People Just Need More Money: The Role of Disposable Income as a Key Moderator of Heart Rate Variability Among Risk Factors in Adolescents from Low-Income Households in Colombia, Nepal, and South Africa
Claire Tait-Doak 2617797
Faculty of Behavioural and Movement Sciences, Vrije Universiteit Amsterdam
P_BTHESEGHG – B-thesis Genes, Cognition and Behaviour
Dr. Martin Gevonden
August 14th, 2025
Word Count: 5939
Introduction
The global burden of depression is a leading cause of disability, yet diagnosis and treatment remain difficult. Paradigms are shifting from purely medicalised models toward understanding genes, risk factors, and physiology in depression. This study examined psychosocial risk exposures and heart rate variability (HRV) in adolescents from low-income countries, and how effects differ between those with and without depression.
Method
A sample of 1,107 participants aged 13-15 from Colombia, Nepal, and South Africa completed psychometric tests (MMAP, IDEA-RS) to identify risk factors of abuse (A) and volatile parental/caregiver relationships (PR). HRV was measured over five minutes using one of two devices. Caregivers provided household income and expenditure data to calculate financial constraint (FC).
Results
In the full regression model, FC was significant (B = 0.111, SE = 0.054, p = .04). The 4-way interaction PR × A × FC × D was significant and negative (B = -0.676, SE = 0.283, p = .017), while the 3-way PR × A × FC (with D controlled) was significant and positive (B= 0.508, SE = 0.219, p = .02). Linear regression of D and HRV was not significant (F(1, 1066) = 0.70, p = .403, R² < .001).
Conclusion
Financial constraint was the main predictor of decreased HRV, with other predictors negligible except when moderated by FC in combination with other risks. Adolescents with depression appeared less responsive to the protective effect of lower FC. Overall, predictors explained little variance in HRV, potentially due to limited variable nuance (e.g., abuse type), unmeasured confounders, or absence of higher-wealth comparison groups. These findings provide a foundation for understanding physiological impacts of poverty-related stressors in high-risk adolescent populations beyond Western contexts.
Introduction
The global burden of depression is one of the main causes of illness and disability and is predicted to be the main burden of disease by 2030 (Zhang et al., 2024). This is contrasted by the enduring challenges in prevention and treatment of depression, leading to long term symptoms, treatment-resistant depression, and a constellation of financial, social, medical, and occupational consequences of living with this condition. While women and older adults are most affected by depression, children and adolescents have unique vulnerabilities due to their developmental stage and environmental factors beyond their control.
The challenges in preventing and treating depression have their roots in difficulties identifying definitive causes of depression, and reliably measurable biomarkers of the condition. Depression is most often diagnosed as a latent trait using clinical instruments such as questionnaires. To date, there is no scientific consensus on any direct physiological measurement of depression.
In the past, the chemical imbalance theory of depression was relied on heavily by psychiatrists (Ang et al., 2022) and psychometric testing for depression was considered an indirect measurement of such an imbalance. Psychotropic medications like antidepressants and mood stabilizers became increasingly popular as a first line treatment to combat this “imbalance”. This theory has come under criticism in more recent years due to lack of empirical evidence (Ang et al., 2022).
Although medications remain popular, there has been a growing opposition to their usage due to exposure of early publication bias, and a more recent body of research demonstrating that the previously overinflated effect of antidepressants to be little more than that of the placebo (Kirsch et al., 2008, Kirsch, 2020). Mixed reports from patients and clinicians also question the benefits of these medications versus a cascade of side effects including worsening of depression symptoms, anxiety, suicidality and death, flat effect, decreased libido, and various cardiovascular complaints including SSRI-induced Long QT Syndrome (de Gregorio et al., 2011). Behaviour- and talk-based therapies show mixed response in treating MDD, often predicting positive outcomes only when used as adjunct treatments (van Bronswijk et al., 2019). The limitations of current first-line treatments for depression intensify when placed in a global context. Healthcare systems already struggle with overwhelming demand for mental healthcare, long waiting lists, high costs, and preference for the convenience of medications in the absence of better tailored solutions. These challenges are especially prohibitive for lower-income patients, and in low- and middle-income countries (LMICs), the scarcity of services is even more pronounced.
Contemporary theories of depression have shifted from the belief that the condition is idiopathic, a purely medical issue, or an arbitrarily “psychological” issue. Susceptibility to depression is now understood to be an interplay between genes, risk factors, health, physiology, and cultural context, making it difficult to formulate a unifying theory of depression, or universally applicable treatments.
Adolescent depression carries the risk of reoccurrence, long term depression, anxiety, addiction and other psychosocial problems during adulthood (Chang & Kuhlman, 2022). Adolescents who experience depression with suicidality are more at risk of recurrence and depression that continues into adulthood (Lewinsohn et al., 1994). Since various environmental factors can exacerbate adolescent vulnerability to depression, including social factors, lifetime difficult or traumatic experiences, and various dynamics and conditions found within family systems (DeCarlo Santiago & Wadsworth, 2009), quantifying and addressing these exposures is vital in combatting youth depression before it happens.
Exposure to multiple risk factors is also a concern for depression onset. Wealth status alone introduces both risk and protective factors for mental health, in both children and parents. Children and adolescents from low-income backgrounds are more likely to develop mental health problems than those from higher income backgrounds, and have a marked decrease in grey matter volume compared to developmental norms (Straub et al., 2019). Since research into poverty and health typically focuses on income level rather than financial flexibility, literature about disposable income and family mental health is lacking (Headey, 2008).
