A cross-national study of factors associated with perinatal mental health and wellbeing during the COVID-19 pandemic

Background. Pregnant and postpartum women face unique challenges during the COVID-19 pandemic that may put them at elevated risk of mental health problems. However, few large-scale and no cross-national studies have been conducted to date that investigate modifiable pandemic-related behavioral or cognitive factors that may influence mental health in this vulnerable group. This international study sought to identify and measure the associations between pandemic-related information seeking, worries, and prevention behaviors on perinatal mental health during the COVID-19 pandemic. Methods and Findings. An anonymous, online, cross-sectional survey of pregnant and postpartum women was conducted in 64 countries between May 26 2020 and June 13 2020. The survey, available in twelve languages, was hosted on the Pregistry platform for COVID-19 studies (https://corona.pregistry.com), and advertised predominantly in social media channels and online parenting forums. Participants completed measures on demographic characteristics, COVID-19 exposure and worries, media exposure, COVID-19 prevention behaviors, and mental health symptoms including posttraumatic stress symptoms via the IES-6, anxiety/depression via the PHQ-4, and loneliness via the UCLA-3. Of the 6,894 participants, substantial proportions of women scored at or above the cut-offs for elevated posttraumatic stress (2,979 [43%]), anxiety/depression (2,138 [31%], and loneliness (3,691 [53%]). Information seeking from any source (e.g., social media, news, talking to others) five or more times per day was associated with more than twice the odds of elevated posttraumatic stress and anxiety/depression, in adjusted models. A large majority of women (86%) reported being somewhat or very worried about COVID-19. The most commonly reported worries were related to pregnancy and delivery, including family being unable to visit after delivery (59%), the baby contracting COVID-19 (59%), lack of a support person during delivery (55%), and COVID-19 causing changes to the delivery plan (41%). Greater worries related to children (i.e. inadequate childcare, their infection risk) and missing medical appointments were associated with significantly higher odds of posttraumatic stress, anxiety/depression and loneliness. Engaging in hygiene-related COVID-19 prevention behaviors (face mask-wearing, washing hands, disinfecting surfaces) were not related to mental health symptoms or loneliness. Conclusions. Clinically significant posttraumatic stress, anxiety/depression, and loneliness are highly prevalent in pregnant and postpartum women across 64 countries during the COVID-19 pandemic. Excessive information seeking and worries related to children and medical care are associated with clinically significant symptoms, whereas engaging in hygiene-related preventive measures were not. In addition to screening and monitoring mental health symptoms, reinforcing healthy information seeking, addressing worries about access to medical care and the well-being of their children, and strategies to target loneliness (e.g., online support groups) should be part of intervention efforts for perinatal women. Public and mental health interventions need to explicitly address the impact of COVID-19 on both physical and mental health in perinatal women, as prevention of viral exposure itself does not mitigate the mental health impact of the pandemic.


Introduction
responses from each of the countries with the highest number of COVID-19 cases at the time of 98 recruitment. Interested participants were invited to follow a link to take the survey. The survey 99 collected standard demographic data and included questions that addressed topics such as COVID-100 19 exposure and worries, lifestyle changes, media exposure, protective factors, and mental health. 101 Participants. Women who self-identified as being 18 years or older at the time of the survey and as 102 currently pregnant or having given birth within the past 6 months were eligible to participate. The 103 study was classified exempt by the Harvard Longwood Campus Institutional Review Board (HLC 104 IRB) per the regulations found at 45 CFR 46.104(d)(2) on the basis that it poses no greater than 105 minimal risk and that the recorded information cannot readily identify the subject (directly or 106 indirectly). The total number of participants at the close of enrollment was 7,562 individuals across 107 64 countries. 