Cumulative effects of pre-pandemic vulnerabilities and pandemic-related hardship on psychological distress among pregnant individuals

Abstract

Objective

Our primary objective was to determine whether pre-existing vulnerabilities and resilience factors combined with objective hardship resulted in cumulative (i.e., additive) effects on psychological distress in pregnant individuals during the COVID-19 pandemic. A secondary objective was to determine whether any of the effects of pandemic-related hardship were compounded (i.e., multiplicative) by pre-existing vulnerabilities.

Method

Data are from a prospective pregnancy cohort study, the Pregnancy During the COVID-19 Pandemic study (PdP). This cross-sectional report is based upon the initial survey collected at recruitment between April 5, 2020 and April 30, 2021. Logistic regressions were used to evaluate our objectives.

Results

Pandemic-related hardship substantially increased the odds of scoring above the clinical cut-off on measures of anxiety and depression symptoms. Pre-existing vulnerabilities had cumulative (i.e., additive) effects on the odds of scoring above the clinical cut-off on measures of anxiety and depression symptoms. There was no evidence of compounding (i.e., multiplicative) effects. Social support had a protective effect on anxiety and depression symptoms, but government financial aid did not.

Conclusion

Pre-pandemic vulnerability and pandemic-related hardship had cumulative effects on psychological distress during the COVID-19 pandemic. Adequate and equitable responses to pandemics and disasters may require more intensive supports for those with multiple vulnerabilities.

1. Background

The COVID-19 pandemic created a universal stressor affecting every segment of the population. Government-mandated measures to counter spread of the virus included recommendations for hand and respiratory hygiene, travel restrictions, self-isolation, wearing masks in public, and physical distancing. These measures dramatically changed life for most people, including the ways people worked, socialized, ate, and played. Nevertheless, the nature and severity of pandemic-related hardships were not the same for everyone [1]. For example, lower socioeconomic standing, minority ethnicity, and having a chronic health condition increased risk for infection and severe disease [2], with downstream implications for psychological functioning, including increased fear of COVID-19 [3]. The pandemic exposed, and potentially exacerbated, pre-existing and intersecting vulnerabilities within the population, with detrimental long-term implications for vulnerable segments of the population [4].

Of special interest are pregnant individuals because of the added burden of pandemic-related changes to prenatal care, labour, and delivery. Pregnant individuals have been particularly vulnerable to the negative psychological effects of COVD-19 public health measures [5], both because they represent one of the groups most severely affected by pandemic countermeasures and because pregnancy itself entails significant life changes that require major psychosocial and emotional adjustments [6]. Worldwide, this has resulted in a sharp increase in self-reported anxiety and depression symptoms [7]. This is concerning because prenatal exposure to psychological distress has deleterious effects on fetal and child development [[8], [9], [10], [11], [12], [13], [14], [15]]. The potential for both short-term (e.g., psychological distress) and long-term (e.g., impaired child development) harm resulting from multiple and intersecting vulnerabilities in pregnant individuals calls for a deeper understanding of ways in which the pandemic has affected them.

Previous studies have consistently observed increases in mental health problems following disasters [[17], [18], [19]], highlighting several factors contributing to poor psychological adjustment. The duration and intensity of objective hardship during a disaster strongly predict greater psychological distress across all segments of the population [20]. Pre-existing vulnerabilities (e.g., existing mental illness, low socioeconomic status, minority ethnicity, low social support, younger age, and caring for children) amplify the effects of objective hardship on psychological adjustment [21]. Several early reports suggest that the effects of the COVID-19 pandemic mirror those of previous disasters. For example, younger people, women, and those with the greatest socioeconomic disadvantage have experienced the largest economic and psychological impacts [[23], [24], [25], [26]].

With few exceptions (see for example [22]), prior studies have focused on the cumulative burden of exposures and vulnerabilities and have not formally tested models of compounding effects. One of the ongoing questions about pre-existing vulnerability is whether its effects are cumulative (additive) versus compounding (multiplicative) in the context of external stressors (i.e., objective hardships) [16], such as the COVID-19 pandemic. Disambiguating cumulative and compounding effects has important implications for understanding psychological impairment and resilience.

