Depression and post-traumatic stress symptoms two years post-rape and the role of early counselling: Rape Impact Cohort Evaluation (RICE) study

ABSTRACT

Background: Survivors of sexual violence are at higher risk of adverse mental health outcomes compared to those exposed to other interpersonal traumas.

Objective: To examine the trajectory of both post-traumatic stress disorder (PTSD) and depression as well as the role of early counselling over 24 months among rape survivors.

Method: The South African Rape Impact Cohort Evaluation (RICE) study enrolled women aged 16–40 years attending post-rape care services within 20 days of a rape incident (n = 734), and a comparison group (n = 786) was recruited from primary health care. Women were followed for 24 months; the main study outcomes were depression and PTSD. Reports of early supportive counselling by the exposed group were also included. The analysis included an adjusted joint mixed model with linear splines to account for correlated observations between the outcomes.

Results: At 24 months, 45.2% of the rape-exposed women met the cut-off for depression and 32.7% for PTSD. This was significantly higher than levels found among the unexposed. Although a decline in depression and PTSD was seen at 3 months among the women who reported a rape, mean scores remained stable thereafter. At 24 months mean depression scores remained above the depression cut-off (17.1) while mean PTSD scores declined below the PTSD cut-off (14.5). Early counselling was not associated with the trajectory of either depression or PTSD scores over the two years in rape-exposed women with both depression and PTSD persisting regardless of early counselling.

Conclusion: The study findings highlight the importance to find and provide effective mental health interventions post-rape in South Africa.

Antecedentes: Los sobrevivientes de violencia sexual corren un mayor riesgo de resultados adversos de salud mental en comparación con aquellos expuestos a otros traumas interpersonales.

Objetivo: Examinar la trayectoria tanto del trastorno de estrés postraumático (TEPT) como de la depresión, así como el papel de la terapia temprana durante 24 meses entre las sobrevivientes de violación.

Método: El estudio South African Rape Impact Cohort Evaluation (RICE) inscribió a mujeres de 16 a 40 años que asistían a servicios de atención post-violación dentro de los 20 días posteriores a un incidente de violación (n = 734), y se reclutó un grupo de comparación (n = 786) reclutado de salud primaria. Las mujeres fueron seguidas durante 24 meses; los resultados principales del estudio fueron la depresión y el TEPT. También se incluyeron informes de terapia de apoyo temprana por parte del grupo expuesto. El análisis incluyó un modelo mixto conjunto ajustado con splines lineales para controlar las correlaciones entre los resultados.

Resultados: A los 24 meses, el 45,2% de las mujeres expuestas a violaciones alcanzaron el punto de corte para depresión y el 32,7% para TEPT. Esto fue significativamente más alto que los niveles encontrados en el grupo no expuesto. Aunque se observó una disminución en la depresión y el TEPT a los 3 meses entre las mujeres que denunciaron una violación, las puntuaciones medias se mantuvieron estables a partir de entonces. A los 24 meses, las puntuaciones medias de depresión se mantuvieron por encima del límite de depresión (17.1), mientras que las puntuaciones medias de TEPT descendieron por debajo del límite de TEPT (14.5). La terapia temprana no se asoció con la trayectoria de las puntuaciones de depresión o TEPT durante los dos años en mujeres expuestas a violación, donde depresión y TEPT fueron persistentes independientemente de la terapia temprana.

Conclusión: Los hallazgos del estudio resaltan la importancia de encontrar y proporcionar intervenciones de salud mental efectivas después de una violación en Sudáfrica.

背景：与其他人际创伤暴露者相比，性暴力的幸存者遭受不良心理健康后果的风险更高。

目的：旨在考查强奸幸存者中创伤后应激障碍 (PTSD) 和抑郁的轨迹，以及 24个月以上早期咨询的作用。

方法：南非强奸影响队列评估 (RICE) 研究招募了在强奸事件发生后 20 天内接受强奸后护理服务的 16-40岁女性 (n = 734)，并从初级医护系统招募了一个对照组 (n = 786)。 对妇女进行了 24个月的随访；主要研究结果是抑郁和创伤后应激障碍。暴露组的早期支持性咨询报告也包括在内。分析包括一个调整后具有线性样条以解释结果之间相关观测的联合混合模型。

结果：在 24个月时，45.2% 遭受强奸的女性达到了抑郁的临界值，32.7% 的女性达到了 PTSD 的临界值。这显著高于无暴露群体中发现的水平。尽管报告强奸的女性在 3个月时抑郁和创伤后应激障碍有所下降，但此后平均评分保持稳定。在 24个月时，抑郁平均评分仍高于抑郁临界值 (17.1)，而平均 PTSD 评分下降至低于 PTSD 临界值 (14.5)。 在遭受强奸并存在持续抑郁和 PTSD的女性中，尽管接受早期咨询，早期咨询与两年来抑郁或 PTSD 评分轨迹无关。

结论：研究结果强调了在南非寻找并提供有效强奸后心理健康干预措施的重要性。

KEYWORDS:

• Rape
• depression
• post-traumatic stress disorder
• longitudinal research
• joint modelling
• linear splines

PALABRAS CLAVE:

• Violación
• depresión
• trastorno de estrés postraumático
• investigación longitudinal
• modelado conjunto
• splines lineales

关键词:

• 强奸
• 抑郁
• 创伤后应激障碍
• 纵向研究
• 联合模型
• 线性样条

1. Introduction

Survivors of rape and sexual assault are at higher risk of developing adverse mental health outcomes compared to individuals exposed to other trauma types (Elklit & Christiansen, 2013; Nickerson et al., 2013; Tiihonen Möller et al., 2014). Depression and post-traumatic stress disorder (PTSD) are the most prevalent types of mental health problems among survivors of rape (Dworkin, 2020; Kaminer et al., 2008). Women are more likely to be victims of sexual violence and have a higher relative risk for the development of depression and PTSD compared to men (Campbell et al., 2010; Elklit & Christiansen, 2013; Hyde & Mezulis, 2020).

