Early life risk factors for depression among young adults in the United States general population: attributable risks and gender differences

Abstract

Background.

This study adopts individual and societal-level approaches to examine the contribution of childhood risk factors to major depressive episodes (MDE) in 2,526 American young adults.

Methods.

Nationally representative data from the 2017 U.S. Panel Study of Income Dynamics – Transition into Adulthood Supplement were analyzed using multivariate methods to assess the impact of parental mental illness, childhood adversities, childhood mental disorders, and childhood physical conditions. Adjusted odds ratios and population attributable risk proportions (PARPs) are calculated to estimate the proportion of MDE cases related to risk factors.

Results.

The 12-month prevalence of positive screens for MDE was 25.4%. Approximately 34% of these were attributable to childhood mental disorders, 24% to childhood physical conditions, 21% to childhood adversities, and 16% to parental mental illness. Childhood and parental depression were critical risk factors, both at the individual (odds ratio exceeding 2) and societal (PARP approximately 24%) levels. Gender-specific risk factors were identified, with childhood physical abuse and childhood anxiety disorders constituting risk factors for females, and childhood externalizing disorders and childhood headaches as risk factors for males. Approximately 60% of U.S. young adult MDE cases are attributable to risk factors before age 18.

Limitations.

Possible over reporting of MDE may have biased the associations between predictors and depression.

Conclusions.

Exposure to depression at a young age—one’s own or parental depression—is a robust risk factor for both genders. Policies and interventions focused at alleviating the societal burden of depression should value its generational transmission.

INTRODUCTION

Major depressive episode (MDE) is the most common mental disorder worldwide, with recent 12-month estimates in the United States of 3.0–16.6%.^1,2 By 2030, this condition is projected to be the leading cause of disability.^3,4 MDE has an early age of onset and features considerable functional impairment spanning the personal, social, and employment domains.⁵ Identifying and quantifying early-life risk factors is crucial for public health initiatives aimed at preventing depression.⁶ A plethora of studies have identified depression risk factors, including sociodemographic characteristics (e.g., female gender, recent birth cohorts⁷), distal risk factors (e.g., adverse childhood experiences (ACEs), parental mental illness⁸), and proximal risk factors (e.g., recent stressors).

Although these studies offer valuable insights, there are three major limitations in the current scientific knowledge on depression’s risk factors. First, this body of research, and related studies of major depression, focus on individual risk factors without considering their population-level prevalence. The few studies that did so largely focused on single factors, like early childhood experiences of childhood abuse, parental loss, or household dysfunction that account for approximately half of adult MDE cases.^9,10 However, this approach overlooks the broader domain of distal risk factors, like parental mental illness or physical conditions. Second, it is unlikely that risk factors operate in isolation in their association with MDE. The extent to which different risk factors, or domains of risk factors, effectively cluster together and influence one another is unknown, leaving a gap in our understanding of how risk factors before age 18 contribute to depression in young adulthood. Third, although female gender is associated with higher MDE prevalence and represents a major health disparity,^11,12 analyses of which risk factors differ by gender remains incomplete. Existing work finds that being female and exposure to ACEs are synergistic risk factors in the U.S. population and calls for further studies of gender-based societal inequalities that might contribute to the increased prevalence of depression in women and girls.^13,14

This study examines associations of different risk factor domains in the 2017 U.S. general population under 30 years of age. We use data from the Panel Survey on Income Dynamics – Transition into Adulthood Supplement (PSID-TAS) 2017 wave.^15,16 We address the limitations described above by conducting comprehensive analyses of early risk factors for MDE in young adulthood. We investigate (1) the extent to which parental mental disorders, ACEs, childhood mental disorders, and childhood physical conditions are associated with MDE on the individual level (i.e., among those who have one or more risk factors), (2) the amount of preventable MDE cases when these risk factors could be fully targeted or eliminated, and (3) gender differences in the associations.