Access to key resources is an important determinant of disaster resilience and is expected to also contribute to psychological adjustment during the pandemic. Of primary significance to pregnancy, social support is usually a protective factor in the face of large-scale stressful events [[27], [28], [29], [30]]. But the potential for this kind of socially supported resilience was severely curtailed by pandemic counter measures, which increased the sense of social isolation [31]. Physical distancing policies and restrictions on support persons attending prenatal care appointments or delivery are especially concerning because social support buffers the negative effects of prenatal distress on both the pregnant parent and their child [32,33]. Another key resilience factor is access to basic necessities in the face of economic losses. During the COVID-19 pandemic, the federal government in Canada (and many other governments), provided targeted financial aid to those whose income was impacted by the pandemic. Although its intended purpose was to mitigate financial losses resulting from the pandemic, one might reasonably expect that financial aid could also buffer the effects of hardships on psychological distress.

1.1. Current study

Our primary objective was to determine the nature of the associations between objective hardship and psychological distress among pregnant individuals during the COVID-19 pandemic. Specifically, we set out to determine whether the combination of pre-existing vulnerabilities and pandemic-related hardship result in cumulative (i.e., additive) effects on psychological distress and whether resilience resources (i.e., social support and financial aid) would be protective against psychological distress. A secondary objective was to determine whether pre-existing vulnerabilities and objective hardship have compounding (i.e., multiplicative) effects on psychological distress. These objectives were pre-registered in our study protocol [34]. This cross-sectional report is based upon the initial survey data collected in the Pregnancy during the Pandemic (PdP) study between April 5, 2020 and April 30, 2021.

2. Method

2.1. Participants

Participants were eligible for the PdP study if they were pregnant, ≤35 weeks gestation at study enrollment, ≥17 years old, living in Canada, and able to read and write in English or French [34]. Participants were recruited via ads on Facebook and Instagram in both French and English. Study enrollment, consent, and administration of questionnaires were conducted through Research Electronic Data Capture (REDCap) [35]. Participants did not receive compensation for participating in the intake survey, which took a median time of 25 min to complete. This study received ethics approval (REB20–0500) from the University of Calgary Conjoint Health Research Ethics Board on March 26, 2020. All participants signed the electronic consent form prior to providing any data.

2.2. Measures

2.2.1. 2.2.1 Pandemic objective hardship index (POHI)

The COVID-19 pandemic represents a novel set of circumstances for which there were no previously developed measures. Our measure of pandemic-related hardship adapted previous measures of objective hardship designed for natural disasters. A principled and systematic framework developed by Suzanne King and colleagues identified four major domains of objective hardship: scope (duration/intensity), loss (financial, social or physical), threat (physical and health consequences), and change (adjustments to daily living) [36]. In the context of the COVID-19 pandemic, Scope reflects duration (e.g., duration of quarantine or daycare closures) and intensity (e.g., regional number of infections in the past 2 weeks). Loss reflects, for example, loss of employment or savings, or closure of schools. Threat refers to physical and health-related consequences (e.g., COVID infection or hospitalization of family member with COVID-19). Change captures the adjustments to daily living, prenatal care, work, and social interaction caused by the COVID-19 pandemic (e.g., working from home, altering a birth plan). A Total score is derived by combining scores from the four domains. To account for raw mean differences between domains, and because we had no prior theoretical rationale for differential weighting of the POHI domain scores, we z-score transformed each domain score prior to taking the mean. The POHI Total score is therefore equally weighted by the four domains. To facilitate intuitive comparisons between the domain scores and the Total score, all analyses of the individual domains and POHI Total scores use z-scores. An expanded description of the POHI measure along with detailed descriptive information about each item can be found in Supplemental Table 1. Here we considered the entire duration of the pandemic up to the point at which the participant answered the questions (which varied from 1 to 12 months after pandemic onset).

