, Young et al. (2021), found that although both childhood and cumulative adversity were associated with blunted cortisol patterns, the interaction between childhood and current adversity most strongly predicted flatter diurnal cortisol slopes, indicating the greatest dysregulation of the HPA axis. In contrast, a study of children in England found that cumulative and recent adversity more strongly predicted child emotional and behavioral problems than the sensitive period model (Dunn et al., 2018). These findings highlight the complexity of life course processes and underscore the importance of examining childhood, chronic, and recent adversity simultaneously.

Despite increasing research on adversity and health in low-resourced contexts, such as low -and middle-income countries, existing studies have largely focused on childhood adversity with limited attention to adversity exposure at other points in the life course (Ikram et al., 2022). Among the smaller number of studies in low-resourced contexts that have examined adversity across more than one life stage, the focus has often been on examining adversity trajectories rather than on assessing whether the strength of the association between adversity and health outcomes varies by childhood, chronic, and recent adversity (Panneh et al., 2026). Investigating adversity, internalizing symptoms, and HPA axis activity in low-resourced contexts can provide insight into the generalizability and variability of life course models of adversity, in which there is limited research. Women in these low-resourced settings face gendered risks for adversity exposure, including heightened vulnerability to interpersonal violence, making them an important population to study (Pennington et al., 2018). Among mothers, adversity-related internalizing symptoms and stress physiology have implications not only for women’s health but also for child development, caregiving environments, and intergenerational transmission of risk (Cooke et al., 2019). Understanding these associations among mothers could inform interventions that prevent the accumulation of health risks for both women and their children.

The present study examines how childhood, chronic, and recent adversity exposures are associated with women’s internalizing symptoms and HPA axis activity in a low-resourced context, using data from a cohort of mothers in rural Pakistan, a setting where adversities are high. We first characterize the prevalence and distribution of childhood, chronic, and recent adversity in the study sample. We then examine how childhood, chronic, and recent adversity is associated with women’s depressive symptoms, anxiety symptoms, and cortisol levels, both independently and simultaneously to assess their relative contributions. Based on prior literature, we hypothesize that childhood, chronic, and recent adversity will each be associated with depressive symptoms, anxiety symptoms, and cortisol levels, but at varying strength of associations. We further hypothesize that greater exposure to adversity will be associated with higher levels of depressive and anxiety symptoms, whereas associations with cortisol may differ in direction, with chronic adversity expected to be associated with blunted cortisol levels. Through these findings, we aim to clarify how childhood, chronic, and recent adversity relate to physiological and psychological outcomes in adulthood to inform contextually relevant prevention and intervention strategies.

Methods

Study Sample and Design

Data for the present analyses were drawn from the Bachpan Study, conducted in a rural subdistrict of Rawalpindi, Pakistan (Sikander et al., 2019). Bachpan is an ongoing birth cohort study with a nested cluster-randomized controlled trial of a maternal depression and child development intervention, which enrolled a total of 1,154 women at baseline from 40 village clusters. Women who were aged 18 and older, married, and in their third trimester of pregnancy were eligible to participate. Written informed consent was obtained from all participants, or witnessed oral informed consent if participants did not read/write. The study protocol was approved by the institutional review boards at the University of North Carolina at Chapel Hill and Duke University. Between October 2014 and February 2016, eligible participants in each village cluster were screened for depression (Patient-Health Questionnaire-9 (PHQ-9), ≥10) and recruited by local community health workers, with equal numbers of depressed and non-depressed women enrolled from each cluster. Half of the participating village clusters received a peer-delivered intervention for depression, Thinking Healthy Program Peer-delivered (THPP), an adapted version of the WHO’s Thinking Healthy Program (World Health Organization, 2015). The intervention included individual and group sessions using cognitive behavioral approaches to reduce depression, strengthen social support, and, in later sessions, promote child well-being while the remaining clusters received enhanced usual care. In these analyses we used data from the baseline (pregnancy) and 1, 2, 3, and 6-year waves. Our analytic sample consists of the 900 women interviewed at the 6-year wave. The analytic sample for cortisol was based on the participants who provided hair samples, which were collected in a separate visit after the main 6-year survey (N=810). There were no significant differences in baseline characteristics between the baseline, 6-year, and cortisol samples (Tables S1 and S2).