Impacts of these risk factors can also be interpreted through physiological measures. Household conflict and chaos can impact hypothalamic-pituitary-adrenal (HPA) axis function during critical periods of childhood development, such as HPA sensitization or HPA downregulation (Doom et al., 2018). Childhood adversity is also found to be associated with reduced resting-state Heart Rate Variability (HRV), but moreso when the adverse events happened less recently (Wesarg et al., 2022). Because of newer insights into physiology and mental health, the role of the Autonomic Nervous System (ANS) in maladaptive stress responses and mental and physical health outcomes has been gaining attention.
HRV is a reliable, non-invasive measure of ANS function, capturing both parasympathetic nervous system (PNS) and sympathetic nervous system (SNS) activity, where High Frequency (HF) HRV specifically reflects PNS activity and vagal tone Where heart rate is a simple value of beats per minute (BPM), HRV is the variation between inter-beat intervals (IBI). These minor variations are not an irregularity, but the ANS’s perpetual adjustments of responsivity to activity and stimuli (Shaffer et al 2014; de Geus & Gevonden 2024). The optimal state for vagally-mediated HRV is one that is neither in a persistent state of vigilance nor a rigid, mechanical calm, but ideally between the two: a relaxed but flexible state that allows for appropriate responsivity.
Whilst a robust ‘normal range’ for HRV is not fully established, current consensus agrees that this value should be neither too high nor too low, that it typically decreases with age, and may be interpreted differently depending on the individual. HRV can be observed using a variety of technologies, most commonly electrocardiogram (ECG) and photoplethysmography (PPG). While ECG provides the most direct measurement of HRV, PPG is less accurate but more widely accessible, especially to the ordinary consumer. ECG is superior to PPG in measurement of HRV, but this is not necessarily reflected in the performance of PPG devices versus “research-grade” ECG devices, as compared to the quality of laboratory quality ECG equipment (Sinichi et al., 2025). Despite limitations of HRV measurement, it remains a valid and uninvasive tool for interpretation of ANS activity, and reactivity.
Chronically reduced resting HRV has implications for cardiovascular health and may be a predictor for future complaints. HRV also shows promise as a predictor for MDD (Hartmann et al., 2018; Dell’Acqua et al., 2020). Antidepressants, including tricyclic-antidepressants (TCAs), may result in reduced HRV over time in MDD patients (Licht et. al., 2010), and depression symptom reduction associated with antidepressant treatment does not imply improved HRV (Licht et al., 2008; Kemp et al., 2009). Conversely, HRV focused interventions such as heart rate variability biofeedback (HRVB) and meditation-based interventions show significant improvements both in HRV and depression symptoms (Pizzoli et al., 2021; Bringmann et al., 2022).
These insights could be instrumental in shaping future risk assessment and treatment for depression by focusing on interventions which improve ANS regulation and stress adaptation and exercising more caution in overprescription of treatments that may have little benefit, and even hinder recovery.
The current study aims to explore the relationship between multiple salient risk factors for depression onset in adolescents and HRV as a potential biomarker for depression risk. Participants are from poor urban households in Low Income Countries (LICs), where impact of family financial stress may be especially pronounced. Emotional abuse, parental relationship quality, and household disposable income are the focus of contributing risk factors, individually and combined. A distinction is made between adolescents who do not meet the criteria for depression and those who do against the background of these combined risk factors.
The aim of this current research is to discover what impact these conditions alone may have on HRV, and how the presence of depression symptoms may drive an increased effect. If HRV is affected by the risk factors alone, then it’s measurement in at-risk populations may be an effective tool for guiding prevention strategies. If HRV is not significantly reduced by risk factors alone, but drops notably in adolescents with both depression and these risk exposures, this may help identify vulnerable individuals who would benefit from early HRV-focused interventions, such as HRVB, to increase resilience before depression symptoms present or stress responses become maladaptive. This also underscores the importance of implementing strategies to curtail the impact of risks to adolescent mental health, such as financial interventions or emotional and family supports, before depression symptoms or vagal regulation problems occur.
The current research investigates whether the presence of emotional abuse, negative parental relationship, or the presence of financial constraints within the household have an association with reduced heart rate variability in young adolescents from poor backgrounds in low-income countries. Does this effect increase when these risks are combined, and how does this differ between adolescents who meet the criteria for depression versus those who don’t? We hypothesise that: 1) adolescents aged 13-15 from poor urban households in LICSs who have exposure to childhood adversity, namely emotional abuse, financial constraint, or negative relationship between parents will have a slightly lower HRV than those without exposure to the same risk factors, 2) that those with exposure to more than one of these risk factors will have moderately lower HRV than peers without exposure, and 3) those who meet the criteria for depression will exhibit lower HRV than their non-depressed peers, with the most pronounced reductions seen in those exposed to multiple risk factors.
Methods
Participants
For this study, a sample of 1,582 adolescents aged 13 to 15 was recruited from low-income households in Colombia (Bogota), Nepal (Kathmandu), and South Africa (Cape Town), where the poverty indices are 27%, 25.2%, and 55.5% respectively. (Lund et al., 2023) After preprocessing, 1,107 participants with valid HRV data were retained for the current study.
Measures & Procedure
This study used a between-subjects, cross-sectional design, and the data was sourced from the ALIVE study. (Lund et el., 2023) Data was collected from both the adolescent participant and from their caregiver(s), consisting of biological parents, non-biological parents, and other official
guardians. Consent was provided by both the child and the caregiver. Caregivers provided demographic information and details about household income.
