Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Increased maternal mental health burden in a representative longitudinal community cohort coinciding with COVID-19 lockdown

Increased maternal mental health burden in a representative longitudinal community cohort... AUSTRALIAN JOURNAL OF PSYCHOLOGY 2021, VOL. 73, NO. 4, 578–585 https://doi.org/10.1080/00049530.2021.1956286 Increased maternal mental health burden in a representative longitudinal community cohort coinciding with COVID-19 lockdown a b c a,d,e,f,g a,g b,f Amy Loughman , James Hedley , Craig A. Olsson , Michael Berk , Steven Moylan , Richard Saffery , b,h b,f b,e,i a,g Peter D. Sly , Mimi L.K. Tang , Anne-Louise Ponsonby *, Peter Vuillermin * and the BIS Investigator Group IMPACT – the Institute for Mental and Physical Health and Clinical Translation, School of Medicine, Barwon Health, Deakin University, b c Geelong, Australia; Murdoch Children’s Research Institute, Royal Children’s Hospital Melbourne, Parkville, Australia; Centre for Social and Early Emotional Development, School of Psychology, Deakin University, Burwood, Australia; Orygen, the National Centre of Excellence in Youth Mental Health, Centre for Youth Mental Health, Parkville, Australia; Florey Institute for Neuroscience and Mental Health, f g Melbourne, Australia; Department of Psychiatry, The University of Melbourne, Melbourne, Australia; Barwon Health, Geelong, Australia; h i Child Health Research Centre, The University of Queensland, Saint Lucia, Australia; Department of Paediatrics, The University of Melbourne, Parkville, Australia ABSTRACT ARTICLE HISTORY Received 1 October 2020 Measures to control the COVID-19 pandemic have disrupted social networks and employment Accepted 8 June 2021 security worldwide. Longitudinal data in representative samples are required to understand the corresponding mental health impacts. We aimed to estimate the prevalence of depressive KEYWORDS symptoms in Australian women raising young families during the first Victorian lockdown and Mental health; maternal; to identify risk factors. Participants comprise 347 mothers of children aged 7 (mean age: 32·11 pandemic; depression; years [4·27]), from the Barwon Infant Study (BIS). Mothers had previously completed Edinburgh psychiatry; mental disorders Postnatal Depression Scale (EPDS) at child ages zero, two, four. Following the lock down, mothers again completed EPDS along with questions regarding current household and employment demographics. Depressive symptoms were substantially more prevalent in the lockdown sample than at any prior assessment (EPDS10+; 30·6%); and were particularly high in women with previous poor mental health. Anticipated and actual job loss were twice as common relative to previous assessment (5% to 13%, p = 0 006) and (4% to 10%, p = 0 001) and were associated with depressive symptoms. While further studies are required to confirm causal associations, these findings highlight the need to support mental health during the COVID-19 pandemic, particularly in the context of employment insecurity and previous mental illness. KEY POINTS What is already known about this topic: (1) Emerging reports from convenience samples demonstrate elevated depressive symptoms during the COVID-19 pandemic. (2) Maternal mental health is important for child mental health. (3) Representative, longitudinal data are needed to further improve targeting of policy and health service delivery to prevent a post-COVID-19 mental health crisis. What this topic adds: (1) This early report from a population-derived cohort demonstrates high rates of depression symptomatology in mothers of school aged children following the first COVID-19 lockdown. (2) A past history of depression and current threats to employment are identified as key risk factors for adverse mental health. (3) Our findings are consistent with concerns regarding an increase in mental health burden in the wake of the COVID-19 pandemic but further studies are required to assess causality. Interventions and broader community resources to support the mental health of women of school aged children are required, and should target those with a history of depression and current threats to employment. There is widespread concern about a possible “mental lockdown measures taken globally to combat the health tsunami” resulting from the unprecedented spread of COVID-19 (Royal College of Psychiatrists, CONTACT Peter Vuillermin peter.vuillermin@deakin.edu.au *Co-senior and co-corresponding authors © 2021 Australian Psychological Society AUSTRALIAN JOURNAL OF PSYCHOLOGY 579 2020; Tsirtsakis, 2020). Physical distancing measures in on 24 March 2020. It required all Australians to stay at particular have disrupted social networks which are home unless necessary to access goods or services, for central to emotional regulation and social connection care or compassionate reasons, to attend work or edu- (Aida et al., 2013). Lockdown has also had profound cation where not possible by remote work/learning, consequences on economic security which also plays and for socially distanced outdoor exercise (Sutton, a key role in mental health (Ribeiro et al., 2017). 2020). There was widespread closure of non-essential The mental health impacts of the resulting dis- recreational, cultural and entertainment venues. The rupted family and broader social networks, job Australian government introduced the JobKeeper pro- losses, economic uncertainty, increased childcare gram on 30th of March to support part-time and full- commitments and social isolation are likely to be time workers in certain sectors affected by the considerable. Emerging evidence suggests dispro- lockdown. portionate mental health burden in groups with pre-existing mental health problems (Institute, Measures 2020) and also in women (Wenham et al., 2020). Given finite resources, representative, longitudinal Depressive Symptoms: The Edinburgh Postnatal data are urgently needed to guide government Depression Scale (EPDS) (Cox et al., 1987) was adminis- policy and health service delivery to those most tered at birth, four weeks, two years, four years and in need of support during this pandemic. seven to eight years (the current review). A cut point of 10+was used to indicate pre-clinical (mild) depressive Here we report data on the mental health profile in symptoms (Edmondson et al., 2010) and 13+ was used mothers of primary school aged children participating to indicate a high likelihood of depression (positive in the Barwon Infant Study (BIS), which is a large, birth predictive value of 70–90%) (Buist et al., 2002). Both cohort recruited using an unselected antenatal sam- cut-points have been validated in an Australian sample pling frame in southern Victoria, Australia (Vuillermin (Buist et al., 2002; Eastwood et al., 2011; Matthey et al., et al., 2015). The aims of the study were: (1) to estimate 2006). The EPDS has demonstrated validity and use in the prevalence of maternal depressive symptoms, mothers of older children; none of the items are spe- including thoughts of self-harm, during the first lock- cific to the postnatal period (Cox et al., 1996; down in Australia and (2) to identify those most at risk Woolhouse et al., 2015). of poor mental health across the pandemic on the Self-Harm: Given the context of a life-threatening basis of depression in the postnatal period and con- pandemic, we specifically examined women’s temporaneous contextual factors. responses to the single EPDS item: “The thought of harming myself has occurred to me”. We further exam- Methods ined this in combination with a score of 13 or above on the EPDS. Participants Current Household and Social Context: As listed in The Barwon Infant Study (n = 1074) is a longitudinal Table 3, BIS has detailed pre-pandemic data on house- birth cohort study recruited 2010–2012 with serial hold characteristics (e.g., number of adults and chil- questionnaires in sequential reviews (antenatal, four dren in the house; living with the biological father; weeks, two years, four years and seven to eight parental employment), social support (i.e., contact years – current) with recruitment from two hospitals with extended family and friends), parental stress and capturing 90% of births in a geographic catchment substance use, and the occurrence of major life events area, the demographics of which is representative of in the previous 12 months (e.g., serious illness, injury the broader Australian population (Vuillermin et al., and death; employment and financial upheaval, 2015). Here, participants comprise 347 mothers adapted from the Social Readjustment Rating Scale (mean age: 32·11 years [4·27]), who represent the first (Holmes & Rahe, 1967)). Sociodemographic factors at third (32·3%) of the Barwon Infant Study inception conception are described in Table 1 and were used for cohort to respond to the 2020 online questionnaires, backweighted adjustment by inverse probability during the period 29th April to 22 May 2020. weighting. Pandemic lockdown context