Exposure Measures

The exposure of interest, adversity, was characterized in three ways: childhood, chronic, and recent (Table S3). Together, these measures capture indices of adversity exposure across early development, repeated exposure over time during adulthood, and exposure occurring close to outcome assessment for the mothers, each of which are distinct stressors. Childhood adversity was assessed at the 3-year wave, using a revised 10-item version of the Adverse Childhood Experience (ACE) International Questionnaire (IQ). Women were asked to retrospectively report exposure to adverse experiences related to parent-child relationships, family environment, and violence experienced before the age of 18, resulting in a summed score ranging from 0 to 10, with higher scores representing increased ACE count (World Health Organization, 2020). This measure has been validated in other low- and middle-income countries (LMIC) (Kidman et al., 2019).

Chronic and recent adversity were both derived using the same set of eight indicators: intimate partner violence (IPV; any physical, psychological, or sexual IPV) in the past 12 months, socioeconomic status (SES) assets, financial problems, problems with residence, troubled relations with family or friends, marital unrest, worries about children’s problems, and quarrels with family members. These indicators were selected to reflect the most common adversities experienced by women in this context, which was informed by field experience and prior research conducted by the study team. At each wave, indicators were coded as binary variables (0 = not experienced, 1 = experienced). IPV was coded as any occurrence in the past 12 months. SES assets were evaluated at baseline, 2-year, and 6-years using an adapted index from the Demographic and Health Survey wealth index, which asks questions regarding land/home/animal ownership, drinking water and wastewater sources, and household goods (Maselko et al., 2018). For years in which asset data were not collected (1- and 3-years), the most proximate assessment was assigned (baseline and 2-years), given prior research has found that wealth is stable across short periods of time (Roll et al., 2022). Asset scores were categorized into quintiles, with women in the lowest quintile classified as experiencing SES-related adversity. The remaining six indicators assessed whether the participant or her family had faced each event in the last year. These items were derived from the Life Events and Difficulties Scale (Brown & Harris, 1989) and culturally adapted to the study area (Husain et al., 2000).

Recent adversity was assessed using the eight indicators at the 6-year wave and was calculated as the sum of adversities reported at that time point (range: 0–8), reflecting exposures proximal to the outcome assessment. To measure chronic adversity, we assessed adversities during adulthood. An individual adversity was classified as chronic if it was reported at two or more waves of data collection across baseline, 1-year, 2-year, and 3-year assessments, reflecting persistent stress exposure over time. These chronic indicators were then summed to create a chronic adult adversity index (range: 0–8). We conducted sensitivity analyses alternatively coding chronic adult adversity as the number of waves in which the participants experienced two or more adversities. For the categorical analyses, adversity counts were top-coded at four or more to address small cell sizes at higher levels of exposure.

Outcome Measures

All outcomes of interest, including depressive symptoms, anxiety symptoms, and hair cortisol, were measured during the 6-year data collection. The PHQ-9 was used to measure depressive symptoms, assessing the frequency of symptoms in the past two weeks using a 4-point Likert scale, resulting in a total score ranging from 0 to 27, with higher values reflecting more severe depressive symptoms. Anxiety symptoms were assessed using the Generalized Anxiety Disorder-7 (GAD-7), which similarly measured symptom frequency in the past two weeks using a 4-point Likert scale, resulting in a total score range of 0 to 21. Both the PHQ-9 and GAD-7 measures have indicated high reliability and validity in Pakistan (Ahmad et al., 2017; Gallis et al., 2018). The internal consistency for the PHQ-9 and GAD-7 in this sample was 0.80 and 0.85, respectively. All questionnaires were translated into Urdu.

