key: cord-1049753-isuuay9a authors: Ebert, Cara; Steinert, Janina I title: Prevalence and risk factors of violence against women and children during COVID-19, Germany date: 2021-06-01 journal: Bull World Health Organ DOI: 10.2471/blt.20.270983 sha: db11edaf361a0cb54c74a18850cb8fab23a77778 doc_id: 1049753 cord_uid: isuuay9a OBJECTIVE: To assess the prevalence and exacerbating factors of violence against women and children in Germany during the coronavirus disease 2019 pandemic. METHODS: We conducted a representative online survey with partnered women (18–65 years) between 22 April and 8 May 2020, when participants had been under lockdown for a month. We determined the prevalence of several forms of violence within the previous month using both direct elicitation and a list experiment. We conducted a multivariable logistic regression to assess the impact of pandemic-associated risk factors. FINDINGS: Of our 3818 survey respondents, 118 (3.09%; 95% confidence interval, CI: 2.54 to 3.64) reported incidents of physical conflict, 293 (7.67%; 95% CI: 6.83 to 8.52) reported emotional abuse, and 97 (6.58%; 95% CI: 5.31 to 7.85) of 1474 respondents with children reported child corporal punishment. We estimated that 3.57% (95% CI: −0.33 to 7.46) had non-consensual intercourse with their partner. Our regression analysis revealed an increased risk of physical conflict with home quarantine (odds ratio, OR: 2.38; 95% CI: 1.56 to 3.61), financial worries (OR: 1.60; 95% CI: 0.98 to 2.61), poor mental health (OR: 3.41; 95% CI: 2.12 to 5.50) and young (< 10 years) children (OR: 2.48; 95% CI: 1.32 to 4.64); we obtained similar results for other forms of violence. Awareness and use of pertinent support services was low. CONCLUSION: Our findings of an increased risk of domestic violence during the pandemic should prompt policy-makers to improve the safety of women and children. Interventions to alleviate risks factors and extend support services are required. The World Health Organization (WHO) declared the coronavirus disease 2019 (COVID-19) outbreak a public health emergency of international concern on 30 January 2020. 1 To curb the outbreak, many governments implemented social distancing interventions, such as school closures, requirements of working from home and restricting private contacts. Although social distancing regulations are necessary from a virological perspective, they may have unintended consequences and expose certain segments of the population to other physical and mental health risks. One of the most cited aspects in this regard is the rise in domestic violence against women and children. [2] [3] [4] [5] [6] [7] [8] Empirical evidence from numerous countries, including Argentina, India, Peru and the United States of America, USA, has revealed an increase in the number of help requests to domestic abuse and child protection helplines during the pandemic. [9] [10] [11] [12] Further studies document a rise in domestic violence-related emergency calls to the police in several countries in the European Union and in Mexico, the United Kingdom of Great Britain and Northern Ireland and the USA, [13] [14] [15] [16] [17] and higher admission numbers of abuse-related trauma patients in hospitals in South Africa and the United Kingdom. 18, 19 Systematic reviews and meta-analyses of prospective longitudinal studies have highlighted socioeconomic disadvantage, poor mental health, alcohol misuse by a partner, unplanned pregnancies and a history of childhood abuse as risk factors for domestic violence, while older age has been confirmed as a protective factor. [20] [21] [22] [23] In addition to these general risk factors, several COVID-19-specific mechanisms may increase the risk of domestic violence. First, home con-finement can limit a person's ability to escape potential perpetrators and seek social and professional support. 8, 24 Second, pandemic-induced economic pressures may exert a high level of financial distress and result in pecuniary losses for those on furlough or short-term work schemes, or who have become newly unemployed. Previous studies have revealed significant increases in domestic violence in the wake of economic recessions. [25] [26] [27] Third, the closure of day care centres and schools inflicts a care burden on parents, causing them to renegotiate the distribution of household tasks, creating further potential for conflict. 28 Lastly, social isolation, economic uncertainty and an increased care burden may have detrimental effects on mental health, 29 a central risk factor for domestic violence in normal times. 