METHODS

Respondents and sampling

Data come from the PSID-TAS 2017 which provides a nationally representative sample of young adults in PSID families. With a sample of over 5,000 U.S. households, PSID is the longest running longitudinal household survey globally.^c TAS 2017 interviewed 2,526 young people aged 18–28 with a response rate of 87%. Beginning in 2017, TAS included all age-eligible (18–28 years) PSID sample members, covering information on timing, sequence, and context of transition to adulthood events.¹⁶ Of particular importance to understanding MDE is the health content, including childhood health conditions, a set of questions on ACEs, and questions about parental mental health.¹⁸ Although the median age of MDE onset can vary considerably across countries, epidemiological studies indicate that the average age of onset in the U.S. is 22.7 years, with an interquartile range of 15.1–34.6 years.⁷ Consequently, the PSID-TAS sample presents an appropriate age range for investigating MDE in the U.S. population.

Instruments

The PSID-TAS utilizes two core MDE items from the Composite International Diagnostic Instrument (CIDI):¹⁹ “In the past 12 months, have you had two weeks or longer when nearly every day you felt sad, empty, or depressed for most of the day?” and “In the past 12 months, have you had two weeks or longer when you lost interest in most things like work, hobbies, and other things you usually enjoyed?” Respondents who answered ‘yes’ to both questions were categorized as having MDE in the past 12 months. Although these two items do not encompass the full range of MDE diagnostic criteria, they are the two core items of MDE.²⁰ Sensitivity and specificity analyses show high sensitivity (>90%) but a lower specificity (~67%),²¹ suggesting overestimation of MDE.

We include four risk factor domains: parental mental disorders, ACEs, childhood mental disorders, and childhood physical conditions. The inclusion of these specific risk domains was based on a lifespan conceptual framework, assuming that occurrences in early life may have large effects in adulthood and that the sequence of an earlier occurrence may influence the effect a following occurrence may have on the outcome. This framework has been used previously in identifying targetable risk factors for depression⁹ and suicidality.²²

Parental mental disorders

Five different forms of parental mental illness during respondents’ childhood are captured in the dataset: parental depression, parental anxiety disorder, parental alcohol problems, parental substance use, and other parental mental illness. Parental mental disorder was classified as present if the respondent gave an affirmative response to questions on the core symptoms of that specific mental disorder in the mother or father. For example, for mother’s depression, respondents who indicated they were raised primarily by their mother were asked whether she experienced depression prior to respondent age 18 and how often this was the case. Respondents who replied yes to the first question and ‘all of the time’ or ‘some of the time’ to the second question were classified as having had a mother with depression during the respondents’ childhood prior to age 18.

Adverse childhood experiences (ACEs)

ACEs measured in TAS include: parental physical abuse (e.g., pushing, throwing things, physical harm towards the respondent), parental emotional abuse (of the respondent), parental violent behavior (insults/swearing, pushing, throwing things, domestic violence), forced intercourse, parental criminal behavior (endorsed for respondents’ parents who were in contact with the legal system), and parental divorce prior to respondent age 18.

Childhood mental disorders

Childhood mental disorders included: anxiety disorder, depression, substance abuse disorder, externalizing disorder, and other mental disorders. The presence of childhood disorders was ascertained by the following question: “During your childhood – that is, up to when you reached 18 years of age – did you have any of the following psychological, developmental or behavioral conditions for one month or more?”

Childhood physical conditions

Childhood physical conditions were assessed with a standard chronic disorders checklist adapted from the U.S. Health Interview Survey (1994) collected by the National Center for Health Statistics. Such checklists yield more complete and accurate reports than those based on open-ended questions.²³ There is moderate to good concordance between such reports and medical records.²⁴ The five physical conditions included: allergy, pain conditions, obesity, headaches, and other physical conditions during childhood. The physical conditions were ascertained by the following question: “During your childhood – that is, up to when you reached 18 years of age – did you have any of the following physical health conditions for one month or more?” Those who replied affirmatively were classified as having the condition.

Demographic characteristics

The key demographic characteristics included from TAS are age, sex, living arrangements, working arrangements, race, and whether the respondent was raised by two parents. Age was measured as a continuous variable, based on the respondent’s date of birth. Respondent sex was self-reported as either male or female. Living arrangements classified respondents based on their current living situation: married, never married and cohabiting, never married and no cohabiting, and separated, divorced, or widowed. The working status of the respondent was categorized into working (regardless of full- or part-time status), unemployed, student, or keeping house. Race and ethnicity were self-identified by participants. The TAS race and ethnic group categories included are Black or African American, Hispanic, Latino or Spanish, and White. A broad, general binary variable captured whether the respondent was raised in a household by two parents (any of biological, adoptive, or stepparents) for most of the time before age 18.