Cortisol was measured using a standardized collection protocol to obtain hair samples from women in the study at the 6-year wave (Russell et al., 2012). For each participant who consented to hair collection, about 200 strands of hair were clipped from the posterior hairline by the assessor. Once collected, the samples were stored in aluminum foil, in a plastic bag with a desiccant and kept at room temperature. Cortisol extraction and quantification were conducted by Dresden LabService using liquid chromatography-mass spectrometry (Gao et al., 2013). This approach simultaneously extracts steroid hormones, such as cortisol, from intact (non-pulverized) hair samples using methanol incubation. As one centimeter of hair reflects cumulative HPA axis hormone output for the preceding month (Stalder & Kirschbaum, 2012), the first 2 centimeters of hair used in this analysis captures average cortisol output over the previous two months. This approach reduces the short-term fluctuations in cortisol and provides a more stable estimate of HPA axis activity. All participants who provided hair samples had cortisol concentrations above the detectable limit. Cortisol values were log-transformed to correct for right-skewness and then values greater than three standard deviations above the mean were winsorized to that threshold (N=5).

Covariates

Guided by previous research, several variables were identified as potential confounders and selected as covariates in the models. Baseline covariates include study arm (RCT control, non-depressed, and intervention), age in years, education in years, and family history of mental illness. Covariates at 6-years include current pregnancy status and assessor fixed effects. Cortisol analyses additionally accounted for weather on the day of data collection (maximum temperature and humidity), number of days since last hair wash, whether the participant had recent sweat resulting in hair moistening, whether they used hair products, whether they used medical creams, and the weight of the hair sample (Staufenbiel et al., 2015). Sensitivity analyses were conducted removing these hair-related covariates from the cortisol analyses.

Analytic Strategy

Generalized linear models with a Gaussian family and identity link were used to examine the associations between childhood, chronic adult, and recent adversity and psychological and physiological health outcomes, with robust standard errors accounting for clustering at the village level. Separate models were estimated for childhood (Model 1), chronic adult (Model 2), and recent adversity (Model 3), with an additional combined model which included all three simultaneously (Model 4). Models were estimated separately for each outcome: depressive symptoms (PHQ-9), anxiety symptoms (GAD-7), and hair cortisol. To capture potential variability and non-linearity in the associations between adversity and health outcomes, adversity measures were modeled as categorical variables (0, 1, 2, 3, and 4+).

Sensitivity analyses were conducted with continuous measures of adversity as well as with an alternative operationalization of chronic adult adversity. In the sensitivity analyses, chronic adult adversity was alternatively coded as the number of study waves in which the participant experienced two or more adversities (range: 0-4). Missing data was addressed using stabilized inverse probability censoring weights (IPCW) to account for loss to follow-up and multiple imputation to account for item-level missingness. IPCW were estimated using variables that predicted missingness at the 6-year wave along with baseline confounders of the adversity and exposure-outcome relationship. Item-level missingness was relatively low with most items having less than 6% missingness (Table S4). Little’s MCAR test suggests that the data is not missing completely at random (p-value = 0.02). We assume data is missing at random and thus conducted multiple imputation with variables that predict missingness. Multiple imputation using chained equations was conducted with 10 imputations and 15 iterations per imputation, incorporating all exposures, outcomes, and covariates included in the analytic models. Imputed datasets were analyzed using Rubin’s rules. Stata 19 was used to conduct all data analysis and R Version 4.5.0 was used to conduct multiple imputation. During preparation of this publication, the authors used ChatGPT 5.2 to assist with language refinement and readability of the text. The authors reviewed, edited, and verified all content generated by artificial intelligence.

Results

Descriptive Analyses

Table 1 presents descriptive characteristics of the 900 women in the analytic sample. Descriptive statistics for hair cortisol and hair-related covariates are reported for the sample of 810 women who had hair cortisol data. Sample characteristics were similar across the internalizing symptoms and cortisol analytic samples; more descriptive statistics for the cortisol subsample are provided in the supplement (Table S2). At baseline, women were on average 27 years old (SD=4.4), had a mean of 8 years of education, and 10% reported a family history of mental illness. With respect to study arm, 25% were in the RCT control arm, 24% in the RCT intervention arm, and 51% were in the RCT non-depressed arm. At the 6-year wave, 8% of women were pregnant. Mean depressive symptoms (PHQ-9) were 6.0 (SD=5.6), and mean anxiety symptoms (GAD-7) were 3.9 (SD=4.9). Prior to log-transformation, hair cortisol values ranged from 0.1 to 73.8 (M=6.1, SD=8.1).