2.2.2. Pre-pandemic characteristics

Participants self-reported their household income, education, race/ethnicity, family composition (i.e., marital status and number of previous children), housing instability, food insecurity, income precarity, and other health-related information such as age, and history of anxiety or depression problems prior to current pregnancy (yes/no). Housing instability was defined as having difficulty finding stable housing or moving more than three times in the past two years. Food insecurity was defined as anyone in the household receiving food from a food bank, soup kitchen, or other charitable agency in 2019, or, if prior to the pandemic, they had ‘often’ or ‘sometimes’ experienced a time that food didn't last, and they didn't have money to buy more. Income precarity was defined as being unable to live at their current address and standard of living for more than one month if they lost all their sources of income.

2.2.3. Psychological distress

Depression symptoms were assessed using the Edinburgh Postpartum Depression Scale [37]. A cut-off score of ≥13 was used to identify participants with clinically concerning depression symptoms [37]. General anxiety symptoms were assessed using the Patient-Reported Outcomes Measurement Information System (PROMIS®) Anxiety Adult 7-item short form [38]. Raw scores were converted to T-scores using US general population norms. T-scores ≥60 indicate elevated anxiety symptoms [38,39].

2.2.4. Resilience resources

Perceived support received from a romantic partner was assessed using the Social Support Effectiveness Questionnaire (SSEQ) [40]. This 35-item measure includes items assessing informational, emotional, instrumental, and negative (e.g., support that infringes on self-efficacy or self-esteem) aspects of support. Perceived interpersonal social support was assessed using the 12-item Interpersonal Support Evaluation List (ISEL) [41] to determine appraisal (e.g., advice or problem solving), belongingness (e.g., shared experiences), and tangible support (e.g., help with daily chores). Higher scores indicate greater perceived support.

Participants self-reported whether they had received financial aid from the federal government of Canada. Canadians were eligible to receive $2000/month if they stopped working because of the COVID-19 pandemic and did not earn more than $1000/month within the period of the claim. Duration of financial aid was not recorded. Data were binary (yes/no).

2.3. Data analysis

Prior to data analysis, all records were screened for plausibility and completeness. Records with email addresses that could not be verified (e.g., bounce-back), those with nonsense responses, and those that did not provide data were removed from the database. Data analyses were conducted in IBM SPSS Statistics 27.0 software.

Descriptive statistics and associations with pandemic-related hardship are presented as means, correlations, and estimated marginal means derived from ANCOVAs. In preliminary models we observed that time since the onset of the pandemic was positively associated with Threat, Scope, and POHI Total scores and negatively associated with Loss and Change (Supplemental Fig. 1). All statistical models were therefore adjusted for time since onset of the pandemic to account for changes in hardship over time. Objectives were tested using binary logistic regressions in which the outcome was above (1) or below (0) the clinical cut-off score for the depression or anxiety symptom measures. Secondary analyses to test for multiplicative (interaction) effects were conducted using logistic regressions with the PROCESS macro 4.1 for SPSS [42]. To account for the large number of predictors assessed in our analyses, we set statistical significance at p < 0.001.

3. Results

3.1. Recruitment

Paid study ads had 2,385,344 potential views (‘impressions’) during the recruitment period of April 5, 2020 to April 30, 2021. Ads directed potential participants to the study website, where they could click the link to participate. The study website had 121,989 visits during this period (not all visits would have been potential participants, and some likely visited multiple times). The link to complete the study screening questions was used 28,710 times; 2611 participants screened as ineligible. Of the 26,099 who passed screening, 14,639 did not sign the consent form or provided no data, and 83 declined (said ‘no’) to provide consent. Duplicate entries (n = 265), those who withdrew (n = 420), or provided fake data (n = 6; i.e., multiple nonsense or illogical responses) were removed from the dataset.