Statistical analysis

In the multivariate analyses, we used both individual- and societal-level approaches to study early life risk factors of depression. Descriptive analyses include N, percent, and standard deviations (SD).

Individual-level approaches refer to the difference in the likelihood of an outcome for an individual, given the occurrence of a specific risk factor. To investigate the individual-level associations, we estimated the time-ordered logistic equation on the associations between the risk factors and the odds of 12-month MDE. The strength of the associations is expressed in odds ratios (OR) with 95% confidence intervals (95% CI) where the ORs represent the odds of having the outcome relative to the presence of the independent variable (e.g., childhood depression).

For the societal-level approach, we used population attributable risk proportions (PARPs) simulations that enable identification of independent variables that have the most impact on the studied public health outcomes at the population-level. The PARP can be interpreted as the proportion of MDE in young adulthood that may be reduced when one of the specific risk factors (e.g., childhood depression) would be preventable or fully treated. The variables we constructed for the PARP simulations were binary (i.e., at least one ACE/Childhood mental disorders/parental mental disorders present vs. absence of any).

First, we provide multivariate PARPs for each of the risk factors; these provide insight into the unique population-scale contribution of specific risk factors in the occurrence of MDE.^22,25 Second, because we assume a temporal sequence in which risk factors occur across a young person’s lifespan, we calculated PARPs for the 4 risk factor domains (parental mental illness, ACEs, childhood mental disorders, and childhood physical conditions) in temporal order. We started by investigating the PARP for the most distal domain of risk factors: parental mental illness. Then, the next most proximal risk factor domain (ACEs) was added to the model. This resulted in a series of different models with parental mental illness as the only predicting factor (Model 1), parental mental illness and ACEs (Model 2), parental mental illness, ACEs, and childhood mental disorders (Model 3), and, finally, parental mental illness, ACEs, childhood mental disorders, and childhood physical conditions (Model 4). Some of the PARPs resulted in negative population attributable risk. These can be systematically reassigned to “not applicable” in order to convey that there would be no change in the outcome variables.

Race/ethnicity, age, raised in a two-parent household, and living and working arrangements were entered as covariates in the models. All models (both individual-level and societal-level) include controls for person-years, sociodemographic variables, and the other risk factors. Standard errors were estimated with the Taylor series method 17 in SAS 9.4. Multivariate significance was evaluated with Waldχ2 tests based on design-corrected coefficient variance-covariance matrices. All significance tests were evaluated using .05-level two-sided tests.

Ethics approval

This study is based on secondary data analysis from publicly available PSID-TAS 2017 data from the PSID website (https://simba.isr.umich.edu/data/data.aspx). This study was reviewed and approved for implementation by the Health Sciences and Behavioral Sciences IRB at the University of Michigan (HUM00112878). The procedures used in this study adhere to the tenets of the Declaration of Helsinki.

Data Availability

This research uses publicly available data from the 2017 U.S. Panel Study of Income Dynamics – Transition to Adulthood Supplement (PSID-TAS). Data are located at https://simba.isr.umich.edu/data/data.aspx.

RESULTS

Sample descriptive statistics

The sample comprised 2,526 respondents (1,317 identifying as female), aged 18–28 years at the time of the interview, with a mean age of 22.5 years (SD = 3.5). The participants included 1,173 individuals identifying as black, 299 as Hispanic, and 1,039 as white. The majority were raised by two parents (84.8%) and 38.6% reported that they had at least one parent with a mental disorder. ACEs before age 18 were reported by 44.0% of the respondents. 32.5% reported a childhood mental disorder. 44.9% had a childhood chronic physical condition. 25.8% met the criteria for past year MDE. More details are found in Table 1.

Table 1.

Descriptive Statistics of PSID-TAS 2017 General Population Respondents (18–28 years), U.S., N = 2,526.