108 Measures 109  Participants were asked about the frequency (never, <1x/day, 2-114 4x/day, 5-8x/day, 9-16x/day, and >16x/day) of their interactions with various sources of information, 115 including the news, social media, and interpersonal discussions about COVID-19 using a measure 116 modified from other published studies [24,25]. For analyses, interactions were categorized as never, 117 <1x/day, 2-4x/day, 5+x/day. 118 COVID-19 Worries. Participants were asked to rate their overall level of worry about COVID-19 119 on a Likert Scale ranging from 1 for "not worried at all" to 4 for "very worried" [26]. They were then 120 asked to endorse fifteen specific worries on a list developed for this study. Exploratory factor analyses 121 were conducted to identify domains within the questionnaire, with oblimin rotation on a tetrachoric 122 correlation matrix due to the binary nature of the variables. Details of the factor analyses are presented 123 in the supplementary materials (S2 Text). Worries were categorized into the following domains: 124 social (parents/grandparents unable to visit, family unable to visit, not able to have a baby shower, 125 not able to attend a funeral), COVID-19 infection-related (participant or partner will bring infection 126 home, family or friends will get COVID-19), child-related (no adequate childcare, other children will 127 get COVID-19), delivery-related (partner not present during delivery, changes to delivery plan, 128 unborn baby will get COVID-19, not able to breastfeed), economic (significantly affect economic 129 situation/finances), and missing doctor appointments. 130 COVID-19 prevention behaviors. Participants were asked to endorse seventeen behaviors they had 131 engaged in to protect themselves from COVID-19 from a list developed for this study based on WHO 132 recommendations and media reports. Behaviors were classified into the following categories: 133 hygiene-related (mask-wearing, washing hands, disinfecting surfaces), physical distancing (avoiding 134 public places, restaurants and other people, canceling personal engagements, work or school and 135 working at home), canceling travel (for work or pleasure), stockpiling essential resources (food or  136 water, hand sanitizer, medication), postponing medical care, and prayer. 137 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted December 4, 2020. four-item inventory rated on a four-point Likert-type scale. Items are drawn from the first two items 140 of the 'Generalized Anxiety Disorder-7 scale' (GAD-7) and the 'Patient Health Questionnaire-8' 141 . The overall PHQ-4 score is a sum of the four items (0 = not at all, 1 = several days, 2 = 142 more than half the days, 3 = nearly every day). A PHQ-4 score of >=6 is considered clinically 143 significant. 144 Posttraumatic stress symptoms were assessed via a modified version of the Impact of Events Scale 145 -6 (IES-6). Participants were asked to report how bothered they were by each symptom from 0 (Not 146 at all) to 4 (Extremely) over the past seven days. The symptom statements were modified to assess 147 impact related to the COVID-19 pandemic. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 4, 2020. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 4, 2020. ; https://doi.org/10.1101/2020.12.03.20243519 doi: medRxiv preprint Native/Indigenous (<0%), more than one race (4%) or other/not specified (3%) with a small 201 number of women missing race information but who answered all other questions (1%). The 202 majority of women reported being married (65%), being at least a college graduate (73%), not 203 being an essential or healthcare worker (75%), and having health care coverage (71%). With 204 regard to COVID-19 exposure, the vast majority of women reported not having been tested for 205 COVID-19 (89%), not having been in contact with an individual who has or had COVID-19 (78%), 206 never having been diagnosed with COVID-19 (98%), and not being an essential or healthcare 207 worker (75%). 208 Mental health and loneliness. Substantial proportions of women scored at or above the risk cut- CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 4, 2020. ; https://doi.org/10.1101/2020.12.03.20243519 doi: medRxiv preprint Results for multivariable models of socio-demographic and COVID-19 exposure variables in 220 relation to IES-6, PHQ4 and UCLA-3 are presented in the supplementary materials (see Table A  221 in S1 Text). In fully adjusted models, only age was consistently associated with reduced risk of 222 clinically significant outcomes across measures at p<.001. Only "maybe" having been in contact 223 with someone with COVID-19 compared to "never" was consistently associated with increased 224 risk of clinically significant outcomes across measures at p<.001. 225 Information Seeking and Mental Health.  For all types of information seeking, there was a dose-response relation between information 236 seeking frequency and meeting the clinical threshold for both the IES-6 and PHQ-4, in fully 237 adjusted models. In fact, participants engaging in any type of information seeking five or more 238 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 4, 2020. ; https://doi.org/10.1101/2020.12.03.20243519 doi: medRxiv preprint times per day had more than twice the odds of meeting the clinical threshold for the IES-6 and 239 PHQ-4 ( Table 2) compared to those who did not engage in that type of information seeking. There 240 was also a dose-response relation between number of types of information seeking 5+ times a day 241 and meeting the clinical threshold on the IES-6 and PHQ-4, with individuals seeking information 242 via multiple sources showing more than twice the odds of distress. 