Statistical analysis The Stage 2 lockdown associated with the first wave of We show the proportion of mothers reporting (1) COVID-19 infection in Australia commenced in Victoria thoughts of self-harm (EPDS Item 10), (2) EPDS 10+, 580 A. LOUGHMAN ET AL. Table 1. Sociodemographic characteristics of 2020 review sample and inception cohort. Inception cohort Current review Characteristics N = 1,064 N = 346 p-value* At current review Days since lockdown on 24 March 2020, mean (SD) - 42.42 (6.44) - Maternal age, mean (SD) - 40.79 (4.33) - Child age, mean (SD) - 7.95 (0.85) - Child female - 167 (48) - At conception Maternal age, mean (SD) 31.32 (4.79) 32.08 (4.30) <0.001 Paternal age, mean (SD) 33.49 (5.87) 33.44 (5.29) 0.8 Maternal uni degree or above, n (%) 541 (51) 216 (62) <0.001 Paternal uni degree or above, n (%) 360 (34) 141 (41) 0.01 Most socio-economically disadvantaged tertile postcode, n (%) 352 (33) 93 (27) 0.003 Regional/rural postcode, n (%) 282 (27) 90 (26) 0.8 Mother born in Australia, n (%) 957 (90) 315 (91) 0.4 Both parents born in Australia, n (%) 834 (78) 284 (82) 0.2 Number of older siblings in the same house, mean (SD) 0.88 (0.95) 0.77 (0.88) 0.01 Maternal history of asthma, n (%) 320 (30) 106 (31) 0.03 * p-values comparing current review to inception cohort using Pearson’ chi-squared test for categorical characteristics, and Student’s t-test for continuous characteristics Figure 1. Proportion of responses indicating (a) Thoughts of self-harm, (b) EPDS score 10+, (c) EPDS score 13+, (d) EPDS score 13 +and thoughts of self-harm (unadjusted analyses). (3) EPDS 13+, and (4) thoughts of self-harm with EPDS disadvantage of postcode, remoteness of postcode 13+. To assess any potential selection bias in the and household size. We have used the same weights respondent sample included in this report, we also in our previous work (Allen et al., 2013) see present results unadjusted as well as backweighted Supplementary Figure 1 for the distribution of inverse using inverse probability weighting (Little & Rubin, probability weights used. All estimates are presented 2019). Backweighting was done in terms of age, history with 95% confidence intervals of asthma (a primary outcome of interest in the BIS To examine the potential role of previous mental cohort during recruitment), socio-economic health difficulties, we assessed associations between AUSTRALIAN JOURNAL OF PSYCHOLOGY 581 Table 2. Subgroup analyses of current EPDS scores by past depressive symptoms. Subgroup Outcome, % (95%CI), EPDS <10 at 2 and 4 years EPDS 10+ at 2 or 4 years EPDS 10+ at 2 and 4 years EPDS score 10+ 23.4 (17.6–30.1) 44.2 (30.5–58.7) 82.6 (61.2–95.0) EPDS score 13+ 11.5 (7.3–16.8) 23.1 (12.5–36.8) 56.5 (34.5–76.8) Thoughts of self-harm 2.6 (0.9–6.0) 11.5 (4.4–23.4) 30.4 (13.2–52.9) Adjusted for characteristics at conception reported in Table 1 Table 3. Univariate associations between EPDS scores and household, family, community and employment characteristics. EPDS score 10+ EPDS score 13+ Thoughts of self-harm Characteristics RR 95% CI p RR 95% CI p RR 95% CI p Days since lockdown started (24 March 2020) 0.98 0.96–1.01 0.2 0.98 0.94–1.02 0.3 0.99 0.93–1.05 0.7 Household characteristics Number of adults in household 1.10 0.87–1.38 0.4 0.75 0.38–1.50 0.4 0.46 0.19–1.12 0.09 Number of children in household 1.09 0.89–1.33 0.4 1.13 0.80–1.58 0.5 1.63 0.93–2.85 0.08 Number of people in household (adults and children) 1.07 0.96–1.18 0.2 0.96 0.83–1.10 0.5 0.98 0.86–1.12 0.8 In a relationship 1.04 0.61–1.78 0.9 0.69 0.35–1.38 0.3 0.36 0.15–0.86 0.02 Lives with biological father of the child/ren 0.42 0.35–0.50 <0.001 0.59 0.33–1.05 0.07 0.45 0.18–1.10 0.08 Most socio-economically disadvantaged tertile 1.02 0.69–1.50 0.9 1.25 0.72–2.19 0.4 1.72 0.75–3.96 0.2 Lives in a regional area 0.89 0.65–1.24 0.5 0.90 0.54–1.50 0.7 0.71 0.29–1.72 0.4 Employment 0.2 0.5 0.5 Employed/job keeper, partner employed/job keeper ref ref ref Employed/job keeper, partner unemployed/no partner 3.07 0.50–18.85 1.38 0.21–9.29 1.07 0.16–7.15 Not working, partner employed/job keeper 1.42 0.20–9.82 0.56 0.06–5.08 - Not working, partner unemployed/no partner 2.54 0.42–15.21 1.47 0.25–8.76 0.62 0.11–3.52 Pre-COVID-19 social support See parents at least once per week 0.89 0.64–1.25 0.5 1.15 0.64–2.06 0.6 0.62 0.26–1.46 0.3 See inlaws at least once per week 1.51 1.11–2.06 0.008 1.60 0.98–2.60 0.06 1.41 0.63–3.15 0.4 See other family at least once per week 1.26 0.93–1.70 0.1 1.25 0.78–2.00 0.4 0.61 0.25–1.51 0.3 See friends at least once per week 1.22 0.85–1.77 0.3 0.95 0.56–1.61 0.8 0.67 0.29–1.52 0.3 Stress and substance abuse Perceived stress scale score 1.11 1.06–1.15 <0.001 1.13 1.05–1.22 <0.001 1.17 1.01–1.36 0.04 Alcohol more than once per week 0.98 0.71–1.34 0.9 1.39 0.86–2.25 0.2 0.94 0.42–2.11 0.9 Binge drinking ever 0.95 0.69–1.30 0.8 1.25 0.75–2.08 0.4 0.70 0.31–1.60 0.4 Smoking (at 4 years) 0.49 0.14–1.67 0.3 - - Major life events in the past 12 months Suffered a serious illness, injury, or assault 1.35 0.82–2.21 0.2 1.18 0.49–2.86 0.7 0.69 0.10–4.66 0.7 Close relative with illness, injury or assault 1.02 0.70–1.49 0.9 1.00 0.54–1.83 0.9 0.74 0.24–2.34 0.6 Parent, partner, or child died 2.83 1.18–6.78 0.02 1.39 0.30–6.39 0.7 2.77 0.62–12.44 0.2 Close family friend or relative died 1.43 1.04–1.96 0.03 1.61 0.96–2.70 0.07 0.89 0.28–2.82 0.8 Broke off a steady romantic relationship 1.39 0.79–2.44 0.3 2.51 1.37–4.60 0.003 3.69 1.36–10.05 0.01 Serious problem with a friend, neighbour, or relative 1.48 1.04–2.09 0.03 0.88 0.39–1.98 0.8 0.77 0.19–3.06 0.7 Seeking work unsuccessfully for more than a month 1.36 0.80–2.30 0.3 0.80 0.23–2.86 0.7 1.07 0.17–6.89 0.9 Major financial crisis 1.52 0.99–2.35 0.06 2.26 1.15–4.42 0.02 4.38 1.93–9.91 <0.001 Problems with police and court appearance - - - Something valuable lost or stolen 3.34 1.84–6.04 <0.001 - 31.45 5.34–185.26 <0.001 Someone in household had an alcohol problem 1.05 0.39–2.84 0.9 1.10 0.20–5.94 0.9 2.51 0.52–12.18 0.3 Someone in household had a drug problem - - - Serious disappointment in work or career 1.44 1.03–1.99 0.03 1.70 1.03–2.81 0.04 1.00 0.37–2.70 0.9 Thought would lose job soon 1.78 1.27–2.51 <0.001 2.23 1.34–3.68 0.002 5.25 2.26–12.19 <0.001 Lost job (not from choice) 1.29 0.87–1.90 0.2 2.10 1.25–3.53 0.005 3.39 1.54–7.45 0.002 Analyses are adjusted for 2 year or 4 year EPDS score 10+ previous postnatal depressive symptoms (no ele- Binomial regression was used to estimate risk vated score, elevated at two or four years, elevated ratios for the four dichotomous depression out- at both two and four years) and the current EPDS comes, prior mental health status, and contempor- score. ary contextual factors and major life events such as To examine the potential role of current context, we job loss (Wacholder, 1986). Population attributable assessed associations between household, family, fractions (Greenland & Drescher, 1993) were calcu- community and employment characteristics relevant lated for associations between major life events to the pandemic and depression, adjusting for an that had (i) increased in prevalence from child EPDS 10+at the two year or four year reviews (termed age four to age seven-eight, and were also (ii) past depressive symptoms). associated with depressive symptoms in Table 4. 582 A. LOUGHMAN ET AL. Table 4. Major life events that increased in prevalence at 7 year review. Due to Prevalence, n (%) Population attributable fraction, % (95% CI) COVID-19 Thoughts of Year 4 Year 7 N (%) EPDS score 10+ EPDS score 13+ self-harm Major life event N = 323 N = 346 p % 95% CI % 95% CI % 95% CI Parent, partner, or child died 12 (4) 5 (1) 0.02 1 (<1) 2.53 −0.95,5.89 0.58 −2.61,3.66 3.14 −4.24,9.99 Close family friend or relative died 79 (25) 64 (18) 0.06 2 (<1) 6.82 −0.20,13.34 9.39 −3.03,20.31 −1.89 −21.07,14.25 Broke off a steady romantic relationship 5 (2) 9 (3) 0.6 0 (0) 1.06 −1.06,3.14 4.05 0.12,7.83 7.62 −2.57,16.80 Serious problem with a friend, neighbour, 22 (7) 32 (9) 0.2 1 (<1) 4.83 −0.18,9.59 −1.30 −9.57,6.35 −2.72 −16.55,9.47 or relative Major financial crisis 17 (5) 17 (5) 0.9 7 (2) 1.80 −0.46,4.01 4.20 −0.80,8.95 10.39 −0.08,19.77 Something valuable lost or stolen 7 (2) 8 (2) 0.8 0 (0) 5.51 0.84,9.95 22.91 21.86,23.95 47.99 −34.26,79.85 Serious disappointment in work or career 34 (11) 53 (15) 0.07 19 (5) 7.17 −0.20,14.00 11.23 −1.78,22.58 −0.05 −21.54,17.64 Thought would lose job soon 17 (5) 44 (13) 0.006 32 (9) 10.10 3.07,16.62 15.29 2.95,26.06 40.35 11.14,59.96 Lost job (not from choice) 12 (4) 35 (10) 0.001 28 (8) 2.74 −2.01,7.26 9.79 0.61,18.12 19.30 −0.06,34.91 Population attributable fraction (PAF) estimates are provided for factors associated with outcomes in Table 3 and indicate how much, in the current wave, these outcomes could potentially be attributed to factors associated with these life events, if the significant associations shown in Table 3 were causal in nature. Adjusted for past depression history. Results with diagnosable depression (EPDS score 13+), com- pared