A recent meta-analysis of prospective studies investigating the prevalence of PTSD in sexual assault survivors reported a prevalence rate of 74.6% for PTSD 1-month post-assault and a prevalence of 41.5% at 12-months post-rape (Dworkin et al., 2021). A meta-analysis comparing participants who reported a sexual assault at some point in their lifetime to those who have not reported a sexual assault showed a 12-month prevalence of 18% and a lifetime prevalence of 36% for PTSD. The lifetime odds of meeting PTSD criteria were 7.57 times higher in the group that experienced sexual assault. For depression, the 12-month prevalence was 24% and the lifetime prevalence was 39%. The lifetime odds of meeting criteria for depression were 3.1 times higher in the group that experienced sexual assault (Dworkin, 2020). Few studies have prospectively followed recently exposed rape survivors and their symptom persistence and recovery beyond 12-month post-rape (Dworkin, 2020; Dworkin et al., 2017). Most previous studies were conducted in high-income countries with limited information available on the prevalence of PTSD and depression in low- and middle-income countries (LMIC), characterised by resource constraints.

Effective and accessible mental health care post-rape has been identified as a critical gap in point-of-care services for rape survivors in LMIC (Abrahams & Gevers, 2017; Keynejad et al., 2020). In South Africa, immediate crisis counselling is offered to rape survivors who attend post-rape services i.e. Thuthuzela Care Centres (TCCs) who in partnership with non-government organisations (NGOs) provide crisis counselling. The same NGO also offer rape survivors long-term supportive counselling (Abrahams & Gevers, 2017). A concerted effort to provide holistic care to rape survivors has been made in the last decade through Thuthuzela Care Centres (TCCs) which are one-stop service centres that provide comprehensive services to victims of gender-based violence (GBV), including rape. The primary aim of the TCCs includes providing health care, reducing secondary victimisation (e.g. police or medical staff not recognising the rape as a crime, intimidation by family members or friends of the perpetrator, misconception around rape and stigmatising attitudes from significant others and the community), increasing conviction of perpetrators and decreasing time in the finalisation of legal cases. Mental health support at TCCs is divided by that offered at the time of reporting (crisis counselling) and long-term psychosocial support (in the follow-up period). Typically, auxiliary social workers (lay counsellors) provide immediate crisis counselling at the time of reporting the rape while trained social workers provide support in the follow-up period.

While counselling is available at TCCs, not much is known about its uptake and impact, especially when working with severe and stigmatised trauma types such as rape (Abrahams & Gevers, 2017). The 2016 audit of the TCC services confirmed that mental health care has not been sufficiently prioritised (Jordaan et al., 2016). A rapid appraisal of mental health care post-rape in the Western Cape region showed that the low levels of mental health care provision were driven by limited resources and poor capacity (e.g. minimal training and knowledge of the mental health needs of women who have experienced rape and addressing issues such as stigma, guilt, victim blaming and the need for prolonged support and care), resulting in few survivors receiving more than acute mental health support and poor continuity of care (Abrahams & Gevers, 2017).

A body of research underpins evidence-based guidelines for psychological interventions in the management of PTSD symptoms that can be initiated within a month of trauma (Bisson et al., 2019). A smaller body of research focuses on early interventions post-sexual assault, with a meta-analysis of seven studies showing that early interventions (within 3 months of the sexual violence) can be effective in alleviating the severity of PTSD (Oosterbaan et al., 2019; Short et al., 2020). However, another review showed mixed results for early interventions (Dworkin & Schumacher, 2018). Trauma-focused cognitive behavioural therapy (CBT), e.g. prolonged exposure therapy and cognitive processing therapy, has been found to be an effective early intervention for reducing adverse mental health outcomes associated with trauma exposure (Bisson et al., 2019; Keynejad et al., 2020; Yehuda et al., 2015). However, CBT is typically resource intensive requiring tertiary-level educated mental health professionals which is not currently offered in the public health sector in South Africa (Yehuda et al., 2015).

The Rape Impact Cohort Evaluation (RICE) study, located in Durban KwaZulu Natal Province, South Africa, provided an opportunity to first explore the prevalence, risk and symptom trajectory of depression and PTSD in a group of women who have experienced a rape, in comparison with women who have never experienced a rape, over a two year period. Second, the mediating role of early person-centred supportive counselling (within 6 months post-rape), offered at rape crisis centres, on long-term PTSD and depression symptom trajectories is explored. Third, the role of demographic factors (age, education, employment, residential area, relationship status, food security and community support), HIV status, childhood trauma, lifetime trauma, intimate partner violence, alcohol use and other psychosocial risk and protective factors (resilience, social support, perceived stress) is investigated in relation to depression and PTSD symptoms in a longitudinal naturalistic design.

2. Methods

2.1. Study design and setting

The Rape Impact Cohort Evaluation (RICE) study was conducted in and around the city of Durban in Kwa-Zulu Natal province, South Africa. The RICE study was a comparative cohort study of women exposed to non-consensual, penile-vaginal penetrative rape and a comparison group of women who have never been exposed to non-partner or intimate partner rape (Abrahams et al., 2017).