30, 31 From a representative sample of women surveyed online, we aim to estimate the prevalence of violence against women and children in Germany during the COVID-19 pandemic. We also aim to determine the pandemic-related and general risk factors that contribute to an increased risk of some types of violence. We conducted our online survey between 22 April and 8 May 2020, when all states in Germany were enforcing strict policies to contain the spread of the pandemic. From 10 March 2020 onwards, schools, kindergartens, stores, restaurants and other public places were closed, and social contacts were limited to a minimum. We enrolled and interviewed 3818 partnered women aged 18-65 years (all of whom provided written electronic consent) via the survey firm respondi Objective To assess the prevalence and exacerbating factors of violence against women and children in Germany during the coronavirus disease 2019 pandemic. Methods We conducted a representative online survey with partnered women (18-65 years) between 22 April and 8 May 2020, when participants had been under lockdown for a month. We determined the prevalence of several forms of violence within the previous month using both direct elicitation and a list experiment. We conducted a multivariable logistic regression to assess the impact of pandemicassociated risk factors. Findings Of our 3818 survey respondents, 118 (3.09%; 95% confidence interval, CI: 2.54 to 3.64) reported incidents of physical conflict, 293 (7.67%; 95% CI: 6.83 to 8.52) reported emotional abuse, and 97 (6.58%; 95% CI: 5.31 to 7.85) of 1474 respondents with children reported child corporal punishment. We estimated that 3.57% (95% CI: −0.33 to 7.46) had non-consensual intercourse with their partner. Our regression analysis revealed an increased risk of physical conflict with home quarantine (odds ratio In designing our study, we considered how domestic violence is a sensitive and stigmatized phenomenon and therefore prone to social desirability bias in self-reports. [33] [34] [35] To tackle this issue, we adopted a two-pronged approach of (i) direct elicitation about less severe forms of violence, namely verbal and physical conflict with, or emotional abuse from, a partner, or corporal punishment of children; and (ii) indirect elicitation through double list experiments to measure sexual violence and more severe forms of physical violence against women and children. In single list experiments, respondents are randomly assigned to one of two lists; one list consists of four innocuous statements (reference group) and the other list includes these same four statements plus an additional sensitive item (experimental group). Respondents are then asked to specify the number of presented statements that apply to them (e.g. "3 out of 5"), allowing the researcher to establish the prevalence of violence by comparing the average total number between the reference and the experimental group. To increase statistical power, 36 we employed a double list experiment in which all respondents were presented with two distinct lists per outcome of interest, one with and one without the sensitive item. Our survey took 15-20 minutes to complete, and respondents received a small financial incentive (an online shopping voucher) to participate. Since face-to-face debriefings after completion of the interview were not possible, we provided respondents with information about selected domestic abuse helplines and email contacts (see details in data repository). 37 Wherever possible, we adapted measures of violence from previous surveys conducted in Germany to ensure contextual relevance. We piloted individual violence questions and constructed list experiments in waves 6 and 12 of the German COVID-19 Snapshot Monitoring study, 38 and cross-correlated these with other measures of violence as well as presumed predictors for validation purposes. We also included a social desirability scale validated in the German context to assess its effect on the reporting rates of violence. 39 We provide all constructed variables in the online data repository. 37 Our survey also elicited information on COVID-19-specific stressors that may exacerbate violence risk. We used the validated short version of the depression and anxiety scale (Patient Health Questionnaire 4) to capture the current mental health status of respondents and their partners. 