to 11·5% (95% CI: 7·3–16·8) of those who did not Table 1 outlines the sociodemographic characteristics have past depressive symptoms (p < 0 · 001). Those of the 2020 review sample and the inception cohort at with past depressive symptoms were therefore more birth. While they are similar in most respects, the cur- likely to have depressive symptoms in the 2020 lock- rent lockdown sample had a higher proportion of down. See Table 2 for subgroup analyses of current mothers with university level education (63% vs 51%), symptoms by past depressive symptoms. were less likely to live in a disadvantaged postcode, Of the household and related social contextual fac- and had fewer older siblings in the house at birth than tors considered, an important factor was living with the full inception cohort. Questionnaires were com- the biological father (of the child/ren), associated with pleted an average of 42 days (SD: 6 days) since lock- reduced risk of EPDS 10+ (RR 0·42, 95% CI: 0·35–0·50, down began. No families reported positive COVID-19 p < 0 · 001), with weaker evidence of associations with tests. EPDS 13+and thoughts of self-harm (RR 0.59, 95% CI: 0.33–1.05, p = 0.07; RR 0.45, 95% CI 0.18–1.10, p = 0.08). The numbers of other household residents were not Increased prevalence of depressive symptoms in significantly associated with EPDS score. The death of the current review a “parent, partner or child” or “close family friend or Depressive symptoms were more prevalent in the relative” were both strongly associated with EPDS 10+ current lockdown sample than at any prior wave of (RR 2·83, 95% CI: 1.18–7.68, p = 0.02; RR 1·43, 1.04–1.96, assessment (antenatal, four week, two year or p = 0.03), however these associations were not consis- four year reviews), with 30.6% (95% CI: 25·8–35·8) tent across EPDS 13+and thoughts of self-harm, or of the sample scoring EPDS 10+and 16·2% (95% CI: following adjustment for backweighting. 12·5–20·5) scoring EDPS 13+ (Panels 2 and 3, The prevalence of recent adverse life events, parti- Figure 1). Furthermore, thoughts of self-harm were cularly pertaining to employment or finance present in 7.2% (95% CI: 4··7–10·5) of the sample increased from child age four to age seven-eight. (Panel 1, Figure 1). Thus, there was a high burden of The proportion of respondents concerned about depressive symptoms and 1 in 14 had thoughts of imminent job loss more than doubled (5% to 13%, self-harm. p = 0 · 006) as did the proportion who had already lost their jobs (4% to 10%, p = 0 · 002; see Table 4). Most respondents stated that these two employ- Previous mental health difficulties, demographic ment-related events had occurred due to COVID-19 and lockdown characteristics associated with (73% and 80%, respectively). Both threat of job loss elevated depressive symptoms and actual job loss were positively associated with At child age seven to eight years, coinciding with lock- EPDS 13+ (RR 2·23, 95% CI: 1·34–3·68, p = 0 · 002; RR down, 56·5% (95% CI: 34·5–76·8) of the respondents 2·10, 95% CI: 1·25–3.53, p = 0 · 005; respectively) and who had past depressive symptoms at two prior time- with thoughts of self-harm (RR 5·25, 95% CI: 2·26– points reported current symptom burden concordant 12·19, p < 0.001; 3·39, 1·54–7·45, p = 0 · 002, AUSTRALIAN JOURNAL OF PSYCHOLOGY 583 respectively). The population attributable fractions planning, more than half of those who had previously from these employment factors indicate that the experienced depressive symptoms reported depres- impacts on depression could be considerable (threat sive symptoms concordant with diagnosable depres- of job loss: 15.29%, 95% CI: 2.95–26.06; actual job sion following the lockdown. loss: 9.79, 95% CI: 0.61–18.12) although the estimates Of the household, social and adverse life events are imprecise. Overall, as expected, depressive symp- factors we considered, currently living with the biolo- toms were positively associated with higher per- gical father of the child/ren was protective. Threats to ceived stress scores. job security, commonly reported by participants as Backweighting by inverse probability weighting did occurring due to the pandemic, were associated with not materially change the estimates of prevalence, higher depressive symptoms. These work-related subgroup analyses or associations with current house- major life events increased in prevalence relative to hold, social and adverse life events characteristics the four-year review. However, the greatest risk for reported here, with the exception of death of reporting either EPDS 10+or 13+was in those mothers a “parent, partner or child” or “close family friend or with previous mental health difficulties, echoing other relative” (Supplementary Tables 1–4). reports of vulnerability in this group during times of stress (Milgrom et al., 2008). This study did not include questions regarding domestic violence, which has Discussion since emerged as highly relevant to mental health Our results indicate that around one in three mothers and wellbeing during COVID-19 lockdown restrictions with primary school aged children experienced ele- (Neil, 2020). vated pre-clinical depressive symptomology (EPDS 10 Emerging reports of psychological symptoms in +), close to 2 in 10 (16%) experienced more severe adults from the general population completing online depression (EDPS 13+), and 1 in 14 experienced surveys from Australia, the UK and China suggest thoughts of self-harm. There are no published data elevated symptoms relative to population norms dur- on depression in mothers of early primary school ing COVID-19 lockdowns (Newby, 2020; Shevlin et al., aged children, from prior to and/or during the pan- 2020; Wang et al., 2020). A meta-analysis of eight demic, with which our results can be compared. This studies of pregnancy or postpartum women from limits our capacity to infer that the observed increase different countries during the pandemic showed in adverse maternal mental health is in fact caused by strong evidence for increased anxiety symptoms, the COVID-19 lockdown. and weaker evidence for increased depressive symp- At more severe clinical levels of depression, as indi- toms (Hessami et al., 2020). None of the above studies cated by EPDS 13+scores, the prevalence we report is were in representatively sampled or pre-existing similar to that reported in the pre-COVID literature. The cohorts, leaving open the possibility of selection most comparable estimates come from a large, popu- bias. Selection bias into the initial cohort or early lation-based Australian study conducted in 2014 which responders in the current wave have been investi- reported approximately 14% of first-time mothers had gated using inverse probability weighting and do EPDS scores of 13+when their children were four-years not appear to be substantial. However, the absence old, and again when their children were ten (using the of the counterfactual (i.e., those not exposed to lock- equivalent cut-off score on the Center for down) precludes causal inference about the effects of Epidemiologic Studies Depression Scale at this later lockdown. It is also unclear whether the prevalence of follow-up) (Brown et al., 2020; Woolhouse et al., pre-clinical depressive symptoms (EPDS 10+) requires 2015). The prevalence of EDPS 13+scores in parents intervention or reflects understandable adjustment of primary school aged children during lockdown was difficulties in the context of social adversity, which only marginally higher (16·2%). However, we observed are likely to be transient and self-limiting. Ongoing a clear increase between the four and the seven-eight research regarding the mental health of men as well year reviews for both pre-clinical and clinical level as women who are not mothers, and across other age symptoms and a parallel increase in adverse life events groups and geography during the COVID-19 pan- such as job insecurity, indicating a potential effect of demic will be required to generalise and extend the COVID-19 lockdown. Nonetheless, it is impossible upon these findings. to make robust causal inferences in the absence of The Australian government has attempted to pro- counterfactual data i.e., measures conducted among tect household incomes through the JobKeeper pro- women who were not exposed to the COVID-19 lock- gram. Even in this context, the impact of short-term down. Of particular importance to clinical services work-related events on mental health appears 584 A. LOUGHMAN ET AL. substantial and these findings need to be taken into References account for future policy decisions as the COVID-19 Aida, J., Kawachi, I., Subramanian, S., & Kondo, K. (2013). pandemic continues. Our findings also indicate the Disaster, social capital, and health. In I. Kawachi, S. Takao, need to make mental health services accessible to & S. V. Subramanian (Eds.), Global perspectives on social those with a mental health history. In countries that capital and health (pp. 167–187). Springer. Allen, K. J., Koplin, J. J., Ponsonby, A.