3.2. Missing data

The total sample size for this analysis was 10,669, however data on objective hardship were missing from 762 participants (7.1%). Although the analytic sample differed statistically from the sample with missing data on sociodemographic characteristics (Table 1 ), the differences were small and not clinically relevant. Therefore, we used a listwise deletion approach. Previous simulation studies suggest that parameter estimates are unbiased using listwise deletion when <10% of the cases have missing data [43].

Table 1.

Sample characteristics and comparison to those not included in analyses.

	Analytic Sample (n = 9907)		Missing Data Sample (n = 762)		Group Difference
	M (SD)	Range	M (SD)	Range	F	p-value
Participant Age (years)	31.9 (4.4)	17.0–49.7	31.3 (4.8)	18.1–45.3	11.34	<0.001
	n	%	n	%	Chi-Square	p-value
Trimester at study enrollment					5.23	0.07
First	2328	23.5	183	24.4
Second	4578	46.3	371	49.4
Third	2981	30.2	197	26.2
Marital Status					16.71	<0.001
Single/Separated/Divorced/Widowed	434	4.4	44	6.6*
Married	6113	62.0	364	54.7*
Cohabitating	3315	33.6	257	38.6*
Education					80.92	<0.001
Graduate Degree (MSc, MD, etc)	2555	25.9	108	16.2*
Bachelor's degree	3855	39.1	231	34.6*
Completed Trade/Technical Degree	2523	25.6	219	32.8*
Completed High School Diploma	810	8.2	85	12.7*
Less Than High School Diploma	118	1.2	24	3.6*
Ethnicity					48.49	<0.001
White (Caucasian)	8286	84.6	505	82.7
First Nations/Inuit	128	1.3	18	2.9*
Metis	141	1.4	11	1.8
Black	121	1.2	10	1.6
West Asian	100	1.0	10	1.6
South Asian	265	2.7	12	2.0
Southeast Asian	149	1.5	12	2.0
East Asian	170	1.7	3	0.5*
Hispanic/Latinx	192	2.0	20	3.3*
Biracial or other	245	2.5	10	1.6
Missing
Annual Household Income					79.27	<0.001
$200,000+/Year	951	9.7	45	7.2*
$175,000 – $199,999/Year	622	6.3	28	4.5
$150,000 – $174,999/Year	1084	11.0	47	7.5*
$125,000 – $149,999/Year	1288	13.1	71	11.4
$100,000 – $124,999/Year	1849	18.8	99	15.8
$70,000 – $99,999/Year	1956	19.9	128	20.5
$40,000 – $69,999/Year	1309	13.3	108	17.3*
$20,000 – $39,999/Year	553	5.6	59	9.4*
Less Than $20,000/Year	223	2.3	40	6.4*
Number of Children					32.33	<0.001
0	5654	57.2	396	52.2*
1	2949	29.8	245	32.3
2	962	9.7	68	9.0
3	237	2.4	29	3.8*
4	87	0.9	20	2.6*
Food Insecurity	861	8.8	100	16.1*	37.95	<0.001
Housing Insecurity	729	7.4	72	11.7*	14.73	<0.001
Income Precarity	1025	10.4	101	16.9*	24.80	<0.001

* asterisk indicates row proportions that differ at p < 0.05.

3.3. Sample characteristics

The sample comprises a diverse range of sociodemographic characteristics, with a range of gestational ages and representation from every Canadian province and territory (Table 1). Most participants self-identified as White (84.6%), educated (65% held a university degree), and economically stable (∼ 10% had income precarity).

3.4. Descriptive findings for the pandemic objective hardship index (POHI)

Descriptive data for the POHI measure are reported in Supplemental Table 2. Bivariate correlations among the four individual POHI domains were positive but small, (Pearson r = 0.01–0.15) except for a moderate association between Loss and Change (Pearson r = 0.40). In contrast, associations between individual domain scores and the POHI total score were moderate to strong (Pearson r = 0.53–0.70). These findings suggest this index measures dissociable aspects of pandemic-related hardship that can be combined into a coherent whole.