Due to substantial caregiver dropout, data about household finances were missing for many participants. Therefore, income data from the three countries was standardised using Purchasing Power Parity (PPP) in order to pool data from the three countries into one sample
| Figure 1. Full analysis model of clustered risk factors Financial Constraint (FC), Abuse (A) and Parent’s Relationship (PR) and their individual and combined effects on heart rate variability (HRV). Participants are grouped by depression condition = yes/no |
for analysis. Information provided by caregivers about household income and expenditure was used to calculate the household disposable income after fixed expenses, representing the degree of financial freedom or constraint within the family unit (FC). This value can be either positive or negative, representing net surplus or deficit, respectively. All analyses in this study are conducted as one sample, and not by individual country.
The 25-item Revised Child Anxiety and Depression Scale from the Measurement of Mental Health among Adolescents at the Population level (MMAP) instrument was completed by participants (Carvajal-Velez et al., 2023). The MMAP is scored on a 4-point Likert scale ranging from Never (0) to Always (3). Participants were grouped by whether or not they met the criteria for depression (D) at the time of testing.
The participants also complete the Identifying Depression Early in Adolescence Risk Score (IDEA-RS) which is a measure of factors which can contribute to risk of developing depression in the future (Kieling et al., 2021). There are eight items in the questionnaire, seven of which have a positive or negative response, and one has a list of five possible responses, where multiple choices are possible. In the current study, two items from the questionnaire were used as key predictors of childhood adversity: Parent’s Relationship (PR) and Emotional Abuse (A). For PR, question number 6 was used: “Relationship between mother and father: What is the relationship between your mother and father? Are they emotionally supportive and regularly communicate with each other; or do they argue, fight, or live apart?”; and for A, the emotional abuse component of question (8) is used: “Experience of any abuse: Have you ever experienced emotional, physical, sexual, or other forms of abuse from anyone (adults, peers, others)?”
During the data collection procedure, participants were fitted with one of two heart rhythm devices employed for this study. The Polar H10 chest strap (Polar Electro Oy) is a single-lead ECG device that records at a sampling rate of 130Hz and is worn with a chest strap at the sternum. The Inner Balance Coherence Plus by HeartMath (HeartMath, Inc) is a PPG device which uses LED photodetectors to record at a sampling rate of 500Hz and is worn attached to the earlobe using an ear clip. The device was fitted during administration of the questionnaires, and recording took place after an interval to ensure basal values of Resting HRV. The HRV Resting Phase was recorded over a 5-minute period, during which the participant assumed a neutral posture according to standard protocol, sitting with their knees at 90 degrees, hands resting on their thighs facing upward, eyes closed, and instructed to breathe normally and spontaneously.
Data Analysis
Data analysis was performed using 2024.04.2 Build 764. Preprocessing of variables to create predictor specific values was performed prior to analyses. Due to missing income data from caregiver dropout, two multiple regression models were conducted for confirmatory analysis. A reduced model including all predictor and outcome variables except the income variable FC was performed using the full sample (N = 1,107) to measure effect for other key predictors using all available HRV data. The full model was subsequently performed using all variables pertaining to our hypothesis in the reduced sample (N = 649). Independent sample means tests were performed between the full sample and reduced sample for all variables included in both models to ensure sample representativeness when interpreting results. Exploratory analyses were also performed to measure relationships between the key variables to support interpretation of confirmatory analyses. The alpha level was set at .05 for all statistical tests.
Age, Body Mass Index (BMI), and gender are added as covariates in multiple regression analyses to control for confounding effects of these variables on HRV. Gender is coded as two categories: Male (0); and Female, non-binary, and other (1). The variable Financial Constraint (FC) was calculated by subtracting the total household expenditure from the total household income, following standardisation to PPP. To eliminate outliers, FC values were converted to z-scores and Windsorized at ±3 standard deviations. The variable Parent’s Relationship (PR) was coded as “0” for positive relationship and “1” for negative relationship. “Don’t know” responses were treated as missing values. Likewise, Abuse (A) was coded as “0” when no abuse is present and “1” when experience of abuse is present at some point in the child’s lifetime. Values for Depression (D) were retained as a continuous variable for exploratory analysis. For confirmatory analysis, a separate categorical variable was created dividing participants into those who met the criteria for depression and those who did not.
In the confirmatory analysis, a multiple regression was performed to test direct and interaction effects of predictors on HRV (RMSSD, log-transformed), controlling for age, BMI, and gender. The model includes testing of all three hypotheses: the effect of each risk factor on HRV; the effect of combined risk factors on HRV; and the difference in the impact of these risk factors on HRV for depressed versus non-depressed participants. In the reduced model, which excludes income data, the predictors were: PR, A, D, PR × D, A × D, PR × A, and PR × A × D; where the outcome variable was HRV. The full model inclusive of income was: PR, A, FC, D, PR × D, A × D, FC × D, PR × A, PR × FC, A × FC, PR × A × FC, and PR × A × FC × D; where the outcome variable was HRV. Assumption testing prior to analysis was conducted to check for linearity, homoscedasticity, normality of residuals, and multicollinearity. Exploratory analyses using linear regression were performed to measure the relationship between HRV and D, as well as that of PR, A, and FC each on D.
Ethics
Ethical approval for the study was obtained from King’s College London (KCL)’s Health Faculty Research Ethics Subcommittee under reference HR/DP-23/24-38680. Approvals were also obtained by each of the data collection sites’ Institutional Review Boards: The Faculty of Health Sciences’ Human Research Ethics Committee at the University of Cape Town (South Africa) (reference number HREC315/2022), Innovations for Poverty Action’s Institutional Review Board (Colombia) (protocol number 4062), and the Ethical Review Board of the Nepal Health Research Council (Nepal) (protocol registration number 661/2023). Participation required both adolescent assent and primary caregiver consent, in compliance with national ethical guidelines in each country.