	Sample		Females		Males
	(N)	Weighted Percentage	(N)	Weighted Percentage	(N)	Weighted Percentage
Age (continuous)	2526	mean age + SD	1317	mean age + SD	1209	mean age + SD
Gender
Male	1209	50.6 (1.45)			1209	100 (0.00)
Female	1317	49.4 (1.45)	1317	100 (0.00)
Living arrangements
Married	252	14.8 (1.09)	151	17.2 (1.60)	101	12.4 (1.45)
Never married & cohabiting	436	15.1 (0.98)	238	17.9 (1.49)	198	12.3 (1.27)
Never married & no cohabiting	1767	67.3 (1.36)	879	61.2 (1.96)	888	73.2 (1.85)
Separated/divorced/widowed	70	2.9 (0.48)	48	3.8 (0.73)	22	2.1 (0.64)
Working arrangements
Keeping house	108	6.0 (0.76)	86	9.4 (1.16	22	2.7 (0.81)
Student	196	10.3 (0.94)	94	8.8 (1.18)	102	11.7 (1.44)
Unemployed	366	12.8 (0.99)	148	9.4 (1.16)	218	16.1 (1.59)
Working	1855	71.0 (1.37)	989	72.5 (1.85)	866	69.5 (2.01)
Race
Black	1173	15.2 (0.53)	630	16.3 (1.26)	543	14.2 (1.19)
Hispanic	299	19.8 (1.27)	151	18.2 (1.67)	148	21.3 (1.91)
White	1039	65.0	536	65.5	518	64.5
Raised by two parents	2022	84.8 (1.07)	1046	85.8 (1.38)	976	83.9 (1.63)

Individual-level effects of early life risk factors on MDE in early adulthood

Individual-level effects were evaluated for various parental and early childhood factors and then considered for each gender. MDE was more prevalent among respondents who were not married (ORs range 3.09–4.06). Of the early life risk factors, strong associations were found for childhood depression (OR = 2.75) and parental depression (OR = 2.06) (Table 2). Other significant risk factors included childhood headaches, childhood externalizing disorders, and parental anxiety (ORs between 1.49–1.88). Other early childhood adversities such as parental emotional abuse, parental violence, parental criminal behavior, or forced intercourse did not reach statistical significance.^d We also found a clear gradient between number of childhood adversities and the likelihood of MDE, with ORs ranging from 1.52 (95% CI 1.05–2.2.) for one adversity, to 1.80 (95% CI 1.16–2.80) for two adversities, 2.30 (95% CI 1.37–3.85) for three adversities, and 1.93 (95% CI 0.79–4.68) for 3 or more reported adversities.

Table 2.

Prevalence of Early Age Risk Factors and Individual-Level Predictors for Positive Major Depressive Episode Screen in Young Adulthood, PSID-TAS 2017 General Population Respondents, U.S.