-L., Gurrin, L. C., Wake, M., have experienced different COVID-19 infection rates, Vuillermin, P., Martin, P., Matheson, M., Lowe, A., Robinson, stricter lockdowns and different policy responses, the M., Tey, D., Osborne, N. J., Dang, T., Tan, H.-T. T., Thiele, L., effects on mental health may be greater than those Anderson, D., Czech, H., Sanjeevan, J., Zurzolo, G., . . . reported here. Further, the mental health burden is Dharmage, S. C. (2013). Vitamin D insufficiency is asso- likely to have increased in the context of the second ciated with challenge-proven food allergy in infants. Journal of Allergy and Clinical Immunology, 131(4), 1109– wave and stricter lockdown in Victoria from July to 1116.e1106. https://doi.org/10.1016/j.jaci.2013.01.017 October 2020. Brown, S. J., Mensah, F., Giallo, R., Woolhouse, H., Hegarty, K., Here we show the first data from a representative Nicholson, J. M., & Gartland, D. (2020). Intimate partner sample of Australian mothers, revealing a strong pat- violence and maternal mental health ten years after a first tern of increasing depressive symptoms as their chil- birth: An Australian prospective cohort study of first-time dren age from birth to seven/eight years of age, the mothers. Journal of Affective Disorder, 262, 247–257. https://doi.org/10.1016/j.jad.2019.11.015 latter being coincident with the first COVID-19 lock- Buist, A. E., Milgrom, J., Barnett, B. E. W., Pope, S., Condon, J. T., down. We show that mothers most in need of support Ellwood, D. A., Boyce, P. M., Austin, M.-P. V., & Hayes, B. A. at child age seven/eight are those with prior depres- (2002). To screen or not to screen — That is the question in sion. Whilst causality cannot be inferred, findings coin- perinatal depression. Medical Journal of Australia, 177(S7), cide with the first lockdown due to COVID-19, and may S101–S105. https://doi.org/10.5694/j.1326-5377.2002. tb04866.x reflect the impact of physical distancing, and other Cox, J. L., Chapman, G., Murray, D., & Jones, P. (1996). measures, to control the spread of disease. Our find - Validation of the Edinburgh Postnatal Depression Scale ings echo established calls for increased mental health (EPDS) in non-postnatal women. Journal of Affective support for women across the early childhood period, Disorders, 39(3), 185–189. https://doi.org/10.1016/0165- and highlight the importance of supporting parents of 0327(96)00008-0 school age children. We further suggest that measures Cox, J. L., Holden, J. M., & Sagovsky, R. (1987). Detection of postnatal depression: Development of the 10-item taken to deal with the COVID-19 pandemic may inten- Edinburgh Postnatal Depression Scale. The British Journal sify risk in this group and require additional support, of Psychiatry, 150(6), 782–786. https://doi.org/10.1192/bjp. including targeted programs for those at increased risk 150.6.782 (Galea et al., 2020). Increasing the availability of com- Eastwood, J. G., Phung, H., & Barnett, B. (2011). Postnatal munity supports more broadly, such as social support depression and socio-demographic risk: Factors associated with Edinburgh Depression Scale scores in a metropolitan community groups, telephone counselling services area of New South Wales, Australia. Australian and New and other neighbourhood initiatives may also be rele- Zealand Journal of Psychiatry, 45(12), 1040–1046. https:// vant preventative action for mental health during the doi.org/10.3109/00048674.2011.619160 pandemic. Edmondson, O. J., Psychogiou, L., Vlachos, H., Netsi, E., & Ramchandani, P. G. (2010). Depression in fathers in the postnatal period: Assessment of the Edinburgh Postnatal Disclosure statement Depression Scale as a screening measure. Journal of Affective Disorder, 125(1–3), 365–368. https://doi.org/10. No potential conflict of interest was reported by the author(s). 1016/j.jad.2010.01.069 Galea, S., Merchant, R. M., & Lurie, N. (2020). The mental health consequences of COVID-19 and physical distancing: The Funding need for prevention and early intervention. JAMA Internal Medicine, 180(6), 817–8. https://doi.org/10.1001/jamain This work was supported by the Deakin University; Murdoch ternmed.2020.1562 Children’s Research Institute; National Health and Medical Greenland, S., & Drescher, K. (1993). Maximum likelihood Research Council; The Shepherd Foundation; Barwon estimation of the attributable fraction from logistic Health; Perpetual Trustees. models. Biometrics, 49(3), 865–872. Hessami, K., Romanelli, C., Chiurazzi, M., & Cozzolino, M. (2020). COVID-19 pandemic and maternal mental health: ORCID A systematic review and meta-analysis. Journal of Maternal Fetal and Neonatal Medicine, 1–8. https://doi.org/10.1080/ Amy Loughman http://orcid.org/0000-0002-0257-1443 14767058.2020.1843155 Michael Berk http://orcid.org/0000-0002-5554-6946 AUSTRALIAN JOURNAL OF PSYCHOLOGY 585 Holmes, T. H., & Rahe, R. H. (1967). The social readjust- Shevlin, M., McBride, O., Murphy, J. (2020). Anxiety, depres- ment rating scale. Journal of Psychosomatic Research, 11 sion, traumatic stress, and COVID-19 related anxiety in the (2), 213–218. https://doi.org/10.1016/0022-3999(67) UK general population during the COVID-19 pandemic. 90010-4 BJPsych Open, 6(6), e125–e. Institute, B. D. (2020). Mental health ramifications of COVID-10: Stay at Home Directions (No 6), Section 200 C.F.R. (2020). The Australian context. Report. Sutton, B. (2020). Directions issued by Victoria’s Chief Health Little, R. J., & Rubin, D. B. (2019). Statistical analysis with Officer . Victorian State Government. https://www.dhhs.vic. missing data (Vol. 793). John Wiley & Sons. gov.au/victorias-restriction-levels-covid-19 Matthey, S., Henshaw, C., Elliott, S., & Barnett, B. (2006). Tsirtsakis, A. (2020). Fears of post-pandemic ‘tsunami of health Variability in use of cut-off scores and formats on the problems’. RACGP. https://www1.racgp.org.au/newsgp/clini Edinburgh Postnatal Depression Scale: Implications for clin- cal/gps-fear-post-pandemic-tsunami-of-health-problems ical and research practice. Archives of Womens Mental Health, Vuillermin, P., Saffery, R., Allen, K. J., Carlin, J. B., Tang, M. L., 9(6), 309–315. https://doi.org/10.1007/s00737-006-0152-x Ranganathan, S., Burgner, D., Dwyer, T., Collier, F., Jachno, K., Milgrom, J., Gemmill, A. W., Bilszta, J. L., Hayes, B., Barnett, B., Sly, P., Symeonides, C., McCloskey, K., Molloy, J., Forrester, M., Brooks, J., Ericksen, D., & Buist, A. (2008). Antenatal risk & Ponsonby, A.-L. (2015). Cohort profile: The Barwon infant factors for postnatal depression: A large prospective study. study. International Journal of Epidemiology, 44(4), Journal of Affective Disorders, 108(1–2), 147–157. https:// 1148–1160. https://doi.org/10.1093/ije/dyv026 doi.org/10.1016/j.jad.2007.10.014 Wacholder, S. (1986). Binomial regression in GLIM: Neil, J. (2020). Domestic violence and COVID 19. Australian Estimating risk ratios and risk differences. American Journal for General Practitioners. https://www1.racgp.org. Journal of Epidemiology, 123(1), 174–184. https://doi. au/ajgp/coronavirus/domestic-violence-and-covid-19 org/10.1093/oxfordjournals.aje.a114212 Newby, J. M. (2020). Acute mental health responses during the Wang, C., Pan, R., Wan, X., Tan, Y., Xu, L., Ho, C. S., & Ho, R. C. COVID-19 pandemic in Australia. medRxiv (Preprint). (2020). Immediate psychological responses and associated medRxiv. https://doi.org/10.1101/2020.05.03.20089961 factors during the initial stage of the 2019 coronavirus Ribeiro, W. S., Bauer, A., Andrade, M. C. R., York-Smith, M., disease (COVID-19) epidemic among the general popula- Pan, P. M., Pingani, L., . . . Evans-Lacko, S. (2017). Income tion in China. International Journal of Environmental inequality and mental illness-related morbidity and resili- Research and Public Health, 17(5), 1729. https://doi.org/ ence: A systematic review and meta-analysis. The Lancet 10.3390/ijerph17051729 Psychiatry, 4(7), 554–562. https://doi.org/10.1016/S2215- Wenham, C., Smith, J., & Morgan, R. (2020). COVID-19: The 0366(17)30159-1 gendered impacts of the outbreak. The Lancet, 395 Royal College of Psychiatrists. (2020). Psychiatrists see alarm- (10227), 846–848. https://doi.org/10.1016/s0140- ing rise in patients needing urgent and emergency care and 6736(20)30526-2 forecast a ‘tsunami’ of mental illness [Press release]. https:// Woolhouse, H., Gartland, D., Mensah, F., & Brown, S. J. (2015). www.rcpsych.ac.uk/news-and-features/latest-news/detail/ Maternal depression from early pregnancy to 4 years post- 2020/05/15/psychiatrists-see-alarming-rise-in-patients- partum in a prospective pregnancy cohort study: needing-urgent-and-emergency-care?searchTerms Implications for primary health care. BJOG, 122(3), =tsunami 312–321. https://doi.org/10.1111/1471-0528.12837 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Australian Journal of Psychology Taylor & Francis

Increased maternal mental health burden in a representative longitudinal community cohort coinciding with COVID-19 lockdown

Loading next page...