Results
Descriptives
Across the three countries, 1582 participants were included in the study. After excluding participants with unusable HRV data, the total number of valid participants was 1107, comprising Colombia (N = 417), Nepal (N = 372), and South Africa (N = 318); this is referred to hereafter as the full sample. Due to a large amount of caregiver dropout, many participants were missing income data. A reduced sample of 649 participants with complete income data was used for analyses that specifically required income.
In the full sample, participant age ranged from 13 to 15 years (M = 13.95, SD = 0.77) with 13 year olds comprising 32.0%, 14 year olds 40.7%, and 15 year olds 27.4%. 469 participants identified as male, 631 as female, four as non-binary, and three as another gender. There were no missing values. For the purpose of analysis, participants were sorted into two gender categories: Male (42.4%) and Female, Non-binary, or Other (57.6%).
The reduced sample had the same range and similar distribution of age (M = 13.9, SD = 0.78). The gender distribution was also similar with those identifying as Male comprising 44.7% and those as Female, Non-binary, or Other was 55.3%, three of whom had selected other. The distribution of participants per country was affected by caregiver dropout, especially in Nepal (N = 174) and South Africa (N = 129), with fewer losses in Colombia (N = 346).
Summary of Variables
The number of participants who met the criteria for depression (D) according to the MMAP screening questionnaire was 616 (55.6%) as compared to 491 (44.4%) who did not, where 39 values were missing (3.5%) due to incomplete scores. 605 participants reported that their parents or caregivers had a positive relationship (PR), where 468 participants reported a negative relationship.
| Figure 2. Participants per country and total, categorised by: all participants, participants with valid HRV data, participants with valid HRV data and where income data is included. |
36 participants who responded “Don’t Know” or not at all were treated as missing values (3.1%). 472 participants reported experience of abuse (A) during their lifetime, where 624 did not have any experience of abuse, with 11 missing values (1%).
The range of values for RMSSD (N = 1107) was from 5.18 to 223.88 ms, where M = 49.33, Median = 42.94, and SD = 28.62. The log-transformed values of the same had a range from 1.65 to 5.41, where M = 3.74, Median = 3.76, and SD = 0.57. HRV data were collected using the Polar H10 Cheststrap in 618 participants (55.8%) and the HeartMath PPG device for the other 489 participants (44.2%).
The PPP transformed income (Purchasing Power Parity) across the reduced sample had an income range from 0 to 14,598.54 (M = 992.53, SD = 803.18), and an expenditure range from 0 to 12,262.77 (M = 806.46, SD = 652.12). Financial Constraint (FC) then had a range from -4671.53 to 6291.834 (M = 130.08, SD = 547.37). After z-score transformation and Winsorization at ±3 SD to control for extreme outliers, FC had a range from -3 to 3 where M = -0.012 and SD = 0.78.
Confirmatory Analysis
To test the key variables except income utilising all participants, a multiple linear regression model (Reduced Model) was run with HRV as the outcome variable, and PR, A, and D as predictor variables. A total of 1050 participants were included in this model, with 46 observations deleted due to missingness. The model only explained 1.3% of the variance (R² = 0.013, Adjusted R² = 0.004) and, overall was not significant, F(10, 1050) = 1.38, p = .182. The only significant predictors were two control variables BMI (B = 0.013, SE = 0.005, p = .008), and Gender (B = -0.08, SE = 0.038, p = .032). None of the interaction terms were significant.
The full model (FM) multiple linear regression using the reduced sample of participants with income data was run with HRV as the outcome variable, and PR, A, FC, and D as predictor variables. A total of 586 participants were included in this model, with 502 observations deleted largely due to missing income data. The model explained 5.27% of the variance (R² = 0.053, Adjusted R² = 0.024) and was slightly significant, F(18, 586) = 1.81, p = .021. The predictor variable FC was significant (B = 0.111, SE = 0.054, p = .04). Again, the control variable BMI was significant (B = 0.027 , SE = 0.008, p = <.001), and Gender slightly so(B = -0.134, SE = 0.052, p = .01). A significant negative effect was found in the full 4-way interaction term PR × A × FC × D (B = -0.676, SE = 0.283, p = .017), and a similarly significant effect in the 3-way interaction term PR x A x FC, with D controlled, albeit in a positive direction (B = 0.508, SE = 0.219, p = .02).
Exploratory Analysis
A simple linear regression analysis to measure the relationship between D and HRV yielded no significant results in either direction, where F(1, 1066) = 0.70, p = .403, and did not explain a significant proportion of variance, where R² < .001, Adjusted R² < 0.
A simple linear regression for each of the relevant risk factors and D yielded small effect sizes despite statistical significance in two cases: for PR predicting D, F(1, 1036) = 39.61, p < .001, and R² = .0368, Adjusted R² = .036; for A predicting D, F(1, 1057) = 126, p = < .001 , and R² = .107, Adjusted R² = .106; and for FC predicting D, F(1, 607) = 1.45, p = .228, and R² = .002, Adjusted R² < .001.