243 Information seeking was less strongly and consistently related to meeting the risk threshold on the 244 UCLA-3. Checking news, social media, and mass communications increased the odds of 245 loneliness between 38 to 48% ( Table 2) for women doing so 5+ times a day compared to women 246 who engaged in information seeking less than once a day. Discussing with others was only 247 associated with increased loneliness when done 2-4x per day. 248 Worries and Mental Health. The prevalence of reported COVID-19 related worries is presented 249 in Fig 3 and  Using domains obtained from the factor analysis (see S2 Text), Table 3 presents the association  262 between mental health outcomes and the six COVID-19 worry categories. Child-related worries 263 and medical care worries were each associated with significantly higher odds of exceeding the 264 clinical threshold on the IES-6, PHQ4 and UCLA-3. Social worries were only associated with 265 higher odds of loneliness. Economic worries were associated with increased odds of meeting the 266 clinical threshold for IES-6 and PHQ4; infection and delivery worries were only significantly 267 associated with increased odds of meeting the clinical threshold on the IES-6. The prevalence of 268 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 4, 2020. ; https://doi.org/10.1101/2020.12.03.20243519 doi: medRxiv preprint PTSD, depression/anxiety, and loneliness consistently increased for every additional worry 269 category endorsed (Fig 4). 270   Table C in S1 Text. The 284 majority of women engaged in COVID-19 prevention behaviors recommended by public health 285 experts: 93% washed/sanitized hands several times per day, 85% wore a face mask, 83% avoided 286 crowds, 70% avoided eating in restaurants, 67% avoided contact with high-risk people, 64% 287 disinfected surfaces, and 55% canceled or postponed activities. This pattern of engagement was 288 largely consistent across pregnancy stage. Results of logistic regression models for IES-6, PHQ-4 289 and UCLA-3 in relation to types of COVID-19 prevention behaviors are presented in Table 4. 290 291 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 4, 2020. ; https://doi.org/10.1101/2020.12.03.20243519 doi: medRxiv preprint   Hygiene and travel behaviors were not related to clinically elevated IES-6, PHQ-4, or UCLA-3. 304 However, distancing and stockpiling behaviors as well as canceling doctor's appointments and 305 prayer were associated with increased odds of meeting the clinical threshold on the IES-6. None 306 of the COVID-19 related prevention behaviors were associated with increased odds of meeting the 307 PHQ-4 clinical threshold. Only canceling doctor's appointments was associated with increased 308 loneliness. 309 310 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

311
In this large-scale anonymous, online, international cross-sectional survey of pregnant 312 and postpartum women from 64 countries during the COVID-19 pandemic, the prevalence of 313 clinically elevated mental health symptoms were high, with 43%, 31%, and 54% of women 314 exceeding clinical risk thresholds for PTSD, depression/anxiety, and loneliness, respectively. 315 Excessive pandemic-related information seeking was the strongest correlate of adverse mental 316 health outcomes. Across any modality or type of media, information seeking five or more times 317 per day, was associated with a more than two-fold increased risk of clinical elevations in 318 posttraumatic stress and depression/anxiety. Moreover, there was no evidence that discussion 319 about COVID-19, whether with another person or in social media, reduced loneliness. The vast 320 majority of women in our study reported feeling somewhat to very worried about COVID-19, with 321 large proportions reporting pregnancy or birth specific worries such as lack of social support 322 around delivery. Child-related worries (inadequate childcare, other children will get COVID-19) 323 and worries about missing doctor appointments were independently associated with increased odds 324 of clinical elevations in all outcomes. Most women reported following public health science 325 supported COVID-19 hygiene behaviors (e.g., face mask wearing, washing hands). Such behaviors 326 were not associated with adverse mental health symptoms or loneliness. 