Table 1.

Sample Descriptives

Variables	Mean(SD)/Proportion
Outcomes (6-year)
PHQ-9 (0-27)	5.96 (5.62)
GAD-7 (0-21)	3.88 (4.87)
Cortisol (0.1-73.8) (untransformed)a	6.14 (8.13)
Adversity
Childhood adversity (0-10)	1.23 (1.47)
0	0.41
1	0.26
2	0.16
3	0.10
4+	0.08
Chronic adult adversity (0-8)	1.95 (1.76)
0	0.22
1	0.28
2	0.20
3	0.14
4+	0.17
Recent adversity (0-8)	2.60 (1.95)
0	0.14
1	0.20
2	0.23
3	0.13
4+	0.30
Baseline Covariates
Study arm
RCT Control	0.25
RCT Intervention	0.24
RCT Non-depressed	0.51
Age in years (18-45)	26.68 (4.43)
Education in years (0-18)	7.71 (4.46)
Family history of mental illness	0.10
6-Year Covariates
Current pregnancy	0.08
Hair Covariates (N = 810)
Number of days since last hair wash (1-15)	2.34 (1.93)
Sweat since last hair wash	0.49
Use hair products	0.92
Use hair creams	0.02
Weight of hair sample (4-7.5)	7.45 (0.33)
Weather
Max temperature(°C) (11.4-43.2)	29.98 (7.30)
Max humidity (proportion) (0.3-0.9)	0.72 (0.13)

Notes: Multiply imputed and inverse probability weighted descriptives are reported.

^aBased on an N of 810 in the cortisol sample. Baseline sample descriptives for the cortisol sample can be seen in Supplement Table S2.

Women experienced a mean of 1.2 childhood adversities (SD=1.5), 2.0 chronic adult adversities (SD=1.8), and 2.6 recent adversities (SD=2.0). Approximately 60% reported at least one childhood adversity. Chronic adult adversities were also common, with 78% of the sample having at least one adversity across two or more waves. Intimate partner violence (IPV; 59%) and financial problems (45%) were the most persistently reported adversities. At the 6-year wave, when recent adversities were measured, worry about children’s problems (64%), financial problems (56%), and IPV (42%) were the most reported recent adversities.

Correlations among the main study variables (exposures and outcomes) are shown in Table 2. Childhood, chronic adult, and recent adversity were all weakly and negatively correlated with cortisol (r range −0.04 to −0.02). Meanwhile, chronic adult adversity (r = 0.37) and recent adversity (r = 0.43) were moderately correlated with depressive symptoms (PHQ-9), whereas childhood adversity showed a weaker correlation (r = 0.11). A similar pattern was observed for anxiety symptoms (GAD-7), with stronger correlations for chronic adult (r = 0.41) and recent adversity (r = 0.46) compared to childhood adversity (r = 0.14). PHQ-9 and GAD-7 were highly correlated (r = 0.75). Childhood, chronic adult, and recent adversity were moderately correlated with each other, with chronic adult and recent adversity being the most highly correlated (r = 0.54) followed by childhood and chronic adult (r = 0.26) and childhood and recent (r =0.14). These patterns suggest that while childhood, chronic adult, and recent adversity are interrelated, they are capturing different dimensions of exposure.

Table 2.

Correlations for the Main Study Variables

	Cortisol	PHQ-9	GAD-7	Childhood adversity	Chronic adult adversity	Recent adversity
Cortisol	1.00
PHQ-9	0.02	1.00
GAD-7	−0.01	0.75	1.00
Childhood adversity	−0.02	0.11	0.14	1.00
Chronic adult adversity	−0.03	0.37	0.41	0.26	1.00
Recent adversity	−0.04	0.43	0.46	0.14	0.54	1.00