 
/lp/taylor-francis/increased-maternal-mental-health-burden-in-a-representative-1pTmztAZ7u

References (32)

Publisher
Taylor & Francis
Copyright
© 2021 Australian Psychological Society
ISSN
1742-9536
eISSN
0004-9530
DOI
10.1080/00049530.2021.1956286
Publisher site
See Article on Publisher Site

Abstract

AUSTRALIAN JOURNAL OF PSYCHOLOGY 2021, VOL. 73, NO. 4, 578–585 https://doi.org/10.1080/00049530.2021.1956286 Increased maternal mental health burden in a representative longitudinal community cohort coinciding with COVID-19 lockdown a b c a,d,e,f,g a,g b,f Amy Loughman , James Hedley , Craig A. Olsson , Michael Berk , Steven Moylan , Richard Saffery , b,h b,f b,e,i a,g Peter D. Sly , Mimi L.K. Tang , Anne-Louise Ponsonby *, Peter Vuillermin * and the BIS Investigator Group IMPACT – the Institute for Mental and Physical Health and Clinical Translation, School of Medicine, Barwon Health, Deakin University, b c Geelong, Australia; Murdoch Children’s Research Institute, Royal Children’s Hospital Melbourne, Parkville, Australia; Centre for Social and Early Emotional Development, School of Psychology, Deakin University, Burwood, Australia; Orygen, the National Centre of Excellence in Youth Mental Health, Centre for Youth Mental Health, Parkville, Australia; Florey Institute for Neuroscience and Mental Health, f g Melbourne, Australia; Department of Psychiatry, The University of Melbourne, Melbourne, Australia; Barwon Health, Geelong, Australia; h i Child Health Research Centre, The University of Queensland, Saint Lucia, Australia; Department of Paediatrics, The University of Melbourne, Parkville, Australia ABSTRACT ARTICLE HISTORY Received 1 October 2020 Measures to control the COVID-19 pandemic have disrupted social networks and employment Accepted 8 June 2021 security worldwide. Longitudinal data in representative samples are required to understand the corresponding mental health impacts. We aimed to estimate the prevalence of depressive KEYWORDS symptoms in Australian women raising young families during the first Victorian lockdown and Mental health; maternal; to identify risk factors. Participants comprise 347 mothers of children aged 7 (mean age: 32·11 pandemic; depression; years [4·27]), from the Barwon Infant Study (BIS). Mothers had previously completed Edinburgh psychiatry; mental disorders Postnatal Depression Scale (EPDS) at child ages zero, two, four. Following the lock down, mothers again completed EPDS along with questions regarding current household and employment demographics. Depressive symptoms were substantially more prevalent in the lockdown sample than at any prior assessment (EPDS10+; 30·6%); and were particularly high in women with previous poor mental health. Anticipated and actual job loss were twice as common relative to previous assessment (5% to 13%, p = 0 006) and (4% to 10%, p = 0 001) and were associated with depressive symptoms. While further studies are required to confirm causal associations, these findings highlight the need to support mental health during the COVID-19 pandemic, particularly in the context of employment insecurity and previous mental illness. KEY POINTS What is already known about this topic: (1) Emerging reports from convenience samples demonstrate elevated depressive symptoms during the COVID-19 pandemic. (2) Maternal mental health is important for child mental health. (3) Representative, longitudinal data are needed to further improve targeting of policy and health service delivery to prevent a post-COVID-19 mental health crisis. What this topic adds: (1) This early report from a population-derived cohort demonstrates high rates of depression symptomatology in mothers of school aged children following the first COVID-19 lockdown. (2) A past history of depression and current threats to employment are identified as key risk factors for adverse mental health. (3) Our findings are consistent with concerns regarding an increase in mental health burden in the wake of the COVID-19 pandemic but further studies are required to assess causality. Interventions and broader community resources to support the mental health of women of school aged children are required, and should target those with a history of depression and current threats to employment. There is widespread concern about a possible “mental lockdown measures taken globally to combat the health tsunami” resulting from the unprecedented spread of COVID-19 (Royal College of Psychiatrists, CONTACT Peter Vuillermin peter.vuillermin@deakin.edu.au *Co-senior and co-corresponding authors © 2021 Australian Psychological Society AUSTRALIAN JOURNAL OF PSYCHOLOGY 579 2020; Tsirtsakis, 2020). Physical distancing measures in on 24 March 2020. It required all Australians to stay at particular have disrupted social networks which are home unless necessary to access goods or services, for central to emotional regulation and social connection care or compassionate reasons, to attend work or edu- (Aida et al., 2013). Lockdown has also had profound cation where not possible by remote work/learning, consequences on economic security which also plays and for socially distanced outdoor exercise (Sutton, a key role in mental health (Ribeiro et al., 2017). 2020). There was widespread closure of non-essential The mental health impacts of the resulting dis- recreational, cultural and entertainment venues. The rupted family and broader social networks, job Australian government introduced the JobKeeper pro- losses, economic uncertainty, increased childcare gram on 30th of March to support part-time and full- commitments and social isolation are likely to be time workers in certain sectors affected by the considerable. Emerging evidence suggests dispro- lockdown. portionate mental health burden in groups with pre-existing mental health problems (Institute, Measures 2020) and also in women (Wenham et al., 2020). Given finite resources, representative, longitudinal Depressive Symptoms: The Edinburgh Postnatal data are urgently needed to guide government Depression Scale (EPDS) (Cox et al., 1987) was adminis- policy and health service delivery to those most tered at birth, four weeks, two years, four years and in need of support during this pandemic. seven to eight years (the current review). A cut point of 10+was used to indicate pre-clinical (mild) depressive Here we report data on the mental health profile in symptoms (Edmondson et al., 2010) and 13+ was used mothers of primary school aged children participating to indicate a high likelihood of depression (positive in the Barwon Infant Study (BIS), which is a large, birth predictive value of 70–90%) (Buist et al., 2002). Both cohort recruited using an unselected antenatal sam- cut-points have been validated in an Australian sample pling frame in southern Victoria, Australia (Vuillermin (Buist et al., 2002; Eastwood et al., 2011; Matthey et al., et al., 2015). The aims of the study were: (1) to estimate 2006). The EPDS has demonstrated validity and use in the prevalence of maternal depressive symptoms, mothers of older children; none of the items are spe- including thoughts of self-harm, during the first lock- cific to the postnatal period (Cox et al., 1996; down in Australia and (2) to identify those most at risk Woolhouse et al., 2015). of poor mental health across the pandemic on the Self-Harm: Given the context of a life-threatening basis of depression in the postnatal period and con- pandemic, we specifically examined women’s temporaneous contextual factors. responses to the single EPDS item: “The thought of harming myself has occurred to me”. We further exam- Methods ined this in combination with a score of 13 or above on the EPDS. Participants Current Household and Social Context: As listed in The Barwon Infant Study (n = 1074) is a longitudinal Table 3, BIS has detailed pre-pandemic data on house- birth cohort study recruited 2010–2012 with serial hold characteristics (e.g., number of adults and chil- questionnaires in sequential reviews (antenatal, four dren in the house; living with the biological father; weeks, two years, four years and seven to eight parental employment), social support (i.e., contact years – current) with recruitment from two hospitals with extended family and friends), parental stress and capturing 90% of births in a geographic catchment substance use, and the occurrence of major life events area, the demographics of which is representative of in the previous 12 months (e.g., serious illness, injury the broader Australian population (Vuillermin et al., and death; employment and financial upheaval, 2015). Here, participants comprise 347 mothers adapted from the Social Readjustment Rating Scale (mean age: 32·11 years [4·27]), who represent the first (Holmes & Rahe, 1967)). Sociodemographic factors at third (32·3%) of the Barwon Infant Study inception conception are described in Table 1 and were used for cohort to respond to the 2020 online questionnaires, backweighted adjustment by inverse probability during the period 29th April to 22 May 