Assumption Testing
Independent-samples t tests and chi-square tests indicated that the full sample (N = 1,107) and reduced sample (N = 649) were comparable on most variables, including BMI, D, PR, and A (all ps > .05). Small but statistically significant differences were observed for Age (M = 13.91 vs. 14.02), t(1001.9) = -2.49, p = .013, and HRV (M = 3.70 vs. 3.80), t(1044) = -3.26, p = .001. Assumption checks supported homoscedasticity (Breusch-Pagan ps > .35), normality (Shapiro-Wilk ps > .01), and acceptable multicollinearity (VIFs < 10).
Discussion
The goal of this study was to explore the impact of each of three stress and trauma based risk factors on HRV in a population of adolescents aged 13-15 from low-income urban backgrounds in Colombia, Nepal, and South Africa; to determine if these risk factors combine to negatively impact HRV more than each factor alone; and to test whether this effect is more pronounced in participants who meet the criteria for depression.
The reduced model which excluded FC to include all participants with valid HRV data (N = 1050) explained only 1.3% of the variance in HRV, and was not statistically significant overall. Whereas, once FC was included in the full model (N = 586) the explained variation then increased to 5.27%, and was statistically significant.
Our first hypothesis was only partially supported, where FC was a significant predictor of HRV, but neither PR, A, nor D were significant. Our second hypothesis was also partially supported, but only where multiple risk factors (FC, PR, and A) were present, rather than a linear cumulative effect. Our third hypothesis was not supported, in that the addition of D reversed the effect of the three risk factors combined, rather than emphasising it. In our exploratory analysis, it transpired that D and HRV did not share a relationship at all. Overall, Financial Constraint emerged as the key predictor in the variance of HRV in our sample. However, given the small proportion of variance explained, these findings should be interpreted as statistically significant but of limited practical significance.
Interpretation Of Findings
Income
In the current study, there was evidence that when families in poor urban households have more disposable income, their adolescent children have a marked increase in HRV (~33ms to ~66ms), although this is subject to complex interpretation when other risk factors are present. Whilst literature on disposable income and health is sparse, there is a growing interest in this metric as the cost of living crisis deepens and material standard of living emerges as a more meaningful
| Figure 3.Full Model – Financial Constraint Predicting HRV, controlling for all other variables |
measure of wealth than socioeconomic status (SES) alone. (Headey 2008; Nagasu et al., 2019) However, there is a gap in literature specifically about disposable income and HRV. The effect of SES on stress pathways has been broadly studied, where lower income populations have notably higher cardiovascular risks and mortality across domains (Schultz et al., 2018). Children from low SES backgrounds are vulnerable to increased cortisol during childhood (Zhu et al., 2019) as well as delayed blood-pressure recovery in adolescence (Evans et al., 2013). The current study begins to connect family experiences of financial stress to HRV in children by demonstrating that there is a meaningful difference in HRV between children from families with more disposable income versus those with less, or in debt, just within the bracket of poverty alone. This tells us that HRV may be mediated by standard of living alone, as compared to the broader metric of SES.
Depression
Most surprisingly of all, there was no association found between D and HRV in the current study, against a backdrop of compelling literature to the contrary in young people (Kumar et al., 2024), in the case of current and past depression (Licht et al., 2008), and impact on HRV has been observed to be similar in those with past depression and those at risk of developing depression (Dell’Acqua et al., 2020). The findings seen in much of the literature about HRV and depression, predominantly conducted in western countries and majority white populations, do not hold up for the current sample. This indicates that there may be important contextual differences may explain this discrepancy. This finding also challenges the prospect that HRV could act as a biomarker for depression and possible early intervention strategy before symptoms present, which may be due to limited reliability of the metric, or it may indicate that HRV as an index of parasympathetic regulation is not universally applicable to different populations, nor as a valid indicator of depression itself. However, the addition D to the interaction between FC, A, and PR reversed the effect of these predictors alone, suggesting that greater disposable income is not necessarily protective for adolescents with compound risk factors. This is further complicated by the lack of meaningful effect of any of the risk factors on D, indicating that the prevalence of depression in the sample is not explained by these vulnerabilities alone. Consequently, addressing these specific issues in vulnerable populations is unlikely to be sufficient for preventing either depression or cardiovascular concerns such as low HRV.
| Figure 4.Full Model 4 Way Interaction – Financial Constraint, Abuse, Parent’s Relationship, & Depression Predicting HRV |
Trauma
In this study, no association was observed between abuse and HRV alone, although in interactions, the protective ability of better financial security was inconsistent where abuse was paired with different types of parental relationship, meaning no conclusive pattern emerged about a relationship between experience of abuse and parasympathetic regulation. There is already evidence to suggest that the impact of childhood adversity is largely dependent on recency of adverse events and type of adversity (Wesarg et al., 2022), and that there is a potential for minor long-term changes in ANS regulation after childhood maltreatment, but this effect is more specific to neglect than other types of maltreatment (Stürmer et al., 2025). A lack of distinction made in the current study between abuse type, recency, and relationship to abuser suggests that these aspects need to be studied with more nuance. In the case of parental relationship, this is especially important as we cannot determine from the available data if participant’s experiences of abuse stem from their parent’s or caregivers, nor do we have enough information to correlate this with the relationship between the child’s primary caregivers. Overall, the relationship between abuse, parental relationship quality, and HRV appears to be highly context-dependent, with no consistent pattern across the sample.
Limitations
The study focuses on poor urban populations in Columbia, Nepal, and South Africa, which distinguishes it from the majority of literature on both HRV and mental health in general, especially in light of contradictory results to the main body of research. However, this study acted as a within-group comparison of people on a spectrum of poverty, without a contextually appropriate comparison group of non-poverty status where mean baseline HRV may be different to that of their low-income peers. Existing literature on populations from majority white, western backgrounds are a poor substitute for this given the existing evidence for differences in HRV based on ethnicity within a western context (Jennings et al., 2015; Hill et al., 2015).