2.2. Participants and procedure

The primary aim of the study was to determine HIV acquisition in adult rape-exposed and unexposed women. Participants were cisgender women between 16 and 40 years of age. Children under 16 were excluded because they constitute a vulnerable group that requires additional ethical safeguards. The upper age limit of 40 was chosen because the HIV incidence is much lower in women over the age of 40 years. Participants in both groups were excluded if they had an intellectual disability or were more than 14 weeks pregnant or lactating, as any HIV-positive pregnant and lactating women at the time of study recruitment were managed under different clinical protocols. The majority of women were followed up to 24 months while the minimum follow-up period was 12 months and the maximum period was 36 months. Interviews were scheduled at 3 monthly intervals in the 1st year (3, 6, 9, 12 months) and at six monthly intervals thereafter (18, 24 months).

2.2.1. Rape-exposed women and typical services offered at post-rape care services

The rape-exposed women were care-seeking women, recruited from four post-rape care services (including three TCCs). Women may have reported the rape at a police service or they reported directly to the post-rape services. Typically, the first point of care is assistance from a trained auxiliary social worker who provides initial support and assists the rape survivor through the processes and procedures. Staff at the post-rape services are mainly nurses or trained social workers. If women agree a medical examination is completed according to a standard protocol for a sexual assault forensic exam. Women are offered an HIV test and if they are negative and the rape occurred within 72 h, they are offered HIV pre-exposure prophylaxis. A police investigating officer will also take an official statement if not yet done. At completion, women are provided with a date for a follow-up appointment for medical care related to HIV and sexually transmitted infections (STIs). Throughout all the procedures, the auxiliary social worker remains with the woman to provide supportive counselling and all women are offered the option to return for longer-term person-centred counselling.

The RICE fieldwork staff were present at the rape crisis centres and worked with the staff to identify women who meet our eligibility criteria which included having experienced a rape within the last 20 days. A 20-day window was chosen to allow for baseline confirmation and documentation of HIV status and assessment of acute stress reactions post-rape. Women who were interested in participating in the study were provided with additional information about the study, their contact details were recorded, and they were invited to the RICE research site to complete the research informed consent procedures and the baseline interviews. Women who were visibly distressed e.g. excessive crying, severe withdrawal or a dissociative state, were flagged by rape services staff and the fieldwork staff were asked not to approach the rape survivor for participation in the study. If there was an imminent risk for suicide/self-harm or the presence of psychosis or severe dissociation, women were referred to higher care or admitted to the psychiatry division at the closest hospital in the vicinity of the rape crisis centre.

We defined rape in our study according to the South African Sexual Offences Act of 2007. The Act defines a wide range of sexual offences with rape defined as the genital, anal or oral penetration of the victim (without his/her consent) by the perpetrator (Criminal Law (Sexual Offences and Related Matters) Amendment Act 32 of 2007, 2007). Adult rape can be described according to the perpetrator e.g. acquaintance rape also known as date rape (perpetrator is known to the victim), stranger rape (perpetrator not known to victim), spousal rape (perpetrator is a marital partner), or rape according to the circumstances such as compelled rape (when the perpetrator forces victims to have sex with each other), gang rape (multiple perpetrators) and corrective rape (rape as punishment for homosexuality). The absence of consent is the underlying theme of rape in the South African law, i.e. the rape survivor did not consent to the act (Criminal Law (Sexual Offences and Related Matters) Amendment Act 32 of 2007, 2007). For the purpose of this study, we limited the definition of rape to non-consensual penile-vaginal penetration which is the most commonly reported form of rape in South Africa. We were not allowed to collect information on the rape event (index rape) including the perpetrator. We can therefore not identify if the perpetrator was an intimate partner by a non-partner (non-partner rape).

2.2.2. Rape-unexposed women

The unexposed women were recruited from public health primary healthcare services in close proximity of the rape care services. Primary care clinics (i.e. family planning and well-baby clinics) were chosen as the recruitment site since the majority of women attending appointments at these clinics were within the 16–40-year age range and were sexually active which made them susceptible to contracting HIV. Women were approached by the fieldwork staff in the waiting room of the clinics and were offered participation in the study. Women expressing interest in participation were provided with further information about the study procedures and were screened for eligibility. Screening for lifetime rape included ever experiencing non-partner and intimate partner rape which were defined as non-consensual, penile-vaginal penetrative rape or forced sex. The contact details of eligible women were recorded, and they were invited to the RICE research site to complete informed consent procedures and the baseline interview. Eligible and ineligible women were provided with a list of service providers who offer support to victims of sexual assault and violence.

2.3. Counselling

Women attending rape crisis centres were invited to return to the centre for supportive person-centred counselling. The purpose of the counselling was to provide support to rape survivors, to address stigma, self-blame, shame and other unsettling emotions and cognitions, and to encourage healing and empowerment. The counselling was offered typically by Auxiliary Social Workers who completed a non-degree two-year Social Work training (this training was initiated to respond to the huge gap in psycho-social care in the country). The specialised training to support rape survivors included weekly sessions of 3.5 h over a period of 5–6 months. The training focussed on personal growth and interpersonal skills (self-concept, group and relationship dynamics, leadership, feelings and communication, values and attitudes, death and grieving and spirituality and religion) as well as basic counselling skills (empathy, listening and reflecting skills, reframing, problem management, action and behavioural change, substance abuse, domestic violence, crisis care and suicide prevention, child abuse, stress, self-care and referral procedures). The courses were followed by a 2-month internship in which trauma-specific counselling skills were explored and supervision was provided.

Rape survivors were given appointments to return to the post-rape care service at one week, three weeks and three months, primarily for HIV status monitoring. During these visits, they were often reminded about the availability of counselling, if they have not already seen the counsellor for follow-up visits, and especially if they have difficulty coping in the aftermath of the rape.