3.5. Associations between objective hardship and psychological distress

All POHI domains were associated with anxiety or depression symptoms, except Scope (Table 2 ). The odds ratio for scoring above the clinical cut-off on the anxiety and depression symptom measures were 1.96 (95% CI = 1.83, 2.10) and 2.29 (95% CI = 2.13, 2.47), respectively, for each standard deviation increase in POHI Total score. Comparing individuals with the highest (99th percentile) vs lowest (1st percentile) POHI Total score, there was a 10-fold increase in the odds of scoring above the cut-off on anxiety and an 11-fold increase in the odds of scoring above the cut-off on depression.

Table 2.

Adjusted logistic regression results for pandemic objective hardship predicting scores above the clinical cut off for depression and anxiety symptoms.

Predictor	Coefficient	aOR	p-value	95%C.I. for aOR
Depression symptoms
POHI Total	0.83	2.29	<0.001	2.13–2.47
Change	0.44	1.55	<0.001	1.49–1.63
Loss	0.45	1.57	<0.001	1.51–1.64
Threat	0.25	1.28	<0.001	1.23–1.34
Scope	0.07	1.08	0.002	1.03–1.13
Anxiety symptoms
POHI Total	0.67	1.96	<0.001	1.83–2.10
Change	0.40	1.49	<0.001	1.43–1.56
Loss	0.34	1.41	<0.001	1.35–1.47
Threat	0.22	1.24	<0.001	1.19–1.29
Scope	0.04	1.04	0.08	1.02–1.12

aOR – adjusted odds ratio (model is adjusted for time since onset of the pandemic); C.I. – confidence interval; ns – non-significant. Depression symptom scores are from the Edinburgh Postnatal Depression Scale and are dichotomized at the clinical cut score of ≥13. Anxiety symptom scores are from the PROMIS anxiety and are dichotomized at the clinical cut score of T ≥ 60. All predictors were modeled separately. P < 0.001 is considered statistically significant.

3.6. Relations between objective hardship and pre-pandemic characteristics

Associations between pre-pandemic vulnerabilities and objective hardship are graphically displayed in Supplemental Fig. 2, and results of ANCOVAs testing associations between pre-pandemic vulnerabilities and objective hardship are reported in Supplemental Table 3.

3.6.1. Socioeconomic characteristics

Lower income, lower education, housing insecurity, food insecurity, income precarity, single marital status were associated with greater Threat, Loss, and Total scores but lower Change scores (except for income precarity and food insecurity were not associated with Change). Older age (>26 years) was associated with higher Threat and Loss but lower Change and Scope and was not associated with the Total.

3.6.2. Race/ethnicity

Ethnicity was associated with all components of POHI, except Change. East Asians reported the lowest POHI scores, whereas all other ethnic groups had greater hardship compared to White participants.

3.6.3. Previous mental health problems

A history of anxiety or depression prior to the pandemic was positively associated with all POHI measures, however the effect for Scope was not statistically significant.

3.6.4. Parity

Participants who had other children reported higher Loss, Change, Scope, and Total but not Threat.

3.6.5. Location

Geographic location was associated with differences in all components of POHI. Those living in Quebec and Ontario had the highest overall hardship scores whereas those in northern and eastern Canada had the lowest scores.

To limit the number of models tested, all further analyses focussed on the POHI Total.

3.7. Additive effects of pre-existing vulnerabilities and hardship in relation to psychological distress

Scoring above the cut-off on either anxiety or depression symptom measures was associated with lower income, less education, income precarity, food insecurity, younger participant age, being separated/divorced/single, or having a prior history of mental health problems (Table 3 ). Additionally, minority ethnicity was associated with scoring above the clinical cut-off for depression and having previous children was associated with scoring above the clinical cut-off for anxiety. Note that POHI Total was a significant contributor to each model, and the effects for each pre-existing vulnerability reported in Table 3 are in addition to the effects of objective hardship.

Table 3.

Odds ratios for additive effects of pandemic hardship and pre-existing vulnerabilities and resilience factors on depression and anxiety symptoms.