2020. weighting. Pandemic lockdown context Statistical analysis The Stage 2 lockdown associated with the first wave of We show the proportion of mothers reporting (1) COVID-19 infection in Australia commenced in Victoria thoughts of self-harm (EPDS Item 10), (2) EPDS 10+, 580 A. LOUGHMAN ET AL. Table 1. Sociodemographic characteristics of 2020 review sample and inception cohort. Inception cohort Current review Characteristics N = 1,064 N = 346 p-value* At current review Days since lockdown on 24 March 2020, mean (SD) - 42.42 (6.44) - Maternal age, mean (SD) - 40.79 (4.33) - Child age, mean (SD) - 7.95 (0.85) - Child female - 167 (48) - At conception Maternal age, mean (SD) 31.32 (4.79) 32.08 (4.30) <0.001 Paternal age, mean (SD) 33.49 (5.87) 33.44 (5.29) 0.8 Maternal uni degree or above, n (%) 541 (51) 216 (62) <0.001 Paternal uni degree or above, n (%) 360 (34) 141 (41) 0.01 Most socio-economically disadvantaged tertile postcode, n (%) 352 (33) 93 (27) 0.003 Regional/rural postcode, n (%) 282 (27) 90 (26) 0.8 Mother born in Australia, n (%) 957 (90) 315 (91) 0.4 Both parents born in Australia, n (%) 834 (78) 284 (82) 0.2 Number of older siblings in the same house, mean (SD) 0.88 (0.95) 0.77 (0.88) 0.01 Maternal history of asthma, n (%) 320 (30) 106 (31) 0.03 * p-values comparing current review to inception cohort using Pearson’ chi-squared test for categorical characteristics, and Student’s t-test for continuous characteristics Figure 1. Proportion of responses indicating (a) Thoughts of self-harm, (b) EPDS score 10+, (c) EPDS score 13+, (d) EPDS score 13 +and thoughts of self-harm (unadjusted analyses). (3) EPDS 13+, and (4) thoughts of self-harm with EPDS disadvantage of postcode, remoteness of postcode 13+. To assess any potential selection bias in the and household size. We have used the same weights respondent sample included in this report, we also in our previous work (Allen et al., 2013) see present results unadjusted as well as backweighted Supplementary Figure 1 for the distribution of inverse using inverse probability weighting (Little & Rubin, probability weights used. All estimates are presented 2019). Backweighting was done in terms of age, history with 95% confidence intervals of asthma (a primary outcome of interest in the BIS To examine the potential role of previous mental cohort during recruitment), socio-economic health difficulties, we assessed associations between AUSTRALIAN JOURNAL OF PSYCHOLOGY 581 Table 2. Subgroup analyses of current EPDS scores by past depressive symptoms. Subgroup Outcome, % (95%CI), EPDS <10 at 2 and 4 years EPDS 10+ at 2 or 4 years EPDS 10+ at 2 and 4 years EPDS score 10+ 23.4 (17.6–30.1) 44.2 (30.5–58.7) 82.6 (61.2–95.0) EPDS score 13+ 11.5 (7.3–16.8) 23.1 (12.5–36.8) 56.5 (34.5–76.8) Thoughts of self-harm 2.6 (0.9–6.0) 11.5 (4.4–23.4) 30.4 (13.2–52.9) Adjusted for characteristics at conception reported in Table 1 Table 3. Univariate associations between EPDS scores and household, family, community and employment characteristics. EPDS score 10+ EPDS score 13+ Thoughts of self-harm Characteristics RR 95% CI p RR 95% CI p RR 95% CI p Days since lockdown started (24 March 2020) 0.98 0.96–1.01 0.2 0.98 0.94–1.02 0.3 0.99 0.93–1.05 0.7 Household characteristics Number of adults in household 1.10 0.87–1.38 0.4 0.75 0.38–1.50 0.4 0.46 0.19–1.12 0.09 Number of children in household 1.09 0.89–1.33 0.4 1.13 0.80–1.58 0.5 1.63 0.93–2.85 0.08 Number of people in household (adults and children) 1.07 0.96–1.18 0.2 0.96 0.83–1.10 0.5 0.98 0.86–1.12 0.8 In a relationship 1.04 0.61–1.78 0.9 0.69 0.35–1.38 0.3 0.36 0.15–0.86 0.02 Lives with biological father of the child/ren 0.42 0.35–0.50 <0.001 0.59 0.33–1.05 0.07 0.45 0.18–1.10 0.08 Most socio-economically disadvantaged tertile 1.02 0.69–1.50 0.9 1.25 0.72–2.19 0.4 1.72 0.75–3.96 0.2 Lives in a regional area 0.89 0.65–1.24 0.5 0.90 0.54–1.50 0.7 0.71 0.29–1.72 0.4 Employment 0.2 0.5 0.5 Employed/job keeper, partner employed/job keeper ref ref ref Employed/job keeper, partner unemployed/no partner 3.07 0.50–18.85 1.38 0.21–9.29 1.07 0.16–7.15 Not working, partner employed/job keeper 1.42 0.20–9.82 0.56 0.06–5.08 - Not working, partner unemployed/no partner 2.54 0.42–15.21 1.47 0.25–8.76 0.62 0.11–3.52 Pre-COVID-19 social support See parents at least once per week 0.89 0.64–1.25 0.5 1.15 0.64–2.06 0.6 0.62 0.26–1.46 0.3 See inlaws at least once per week 1.51 1.11–2.06 0.008 1.60 0.98–2.60 0.06 1.41 0.63–3.15 0.4 See other family at least once per week 1.26 0.93–1.70 0.1 1.25 0.78–2.00 0.4 0.61 0.25–1.51 0.3 See friends at least once per week 1.22 0.85–1.77 0.3 0.95 0.56–1.61 0.8 0.67 0.29–1.52 0.3 Stress and substance abuse Perceived stress scale score 1.11 1.06–1.15 <0.001 1.13 1.05–1.22 <0.001 1.17 1.01–1.36 0.04 Alcohol more than once per week 0.98 0.71–1.34 0.9 1.39 0.86–2.25 0.2 0.94 0.42–2.11 0.9 Binge drinking ever 0.95 0.69–1.30 0.8 1.25 0.75–2.08 0.4 0.70 0.31–1.60 0.4 Smoking (at 4 years) 0.49 0.14–1.67 0.3 - - Major life events in the past 12 months Suffered a serious illness, injury, or assault 1.35 0.82–2.21 0.2 1.18 0.49–2.86 0.7 0.69 0.10–4.66 0.7 Close relative with illness, injury or assault 1.02 0.70–1.49 0.9 1.00 0.54–1.83 0.9 0.74 0.24–2.34 0.6 Parent, partner, or child died 2.83 1.18–6.78 0.02 1.39 0.30–6.39 0.7 2.77 0.62–12.44 0.2 Close family friend or relative died 1.43 1.04–1.96 0.03 1.61 0.96–2.70 0.07 0.89 0.28–2.82 0.8 Broke off a steady romantic relationship 1.39 0.79–2.44 0.3 2.51 1.37–4.60 0.003 3.69 1.36–10.05 0.01 Serious problem with a friend, neighbour, or relative 1.48 1.04–2.09 0.03 0.88 0.39–1.98 0.8 0.77 0.19–3.06 0.7 Seeking work unsuccessfully for more than a month 1.36 0.80–2.30 0.3 0.80 0.23–2.86 0.7 1.07 0.17–6.89 0.9 Major financial crisis 1.52 0.99–2.35 0.06 2.26 1.15–4.42 0.02 4.38 1.93–9.91 <0.001 Problems with police and court appearance - - - Something valuable lost or stolen 3.34 1.84–6.04 <0.001 - 31.45 5.34–185.26 <0.001 Someone in household had an alcohol problem 1.05 0.39–2.84 0.9 1.10 0.20–5.94 0.9 2.51 0.52–12.18 0.3 Someone in household had a drug problem - - - Serious disappointment in work or career 1.44 1.03–1.99 0.03 1.70 1.03–2.81 0.04 1.00 0.37–2.70 0.9 Thought would lose job soon 1.78 1.27–2.51 <0.001 2.23 1.34–3.68 0.002 5.25 2.26–12.19 <0.001 Lost job (not from choice) 1.29 0.87–1.90 0.2 2.10 1.25–3.53 0.005 3.39 1.54–7.45 0.002 Analyses are adjusted for 2 year or 4 year EPDS score 10+ previous postnatal depressive symptoms (no ele- Binomial regression was used to estimate risk vated score, elevated at two or four years, elevated ratios for the four dichotomous depression out- at both two and four years) and the current EPDS comes, prior mental health status, and contempor- score. ary contextual factors and major life events such as To examine the potential role of current context, we job loss (Wacholder, 1986). Population attributable assessed associations between household, family, fractions (Greenland & Drescher, 1993) were calcu- community and employment characteristics relevant lated for associations between major life events to the pandemic and depression, adjusting for an that had (i) increased in prevalence from child EPDS 10+at the two year or four year reviews (termed age four to age seven-eight, and were also (ii) past depressive symptoms). associated with depressive symptoms in Table 4. 582 A. LOUGHMAN ET AL. Table 4. Major life events that increased in prevalence at 7 year review. Due to Prevalence, n (%) Population attributable fraction, % (95% CI) COVID-19 Thoughts of Year 4 Year 7 N (%) EPDS score 10+ EPDS score 13+ self-harm Major life event N = 323 N = 346 p % 95% CI % 95% CI % 95% CI Parent, partner, or child died 12 (4) 5 (1) 0.02 1 (<1) 2.53 −0.95,5.89 0.58 −2.61,3.66 3.14 −4.24,9.99 Close family friend or relative died 79 (25) 64 (18) 0.06 2 (<1) 6.82 −0.20,13.34 9.39 −3.03,20.31 −1.89 −21.07,14.25 Broke off a steady romantic relationship 5 (2) 9 (3) 0.6 0 (0) 1.06 −1.06,3.14 4.05 0.12,7.83 7.62 −2.57,16.80 Serious problem with a friend, neighbour, 22 (7) 32 (9) 0.2 1 (<1) 4.83 −0.18,9.59 −1.30 −9.57,6.35 −2.72 −16.55,9.47 or relative Major financial crisis 17 (5) 17 (5) 0.9 7 (2) 1.80 −0.46,4.01 4.20 −0.80,8.95 10.39 −0.08,19.77 Something valuable lost or stolen 7 (2) 8 (2) 0.8 0 (0) 5.51 0.84,9.95 22.91 21.86,23.95 47.99 −34.26,79.85 Serious disappointment in work or career 34 (11) 53 (15) 0.07 19 (5) 7.17 −0.20,14.00 11.23 −1.78,22.58 −0.05 −21.54,17.64 Thought would lose job soon 17 (5) 44 (13) 0.006 32 (9) 10.10 3.07,16.62 15.29 2.95,26.06 40.35 11.14,59.96 Lost job (not from choice) 12 (4) 35 (10) 0.001 28 (8) 2.74 −2.01,7.26 9.79 0.61,18.12 19.30 −0.06,34.91 Population attributable fraction (PAF) estimates are provided for factors associated with outcomes in Table 3 and indicate how much, in the current wave, these outcomes could potentially be attributed to factors associated with these life events, if the significant associations shown in Table 3 were causal in nature. Adjusted for past depression history. Results with diagnosable