3. Ethical considerations

Project staff received extensive training in safety and ethical principles applicable when working with vulnerable populations such as victims of rape. The study was introduced as the Women’s Health and Well-Being Study to protect women from being identified as rape survivors. The project staff received training from the National Prosecuting Authority on what may or may not be discussed by the rape survivor. Staff did not ask rape-exposed women about the circumstances of the rape (e.g. how many men were involved, forms of violence used, where the rape occurred, who the perpetrator was) in an effort to protect the integrity of the legal process. Women were compensated for their travel expenses to and from the research site. They also received compensation for their time in the form of R80 for completing the baseline interview and an incremental addition of R20 for each follow-up visit. Ethical approval was granted by the South African Medical Research Council Ethics Committee and the study protocol has been published (Abrahams et al., 2017).

3.1 Measures

3.1.1. Demographic questionnaire

Sociodemographic characteristics of the sample were assessed at baseline. These included information on age; education level (completed 12 years of schooling vs not completed); employment status (employed vs not employed); area of residence (informal, settlement/formal, township/rural-semi-rural); relationship status (no relationship, non-cohabiting relationship, cohabitating/married); household food security (sometimes or often go without food, never/seldom go without food); community support (difficult to find R200 ($13) in an emergency vs. easy). The latter two variables were used as proxies to measure socio-economic status.

3.1.2. HIV testing

A rapid HIV test was completed at baseline and at each return visit. Positive tests were confirmed by an ELISA (enzyme-linked immunoassay) HIV viral load tests.

3.1.3. Rape, sexual assault and other traumas

All women in the rape-exposed group experienced a non-consensual, penile-vaginal penetrative rape. They were enrolled into the study within 20 days of the rape event and completed the baseline assessment. As per agreement with the NPA, we did not ask any further questions around the characteristics of the rape. In addition, the rape-exposed women could have reported previous experience of rape in childhood, partner rape or non-partner rape. None of the women in the unexposed group ever experienced a rape as an adult or a child. However, women in both groups could have experienced other sexual traumas e.g. unwanted childhood sexual experiences, unwanted intimate partner sexual experiences or attempted rape.

By nature of the study inclusion criteria, all women in the rape-exposed group were exposed to a DSM-IV criterion A qualifying trauma type. Exposure to a DSM-IV criterion A trauma type in the unexposed group was measured using the Life Events Checklist (LEC), the Childhood Trauma Questionnaire (CTQ) and items designed to measure the prevalence of intimate partner violence (IPV) and non-partner sexual violence. The study inclusion and exclusion are presented in Figure 1 along with (1) the number of participants exposed to at least one trauma type; (2) the mean number of different lifetime trauma types experienced, (3) the number of participants exposed to sexual traumas; (4) the number of participants exposed to childhood trauma i.e. neglect, emotional abuse, physical abuse and domestic violence; (5) the number of participants exposed to intimate partner emotional abuse, financial abuse or physical abuse; and (6) the number of participants exposed to a traumatic event measure by the Life Events Checklist.

Figure 1. Recruitment criteria and baseline trauma exposure stratified by rape exposed or rape unexposed group.

3.1.4. Life Events Checklist (LEC)

The LEC trauma types were adapted to the South African context (Weathers et al., 2013). The trauma types were (1) imprisonment, (2) civil unrest/war, (3) serious injury, (4) being close to death, (5) murder of family or friend, (6) unnatural death of family or friend, (7) murder of stranger, (8) torture, (9) robbed/carjacked at gunpoint/ knifepoint, and (10) kid-napping. Responses were recorded as 0 (‘no, have not experienced this trauma’) or 1 (‘yes, have experienced this trauma’). Total scores rage between 0 and 10.

3.1.5. Childhood Trauma Questionnaire (CTQ)

An adjusted version of the CTQ was used to measure experiences of childhood trauma before the age of 18 (Bernstein et al., 2003; Jewkes et al., 2010). The measure contained 14 items which measured neglect, emotional abuse, domestic violence, sexual abuse and physical abuse. Neglect was measured using two original items from the CTQ (item 1 and 4) and two new items ‘I lived in different households at different times’ and ‘I spent time outside the home and none of the adults knew where I was’. Emotional abuse was measured using two original items of the CTQ (item 3 and 14). Domestic violence was measured using a new item ‘I saw or heard my mother being beaten by her husband or boyfriend’. Sexual abuse was measured using one original item from the CTQ (item 24, item used to define unwanted childhood sexual experiences) and three new items ‘I had sex with a man who was more than 5 years older than me’, ‘I had sex with someone because I was threatened or frightened or forced’, and ‘I was forced to have sex against my will by a boyfriend’. The latter two sexual abuse items were used to define child rape. Physical abuse was measured using two of the original CTQ items (item 11 and 12) and a new item i.e. ‘I was beaten or physically punished at school by a teacher or headmaster’. Responses were measured on a 4-point Likert scale with response options ranging between 1 (never) and 4 (very often). The Cronbach alpha score was 0.77 which indicated acceptable reliability.

3.1.6. Intimate partner violence (IPV)

Lifetime and past year exposure to intimate partner violence (IPV) was based on measures used in South African studies and the WHO multi-country study (Garcia-Moreno et al., 2006; Machisa et al., 2011). Seven items measured emotional IPV e.g. ‘has a current or previous husband or boyfriend ever belittled or humiliated you in from of other people’. Four items measured economic abuse e.g. ‘has a current or previous husband or boyfriend ever prohibited you from getting a job, going to work, trading, earning money or participating in income generation projects’. Five items measured physical abuse e.g. ‘has a current or previous husband or boyfriend ever hit you with a fist or something else which could hurt you’. Four items were used to measure sexual IPV ‘has a current or previous husband or any boyfriend ever physically forced you to have sex when you did not want to’, ‘have you ever had sex with a current or previous husband or any boyfriend because you were afraid what he might do’, ‘has your current or previous husband or any boyfriend ever forced you to watch pornography when you did not want to’, and ‘has a current or previous husband or any boyfriend ever forced you to do something else sexual that you did not want to do’. The former two sexual IPV items were used to define partner rape and the latter two defined partner unwanted sexual experiences. Women in the rape-exposed group were asked not to refer to the recently reported rape when completing the sexual IPV items.