Predictor	Depression Symptoms		Anxiety Symptoms
	aOR	p-value	aOR	p-value
Education	0.75	<0.001	0.82	<0.001
POHI Total	2.20	<0.001	1.89	<0.001
Income	0.87	<0.001	0.90	<0.001
POHI Total	2.15	<0.001	1.86	<0.001
Ethnicity	1.29	<0.001	1.15	0.01
POHI Total	2.26	<0.001	1.94	<0.001
Number of children	0.94	0.02	0.90	<0.001
POHI Total	2.34	<0.001	2.03	<0.001
Marital status	2.05	<0.001	1.75	<0.001
POHI Total	2.28	<0.001	1.95	<0.001
Food insecurity	2.93	<0.001	2.48	<0.001
POHI Total	2.17	<0.001	1.87	<0.001
Housing instability	2.41	<0.001	2.42	<0.001
POHI Total	2.20	<0.001	1.90	<0.001
Income precarity	1.84	<0.001	1.87	<0.001
POHI Total	2.25	<0.001	1.92	<0.001
Participant age	0.95	<0.001	0.96	<0.001
POHI Total	2.36	<0.001	2.00	<0.001
Mental health history	3.06	<0.001	3.53	<0.001
POHI Total	2.22	<0.001	1.89	<0.001
Partner support	0.96	<0.001	0.97	<0.001
POHI Total	2.09	<0.001	1.75	<0.001
Interpersonal support	0.92	<0.001	0.93	<0.001
POHI Total	2.16	<0.001	1.81	<0.001
Financial aid	0.95	0.32	0.88	0.01
POHI Total	2.32	<0.001	2.01	<0.001

aOR – odds ratios are adjusted for time since onset of the pandemic. POHI Total – Pandemic Objective Hardship Index Total score. Depression symptom scores are from the Edinburgh Postnatal Depression Scale and are dichotomized at the clinical cut score of ≥13. Anxiety symptom scores are from the PROMIS anxiety and are dichotomized at the clinical cut score of T ≥ 60. P < 0.001 is considered statistically significant.

Given that all pre-existing vulnerabilities contributed individually to the burden of psychological distress, even after accounting for the effects of pandemic-related hardship, we conducted multivariable logistic regression including all pre-existing vulnerabilities simultaneously (Table 4 ). These analyses were intended to further illuminate the cumulative burden of pre-existing vulnerabilities by estimating the unique contribution of each vulnerability, above and beyond other vulnerabilities. For depression symptoms, there were independent additive effects of objective hardship, education, ethnicity, food insecurity, housing instability, and mental health history. For anxiety symptoms, there were additive effects for objective hardship, food insecurity, housing instability, and mental health history.

Table 4.

Multivariable logistic regression results for objective hardship and pre-existing vulnerability predicting scores above the clinical cut-off for depression and anxiety symptoms.

Predictor	Depression Symptoms		Anxiety Symptoms
	aOR	p-value	aOR	p-value
POHI Total	2.09	<0.001	1.83	<0.001
Education	0.89	<0.001	0.98	0.35
Income	0.97	0.02	0.98	0.20
Ethnicity	1.33	<0.001	1.22	0.001
Number of children	0.95	0.08	0.93	0.01
Marital status	1.17	0.17	1.09	0.46
Food insecurity	1.63	<0.001	1.44	<0.001
Housing instability	1.59	<0.001	1.73	<0.001
Income precarity	1.14	0.11	1.26	0.004
Participant age	0.99	0.05	0.98	0.005
Mental health history	2.76	<0.001	3.30	<0.001

aOR – odds ratios are adjusted for all other variables in Table 4, plus time since onset of the pandemic. POHI Total – Pandemic Objective Hardship Index Total score. Depression symptom scores are from the Edinburgh Postnatal Depression Scale and are dichotomized at the clinical cut score of ≥13. Anxiety symptom scores are from the PROMIS anxiety and are dichotomized at the clinical cut score of T ≥ 60.