	Sample			Females			Males
	(N)	Predictor Prevalence (Weighted %, SD)	Adjusted OR (95% CI)	(N)	Predictor Prevalence (Weighted %, SD)	Adjusted OR (95% CI)	(N)	Predictor Prevalence (Weighted %, SD)	Adjusted OR (95% CI)
Parental mental disorders
Parental depression	443	19.2 (1.13)	2.06 (1.41–3.00)	227	20.6 (1.62)	2.12 (1.28–3.54)	216	17.9 (1.57)	1.87 (1.07–3.28)
Parental alcohol problems	313	13.6 (1.00)	0.80 (0.52–1.23)	162	14.4 (1.45)	0.64 (0.35–1.20)	151	12.8 (1.39)	0.95 (0.52–1.72)
Parental substance abuse	160	5.8 (0.64)	0.59 (0.30–1.15)	82	6.1 (0.96)	0.80 (0.33–1.97)	78	5.5 (0.86)	0.28 (0.11–0.75)
Parental anxiety disorder	497	23.4 (1.22)	1.49 (1.02–2.17)	280	25.9 (1.75)	1.12 (0.67–1.88)	217	20.9 (1.71)	2.43 (1.36–4.33)
Other parental emotional problems	128	6.9 (0.74)	0.53 (0.27–1.06)	80	8.9 (1.19)	0.45 (0.21–0.98)	48	4.9 (0.89)	0.94 (0.29–3.00)
Adverse childhood experiences (ACEs), below 18 years
Physical abuse	405	13.5 (0.98)	1.61 (1.00–2.59)	175	11.2 (1.22)	2.32 (1.17–4.63)	230	15.7 (1.52)	1.18 (0.57–2.43)
Emotional abuse	528	20.9 (1.17)	1.55 (0.99–2.41)	256	20.2 (1.56)	1.23 (0.67–2.26)	272	21.7 (1.73)	1.86 (0.96–3.63)
Parental violent behavior	667	28.9 (1.32)	1.06 (0.71–1.56)	363	30.7 (1.85)	0.94 (0.54–1.65)	304	27.1 (1.88)	1.34 (0.75–2.39)
Parental criminal behavior	199	6.5 (0.70)	1.23 (0.63–2.39)	93	6.6 (1.01)	1.35 (0.49–3.74)	106	6.4 (0.98)	1.10 (0.48–2.54)
Parental divorce	290	11.0 (0.88)	1.05 (0.67–1.64)	148	11.8 (1.25)	0.94 (0.49–1.83)	142	10.1 (1.23)	1.25 (0.63–2.45)
Forced intercourse	117	4.3 (0.54)	1.00 (0.52–1.92)	89	6.8 (0.94)	0.83 (0.40–1.68)	28	2.0 (0.54)	2.42 (0.87–6.73)
Childhood mental disorders
Childhood anxiety	417	20.9 (1.21)	1.32 (0.87–2.01)	264	25.9 (1.80)	1.90 (1.14–3.17)	153	15.9 (1.59)	0.76 (0.36–1.63)
Childhood depression	381	18.0 (1.14)	2.75 (1.81–4.19)	221	20.0 (1.64)	2.41 (1.39–4.18)	160	16.0 (1.57)	3.62 (1.88–6.94)
Childhood substance abuse	86	3.7 (0.52)	1.72 (0.86–3.47)	39	3.8 (0.75)	1.45 (0.50–4.21)	47	3.6 (0.77)	2.67 (1.10–6.44)
Childhood externalizing disorders	286	12.0 (0.96)	1.88 (1.17–3.00)	96	8.6 (1.19)	1.45 (0.69–3.07)	190	15.3 (1.49)	2.12 (1.73–3.82)
Other childhood mental disorders	68	3.3 (0.54)	1.09 (0.51–2.35)	27	2.2 (0.56)	0.71 (0.23–2.18)	41	4.4 (0.92)	1.40 (0.47–4.18)
Childhood physical conditions
Allergy	397	14.9 (1.0)	1.38 (0.89–2.15)	234	16.2 (1.46)	1.28 (0.75–2.18)	163	13.7 (1.41)	1.85 (0.90–3.81)
Pain	166	7.6 (0.8)	1.46 (0.78–2.72)	118	10.7 (1.26)	1.31 (0.63–2.74)	48	4.5 (0.90)	2.60 (0.89–7.62)
Obesity	305	13.2 (1.0)	1.37 (0.89–2.11)	166	11.2 (1.21)	1.36 (0.78–2.37)	139	15.1 (1.58)	1.66 (0.87–3.15)
Headaches	291	13.3 (1.0)	1.88 (1.26–2.81)	190	17.5 (1.57)	1.77 (1.06–2.95)	101	9.1 (1.24)	2.70 (1.34–5.44)
Other physical conditions	412	18.2 (1.1)	0.98 (0.68–1.40)	189	15.3 (1.42)	0.86 (0.50–1.48)	223	21.1 (1.69)	1.06 (0.64–1.76)
% concordance			73.2%			72.4%			75.3%
Somers’ D			0.47			0.45			0.51
C-statistic			0.73			0.73			0.76

All significance tests were evaluated using .05-level two-sided tests. Significant results are in bold typeface.

Analysis revealed gender-specific associations between parental and early childhood factors and the prevalence of MDE in young adults. In the entire sample, female gender was associated with a higher prevalence of positive MDE screens (OR = 1.40 [95% CI = 1.03–1.92]). After disaggregation by gender, two risk factors were consistently associated with MDE: childhood depression (OR = 2.41 for females and 3.62 for males) and parental depression (OR = 2.12 for females and 1.87 for males). Parental physical abuse was a specific risk factor for MDE among females (OR = 2.32), as was childhood anxiety (OR = 1.90). By contrast, childhood headaches, childhood substance abuse disorder, parental anxiety, and childhood externalizing disorders were specific risk factors of MDE for males in young adulthood (OR of 2.70, 2.67, 2.43, and 2.12, respectively). Also, we found that males who reported parental substance abuse (prevalence: 5.5%) had lower odds of MDE (OR = 0.28).