Ethnicity was also not a focus in this study, which aimed to explore poverty and risk factors against a background of specifically non-Western adolescents in poor, urban households. However, Sinichi et al. (2025) found a statistically significant difference in mean HRV between participants from each country within the same sample used here. Together with findings in the current research that extremes of financial burden are impactful on HRV in this population of adolescents alone, a further understanding of which adolescents are most vulnerable to this effect can be, in part, explored through the lens of ethnic background.
The measurement of heart rate variability in this study was performed using equipment of verified quality (Sinichi et al., 2024; Hinde et al., 2021). Despite this, there are limitations to the use of more than one device type across the three populations. The decision to do so was due to cultural challenges in using the Polar H10 chest strap device, where the HeartMath earlobe device was preferable. The combination of ECG and PPG measurement presents differences in quality and validity of data that are then pooled into one analysis. PPG also presents with the additional challenge of using LED photodetectors, which may be less effective where melanin concentration in skin is higher, making it a less reliable technology for ethnically diverse populations. This accounts for one possible explanation for loss of HRV data in this study.
Future Directions
This study provides a snapshot into how adolescents from poor backgrounds with and without risk factors present with depression and how these influences might correlate with HRV. This assumes that depression and stress impact PNS regulation in adolescence similar to what we currently understand for adults, and vice versa, what we know about this phenomena in adults does not take childhood experiences into account. These are young participants, and the broader consequences of childhood trauma and stress may not become apparent in the short term. Connecting the current study participants with their future outcomes could begin to fill in those gaps in understandings and dispel false assumptions that childhood psychophysiology follows the same logic as in adulthood, and also help to make predictions about childhood risk factors and future psychophysiological risks. Possible research questions would include: Do adolescents in poor urban backgrounds who currently have depression but healthy HRV continue to have depression in adulthood, and is it accompanied by a decrease in HRV, and significantly lower compared to non-depressed peers? Do participants who have lower HRV in adolescence in the presence of higher financial strain continue to have persistent low HRV in adulthood compared to those who had more financial freedom during adolescence? And are they more likely to develop problems with depression in adulthood as compared to peers who’s families had more disposable income?
Other future considerations include comparison studies in non-impoverished adolescents in the same countries, to determine if there is a measurable difference between adolescents with trauma-based risk factors within the same culture who are not exposed to the additional risks of poverty. Ethnicity focused studies will also be a required future direction for both mental health and psychophysiology, especially when it comes to the instruments used, whether that be appropriate physiological measurement tools, or mental health and trauma screening tools that are culturally valid and reliable. Exploring HRV specifically in a racial context also requires differentiating between, for example, black participants in predominantly black populated countries versus black participants in white dominated western countries where effects of marginalization can impact health and mental health.
Conclusion
After exploring the relationships between childhood risk factors for mental illness and how this might present as differences in heart rate variability, of the variables included, high financial constraint (<0) emerged as the main predictor of decreased HRV in adolescents in poverty in Colombia, Nepal, and South Africa. All other predictors had negligible impact on HRV, except when moderated by financial constraint, especially when risk factors were combined. Adolescents with current depression may be less responsive to the protective qualities of lower financial constraint (>0). However, these effects contained contradictions, which further highlights the complexity of determining the impact of adverse circumstance on children’s parasympathetic flexibility and mental health, and the need for more nuance in how these risks truly manifest, and relate to each other. These investigations, however, did not explain the majority of variance in HRV in this population, which may be due to true lack of impact of the examined stressors; lack of explanatory power variables which require more nuance, such as types of abuse and caregiver relationship; or absence of contextually appropriate comparison groups in higher wealth brackets with similar adverse circumstances. These findings serve as a foundational study in high risk populations of adolescents who’s mental and physiological health is subject to factors and influences not easily explained or measured by western instruments and norms.
Appendix A – Tables
Appendix B: Figures
| Figure 5.Full Model 3 way Interaction – Financial Constraint, Abuse, & Parent’s Relationship |
| Figure 6.Linear Regression – HRV & Depression |
References
Ang, B., Horowitz, M., & Moncrieff, J. (2022). Is the chemical imbalance an ‘urban legend’? An exploration of the status of the serotonin theory of depression in the scientific literature. SSM – Mental Health, 2, 100098. https://doi.org/10.1016/j.ssmmh.2022.100098
Bringmann, H. C., Bogdanski, M., Seifert, G., & Voss, A. (2022). Impact of meditation-based lifestyle modification on HRV in outpatients with mild to moderate depression: An exploratory study. Frontiers in Psychiatry, 13, 808442. https://doi.org/10.3389/fpsyt.2022.808442
Carvajal-Velez, L., Ahs, J. W., Requejo, J. H., Kieling, C., Lundin, A., Kumar, M., Luitel, N. P., Marlow, M., Skeen, S., Tomlinson, M., & Kohrt, B. A. (2023). Measurement of mental health among adolescents at the population level: A multicountry protocol for adaptation and validation of mental health measures. Journal of Adolescent Health: Supplement, 72(1), S27-S33. https://doi.org/10.1016/j.jadohealth.2021.11.035
Chang, K., & Kuhlman, K. R. (2022). Adolescent-onset depression is associated with altered social functioning into middle adulthood. Scientific Reports, 12(1), 17943. https://doi.org/10.1038/s41598-022-22131-1
de Geus, E. J. C., & Gevonden, M. J. (2024). Acquisition and analysis of ambulatory autonomic nervous system data. In M. R. Mehl, M. Eid, C. Wrzus, G. M. Harari, & U. W. Ebner-Priemer (Eds.), Mobile sensing in psychology: Methods and applications (pp. 129-167). The Guilford Press.