3.8. Protective effects of resilience factors

Protective effects were observed for both partner and interpersonal social support (Table 3). Independent of POHI Total, partner and interpersonal support were associated with decreased odds of scoring above the clinical cut-off on measures of depression and anxiety. In contrast, government aid did not contribute to reduction in the odds of scoring above the clinical cut-off for either anxiety or depression.

3.9. Multiplicative effects of pre-existing vulnerabilities and hardship in relation to psychological distress

Interaction terms between each of the pre-existing vulnerabilities and objective hardship were calculated to determine if the effects on psychological distress were multiplicative. None of the interactions between POHI Total and pre-existing vulnerabilities reached statistical significance (Supplemental Table 4). Similarly, none of the interaction terms between POHI Total and resilience factors was statistically significant.

4. Discussion

Here, we adapted an objective hardship measure from previous disaster studies and implemented it in a prospective pregnancy cohort study during the first year of the COVID-19 pandemic. Our data provide substantial evidence suggesting that hardship experiences during the pandemic increased the odds of scoring above the clinical cut-off on standardized measures of anxiety and depression symptoms. Additionally, there was clear evidence that pre-existing vulnerabilities additively increased the odds that experiences of pandemic-related hardship resulted in clinically elevated depression and anxiety scores. Social support reduced the odds that objective hardship would result in elevated psychological distress, but government financial aid did not. The findings suggest that pandemic-related hardship increased the risk for psychological distress among all segments of the pregnant population. Individuals with greater pre-pandemic vulnerability experienced greater overall risk for psychological distress because of the cumulative (i.e. additive) effect of pre-pandemic vulnerability and pandemic-related hardship.

Pandemic hardship was related to all sociodemographic variables, except participant age. This strongly supports previous research showing inequities in pandemic-related hardship [[23], [24], [25], [26]]. Pandemic hardships may have widened socioeconomic disparities in pregnancy, with the greatest burden of hardship born by those with the greatest vulnerabilities. This is not unique to the COVID-19 pandemic, as policies that contributed to inequities prior to 2020 also contributed to inequalities in pandemic hardships [44]. Accessible and equitable prenatal care and mental health support as well as increased efforts to address substandard housing, food insecurity, and poverty [45,46] are essential to reducing disproportionate hardships among vulnerable groups and mitigating the long-term impacts of the pandemic on parents and children.

Pregnant individuals with the lowest income and education experienced more pandemic-related losses and threat compared to those with higher income or education. This likely reflects both the greater physical risks that come from working or living in close physical proximity to others, and the higher rate of job loss among lower income and lower education individuals [47]. In contrast to the losses and threats born by those at the lower end of the socioeconomic distribution, pregnant individuals with higher education and income faced greater change-related stressors, some of which may be related to increases in work hours or working from home which is seldom an option for lower-paying jobs [48]. This is also consistent with previous studies suggesting that individuals in higher income groups experienced more changes because of their more extensive social and leisure activities [26]. Furthermore, individuals with higher education and income tend to have more established social circumstances with, for example, children whose schooling or daycare was disrupted, leading to increased Change scores. These findings highlight important ways that individuals in different socioeconomic strata experienced the pandemic differently, with potential implications for targeted interventions.

In our study, Threat, Loss, and Change were more related to prenatal psychological distress than Scope. These three domains reflect individual-level exposures whereas scope reflects more region-level exposures (e.g., infection rates, restrictions), suggesting that individual circumstances are more important than regional exposures for psychological distress. This highlights the importance of mental health prevention and intervention measures that target individual-level realities rather than focusing on regional exposures.