depression (EPDS score 13+), com- pared to 11·5% (95% CI: 7·3–16·8) of those who did not Table 1 outlines the sociodemographic characteristics have past depressive symptoms (p < 0 · 001). Those of the 2020 review sample and the inception cohort at with past depressive symptoms were therefore more birth. While they are similar in most respects, the cur- likely to have depressive symptoms in the 2020 lock- rent lockdown sample had a higher proportion of down. See Table 2 for subgroup analyses of current mothers with university level education (63% vs 51%), symptoms by past depressive symptoms. were less likely to live in a disadvantaged postcode, Of the household and related social contextual fac- and had fewer older siblings in the house at birth than tors considered, an important factor was living with the full inception cohort. Questionnaires were com- the biological father (of the child/ren), associated with pleted an average of 42 days (SD: 6 days) since lock- reduced risk of EPDS 10+ (RR 0·42, 95% CI: 0·35–0·50, down began. No families reported positive COVID-19 p < 0 · 001), with weaker evidence of associations with tests. EPDS 13+and thoughts of self-harm (RR 0.59, 95% CI: 0.33–1.05, p = 0.07; RR 0.45, 95% CI 0.18–1.10, p = 0.08). The numbers of other household residents were not Increased prevalence of depressive symptoms in significantly associated with EPDS score. The death of the current review a “parent, partner or child” or “close family friend or Depressive symptoms were more prevalent in the relative” were both strongly associated with EPDS 10+ current lockdown sample than at any prior wave of (RR 2·83, 95% CI: 1.18–7.68, p = 0.02; RR 1·43, 1.04–1.96, assessment (antenatal, four week, two year or p = 0.03), however these associations were not consis- four year reviews), with 30.6% (95% CI: 25·8–35·8) tent across EPDS 13+and thoughts of self-harm, or of the sample scoring EPDS 10+and 16·2% (95% CI: following adjustment for backweighting. 12·5–20·5) scoring EDPS 13+ (Panels 2 and 3, The prevalence of recent adverse life events, parti- Figure 1). Furthermore, thoughts of self-harm were cularly pertaining to employment or finance present in 7.2% (95% CI: 4··7–10·5) of the sample increased from child age four to age seven-eight. (Panel 1, Figure 1). Thus, there was a high burden of The proportion of respondents concerned about depressive symptoms and 1 in 14 had thoughts of imminent job loss more than doubled (5% to 13%, self-harm. p = 0 · 006) as did the proportion who had already lost their jobs (4% to 10%, p = 0 · 002; see Table 4). Most respondents stated that these two employ- Previous mental health difficulties, demographic ment-related events had occurred due to COVID-19 and lockdown characteristics associated with (73% and 80%, respectively). Both threat of job loss elevated depressive symptoms and actual job loss were positively associated with At child age seven to eight years, coinciding with lock- EPDS 13+ (RR 2·23, 95% CI: 1·34–3·68, p = 0 · 002; RR down, 56·5% (95% CI: 34·5–76·8) of the respondents 2·10, 95% CI: 1·25–3.53, p = 0 · 005; respectively) and who had past depressive symptoms at two prior time- with thoughts of self-harm (RR 5·25, 95% CI: 2·26– points reported current symptom burden concordant 12·19, p < 0.001; 3·39, 1·54–7·45, p = 0 · 002, AUSTRALIAN JOURNAL OF PSYCHOLOGY 583 respectively). The population attributable fractions planning, more than half of those who had previously from these employment factors indicate that the experienced depressive symptoms reported depres- impacts on depression could be considerable (threat sive symptoms concordant with diagnosable depres- of job loss: 15.29%, 95% CI: 2.95–26.06; actual job sion following the lockdown. loss: 9.79, 95% CI: 0.61–18.12) although the estimates Of the household, social and adverse life events are imprecise. Overall, as expected, depressive symp- factors we considered, currently living with the biolo- toms were positively associated with higher per- gical father of the child/ren was protective. Threats to ceived stress scores. job security, commonly reported by participants as Backweighting by inverse probability weighting did occurring due to the pandemic, were associated with not materially change the estimates of prevalence, higher depressive symptoms. These work-related subgroup analyses or associations with current house- major life events increased in prevalence relative to hold, social and adverse life events characteristics the four-year review. However, the greatest risk for reported here, with the exception of death of reporting either EPDS 10+or 13+was in those mothers a “parent, partner or child” or “close family friend or with previous mental health difficulties, echoing other relative” (Supplementary Tables 1–4). reports of vulnerability in this group during times of stress (Milgrom et al., 2008). This study did not include questions regarding domestic violence, which has Discussion since emerged as highly relevant to mental health Our results indicate that around one in three mothers and wellbeing during COVID-19 lockdown restrictions with primary school aged children experienced ele- (Neil, 2020). vated pre-clinical depressive symptomology (EPDS 10 Emerging reports of psychological symptoms in +), close to 2 in 10 (16%) experienced more severe adults from the general population completing online depression (EDPS 13+), and 1 in 14 experienced surveys from Australia, the UK and China suggest thoughts of self-harm. There are no published data elevated symptoms relative to population norms dur- on depression in mothers of early primary school ing COVID-19 lockdowns (Newby, 2020; Shevlin et al., aged children, from prior to and/or during the pan- 2020; Wang et al., 2020). A meta-analysis of eight demic, with which our results can be compared. This studies of pregnancy or postpartum women from limits our capacity to infer that the observed increase different countries during the pandemic showed in adverse maternal mental health is in fact caused by strong evidence for increased anxiety symptoms, the COVID-19 lockdown. and weaker evidence for increased depressive symp- At more severe clinical levels of depression, as indi- toms (Hessami et al., 2020). None of the above studies cated by EPDS 13+scores, the prevalence we report is were in representatively sampled or pre-existing similar to that reported in the pre-COVID literature. The cohorts, leaving open the possibility of selection most comparable estimates come from a large, popu- bias. Selection bias into the initial cohort or early lation-based Australian study conducted in 2014 which responders in the current wave have been investi- reported approximately 14% of first-time mothers had gated using inverse probability weighting and do EPDS scores of 13+when their children were four-years not appear to be substantial. However, the absence old, and again when their children were ten (using the of the counterfactual (i.e., those not exposed to lock- equivalent cut-off score on the Center for down) precludes causal inference about the effects of Epidemiologic Studies Depression Scale at this later lockdown. It is also unclear whether the prevalence of follow-up) (Brown et al., 2020; Woolhouse et al., pre-clinical depressive symptoms (EPDS 10+) requires 2015). The prevalence of EDPS 13+scores in parents intervention or reflects understandable adjustment of primary school aged children during lockdown was difficulties in the context of social adversity, which only marginally higher (16·2%). However, we observed are likely to be transient and self-limiting. Ongoing a clear increase between the four and the seven-eight research regarding the mental health of men as well year reviews for both pre-clinical and clinical level as women who are not mothers, and across other age symptoms and a parallel increase in adverse life events groups and geography during the COVID-19 pan- such as job insecurity, indicating a potential effect of demic will be required to generalise and extend the COVID-19 lockdown. Nonetheless, it is impossible upon these findings. to make robust causal inferences in the absence of The Australian government has attempted to pro- counterfactual data i.e., measures conducted among tect household incomes through the JobKeeper pro- women who were not exposed to the COVID-19 lock- gram. Even in this context, the impact of short-term down. Of particular importance to clinical services work-related events on mental health appears 584 A. LOUGHMAN ET AL. substantial and these findings need to be taken into References account for future policy decisions as the COVID-19 Aida, J., Kawachi, I., Subramanian, S., & Kondo, K. (2013). pandemic continues. Our findings also indicate the Disaster, social capital, and health. In I. Kawachi, S. Takao, need to make mental health services accessible to & S. V. Subramanian (Eds.), Global perspectives on social those with a mental health history. In countries that capital and health (pp. 167–187). Springer. Allen, K. J., Koplin, J. J., Ponsonby, A.