Societal-level effects of early life risk factors on MDE in early adulthood

Table 3 shows PARPs of 12-month MDE by any of the risk factors. The range of significant PARPs is 4.6% through 14.4%. Childhood depression (PARP = 14.4%), parental depression (PARP = 9.9%), parental anxiety disorder (PARP = 6.7%), and parental physical abuse (PARP = 4.6%) were individual risk factors yielding the highest attributable risks. These findings suggest that exposure to depression in childhood (either through parents or by one’s own experience) may be accountable for one fourth of MDE in young adulthood.

Table 3.

Multivariate Population Attributable Risk Proportions of Early Life Risk Factors of Positive Screens for Major Depressive Disorder in Young Adulthood, PSID-TAS 2017 General Population Respondents, U.S.

	Sample		Females		Males
	PARP	SD	PARP	SD	PARP	SD
Parental mental disorders
Parental depression	9.9	0.5	9.9	0.7	8.4	0.7
Parental alcohol problems	NA	NA	NA	NA	NA	NA
Parental substance abuse	NA	NA	NA	NA	NA	NA
Parental anxiety disorder	6.6	0.4	1.8	0.1	13.9	1.0
Other parental emotional problems	NA	NA	NA	NA	NA	NA
Adverse childhood experiences (ACEs)
Parental physical abuse	4.6	0.3	6.2	0.7	2.0	0.2
Parental emotional abuse	6.4	0.3	2.7	0.2	10.0	0.8
Parental violent behavior	1.0	0.0	NA	NA	5.2	0.4
Parental criminal behavior	0.8	0.0	1.2	0.2	0.4	0.1
Parental divorce	0.3	0.0	NA	NA	1.4	0.2
Forced intercourse	NA	NA	NA	NA	1.4	0.4
Childhood mental disorders
Childhood anxiety	4.4	0.2	11.7	0.7	NA	NA
Childhood depression	14.4	0.7	12.6	0.8	17.0	1.4
Childhood substance abuse	1.5	0.2	1.0	0.2	2.7	0.6
Childhood externalizing disorders	5.2	0.0	2.1	0.3	8.2	0.8
Other childhood mental disorders	0.2	0.0	NA	NA	1.0	0.2
Childhood physical conditions
Allergy	3.2	0.2	2.6	0.2	5.5	0.6
Pain	2.0	0.1	2.0	0.2	2.8	0.5
Obesity	2.7	0.2	2.2	0.2	4.7	0.5
Headaches	5.7	0.4	6.3	0.5	5.9	0.8
Other physical conditions	NA	NA	NA	NA	0.7	0.1

Note: NA refers to negative population attributable risk proportions.

Despite few differences when analyzing the four risk factor domains for gender separately, our analysis of PARPs revealed gender-specific associations for certain individual risk factors. Childhood mental disorders systematically yielded the highest PARPs for both females and males. When considering individual risk factors for females, we found that mental disorders in childhood (in particular, anxiety or depressive disorder) were among the disorders having the greatest attributable risk for MDE, with PARPs of 11.7% and 12.6%, respectively. Physical abuse by parents accounted for 6.2% of MDE prevalence in female young adults in the U.S. By comparison, in the male sample, we found the highest PARPs for childhood depression (as high as 17.0%) and for parental depression (PARP = 8.4%), suggesting that approximately one-fourth of the MDE cases in male young adults are attributable to exposure to depression. A history of externalizing disorders in childhood accounted for 8.2% of MDE presence in young adulthood but only among males.

Unique contributions of risk factor domains in MDE

In models of the differential attributable risks for MDE (Table 4), the first set of analyses of the full sample find that approximately one-third (31.5%) of U.S. young adult depression cases are associated with a history of parental mental illness (Table 4, Model 1, Full Sample). Entering the next risk factor domain (ACEs) in the model (Table 4, Model 2, Full Sample), the PARP of parental mental illness decreases from 31.5% to 26.9%, suggesting that the effect of parental mental illness is partially mediated by childhood adversities. The next model includes childhood mental disorder (Table 4, Model 3, Full Sample), resulting in a decreased PARP for childhood adversities (from 28.8% to 20.5%) and a further reduced PARP for parental mental illness (from 26.9% to 16.7%). The following model adds childhood physical conditions (Table 4, Model 4, Full Sample) and shows that the PARPs for the already included risk domains are roughly the same. Childhood mental disorders remain the most important risk factor group, accounting for about 34.5% of depression in young adulthood. When all the risk factor domains are accounted for, 59.7% of the MDE cases in the U.S. can be attributed to these four domains.