de Gregorio, C., Morabito, G., Cerrito, M., Dattilo, G., & Oreto, G. (2011). Citalopram-induced long QT syndrome and torsade de pointes: Role for concomitant therapy and disease. International Journal of Cardiology, 148(2), 226-228. https://doi.org/10.1016/j.ijcard.2009.05.060
DeCarlo Santiago, C., & Wadsworth, M. E. (2009). Coping with family conflict: What’s helpful and what’s not for low-income adolescents. Journal of Child and Family Studies, 18(2), 192-202. https://doi.org/10.1007/s10826-008-9219-9
Dell’Acqua, C., Dal Bò, E., Messerotti Benvenuti, S., & Palomba, D. (2020). Reduced heart rate variability is associated with vulnerability to depression. Journal of Affective Disorders Reports, 1, 100006. https://doi.org/10.1016/j.jadr.2020.100006
Doom, J. R., Cook, S. H., Sturza, J., Koinis-Mitchell, D., Yourczek, D., & Miller, A. L. (2018). Family conflict, chaos, and negative life events predict cortisol activity in low-income children. Developmental Psychobiology, 60(3), 364-379. https://doi.org/10.1002/dev.21602
Evans, G. W., Exner-Cortens, D., Kim, P., & Bartholomew, D. (2013). Childhood poverty and blood pressure reactivity to and recovery from an acute stressor in late adolescence: The mediating role of family conflict. Psychosomatic Medicine, 75(7), 691-700. https://doi.org/10.1097/PSY.0b013e31829f9823
Hartmann, R., Schmidt, F. M., Sander, C., & Hegerl, U. (2018). Heart rate variability as indicator of clinical state in depression. Frontiers in Psychiatry, 9, 735. https://doi.org/10.3389/fpsyt.2018.00735
Headey, B. (2008). Poverty is low consumption and low wealth, not just low income. Social Indicators Research, 89(1), 23-39.
Hill, L. K., Hu, D. D., Koenig, J., Sollers, J. J., Kapuku, G., Wang, X., Snieder, H., & Thayer, J. F. (2015). Ethnic Differences in Resting Heart Rate Variability: A Systematic Review and Meta-Analysis. Psychosomatic Medicine, 77(1), 16-25. https://doi.org/10.1097/PSY.0000000000000133
Hinde, K., White, G., & Armstrong, N. (2021). Wearable Devices Suitable for Monitoring Twenty Four Hour Heart Rate Variability in Military Populations. Sensors (Basel, Switzerland), 21(4), 1061. https://doi.org/10.3390/s21041061
Jennings, J. R., Allen, B., Gianaros, P. J., Thayer, J. F., & Manuck, S. B. (2015). Focusing neurovisceral integration: Cognition, heart rate variability, and cerebral blood flow. Psychophysiology, 52(2), 214-224. https://doi.org/10.1111/psyp.12319
Kemp, A. H., Quintana, D. S., Gray, M. A., Felmingham, K. L., Brown, K., & Gatt, J. M. (2010). Impact of depression and antidepressant treatment on heart rate variability: A review and meta-analysis. Biological Psychiatry, 67(11), 1067-1074. https://doi.org/10.1016/j.biopsych.2009.12.012
Kieling, C., Buchweitz, C., Caye, A., Manfro, P., Pereira, R., Viduani, A., Anés, M., Battel, L., Benetti, S., Fisher, H. L., Karmacharya, R., Kohrt, B. A., Martini, T., Petresco, S., Piccin, J., Rocha, T., Rohde, L. A., Rohrsetzer, F., Souza, L., Velazquez, B., … Mondelli, V. (2021). The Identifying Depression Early in Adolescence Risk Stratified Cohort (IDEA-RiSCo): Rationale, methods, and baseline characteristics. Frontiers in Psychiatry, 12, 697144. https://doi.org/10.3389/fpsyt.2021.697144
Kirsch, I. (2020). Challenging received wisdom: Antidepressants and the placebo effect. McGill Journal of Medicine, 11(2). https://doi.org/10.26443/mjm.v11i2.571
Kirsch, I., Deacon, B. J., Huedo-Medina, T. B., Scoboria, A., Moore, T. J., Johnson, B. T., & Hay, P. (2008). Initial Severity and Antidepressant Benefits: A Meta-Analysis of Data Submitted to the Food and Drug Administration. PLoS Medicine, 5(2). https://doi.org/10.1371/journal.pmed.0050045
Lewinsohn, P. M., Clarke, G. N., Seeley, J. R., & Rohde, P. (1994). Major depression in community adolescents: Age at onset, episode duration, and time to recurrence. Journal of the American Academy of Child and Adolescent Psychiatry, 33(6), 809-818. https://doi.org/10.1097/00004583-199407000-00006
Licht, C. M. M., de Geus, E. J. C., van Dyck, R., & Penninx, B. W. J. H. (2010). Longitudinal evidence for unfavorable effects of antidepressants on heart rate variability. Biological Psychiatry, 68(9), 861-868. https://doi.org/10.1016/j.biopsych.2010.06.032
Licht, C. M. M., de Geus, E. J. C., Zitman, F. G., Hoogendijk, W. J. G., van Dyck, R., & Penninx, B. W. J. H. (2008). Association between major depressive disorder and heart rate variability in the Netherlands Study of Depression and Anxiety (NESDA). Archives of General Psychiatry, 65(12), 1358-1367. https://doi.org/10.1001/archpsyc.65.12.1358