40 Violence against women and children during COVID-19, Germany Cara Ebert & Janina I Steinert included additional items to capture the more direct mental health impacts of the pandemic, such as physical anxiety symptoms linked to COVID-19 fears. Because we only interviewed women, we asked respondents to provide an assessment of their partner's mental health status. We captured financial distress through an adapted list of questions used previously in Australia, 41 as well as through reported actual financial losses as a result of the pandemic. We recorded whether respondents had been under home quarantine and, to assess the increased childcare burden caused by closures of day care centres and schools, we enquired about the age of children in the household and hours spent on childcare. Finally, we determined the awareness and use of existing support services for survivors of domestic violence in Germany. We aggregated measurement instruments of mental health, financial concerns and social desirability into continuous-scale scores by using principal component analysis to weight individual items. All scales showed good internal consistency (we obtained Cronbach's α of 0.81, 0.83, 0.84 and 0.69 for respondents' mental health, partners' mental health, financial concerns and social desirability, respectively). We tested hypothesized risk factors using a multivariable logistic regression model and individual-level binary outcome variables from direct elicitation (i.e. less severe forms of violence). We included general predictors of domestic violence, such as women's age and socioeconomic status (as captured by household income, level of education and employment status of both respondent and partner before the pandemic), in our model. We also controlled for household size, whether partners cohabitated and whether the respondent was employed in the health sector or other essential services. In cases where the respondent was not able to evaluate her partner's mental health, we substituted missing values using multiple imputation by chained equations. 42 We conducted all our statistical analyses using the software Stata version 16.0 (StataCorp, College Station, USA). Our study was approved by the ethics committee of the medical faculty at the Technical University of Munich (TUM, IRB 227/20 S). We summarize the sociodemographic characteristics of our study population in Table 1 . Of our 3818 study par-ticipants, 657 (17.21%) reported having been home-quarantined, 91 (2.38%) reported that either they or their partner had lost their job and 1091 (28.58%) were subjected to short-term work or furlough as a result of the pandemic. We observed that 716 (18.75%) respondents revealed worries about their own or their partner's job security, 380 (9.95%) Of our study population, a total of 967 (25.33%; 95% confidence interval, CI: 23.95 to 26.71) and 118 (3.09%; 95% CI: 2.54 to 3.64) women reported verbal and physical conflict, respectively, with their partner during the previous month. We noted that women were also exposed to emotional forms of abuse: 146 (3.82%; 95% CI: 3.22 to 4.43) indicated that they felt threatened by their partner; 85 (2.23%; 95% CI: 1.76 to 2.69) were confined within their homes; and 175 (4.58%; 95% CI: 3.92 to 5.25) reported being controlled in terms of restricted communication with contacts outside their homes. We learned that 97 of 1474 women (6.58%; 95% CI: 5.31 to 7.85) or another household member had corporally punished (one of) their children in the past month ( Table 2 ). The number of respondents who experienced multiple forms of violence is reported in the data repository. 37 The real prevalence of violence was likely underestimated due to misreporting and social desirability bias. The negative association between respondents' sensitivity to social desirability and disclosures of violence depicted in Fig. 1 and in the data repository 37 corroborates this. Based on double list experiment elicitation, we estimated that during the previous month the prevalence of physical violence was 1.53% (95% CI: −2.05 to 5.11), for non-consensual sex it was 3.57% (95% CI: −0.33 to 7.46) and for violence against children it was 1.97% (95% CI: −4.23 to 8.18) ( Table 2) . Compared with households not under quarantine, the risk of physical conflict was more than double in households under home quarantine (odds ratio, OR: 2.38; 95% CI: 1.56 to 3.61). Quarantine was also associated with a significantly higher risk of experiencing emotional abuse during the previous month in terms of being confined to the home (OR: 2.80; 95% CI: 1.70 to 4.60) or controlled (OR: 2.52; 95% CI: 1.79 to 3.54; Table 3 ). We noted an increased risk of some forms of violence for respondents in the highest quintile of the financial concerns scale, in terms of both physical conflict (Table 3) . Apart from a significantly higher odds of verbal conflict (OR: 1.07; 95% CI: 1.02 to 1.12), we found no association between the daily childcare burden, Table 3 ). In robustness checks, we obtained similar results when including the social desirability bias index as an additional control and when estimating risk from only pandemic-specific or general factors (see tables in data repository). 