327 Prevalence rates of psychiatric distress in our study exceed non-pandemic perinatal 328 studies of prenatal and postpartum depression and anxiety [18,19] and most general population 329 estimates during pandemics [37]. Findings are especially notable as a very small proportion of 330 women report COVID-19 infection and most women (> 75%) also reported that they were not 331 exposed to anyone with a known infection. Recent commentaries highlight that women may be 332 particularly vulnerable to the impact of the pandemic's effects and public health measures taken 333 to reduce its effects, e.g., may be more likely to lose employment due to child care demands given 334 school closures [38]. These considerations may be particularly salient for pregnant and postpartum 335 women, for whom pandemic associated worries and the physical distancing required for infection 336 control may exacerbate mental health problems and feelings of loneliness. Given the potential 337 implications for maternal and child health, acknowledging the pandemic's unique impact on 338 pregnant and postpartum women is central to tailoring and scaling public and mental health 339 intervention efforts to address their particular needs. 340 Our findings on information seeking are consistent with others that have found high 341 levels of pandemic-related information-seeking, whether social media or traditional news 342 sources, is associated with increased prevalence of anxiety and depression [24,25]. Our results 343 extend these findings to perinatal women and expand the modalities examined. Strikingly, we 344 found that even discussing COVID-19 with another person five or more times a day was 345 associated with a two-fold increase in posttraumatic stress and depression/anxiety. Discussing 346 COVID-19, whether with another person or in mass communications (e.g. Facebook, Twitter) 347 was not associated with reduced loneliness. In fact, there was evidence of higher loneliness in 348 persons who engaged in such discussions frequently, suggesting that interpersonal engagement 349 with others, while generally beneficial for mental health, may not be protective if strongly 350 focused on the pandemic itself. 351 Degree of worry about COVID-19 was associated with clinically elevated symptoms of 352 posttraumatic stress and depression/anxiety but not loneliness. With regard to specific worries, 353 child-related worries and missing medical appointments were consistently associated elevations 354 in posttraumatic stress, depression/anxiety and loneliness. At the time of this study, we 355 identified only one other study of pregnant or postpartum women that reported on specific 356 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted December 4, 2020. Our study findings should be interpreted in light of some limitations. The cross-sectional 377 nature of the study prevents any causal attributions between the factors we examined. For instance, 378 distancing behaviors and postponing medical care to prevent the contraction of COVID-19 (but 379 not changes to travel plans and hygiene behaviors) were associated with clinically significant 380 posttraumatic stress. Women who experience elevated posttraumatic stress symptoms may be 381 more likely to isolate themselves; social avoidance is part of the syndrome. Social distancing 382 efforts may also perpetuate or exacerbate posttraumatic stress [40]. Additionally, our study 383 findings are based on a convenience sample and thus not-representative of any country or region. 384 Women in our study are likely those more active on social media platforms, which was our primary 385 modality of recruitment. Accordingly, our study likely also does not represent a comprehensive 386 range of concerns and mental health needs of pregnant or postpartum women, such as those with 387 limited internet access. However, given the large sample size, the availability of the survey in 388 multiple languages, and the paucity of perinatal mental health data for women during the COVID-389 19 pandemic, our study provides initial information for future research and intervention efforts. 390 Conclusion 391 Pregnant and postpartum women are reporting clinically significant depression, anxiety, 392 and posttraumatic stress, which are higher than most available general population estimates of 393 psychiatric distress during the pandemic. Such high levels of mental health problems have 394 potential implications for women, and fetal and child health and development. Thus, efforts to 395 screen, monitor, and target a range of mental health symptoms, and other aspects of emotional 396 well-being, such as feeling of loneliness and worries, and coping behaviors (e.g., appropriate 397 information seeking) should be considered. Reinforcing the importance of appropriate 398 information seeking should be a key consideration in intervention efforts. Future studies may 399 also consider examining women's specific concerns and worries related to child well-being and 400 concerns about access to medical care. Finally, public and mental health interventions need to 401 explicitly address both the viral disease risks as well as the mental health risks associated with the 402 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted December 4, 2020. ; https://doi.org/10.1101/2020.12.03.20243519 doi: medRxiv preprint pandemic, as prevention of viral exposure itself may not mitigate the pandemic's mental health 403 impact. 404 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted December 4, 2020. ; https://doi.org/10.1101/2020.12.03.20243519 doi: medRxiv preprint