-L., Gurrin, L. C., Wake, M., have experienced different COVID-19 infection rates, Vuillermin, P., Martin, P., Matheson, M., Lowe, A., Robinson, stricter lockdowns and different policy responses, the M., Tey, D., Osborne, N. J., Dang, T., Tan, H.-T. T., Thiele, L., effects on mental health may be greater than those Anderson, D., Czech, H., Sanjeevan, J., Zurzolo, G., . . . reported here. Further, the mental health burden is Dharmage, S. C. (2013). Vitamin D insufficiency is asso- likely to have increased in the context of the second ciated with challenge-proven food allergy in infants. Journal of Allergy and Clinical Immunology, 131(4), 1109– wave and stricter lockdown in Victoria from July to 1116.e1106. https://doi.org/10.1016/j.jaci.2013.01.017 October 2020. Brown, S. J., Mensah, F., Giallo, R., Woolhouse, H., Hegarty, K., Here we show the first data from a representative Nicholson, J. M., & Gartland, D. (2020). Intimate partner sample of Australian mothers, revealing a strong pat- violence and maternal mental health ten years after a first tern of increasing depressive symptoms as their chil- birth: An Australian prospective cohort study of first-time dren age from birth to seven/eight years of age, the mothers. Journal of Affective Disorder, 262, 247–257. https://doi.org/10.1016/j.jad.2019.11.015 latter being coincident with the first COVID-19 lock- Buist, A. E., Milgrom, J., Barnett, B. E. W., Pope, S., Condon, J. T., down. We show that mothers most in need of support Ellwood, D. A., Boyce, P. M., Austin, M.-P. V., & Hayes, B. A. at child age seven/eight are those with prior depres- (2002). To screen or not to screen — That is the question in sion. Whilst causality cannot be inferred, findings coin- perinatal depression. Medical Journal of Australia, 177(S7), cide with the first lockdown due to COVID-19, and may S101–S105. https://doi.org/10.5694/j.1326-5377.2002. tb04866.x reflect the impact of physical distancing, and other Cox, J. L., Chapman, G., Murray, D., & Jones, P. (1996). measures, to control the spread of disease. Our find - Validation of the Edinburgh Postnatal Depression Scale ings echo established calls for increased mental health (EPDS) in non-postnatal women. Journal of Affective support for women across the early childhood period, Disorders, 39(3), 185–189. https://doi.org/10.1016/0165- and highlight the importance of supporting parents of 0327(96)00008-0 school age children. We further suggest that measures Cox, J. L., Holden, J. M., & Sagovsky, R. (1987). Detection of postnatal depression: Development of the 10-item taken to deal with the COVID-19 pandemic may inten- Edinburgh Postnatal Depression Scale. The British Journal sify risk in this group and require additional support, of Psychiatry, 150(6), 782–786. https://doi.org/10.1192/bjp. including targeted programs for those at increased risk 150.6.782 (Galea et al., 2020). Increasing the availability of com- Eastwood, J. G., Phung, H., & Barnett, B. (2011). Postnatal munity supports more broadly, such as social support depression and socio-demographic risk: Factors associated with Edinburgh Depression Scale scores in a metropolitan community groups, telephone counselling services area of New South Wales, Australia. Australian and New and other neighbourhood initiatives may also be rele- Zealand Journal of Psychiatry, 45(12), 1040–1046. https:// vant preventative action for mental health during the doi.org/10.3109/00048674.2011.619160 pandemic. Edmondson, O. J., Psychogiou, L., Vlachos, H., Netsi, E., & Ramchandani, P. G. (2010). Depression in fathers in the postnatal period: Assessment of the Edinburgh Postnatal Disclosure statement Depression Scale as a screening measure. Journal of Affective Disorder, 125(1–3), 365–368. https://doi.org/10. No potential conflict of interest was reported by the author(s). 1016/j.jad.2010.01.069 Galea, S., Merchant, R. M., & Lurie, N. (2020). The mental health consequences of COVID-19 and physical distancing: The Funding need for prevention and early intervention. JAMA Internal Medicine, 180(6), 817–8. https://doi.org/10.1001/jamain This work was supported by the Deakin University; Murdoch ternmed.2020.1562 Children’s Research Institute; National Health and Medical Greenland, S., & Drescher, K. (1993). Maximum likelihood Research Council; The Shepherd Foundation; Barwon estimation of the attributable fraction from logistic Health; Perpetual Trustees. models. Biometrics, 49(3), 865–872. Hessami, K., Romanelli, C., Chiurazzi, M., & Cozzolino, M. (2020). COVID-19 pandemic and maternal mental health: ORCID A systematic review and meta-analysis. Journal of Maternal Fetal and Neonatal Medicine, 1–8. https://doi.org/10.1080/ Amy Loughman http://orcid.org/0000-0002-0257-1443 14767058.2020.1843155 Michael Berk http://orcid.org/0000-0002-5554-6946 AUSTRALIAN JOURNAL OF PSYCHOLOGY 585 Holmes, T. H., & Rahe, R. H. (1967). The social readjust- Shevlin, M., McBride, O., Murphy, J. (2020). Anxiety, depres- ment rating scale. Journal of Psychosomatic Research, 11 sion, traumatic stress, and COVID-19 related anxiety in the (2), 213–218. https://doi.org/10.1016/0022-3999(67) UK general population during the COVID-19 pandemic. 90010-4 BJPsych Open, 6(6), e125–e. Institute, B. D. (2020). Mental health ramifications of COVID-10: Stay at Home Directions (No 6), Section 200 C.F.R. (2020). The Australian context. Report. Sutton, B. (2020). Directions issued by Victoria’s Chief Health Little, R. J., & Rubin, D. B. (2019). Statistical analysis with Officer . Victorian State Government. https://www.dhhs.vic. missing data (Vol. 793). John Wiley & Sons. gov.au/victorias-restriction-levels-covid-19 Matthey, S., Henshaw, C., Elliott, S., & Barnett, B. (2006). Tsirtsakis, A. (2020). Fears of post-pandemic ‘tsunami of health Variability in use of cut-off scores and formats on the problems’. RACGP. https://www1.racgp.org.au/newsgp/clini Edinburgh Postnatal Depression Scale: Implications for clin- cal/gps-fear-post-pandemic-tsunami-of-health-problems ical and research practice. Archives of Womens Mental Health, Vuillermin, P., Saffery, R., Allen, K. J., Carlin, J. B., Tang, M. L., 9(6), 309–315. https://doi.org/10.1007/s00737-006-0152-x Ranganathan, S., Burgner, D., Dwyer, T., Collier, F., Jachno, K., Milgrom, J., Gemmill, A. W., Bilszta, J. L., Hayes, B., Barnett, B., Sly, P., Symeonides, C., McCloskey, K., Molloy, J., Forrester, M., Brooks, J., Ericksen, D., & Buist, A. (2008). Antenatal risk & Ponsonby, A.-L. (2015). Cohort profile: The Barwon infant factors for postnatal depression: A large prospective study. study. International Journal of Epidemiology, 44(4), Journal of Affective Disorders, 108(1–2), 147–157. https:// 1148–1160. https://doi.org/10.1093/ije/dyv026 doi.org/10.1016/j.jad.2007.10.014 Wacholder, S. (1986). Binomial regression in GLIM: Neil, J. (2020). Domestic violence and COVID 19. Australian Estimating risk ratios and risk differences. American Journal for General Practitioners. https://www1.racgp.org. Journal of Epidemiology, 123(1), 174–184. https://doi. au/ajgp/coronavirus/domestic-violence-and-covid-19 org/10.1093/oxfordjournals.aje.a114212 Newby, J. M. (2020). Acute mental health responses during the Wang, C., Pan, R., Wan, X., Tan, Y., Xu, L., Ho, C. S., & Ho, R. C. COVID-19 pandemic in Australia. medRxiv (Preprint). (2020). Immediate psychological responses and associated medRxiv. https://doi.org/10.1101/2020.05.03.20089961 factors during the initial stage of the 2019 coronavirus Ribeiro, W. S., Bauer, A., Andrade, M. C. R., York-Smith, M., disease (COVID-19) epidemic among the general popula- Pan, P. M., Pingani, L., . . . Evans-Lacko, S. (2017). Income tion in China. International Journal of Environmental inequality and mental illness-related morbidity and resili- Research and Public Health, 17(5), 1729. https://doi.org/ ence: A systematic review and meta-analysis. The Lancet 10.3390/ijerph17051729 Psychiatry, 4(7), 554–562. https://doi.org/10.1016/S2215- Wenham, C., Smith, J., & Morgan, R. (2020). COVID-19: The 0366(17)30159-1 gendered impacts of the outbreak. The Lancet, 395 Royal College of Psychiatrists. (2020). Psychiatrists see alarm- (10227), 846–848. https://doi.org/10.1016/s0140- ing rise in patients needing urgent and emergency care and 6736(20)30526-2 forecast a ‘tsunami’ of mental illness [Press release]. https:// Woolhouse, H., Gartland, D., Mensah, F., & Brown, S. J. (2015). www.rcpsych.ac.uk/news-and-features/latest-news/detail/ Maternal depression from early pregnancy to 4 years post- 2020/05/15/psychiatrists-see-alarming-rise-in-patients- partum in a prospective pregnancy cohort study: needing-urgent-and-emergency-care?searchTerms Implications for primary health care. BJOG, 122(3), =tsunami 312–321. https://doi.org/10.1111/1471-0528.12837

Journal

Australian Journal of PsychologyTaylor & Francis

Published: Oct 2, 2021

Keywords: Mental health; maternal; pandemic; depression; psychiatry; mental disorders

There are no references for this article.