37 We predicted probabilities of less severe forms of violence by risk factor, while holding all other risk factors and covariates constant at means (Fig. 2 and estimated prevalence of violence by risk factors measured on continuous scales in the data repository). 37 As shown in Fig. 2 , we predict the probability of violence for women with hypothetical high-risk (Yes) and low-risk (No) profiles. The predicted probability of emotional and physical forms of violence was almost zero in the low-risk scenario. In the high-risk scenario, we calculated predicted probabilities of 25.17% (95% CI: 12.73 to 37.61) for physical conflict, 21.12% (95% CI: 8.46 to 33.79) to 26.02% (95% CI: 14.94 to 37.10) for emotional abuse, and 23.32% (95% CI: 13.79 to 32.85) for child corporal punishment. We observed similar patterns of risk factors for more severe forms of violence as elicited in the list experiments (data repository). 37 We further investigated the effect of young children in the home separately for women who either worked full-time, part-time or not at all in February 2020 (data repository). 37 The presence of young children was significantly associated with an increased risk of physical conflict and emotional abuse for women working full-time only, and with an increased risk of corporal punishment of children for women working full-time and part-time. We discovered that awareness of domestic violence help services was gen-erally low (Fig. 3) . Likewise, utilization of support services among violence survivors was low (Fig. 4) , ranging from 1.82% (95% CI: 0.37 to 3.27) for both counselling centres and the codeword "Mask 19" in pharmacies to 8.25% (95% CI: 2.67 to 13.82) for the parenting crisis line among women who reported the occurrence of violence against children in their homes. Our survey-based data have the advantage of being more suitable than administrative data for establishing the effect of the pandemic on the prevalence of domestic violence. 29, 43 Our findings of an increased risk of violence with pandemic-induced financial worries or poor mental health of either the respondent or her partner are confirmed by other studies: a survey conducted in Canada during the first COVID-19 lockdown found higher levels of violence among families who were unable to meet current financial obligations, 28 and a survey conducted in the USA in spring 2020 found that parents who had reported depression and anxiety symptoms within the previous two weeks exhibited a greater potential for child abuse. 44 We also observed that one of the most pronounced risk factors was the presence of young children in the home, corroborating a study based in the USA that showed the highest increase in calls to domestic violence helplines from such households. 12 Our observation that the risks of violence and conflict were higher during phases of home quarantine was verified by a study from Argentina, which found a lower prevalence of violence among women whose partners did not have to comply with a stay-at-home order. 11 Quarantine orders increase the time that partners spend together, often in the context of additional pressures such as childcare responsibilities, limited physical space and isolation from support networks outside the home. While the criminological theory of exposure reduction between intimates predicts a decline in violence, physical distancing regulations mechanically increase exposure between partners and thus violence risk. 28, 45 We found that women affected by violence were underutilizing the available support infrastructure. Potential barriers to accessing help services could include perceived stigma or a lack of privacy at home, particularly in the presence of a perpetrator. 