Regression Analyses

Associations between childhood, chronic adult, and recent adversity and depressive symptoms are shown in Table 3. When examined separately (Models 1-3), higher levels of childhood, chronic adult, and recent adversity were each associated with greater depressive symptomology. Childhood adversity was associated with higher PHQ-9 scores only at the highest levels of exposure, four or more adversities. In contrast, chronic adult and recent adversity showed dose-response associations with depressive symptoms, with progressively higher PHQ-9 scores as adversities increased. For example, compared to women reporting zero chronic adult adversities, those experiencing two chronic adult adversities were associated with mean PHQ-9 scores 1.4 (95% CI = 0.5, 2.3) units higher, with larger differences observed for three (coef. = 2.6; 95% CI = 1.2, 3.9), and four or more (coef. = 4.2; 95% CI = 2.9, 5.4). When childhood, chronic adult, and recent adversity were included simultaneously (Model 4), recent adversity remained significantly associated with depressive symptoms as well as chronic adult adversity at four or more adversities. Higher levels of recent adversity continued to show a strong, graded association with PHQ-9 scores, indicating that proximal adversity exposure was most strongly related to current depressive symptomology.

Table 3.

Generalized Linear Models of Adversities and PHQ-9 (N=900)

Variables	Model 1	Model 2	Model 3	Model 4
Childhood adversity (ref. = 0)
1	0.47			0.04
	(−0.28, 1.22)			(−0.73, 0.80)
2	0.39			0.15
	(−0.88, 1.67)			(−1.07, 1.36)
3	0.23			−0.36
	(−1.12, 1.58)			(−1.70, 0.98)
4+	2.54**			1.28
	(0.87, 4.20)			(−0.33, 2.88)
Chronic adult adversity (ref. = 0)
1		0.45		−0.23
		(−0.43, 1.33)		(−1.11, 0.66)
2		1.39**		−0.10
		(0.47, 2.31)		(−1.17, 0.97)
3		2.58***		0.46
		(1.23, 3.94)		(−0.95, 1.86)
4+		4.15***		1.48*
		(2.93, 5.37)		(0.01, 2.94)
Recent adversity (ref. = 0)
1			0.78	0.82
			(−0.02, 1.57)	(−0.03, 1.66)
2			1.81***	1.67**
			(0.87, 2.75)	(0.67, 2.67)
3			3.43***	3.08***
			(2.11, 4.75)	(1.64, 4.53)
4+			5.42***	4.86***
			(4.53, 6.32)	(3.70, 6.03)

Notes: Models control for participant study arm, age in years, education in years, current pregnancy, family history of mental illness, and assessor fixed effect. Analyses use multiply imputed data and are weighted using inverse probability weights. 95% confidence intervals are reported in parentheses.

^***p<0.001,

^**p<0.01,

^*p<0.05

Results for anxiety symptoms (GAD-7), largely mirrored findings for depressive symptoms. For example, chronic adult adversity and recent adversity were associated with higher anxiety symptoms in a dose-response pattern. The one notable exception is that one childhood adversity also showed associations with GAD-7 scores (Table 4; coef. = 0.8; 95% CI = 0.2, 1.5). When childhood, chronic adult, and recent adversity were included simultaneously (Model 4), childhood adversities were no longer associated with anxiety symptoms, and chronic adult adversity was associated with higher GAD-7 scores only at the highest level of exposure (four or more chronic adult adversities; coef. = 1.9; 95% CI = 0.8, 3.1). In contrast, recent adversity remained robustly associated with anxiety symptoms, with increasing levels of recent adversity corresponding to greater symptom severity (1 adversity: coef. = 0.8; 95% CI = 0.2, 1.4; 4+ adversities: coef. = 4.4; 95% CI = 3.6, 5.2).

Table 4.