Licht, C. M. M., de Geus, E. J. C., Zitman, F. G., Hoogendijk, W. J. G., van Dyck, R., & Penninx, B. W. J. H. (2008). Association between major depressive disorder and heart rate variability in the Netherlands Study of Depression and Anxiety (NESDA). Archives of General Psychiatry, 65(12), 1358-1367. https://doi.org/10.1001/archpsyc.65.12.1358
Lund, C., Jordans, M. J. D., Garman, E., Araya, R., Avendano, M., Bauer, A., Bahure, V., Dua, T., Eleftheriou, G., Evans-Lacko, S., García Rodríguez, J. F., Gautam, K., Gevonden, M., Hessel, P., Kohrt, B. A., Krabbendam, L., Luitel, N. P., Roy, S., Seifert Bonifaz, M., et al. (2023). Strengthening self-regulation and reducing poverty to prevent adolescent depression and anxiety: Rationale, approach and methods of the ALIVE interdisciplinary research collaboration in Colombia, Nepal and South Africa. Epidemiology and Psychiatric Sciences, 32, E34. https://doi.org/10.1017/S2045796023000811
Nagasu, M., Kogi, K., & Yamamoto, I. (2019). Association of socioeconomic and lifestyle-related risk factors with mental health conditions: a cross-sectional study. BMC Public Health, 19(1). https://doi.org/10.1186/s12889-019-8022-4
Pizzoli, S. F. M., Marzorati, C., Gatti, D., Monzani, D., Mazzocco, K., & Pravettoni, G. (2021). A meta-analysis on heart rate variability biofeedback and depressive symptoms. Scientific Reports, 11
Schultz, W. M., Kelli, H. M., Lisko, J. C., Varghese, T., Shen, J., Sandesara, P., Quyyumi, A. A., Taylor, H. A., Gulati, M., Harold, J. G., Mieres, J. H., Ferdinand, K. C., Mensah, G. A., & Sperling, L. S. (2018). Socioeconomic status and cardiovascular outcomes: Challenges and interventions. Circulation, 137(20), 2166-2178. https://doi.org/10.1161/CIRCULATIONAHA.117.029652
Sinichi, A., Eleftheriou, G., Ngoc, M. A. H., Kohrt, B. A., Luitel, N., Singh, R., Jacobs, R., Sorsdahl, K., Garcia Jaramillo, S., Dedios Sanguineti, M. C., Lund, C., Jordans, M. J. D., Garman, E., Krabbendam, L., Gevonden, M., … Yarrow, P. (2025, May 30). Rethinking heart rate variability: No evidence of association with self-regulation and psychopathology in a cross-sectional study among adolescents in Colombia, Nepal, and South Africa (OSF preprint). OSF Preprints. https://doi.org/10.31219/osf.io/79c5m_v1
Sinichi, M., Gevonden, M. J., & Krabbendam, L. (2025). Quality in question: Assessing the accuracy of four heart rate wearables and the implications for psychophysiological research. Psychophysiology, 62, e70004. https://doi.org/10.1111/psyp.70004
Straub, J., Brown, R., Malejko, K., Bonenberger, M., Grön, G., Plener, P. L., & Abler, B. (2019). Adolescent depression and brain development: Evidence from voxel-based morphometry. Journal of Psychiatry & Neuroscience, 44(4), 237-245. https://doi.org/10.1503/jpn.170233
Stürmer, M., Klinger-König, J., Vollmer, M., Weihs, A., Frenzel, S., Dörr, M., Kaderali, L., Felix, S. B., Stubbe, B., Ewert, R., Völzke, H., Grabe, H. J., & Krause, E. (2025). Long-term alteration of heart rate variability following childhood maltreatment: Results of a general population study. European Psychiatry, 68(1), e86. https://doi.org/10.1192/j.eurpsy.2025.10040
van Bronswijk, S., Moopen, N., Beijers, L., Ruhe, H. G., & Peeters, F. (2019). Effectiveness of psychotherapy for treatment-resistant depression: A meta-analysis and meta-regression. Psychological Medicine, 49(3), 366-379. https://doi.org/10.1017/S003329171800199X
Wesarg, C., Van den Akker, A. L., Oei, N. Y. L., Wiers, R. W., Staaks, J., Thayer, J. F., Williams, D. P., & Hoeve, M. (2022). Childhood adversity and vagal regulation: A systematic review and meta-analysis. Neuroscience and Biobehavioral Reviews, 143, 104920. https://doi.org/10.1016/j.neubiorev.2022.104920
Zhang, Y., Jia, X., Yang, Y., Sun, N., Shi, S., & Wang, W. (2024). Change in the global burden of depression from 1990-2019 and its prediction for 2030. Journal of Psychiatric Research, 178, 16-22. https://doi.org/10.1016/j.jpsychires.2024.07.054
Zhu, Y., Chen, X., Zhao, H., Chen, M., Tian, Y., Liu, C., Han, Z. R., Lin, X., Qiu, J., Xue, G., Shu, H., & Qin, S. (2019). Socioeconomic status disparities affect children’s anxiety and stress-sensitive cortisol awakening response through parental anxiety. Psychoneuroendocrinology, 103, 96-103. https://doi.org/10.1016/j.psyneuen.2019.01.008