46 In addition, some women might be less comfortable with email counselling than in-person Domestic violence counselling or shelter Support for parents and children CI: confidence interval. Note: We defined a victim as women who have been exposed to at least one form of violence in the past four weeks. The sample size for respondents classified as not being a victim was 3488, except for the category Support for parents and children, which was 1377. The sample size for respondents classified as being a victim was 330, except for the category Support for parents and children, which was 97. counselling, and were therefore alienated from support services as a result of the pandemic. In contrast, we found that awareness and use of help services for children at risk of violence was higher; this could imply that the stigma of seeking support is lower when women are not victims themselves. Our study had several limitations. First, the cross-sectional design of our study meant that we were unable to establish reliable estimates of the extent to which domestic violence has increased as a direct consequence of the pandemic's physical distancing laws. Comparisons with prevalence estimates from surveys conducted before the pandemic are inadequate, as these rely on previous-year or lifetime experiences rather than the past month. Second, the risk factors that we discuss are not causally identified and should therefore be interpreted as associations. This is particularly the case for mental health, which could be both a cause and a consequence of domestic violence. To partly address potential confounding, we controlled for a large number of arguably relevant and simultaneously operating factors. Third, although online surveys have several advantages, such as increased anonymity, they can be prone to selection bias. 47, 48 For example, women with controlling partners might have had difficulties participating in the survey. However, we expect that the association between risk factors and violence is less affected by potential selection bias than the prevalence estimate itself. Fourth, the benefits of the double list experiment in terms of violence disclosures and respondent protection come at the cost of reduced statistical efficiency; results obtained via indirect elicitation were therefore excluded from the regression analysis of risk factors. While we sought to account for possible underreporting of violence by using list experiments and social desirability controls, disclosures may still have been inhibited by fear of reprisal 49 or by posttraumatic amnesia. 50 Fifth, we relied on respondents for an assessment of the mental health status of their partners. However, survivors of violence might perceive their partners' mental condition more negatively than women who are not exposed to violence, which may lead to an upward bias of the coefficient. Our findings of an increased risk of domestic violence in times of crises should prompt policy-makers to improve the safety of women and children. In anticipation of future lockdowns in Germany and other countries, interventions to alleviate risk factors and extend support services for survivors of violence -including emergency childcare centres to reduce parental stress, state-provided financial relief packages to reduce financial concerns, and the provision of easily accessible phone-and internet-based mental health counselling -are urgently required. ■ Prevalencia y factores de riesgo de la violencia contra las mujeres y los niños durante la COVID-19, Alemania Objetivo Evaluar la prevalencia y los factores agravantes de la violencia contra las mujeres y los niños en Alemania durante la pandemia de la enfermedad del coronavirus 2019. Métodos Realizamos una encuesta online representativa con mujeres en pareja (18-65 años) entre el 22 de abril y el 8 de mayo de 2020, cuando las participantes llevaban un mes confinadas. Determinamos la prevalencia de varias formas de violencia en el mes anterior utilizando tanto la elicitación directa como un experimento de lista. Realizamos una regresión logística multivariable para evaluar el impacto de los factores de riesgo asociados a la pandemia. Resultados De las 3.818 encuestadas, 118 (3,09%; intervalo de confianza del 95%, IC: 2,54 a 3,64) informaron de incidentes de conflicto físico, 293 (7,67%; IC del 95%: 6,83 a 8,52) informaron de abuso emocional y 97 (6,58%; IC del 95%: 5,31 a 7,85) de los 1.474 encuestados con hijos informaron de castigos corporales infantiles. Se estimó que el 3,57% (IC 95%: -0,33 a 7,46) tuvo relaciones sexuales no consentidas con su pareja. Nuestro análisis de regresión reveló un mayor riesgo de conflicto físico durante la cuarentena (odds ratio, OR: 2,38; IC del 95%: 1,56 a 3,61), preocupaciones financieras (OR: 1,60; IC del 95%: 0,98 a 2,61), mala salud mental (OR: 3,41; IC del 95%: 2,12 a 5,50) y niños pequeños (< 10 años) (OR: 2,48; IC del 95%: 1,32 a 4,64); obtuvimos resultados similares para otras formas de violencia. El conocimiento y el uso de los servicios de apoyo pertinentes fueron bajos. Conclusión Nuestros resultados sobre un mayor riesgo de violencia doméstica durante la pandemia deberían impulsar a los responsables políticos a mejorar la seguridad de las mujeres