The incidence of emotional, economic, physical and sexual IPV was captured if any of the items in the subcategories were endorsed as happened once or more than once. Both lifetime and 12-month prevalence of IPV were recorded.

3.1.7. Non-partner sexual violence (NPSV)

Experiences of non-partner sexual violence were measured using four items. Two items were used to measure non-partner rape ‘how many times have you been forced or persuaded to have sex against your will by a man who was not your husband or boyfriend’ and ‘how many times were you forced to have sex with a man who was not a husband or boyfriend when you were too drunk or drugged to refuse’. One item measured non-partner attempted rape ‘how many times has a man who was not your husband or boyfriend tried to force or persuade you to have sex against your will but did not succeed’. One item measured non-partner gang rape ‘how many times have you been forced to have sex against your will with more than one man at the same time’. Women in the rape-exposed group were asked not to refer to the recently reported rape when completing the NPSV items. A positive response to one or more of the four questions was indicative of lifetime experiences of NPSV. Only lifetime experiences of NPSV were captured.

3.2. Depression, PTSD and counselling

3.2.1. Davidson Trauma Scale (DTS)

The DTS was used to assess symptoms of PTSD at all time points (Davidson et al., 1997). The 17-items e.g. ‘have you been upset by something which reminded you of the event’ measures both frequency and severity of the 17 symptoms of PTSD according to criteria B to D of the DSM-IV. Responses were recorded on a five-point Likert scale for symptom frequency ranging from 0 (not at all) to 4 (every day) as well as for symptom severity ranging from 0 (not at all distressing) to 4 (extremely distressing). For this analysis, we use only the 17 symptom frequency items since the symptom severity scores were not available for all participants. The correlation between total frequency and total severity scores among participants who had both scores from baseline to 24 months ranged between 0.97 and 0.99 over the follow-up period with frequency and severity scores at each time point demonstrating a strong positive relationship (p ≤ .001). The PTSD frequency items were added to create a score ranging between 0 and 68 with a score of 20 or more considered indicative of PTSD. The baseline DTS severity and cut-off score is referred to as PTSS in this paper since it was measured within 20 d of the rape event occurring (in the rape-exposed group) and therefore does not meet the time-based DSM-IV criteria (criterion E) which stipulates that the distress or disturbance in functioning should be present for more than one month. The scale showed excellent internal consistency at baseline (Cronbach alpha coefficient = 0.96).

Although used widely in South Africa, the psychometric properties of the DTS has not yet been established in this population (Martenyi et al., 2002; Marx et al., 2017; Spies & Seedat, 2014; Stein et al., 2009). However, in this study we followed a rigorous translation and back translation process. Participants were asked how much these symptoms bothered them in the past week. The DTS assessment followed the LEC (see details below) where trauma experiences were assessed. However, we did not anchor the DTS symptoms to a specific trauma type since many women report more than one trauma and often report trauma types that are chronic in nature (e.g. childhood abuse, IPV and high rates of community violence). The general profile of trauma exposure (multiple, chronic and continuous exposures) in our context makes it difficult for women to identify a single trauma type that had the most severe effect on them. In the rape-exposed group, we introduced the DTS to the participants who reported a rape as ‘I am aware that you have had a difficult time in the last while. Please see if you able to answer these questions as best as you can’.

3.2.2. Centre for Epidemiologic Studies Depression Scale (CES-D)

The CES-D was used to assess the current severity of depressive symptoms at all time-points (Radloff, 1977). Reponses to the 20-items e.g. ‘I felt that I could not shake off the blues even with help from family or friends’ were measured on a 4-point Likert scale ranging from 0 (rarely or none of the time) to 3 (most or all the time). A total score was generated ranging from 0-60, with higher scores indicating greater depressive symptoms. A cut-off score of 16 or more is considered indicative of depression. The CES-D has shown good reliability and validity across clinic and community populations and has been validated in South Africa (Carleton et al., 2013; Pretorius, 1991). A baseline Cronbach alpha coefficient of 0.93 indicated excellent reliability.

3.2.3. Counselling

Attendance of post-rape early counselling was assessed by asking the women who experienced a rape if they returned to the post-rape service (at the follow-up visits). Rape-exposed women indicating ‘yes’ at the 3- or 6-month follow-up visit, and those reporting the reason for returning was to attend a counselling session, was grouped as those who received early counselling. We did not ask how many sessions women attended or the nature of the counselling. We defined attending early counselling in this study as ever reporting attending counselling at the 3- and 6-month follow-up visits and this was dichotomised into yes/no. Early counselling excluded counselling received at the time of reporting the rape and counselling reported after the six-month visit.

3.3. Stress, alcohol use, resilience and social support

3.3.1. The Perceived Stress Scale (PSS)

The PSS included 10 items, e.g. ‘In the last month, how often have you found that you could not cope with all the things that you had to do’, assessing individual’s subjective experience of daily stress (Cohen et al., 1983). Responses were measured on a Likert scale ranging from 1 (strongly disagree) to 4 (strongly agree). A total score ranging between 9 and 36 was calculated with higher scores indicating higher perceived stress. A baseline Cronbach alpha coefficient of 0.86 indicated good reliability.