Our analysis of multiplicative models failed to find evidence of moderator effects for pre-existing vulnerabilities, despite adequate sample size [22]. Our findings are consistent with one of the few previous studies to explicitly test models of compounding vulnerabilities, in which the interaction between previous mental health problems and objective hardship exposure among survivors of the September 11th, 2001 terrorist attacks in New York City was not a significant predictor of psychological distress [22]. In contrast, a recent report of pregnant individuals found a statistical interaction between psychological factors like tolerance of uncertainty and self-reported resilience with objective hardship during the COVID-19 pandemic in relation to postpartum anxiety [49]. Taken in the context of the well-established relationship between distress prior to and subsequent to a disaster [21], our findings suggest that pre-existing vulnerabilities cumulatively increase the burden of mental distress, but they don't fundamentally alter the nature of the association between hardship and distress. Experiences of hardship are associated with increased psychological distress across all levels of socioeconomic status, but those occupying lower strata are hardest hit because socioeconomic status and hardship cumulatively affect psychological outcomes.

Our findings agree with other COVID-19 pregnancy studies observing reductions in anxiety and depression as a function of social support [50,51], confirming social support as a resilience factor. However, the lack of evidence for a multiplicative effect of social support contrasts with previous disaster studies showing that social support is an effect modifier for subjective distress and objective hardship on mental health during pregnancy [52,53]. The contrasting findings may reflect the unique circumstances of the pandemic that discouraged and limited social contact. Disasters tend to increase social participation through activities such as volunteering, helping neighbors, and community problem solving [30], but opportunities for this kind of social cohesion and connection were severely curtailed by pandemic restrictions, limiting the social support available to pregnant individuals.

Financial hardship is robustly associated with anxiety and depression [54,55], including in our study. However, there is very little evidence that direct financial support has a protective effect on psychological distress [56], and our data mirror these findings. Nevertheless, ongoing work by the Baby's First Years study suggests that although poverty reduction does not appear to directly affect psychological distress [57], it has positive effects on children's brain development [58], suggesting that there may be long-term psychological benefits of financial aid that are worthy of further study.

4.1. Limitations

All data in this study (except some data included in Scope) were self-reported, and therefore subject to reporting biases. Although participants were diverse, the majority were White, heterosexual (92.3%), cisgender (99.8%), educated and financially stable, which may result in underestimating the true effects of pre-existing vulnerabilities on psychological distress. We note that our sample had relatively few individuals experiencing poverty or income precarity (∼10%), which may have limited our ability to observe effects of financial aid on psychological outcomes. Because data were cross-sectional, we cannot determine the predictive validity of hardship on postpartum mental health, though this is an important aspect of our follow-up surveys.

4.2. Clinical implications

Preparedness for future disasters/pandemics should address inequalities that cumulatively make some individuals more vulnerable to the negative effects of objective hardship. Policies that mitigate spread of the virus should also consider mechanisms to encourage (rather than discourage) natural sources of resilience, for example social support. The data also suggest that policies targeting individual-level vulnerabilities and hardships may be more salutary to mental health outcomes than region-based approaches. Taken together, the findings suggest that accessible and equitable healthcare as well as increased efforts to address substandard housing, food insecurity, and poverty are essential to reducing disproportionate hardships among vulnerable groups and mitigating the long-term impacts of the pandemic on parents and children.

Funding

This research was supported by the Alberta Children's Hospital Research Institute and the Owerko Centre. The funders had no role in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication and the Canadian Institutes of Health Research. Salary funding for L. Tomfohr-Madsen and C. Lebel was provided by thte Canada Research Chairs program.

CRediT authorship contribution statement

G.F. Giesbrecht: Data curation, Formal analysis, Funding acquisition, Methodology, Writing – original draft. M. van de Wouw: Investigation, Visualization, Writing – review & editing. C. Rioux: Methodology, Writing – review & editing. B.P.Y. Lai: Data curation, Writing – review & editing. S. King: Methodology, Writing – review & editing. L. Tomfohr-Madsen: Funding acquisition, Methodology, Writing – review & editing. C. Lebel: Funding acquisition, Methodology, Writing – review & editing.

Declaration of Competing Interest

None.

Appendix A. Supplementary data

Supplementary material: Supplemental fig. 1. POHI scores throughout the first year of the pandemic. Z-scores of POHI (i.e., Total, Threat, Loss, Change, Scope) were plotted against time (in months) throughout study recruitment. The start of the first month is at the start of the study (i.e., 5th of April 2020).