Generalized Linear Models of Adversities and GAD-7 (N=900)

Variables	Model 1	Model 2	Model 3	Model 4
Childhood adversity (ref. = 0)
1	0.83*			0.43
	(0.17, 1.50)			(−0.23, 1.08)
2	0.78			0.55
	(−0.20, 1.76)			(−0.38, 1.48)
3	1.29			0.66
	(−0.03, 2.61)			(−0.59, 1.91)
4+	1.63*			0.40
	(0.06, 3.20)			(−1.06, 1.86)
Chronic adult adversity (ref. = 0)
1		0.35		−0.21
		(−0.24, 0.95)		(−0.80, 0.39)
2		1.53***		0.26
		(0.80, 2.26)		(−0.53, 1.05)
3		2.65***		0.74
		(1.54, 3.75)		(−0.30, 1.79)
4+		4.30***		1.90**
		(3.21, 5.39)		(0.75, 3.05)
Recent adversity (ref. = 0)
1			0.78**	0.75*
			(0.20, 1.36)	(0.15, 1.35)
2			1.20**	0.95*
			(0.40, 2.01)	(0.17, 1.72)
3			2.71***	2.28***
			(1.54, 3.88)	(1.11, 3.44)
4+			5.19***	4.42***
			(4.38, 6.00)	(3.62, 5.23)

Notes: Models control for participant study arm, age in years, education in years, current pregnancy, family history of mental illness, and assessor fixed effect. Analyses use multiply imputed data and are weighted using inverse probability weights. 95% confidence intervals are reported in parentheses.

^***p<0.001,

^**p<0.01,

^*p<0.05

Results for hair cortisol are presented in Table 5. When examined separately, childhood and chronic adult adversity were not significantly associated with cortisol levels. Recent adversity was associated with lower cortisol levels at three and four or more adversities. When childhood, chronic adult, and recent adversity were included simultaneously, none remained significantly associated with cortisol levels (Model 4).

Table 5.

Generalized Linear Models of Adversities and Hair Cortisol (N=810)

Variables	Model 1	Model 2	Model 3	Model 4
Childhood adversity (ref. = 0)
1	−0.14			−0.12
	(−0.30, 0.02)			(−0.28, 0.03)
2	0.10			0.11
	(−0.06, 0.25)			(−0.04, 0.26)
3	−0.18			−0.17
	(−0.40, 0.04)			(−0.38, 0.05)
4+	−0.20			−0.16
	(−0.40, 0.00)			(−0.37, 0.05)
Chronic adult adversity (ref. = 0)
1		0.02		0.06
		(−0.15, 0.20)		(−0.12, 0.23)
2		−0.02		0.06
		(−0.22, 0.17)		(−0.14, 0.26)
3		−0.11		−0.04
		(−0.31, 0.09)		(−0.25, 0.18)
4+		−0.13		−0.03
		(−0.32, 0.06)		(−0.25, 0.19)
Recent adversity (ref. = 0)
1			−0.10	−0.10
			(−0.32, 0.13)	(−0.32, 0.13)
2			−0.18	−0.16
			(−0.39, 0.03)	(−0.37, 0.05)
3			−0.23*	−0.22
			(−0.46, −0.01)	(−0.44, 0.00)
4+			−0.21*	−0.17
			(−0.40, −0.02)	(−0.37, 0.04)

Notes: Models control for participant study arm, age in years, education in years, current pregnancy, family history of mental illness, assessor fixed effect, days since last hair wash, hair moistness since last hair wash, hair product usage, hair cream usage, weight of hair sample, and weather (humidity, temperature) on day of interview. Analyses use multiply imputed data and are weighted using inverse probability weights. 95% confidence intervals are reported in parentheses.

^***p<0.001,

^**p<0.01,

^*p<0.05

Sensitivity Analyses

Sensitivity analyses yielded findings largely consistent with the primary analyses. In analyses modeling adversity as continuous measures, we find similar patterns as the primary analysis (Tables S5–S7). For example, for PHQ-9 and GAD-7, higher adversity is associated with greater internalizing symptoms for all types of adversity when adversity is modeled individually. Chronic adult and recent adversity continue to remain significantly associated with internalizing symptoms in the model with all adversity types combined. Overall, this suggests that adult adversity may be more relevant for internalizing symptoms than childhood adversity. Yet, results still indicate that the association between recent adversity and internalizing symptoms is stronger than it is for childhood and chronic adult adversity. For the cortisol results, the main difference from the primary analyses is that chronic adult adversity on its own is associated with lower cortisol levels. Overall associations for the continuous measures were smaller in magnitude than those observed using categorical adversity indicators. This is likely reflected in the fact that the relationship between adversity and the outcomes was concentrated at the highest levels of adversity, rather than increasing steadily across the full range of exposure.