3.3.2. Alcohol Use Disorders Identification Test-Concise (AUDIT-C)

The three items, e.g. ‘how many drinks containing alcohol do you have on a typical day when you are drinking’ of the AUDIT-C was used to measure alcohol consumption (Saunders et al., 1993). Responses were recorded using a 5-point Likert scale with diverse response options dependent on the individual item. Total scores ranged between 0 and12 with 0 indicating no consumption and a score of 3 or more indicating hazardous alcohol consumption. A baseline Cronbach alpha coefficient of 0.71 indicated acceptable reliability. The AUDIT-C has delivered satisfactorily findings in comparison with the full AUDIT in many studies (Saunders et al., 1993), including those conducted in South Africa (Morojele et al., 2017).

3.3.3. Connor-Davidson Resilience Scale (CD-RISC)

The 25 items e.g. ‘I am able to adapt when changes occur’ of the CD-RISC was used to measure resilience (Connor & Davidson, 2003). Responses were recorded on a 4-point Likert scale and ranged between 25 and 100 with higher scores indicating greater resilience. A baseline Cronbach alpha coefficient of 0.89 indicated good reliability.

3.3.4. Multidimensional Scale of Perceived Social Support (MSPSS)

The 12 items, e.g. ‘There is a special person around when I am in need’ of the MSPSS was used to measure social support. Responses were measured on a 4-point Likert scale ranging between 1 (strongly agree) to 4 (strongly disagree). A total score ranging between 12 and 48 was calculated with higher scores indicating stronger social support (Zimet et al., 1988). A baseline Cronbach alpha coefficient of 0.92 indicated excellent reliability.

3.4. Statistical methods

The Rape Impact Cohort Evaluation (RICE) study, located in Durban KwaZulu Natal Province, South Africa, provided an opportunity to first explore the prevalence, risk and symptom trajectory of depression and PTSD in a group of women who have experienced a rape, in comparison with women who have never experienced a rape, over a two year period. Second, the mediating role of early person-centred supportive counselling (within 6 months post-rape), offered at rape crisis centres, on long-term PTSD and depression symptom trajectories is explored. Third, the role of demographic factors (age, education, employment, residential area, relationship status, food security and community support), HIV status, childhood trauma, lifetime trauma, intimate partner violence, alcohol use and other psychosocial risk and protective factors (resilience, social support, perceived stress) is investigated in relation to depression and PTSD symptoms in a longitudinal naturalistic design.

We started with exploratory analysis to describe baseline study population characteristics by exposure group. We summarised the data using means and standard deviations for continuous variables and frequencies and percentages for categorical variables. We further assessed associations between baseline demographic variables and baseline exposure groups (rape exposed and not) using t-test for continuous variables and Pearson chi-squared test for categorical variables. We similarly also explored the baseline variables across the counselling groups (Table S5). Longitudinal graphical methods were used to explore average profiles of depression and post-traumatic stress symptom scores at the different time periods for both the rape-exposed and unexposed groups. The rape-exposed group was further stratified by whether they attended early counselling at post-rape services or did not attend early counselling

We also assessed the correlation between the two main outcomes (depression and PTSS/PTSD frequency scores) at each time point using the Spearman rank correlation test. The graphical methods were the basis of choosing an appropriate model for the analysis of the mental health status of participants over the 2 years since study entry. The two mental health outcomes Depression and PTSD were analysed using a joint linear mixed model using an unstructured covariance setup to take into account the correlation between the two outcomes. When the dependence in the main outcomes is ignored, the estimated covariates in the model may be biased and inefficient (Bhat, 2015). The linear splines approach was selected after observing a nonlinear trend in the rape-exposed group post enrolment. This approach provides a practical, less complex, easily interpretable and computationally efficient model (Tilling et al., 2014) to make comparisons between exposure groups. We used 4 knots at 96, 191, 384 and 568 days which closely coincided with the 3, 6, 12 and 18 months scheduled visits to especially model the non-linear change in mental health scores in the rape-exposed group over these months compared to the unexposed group. We further stratified the exposed group into those who did and did not attend early counselling, to assess the immediate, medium- and long-term role of early counselling post-rape. The final fitted model used all available data and included baseline variables that showed significant associations with the main outcomes and known potential confounders namely; age, childhood trauma and non-partner sexual violence. We define dropout as only attending baseline visits and the handling of missing data was done under the missing at random (MAR). The missing completely at random (MCAR) mechanism was ruled out by modelling the probability of dropout as a functiom of baseline covariates plus baseline depression & PTSS scores. Sensitivity analysis was performed by comparing the model results obtained from using all available data (MAR assumed) and from incorporating inverse probability weights (IPW). Using IPW involved firstly modelling the probability of dropout for each exposure group using a logistic regression. Baseline variables considered for the prediction of dropout included age, education, employment, previous experiences of trauma using the LEC scale, perceived stress and alcohol use. The inverse probability weight was calculated by weight = 1/ (1-probability of dropout). To incorporate the weights in the full model, the depression and post-traumatic stress symptom scores at baseline were allocated a weight = 1 since the estimated IPW weights were only applicable to the participants who remained in the study post baseline. Separate weights were calculated for every measurement within a subject (observational-level weights). The sensitivity analyses indicated comparable results and thus the final results reported are based on the MAR model.

Post-estimation was conducted on all the models to obtain mean estimates at each time points for each exposure group with 95% confidence intervals. Analysis was conducted using STATA 17.