Table 4.

Multivariate Population Attributable Risk Proportions of Early Life Risk Factor Domains on Major Depressive Episode in Young Adults, PSID-TAS 2017 General Population Respondents, U.S.

	Full Sample				Females				Males
	Parental Mental Illness	Early Childhood Adversity	Childhood Mental Disorder	Childhood Physical Condition	Parental Mental Illness	Early Childhood Adversity	Childhood Mental Disorder	Childhood Physical Condition	Parental Mental Illness	Early Childhood Adversity	Childhood Mental Disorder	Childhood Physical Condition
Parental mental illness	31.5%	26.9%	16.7%	16.7%	23.7%	21.7%	12.5%	12.1%	40.1%	32.7%	21.6%	22.6%
Adverse childhood experiences (ACEs)		28.8%	20.5%	20.9%		15.5%	8.8%	9.2%		45.0%	35.1%	35.1%
Childhood mental disorder			37.0%	34.5%			31.7%	29.6%			43.0%	40.0%
Childhood physical condition				24.5%				20.8%				30.2%
All risk factor domains				59.7%				49.0%				71.9%

Disaggregation by gender shows different unique contributions of risk factor domains for male and female respondents (Table 4). First, considering all risk factor domains, 49.0% of the MDE cases in the female sample can be attributed to the four risk factor domains, compared to 71.9% for male respondents. Second, childhood mental disorders remain the most important risk factor group for both genders; however, the data suggest that depression is more strongly attributable to childhood mental disorders in males than in females (40.0% vs. 29.6%, respectively).

DISCUSSION

This is the first study using representative U.S. data to estimate the proportion of MDE cases in young adulthood that can be attributed to a range of different risk factor domains prior to the age of 18. Whereas previous studies examined the magnitude of associations between individual risk factors and MDE, our analyses considered the prevalence of different risk factor domains, estimating the proportion of MDE cases that could be reduced if the risk factors studied were eliminated, assuming a causal relationship between risk factors and outcomes. Childhood depression is the most important individual risk factor (PARP = 14.4%), but parental depression (PARP = 9.9%), parental anxiety (PARP = 6.6%), and parental physical abuse (PARP = 4.6%) also account for significant proportions of young adult MDE.

The results should be interpreted against the following limitations. First, recall bias may affect the accuracy of adversity recall; rates of not reporting abusive experiences have been estimated in the 20–33% range²⁶ but without consistent evidence for false-positive reports.²⁷ Reports of parental mental illness are also susceptible to recall error: respondents with mental disorders are more likely to report mental disorders in their family compared to those without.²⁸ Also, depressed respondents are more likely to recall negative experiences more frequently than positive or neutral experiences,²⁹ due to negative thinking and rumination.³⁰ In addition, depressed persons’ current negative mood influences the type of memories they recall, i.e., they are more likely to remember events that are consistent with their current negative mood.³¹ Second, although the range of risk factor domains assessed here is large, it is far from exhaustive and did not assess the severity, duration, or sequencing of any of these domains. In addition, we did not include other reported risk factor domains for MDE such as genetic factors or personality features. Sexual abuse was assessed using only one question on forced intercourse; the PSID-TAS did not include a broad assessment of sexually abusive experiences. In addition, some important early life experiences are not included within TAS, such as experiencing homelessness or being removed from the home. Third, we mainly used screening questions to identify depression rather than standardized clinical diagnostic interviews. The use of screening items has shown sufficient reliability in general population studies³² but may create an upward bias in the prevalence of MDE²¹ and, in turn, may have biased both the individual- and population-level associations. A fourth limitation is that we were not able to include stressors in the multivariate models to investigate stress–diathesis approaches for MDE more profoundly.