Additionally, we ran analyses with interactions between the adversity variables to investigate whether the combination of adversities was differentially associated with PHQ-9, GAD-7, and cortisol (Table S8). There were no significant interactions between the adversity variables, suggesting that these adversities may have additive effects. To assess whether study arm potentially moderated the association between adversity and internalizing symptoms and HPA axis functioning, we conducted sensitivity analyses interacting study arm with adversity and observed no significant interaction effect, indicating no moderation of study arm (Tables S9–11). For analyses with the alternative operationalization of chronic adult adversity, defined as the number of study waves in which women experienced two or more adversities (range: 0–4; Table S12), there were also no significant differences from the main results. We also ran sensitivity analyses excluding hair-related covariates (e.g., hair characteristics and environmental conditions) from the main cortisol analyses (Table S13). When these covariates were omitted, the associations between adversity and cortisol were no longer statistically significant, suggesting that cortisol estimates may be sensitive to adjustment for hair-related factors. Lastly, we ran sensitivity analyses excluding the pregnant women from the cortisol analysis and found that the results were consistent with the main results (Table S14).

Discussion

Adversity exposure has been consistently associated with internalizing symptoms and alterations in stress physiology, yet little is known about how childhood, chronic adult, and recent adversity may differentially relate to these outcomes and whether results hold in a low-income country setting (Jones et al., 2018). Using longitudinal data on mothers in rural Pakistan, we find distinct patterns across psychological and physiological outcomes. Although childhood, chronic adult, and recent adversity were each associated with higher depressive and anxiety symptoms when examined separately, recent adversity emerged as the most robust predictor of depressive and anxiety symptoms when all adversity variables were considered simultaneously. In contrast, associations between adversity and cortisol were more modest, with evidence of lower cortisol levels among women experiencing the highest levels of adversity, with these associations attenuated in the combined model. These findings highlight differences in how childhood, chronic adult, and recent adversity relate to internalizing symptoms and stress physiology.

Childhood, chronic adult, and recent adversity were all associated with higher levels of depressive and anxiety symptoms, particularly at greater levels of exposure. These findings are consistent with life course models that emphasize overlapping processes linking adversity to health (Morrissey & Kinderman, 2020). Notably, chronic adult adversity and recent adversity showed a dose-response relationship with depressive and anxiety symptoms, consistent with the accumulation of risk and recency models that emphasize the importance of sustained and proximal stress exposure, respectively (Lian et al., 2024). Women experiencing the highest levels of adversity had regression coefficients that were more than twice as large as those observed for women with only one adversity (relative to those with none), underscoring the potential value of interventions targeting high levels of adversity exposure. Given that high adversity exposure was common in this sample, even partial reductions in adversity burden may yield meaningful mental health benefits. Together, these patterns highlight the importance of considering adversity across the life course when evaluating mental health risk.

We also observed modest evidence that cortisol levels were lower among women experiencing the highest levels of adversity. Lower cortisol levels at high levels of adversity exposure may reflect blunted HPA axis activity, which could be due to repeated or sustained stress over the life course (Fries et al., 2005; Niu et al., 2025). However, associations between adversity and cortisol were generally weaker than those observed for depressive and anxiety symptoms and estimates were sensitive to model specification, as adversity dropped from significance after excluding hair-related covariates. Additionally, neither childhood, chronic adult, nor recent adversity remained statistically significant in the combined models, which may indicate overlapping effects of adversity across the life course. However, this attenuation could also reflect limited statistical power. Given the modest effect sizes and attenuation of associations in the combined models, these findings should be interpreted cautiously, though there was a predominance of negative associations, which is consistent with the blunting hypothesis.

Recent adversities were more strongly associated with depressive and anxiety symptoms than childhood or chronic adult adversities, consistent with prior work showing stronger associations between proximal exposures and anxiety (Seok et al., 2020). One possible explanation is that recent adversities may be particularly salient for current psychological functioning, as individuals evaluate their mental health in the context of stressors that are most immediate in their daily lives. Recent adversities may also reflect the downstream accumulation of earlier exposures (Pearlin, 1999). We did not, however, observe this pattern for cortisol. This could be due to modest cortisol associations overall that weaken even further when multiple, correlated adversity exposures are considered simultaneously. Focused interventions addressing proximal stressors beyond key developmental periods may have measurable psychological benefits, particularly among women experiencing multiple co-occurring adversities.