4. Results

4.1. Data and participant descriptive and bivariate analyses

The RICE study enrolled 1799 participants, 94 non-rape-exposed participants reported experience of rape or forced sex at the baseline interview and were excluded from this analysis. Among the 1705 participants, only those eligible for their 24-month visit were included in this analysis 1520/1705 (89.1%). Among the 1520 participants, 734 (48.3%) were rape-exposed participants and 786 (51.7%) were not exposed to rape. Socio-demographic differences between the two groups at time of enrolment are shown in Table 1. Rape-exposed participants were borderline younger (mean age 25.0 vs 25.3: p = .049), more likely employed (27.0% vs 15.4%: p ≤ .001), more likely to live in a formal township (64.1% vs 78.2%: p ≤ .001) and more likely to report not being in a relationship at time of enrolment (22.0% vs 14.0%: p ≤ .001). The rape-exposed group was also more likely to test positive for HIV at enrolment and reported higher levels of childhood and previous trauma (all p ≤ .001). The women in the rape-exposed group also reported significantly higher levels of all forms of intimate partner and non-partner sexual violence (all p ≤ .01), lower levels of resilience and higher levels of perceived stress (all p ≤ .01). As expected, the women who experienced a recent rape had much higher PTSS and depression scores. Among the non-exposed group 80.8% reported a trauma at enrolment (childhood trauma/non-rape sexual trauma and other forms of trauma), see Figure 1.

Table 1. Description of baseline characteristics by rape exposure at study entry.

	Rape Exposed		Unexposed			Effect size
	n = 734 (48.3%)		n = 786 (51.7%)
	Mean (SD)/n (%)	95% CI	Mean (SD)/n (%)	95% CI	p value	Cohens D/Cramer’s V
Demographics
Age Age range	25.0 (5.8) 16-40	24.6; 25.4	25.5 (5.5) 18-40	25.1; 25.9	.049	0.101
Education: Completed grade 1-11	321 (43.7%)	40.2; 47.4	314 (40.0%)	36.6; 43.4	.135	0.000
Completed grade 12 or >	413 (56.3%)	52.6; 59.8	472 (60.1%)	56.6; 63.4
Employed: Yes	198 (27.0%)	23.9; 30.3	121 (15.4%)	13.0; 18.1	<.001	0.142
No	536 (73.0%)	69.7; 76.1	665 (84.6%)	81.9; 87.0
Area of residence: Urban-Formal	464 (64.1%)	60.5; 67.5	606 (78.2%)	75.1; 81.0	<.001	0.104
Urban -Informal	145 (20.0%)	17.3; 23.1	105 (13.6%)	11.3; 16.1
Rural or semi-rural	115 (15.9%)	13.4; 18.7	64 (8.3%)	6.5; 10.4
Current relationship status Not in a relationship	163 (22.2%)	19.4; 25.4	112 (14.3%)	12.0; 16.9	<.001	0.159
Married/ Cohabiting	57 (7.8%)	6.0; 10.0	61 (7.8%)	6.1; 9.9
Dating (non- cohabiting)	513 (70.0%)	66.6; 73.2	612 (78.0%)	74.9; 80.7
Household food security Sometimes or often go without food	155 (21.1%)	18.3; 24.2	144 (18.3%)	15.8; 21.2	.170	0.035
Never or seldom go without food	579 (78.9%)	75.8; 81.7	642 (81.7%)	78.8; 84.2
Community Support Not difficult to find money in an emergency	119 (16.2%)	13.7; 19.1	100 (12.7%)	10.6; 15.2	.053	0.000
Difficult to find money in case of an emergency	615 (83.8%)	80.9; 86.3	686 (87.3%)	84.8; 89.4
Clinical assessment
HIV Status Negative	358 (48.8%)	45.2; 52.4	464 (59.0%)	55.6; 62.4	<.001	0.103
Positive	376 (51.2%)	47.6; 54.8	322 (41.0%)	37.6; 44.4
Trauma scales
Childhood trauma (score)	16.5 (3.7)	16.2; 16.8	15.7 (2.4)	15.5; 15.8	<.001	0.268
Previous experiences of trauma (score)	2.1 (1.8)	2.0; 2.2	1.2 (1.6)	1.1; 1.3	<.001	0.524
Psychosocial scales
Resilience score	74.2 (6.4)	73.7; 74.6	75.4 (5.9)	74.9; 75.8	<.001	0.190
Social support score	34.6 (5.2)	34.2; 35.0	35.0 (4.2)	34.7; 35.3	.075	0.092
Perceived stressed score	23.6 (5.8)	23.2; 24.0	21.4 (4.9)	21.1; 21.8	<.001	0.405
PTSS score	35.2 (16.1)	34.0; 36.4	7.1 (10.8)	6.4; 7.9	<.001	2.061
CESD score	33.9 (12.8)	32.9; 34.8	13.5 (9.4)	12.8; 14.1	<.001	1.828
Alcohol use
Audit C score	2.1 (2.5)	1.9; 2.3	1.6 (2.4)	1.4; 1.8	<.001	0.207
Intimate partner violence
Emotional IPV (ever)						0.091
No	378 (51.5%)	47.9; 55.1	476 (60.6%)	57.1; 63.9	<.001
Yes	356 (48.5%)	44.9; 52.1	310 (39.4%)	36.1; 42.9
Emotional IPV in the past 12 months						0.118
No	595 (81.4%)	78.4; 84.1	702 (89.7%)	87.3; 91.6	<.001
Yes	136 (18.6%)	15.9; 21.6	81 (10.3%)	8.4; 12.7
Economic IPV (ever)						0.074
No	581 (79.2%)	76.1; 81.9	667 (84.9%)	82.2; 87.2	.004
Yes	153 (20.8%)	18.1; 23.9	119 (15.1%)	12.8; 17.8
Economic IPV in the past 12 months						0.088
No	654 (89.5%)	87.0; 91.5	738 (94.3%)	92.4; 95.7	.001
Yes	77 (10.5%)	8.5; 13.0	45 (5.7%)	4.3; 7.6
Sexual IPV (ever)						0.248
No	583 (79.4%)	76.3; 82.2	752 (95.7%)	94.0; 96.9	<.001
Yes	151 (20.6%)	17.8; 23.7	34 (4.3%)^a	3.1; 6.0
Sexual IPV in the past 12 months						0.196
No	658 (89.7%)	87.2; 91.7	775 (98.7%)	97.6; 99.3	<.001
Yes	76 (10.3%)	8.3; 12.8	10 (1.3%)	0.7; 2.4
Physical IPV- (ever)						0.100
No	357 (48.6%)	45.0; 52.3	461 (58.7%)	55.2; 62.0	<.001
Yes	377 (51.4%)	47.7; 55.0	325 (41.3%)	38.0; 44.8
Physical IPV in the past 12 months						0.117
No	591 (80.6%)	77.6; 83.3	695 (89.0%)	86.6; 91.0	<.001
Yes	142 (19.4%)	16.7; 22.4	86 (11.0%)	10.0; 13.4
Any IPV- (ever)						0.099
No	268 (36.5%)	33.1; 40.1	364 (46.3%)	42.8; 49.8	<.001
Yes	466 (63.5%)	60.0; 66.9	422 (53.7%)	50.2; 57.2
Any IPV in the past 12 months						0.103
No	531 (72.3%)	69.0; 75.5	637 (81.0%)	78.1; 83.6	<.001
Yes	203 (27.7%)	24.5; 31.0	149 (19.0%)	16.4; 21.9
Non partner sexual violence (ever)						0.365
No	485 (66.1%)	62.6; 69.4	745 (94.8%)	93.0; 96.1	<.001
Yes	249 (33.9%)	30.6; 37.4	41 (5.2%)^a	3.9; 7.0