y los niños. Se requieren intervenciones para aliviar los factores de riesgo y ampliar los servicios de apoyo. the -second -meeting -of -the -international -health -regulations -(2005) -emergency -committee -regarding -the -outbreak -of -novel -coronavirus -(2019 -ncov) COVID-19: the gendered impacts of the outbreak Danger in danger: interpersonal violence during COVID-19 quarantine Association of COVID-19 with intimate partner violence. Prim Care Companion CNS Disord Risks to children and young people during covid-19 pandemic policy -brief -covid -19 -and -violence -against -women -and -girls -addressing -the -shadow -pandemic COVID-19: a public health approach to manage domestic violence is needed. Lancet Public Health The hidden disaster of COVID-19: intimate partner violence COVID-19 and the rise of intimate partner violence. World Dev Unintended consequences of lockdowns: COVID-19 and the shadow pandemic 19 -Lockdowns -and -Domestic -Violence -Evidence -from -Two -Studies -in -Argentina .pdf COVID-19 and crime: effects of stay-athome orders on domestic violence The impact of the coronavirus lockdown on domestic violence Hidden violence: how COVID-19 school closures reduced the reporting of child maltreatment A spatiotemporal analysis of the impact of COVID-19 on child abuse and neglect in the city of Changing patterns of domestic abuse during COVID-19 lockdown Covid-19: EU states report 60% rise in emergency calls about domestic violence Trauma patterns during the COVID-19 lockdown in South Africa expose vulnerability of women Increased proportion of physical child abuse injuries at a level I pediatric trauma center during the Covid-19 pandemic Risk and protective factors for intimate partner violence against women: systematic review and meta-analyses of prospective-longitudinal studies A systematic review of risk factors for intimate partner violence Co-occurrence of intimate partner violence and child sexual abuse: prevalence, risk factors and related issues Intimate partner violence victimization and alcohol consumption in women: a systematic review and meta-analysis Family violence and COVID-19: increased vulnerability and reduced options for support Intimate partner violence in the Great Recession Increased incidence of nonaccidental head trauma in infants associated with the economic recession The Great Recession and the risk for child maltreatment Covid-19, family stress and domestic violence: remote work, isolation and bargaining power COVID-19 response measures and violence against children. Bull World Health Organ Risk for domestic violence: factors associated with perpetration and victimization WHO multi-country study on women's health and domestic violence against women. Geneva: World Health Organization Intimate violence and social desirability: a meta-analytic review Effects of social desirability on students' self-reporting of partner abuse perpetration and victimization Parent-child agreement on parent-to-child maltreatment What can we learn with statistical truth serum? Design and analysis of the list experiment Violence against women and children in Germany during the COVID-19 pandemic Open Science Foundation Germany COVID-19 snapshot monitoring (COSMO Germany): Wave II. PsychArchives Eine Kurzskala zur Erfassung des Gamma-Faktors sozial erwünschten Antwortverhaltens: die Kurzskala Soziale Erwünschtheit-Gamma A 4-item measure of depression and anxiety: validation and standardization of the Patient Health Questionnaire-4 (PHQ-4) in the general population The impacts of household financial stress, resilience, social support, and other adversities on the psychological distress of western Sydney parents Multiple imputation of missing values: further update of ice, with an emphasis on interval censoring Measuring and monitoring national prevalence of child maltreatment: a practical handbook. Geneva: World Health Organization Stress and parenting during the global COVID-19 pandemic Explaining the decline in intimate partner homicide: the effects of changing domesticity, women's status, and domestic violence resources The pursuit of gender equality Representativeness in online surveys through stratified samples Does survey format influence self-disclosure on sensitive question items? The fear of reprisal and the failure of victims to report a personal crime Structural brain alterations in female survivors of intimate partner violence We are grateful to the women who participated in this study. We thank Cornelia Betsch, Philipp Sprengholz, Lisa Felgendreff, Sarah Eitze, Jonas Restle, Maximilian Sprengholz, Nina Hansen, Marloes Huis and Tim Büthe.