As this is the first study that investigates the role of different risk domains in depression, our approach is hypothesis-generating and explorative rather than hypothesis-testing and conclusive. Further study may therefore focus on statistical interactions between different risk domains across the lifespan, to confirm or refute our findings and to apply post-hoc adjustments to address experiment-wise errors in testing interactions between risk domains and specific vulnerabilities. The PSID-TAS offers many other analytical possibilities beyond the scope of our analysis. For example, future studies might link to the PSID Main Interview and incorporate family income to understand how experiences of poverty might impact depression. Another focus of further study may be a more profound analysis of the timing, the persistence, and the age of first occurrence of the risk factors, in order to disentangle the gender differences found in PARP estimates.

Our results confirm the high prevalence of childhood anxiety and depression found in past studies. Prevalence has been systematically reported in the 8–20% range,^33,34 so our estimate of 32.5% having ever had a mental disorder are higher than those previously published. Both the prevalence of intra-family ACEs as well as the proportion of respondents with experiences of parental mental illness in the PSID-TAS study are somewhat higher than could be expected based on previous studies.^35–37 Differences across studies in the definitions of intra-family ACEs and in the time periods being reported (past-year and 12-month estimates versus having ever experienced) likely contribute to these observed differences.

This study illuminates the prominent role of generational transmissions of depressive states.^38,39 For both genders, experiences with depression early in life, either through parents or by one’s own experience, are robust risk factors for later MDE. Our PARP analyses show that childhood depression and parental depression accounted for about one fourth of the MDE cases in young adulthood. Even more, respondents with childhood depression who also had depressed parents were 5.7 times more likely to have depression in young adulthood (i.e., a multiplication of the adjusted ORs for childhood depression [2.75] and parental depression [2.06]), with higher odds for males. From a public health perspective, childhood depression is the most important risk factor to consider, with a high prevalence of 18%. Childhood depression shows a nearly threefold higher association with young adulthood MDE, more pronounced for males than for females, and poses a high attributable risk in a societal-level approach. However, we may assume that childhood depression is also a reaction to the presence of adversities,⁴⁰ pointing to the importance of the role abusive experiences may play in the onset of disorders.

In contrast to earlier reports, some of the respondents’ early life experiences like parental divorce,⁴¹ parental alcohol abuse,⁴² or childhood physical abuse⁴³ were not associated with later MDE. Especially remarkable is that we could not find any association between forced intercourse prior to the age of 18 and MDE in young adulthood. One reason may be the relatively late age of onset of depression.² Another potential explanation is that these early life experiences influence depression in adulthood or later in life, and not so much in young adulthood. Another interpretation is more methodological: the associations between risk factors prior to age 18 and the odds of MDE in our study are controlled for a broad range of covariates (like childhood disorders), a broader range than usual in older³⁵ but also newer studies.⁴⁴

CONCLUSION

Understanding patterns of depression and early-life risk factors among young adults is important to generating evidence-based prevention strategies and furthering mental health research on this critical life phase and its multigenerational dynamics. The analysis of PSID-TAS data reveal that 25.8% of young adults reported depressive problems in the previous 12 months and suggest that mental health problems in young adults are directly associated with past experiences. Given the prevalence of young adulthood MDE, public health responses are urgently needed, and effectiveness can be improved through understanding the impact of various early life risk factors and the attendant gender differences. Existing programs highlight the challenges to implementing universal and selective prevention approaches, such as interventions to reduce the effects of parental psychopathology or childhood adversities (e.g., Head Start, WIC nutrition program, and nurse-family partnership programs⁴⁵). Nevertheless, population-based initiatives should not de facto be precluded, and strategies to prevent ACEs have been compiled by the Centers for Disease Control and Prevention (CDC) to aid states and communities.⁴⁶ The CDC guidance identifies the importance of fostering healthy norms around gender, masculinity, and violence to safeguard against violence and adversity. Prevention strategies must more fully account for the gender differences observed, not only differences in risk factors but also in the effectiveness of depression treatments, an area that remains understudied.⁴⁷ Combined prevention strategies, including legislative, individual-level, and societal-level interventions,⁴⁸ have proven successful in sustainably reducing complex clinical conditions at the population level.⁴⁹

HIGHLIGHTS.

• Most U.S. young adult MDE cases are attributable to risk factors before age 18

• Pre-pandemic 12-month prevalence of young adult MDE was 25.4%

• Approximately 34% of cases were attributable to childhood mental disorder

• Identifying depression-prone youth is a crucial step for curative interventions

• Implementing early, developmentally sensitive strategies is vital for prevention