These findings highlight several implications for intervention and future research. The strong associations observed for recent adversity suggest there may be a critical window shortly after adversity exposure during which targeted support could mitigate short-term psychological impact. In low-resourced settings where eliminating all adversities may be unrealistic, these targeted reductions in modifiable stressors, such as intimate partner violence, acute financial stress, or worries about children, may yield meaningful mental health benefits in adulthood, even when earlier adversities cannot be changed. Given evidence that childhood, chronic adult, and recent adversity were all associated with higher levels of depressive and anxiety symptoms, comprehensive wrap-around services that address adversity across the life course are important. There are multiple commonalities between rural Pakistan and other low-resourced contexts such as economic insecurity, limited access to mental health services, and high burdens of stressors, that support the broader relevance of our findings. At the same time, certain contextual factors may be more specific to rural Pakistan and other South Asian settings such as gender norms (e.g., limited female agency, low labor force participation) and family structure (e.g., joint households) which can shape women’s experiences of adversity. Given the study’s focus on mothers, reducing maternal adversities may also have implications for child well-being. Finally, although internalizing symptoms and cortisol were examined as parallel outcomes, future work should investigate how these outcomes are jointly related to adversity over time, such as whether cortisol mediates the relationship between adversity and internalizing symptoms.

Several limitations should be considered when interpreting the findings. First, childhood adversity was assessed retrospectively, which may be subject to recall bias. While retrospective reports of childhood adversity are widely used in life course research (Newbury et al., 2018) depression can result in more negative reports of early life events, which could lead to an overestimation of associations between childhood adversity and later outcomes (Roisman et al., 2014). Second, we used different measures to assess adversity in childhood and adulthood. While this measurement reflects the most salient stressors within each life stage, it also limits comparability across time. Third, SES assets were not available at the 1 and 3-year waves, therefore the most proximate assessment of assets was assigned for those years, baseline and 2-years. Although this approach assumes relative stability in assets over short periods, changes in assets may still have occurred. About half of women in the lowest asset quintile at baseline remained in that quintile three years later, with more limited mobility expected over a one-year period.

Fourth, adversity could not be examined across finer developmental periods or durations and our measure of chronic adversity only captured adversity for a limited window of adulthood. Fifth, top-coding adversity at four or more exposures in the main analyses may have introduced minor distortion in some associations, which may limit the precision of these estimates. Sixth, adversities were evaluated in the survey by asking whether or not the adversity event occurred. By not having the intensity and meaning of events we may be underestimating the association between adversity and the outcomes, particularly for those who experienced the highest number of adversities, by reducing variability in exposure. Future research should evaluate the intensity of the event. An additional limitation is the reliance on self-reported measures for adversity, depressive, and anxiety symptoms could have introduced shared method variance. However, the use of binary event indicators, temporal ordering of the measures, and the inclusion of a biological outcome (hair cortisol) may partially mitigate this concern.

In sum, childhood, chronic adult, and recent adversity exposure were each associated with poorer health, but the strength of these associations varied. Recent stressors were most strongly associated with current depressive and anxiety symptoms when childhood, chronic adult, and recent adversity were all considered simultaneously, whereas associations with cortisol were evident primarily at high exposure levels and consistent with the blunting mechanism. These findings underscore the importance of considering both prior and ongoing adversity when evaluating health risk and suggest that interventions targeting current stressors, alongside efforts to prevent and reduce cumulative adversity, may be especially beneficial.

Public health significance.

This study clarifies how childhood, chronic, and recent adversity relate to psychological and physiological outcomes. Findings suggest that childhood, chronic, and recent adversity are associated with poorer mental health, with recent adversity showing particularly strong associations with current mental health. These results highlight the importance of addressing ongoing stressors, alongside early prevention efforts, to improve health outcomes.