^anon-exposed group reported non-force sex/rape.

Proportions of participants with depression and PTSD at the different time points across the two groups are presented in Figure 2 and Table S3 in the supplementary file. Overall, at 24 months, 45.2% of the rape-exposed group remained above the cut-off for probable clinical depression and 32.7% continued to have high levels of PTSD. These trajectories were significantly different from the group of women who did not report a rape at baseline.

Figure 2. Proportion of women with depression and posttraumatic stress symptom (PTSS) over 2 years.

4.2. Main outcomes description over time

A summary of depression and PTSD frequency scores by exposure group at the different time-points are presented in Figure 3 (see Table S4 for mean scores in supplementary file). The mean depression and PTSD frequency score for the rape-exposed group is non-linear and decreases over time. As expected, scores were high at enrolment for the women who reported a rape (mean CESD score was 33.9 at study entry) and decreased at 3-months but remained above the depression cut off level up to 24 months (mean CESD score was 17.1 at month 24). Mean PTSS/PTSD scores displayed a similar trend with a baseline mean score of 35.2 among rape-exposed participants, declining to 14.5 (below the PTSD frequency cut-off of 20) at 24 months. The mean depression and PTSS/PTSD scores of the unexposed groups remained stable below the clinical levels for all time-points. Mean scores with CI for both depression and PTSD are presented in the Table S4 in the supplementary file.

Figure 3. Mean profile plots of depression and posttraumatic stress symptom (PTSS) over 2 years by exposure group.

Early counselling data were available for 462 participants in the rape-exposed group. Of these participants, 220 (47.6%) reported attending counselling at 3 and/or 6 months follow-up. An analysis of baseline factors between the two counselling groups is presented in the supplemental file (Table S5) and shows very little differences between the two counselling groups across most of the baseline variables. The longitudinal data show there were no differences in PTSD and depression scores among survivors who reported attending counselling and those who did not (see Figure 4 and Table S6).

Figure 4. Early counselling: mean profile plots of depression and posttraumatic stress symptom (PTSS) over 2 years by exposure group and early counselling.

4.3. Final model

We found significant positive correlations between depression and PTSD scores which ranged between 0.5 and 0.8 over the follow-up period (see Table S7 in the supplementary file). Given that these outcomes were highly correlated, a joint linear mixed model was preferred (see Table 2) to estimate the effect of early counselling on the evolution of the two outcomes, while adjusting for baseline age, education, childhood trauma and non-partner sexual violence. Table 2 presents the outcomes of this model and shows mean scores for both depression and PTSS/PTSD at baseline as well as the direction of these trajectories (slopes) over time in exposed and unexposed groups and compares those who received early counselling with those who received no counselling (Table 2 and Figure 4). Overall a significant decline in depression and PTSD scores were found for the period between baseline and 3 months. Survivors who did not obtain early counselling (β = −0.342 (p  ≤ .001)) and women in the early counselling group (β = −0.341 (p ≤ .001)) both demonstrated significant declines in depression scores compared to women in the non-exposed group. A similar decline was observed for PTSD symptoms. Survivors who did not obtain early counselling (β = −0.337 (p ≤ .001)) and women in the early counselling group (β = −0.325 (p ≤ .001)) both demonstrated significant declines in PTSD scores when compared with the unexposed group. No further significant changes in the trajectory of scores were found after 3 months.

Table 2. Final joint linear mixed model with linear splines, adjusted for baseline variables; age, childhood trauma and non-partner sexual violence.

	Depression		PTSD frequency
	Mean estimate (95% CI)	p value	Mean estimate (95% CI)	p value
Intercept	13.4 (12.7, 14.0)	<.001	6.9 (6.1, 7.8)	<.001
Rape exposed with no early counselling (baseline)	32.4 (31.6, 33.3)	<.001	32.8 (31.8, 33.9)	<.001
Rape Exposed with early counselling (baseline)	33.4 (31.9, 34.8)	<.001	34.9 (33.2, 36.6)	.000
	Depression		PTSD frequency
	Coefficient (95% CI)^a	p value	Coefficient (95% CI)^a	p value
Unexposed
0-3 months spline	−0.131 (−0.157, −0.105)	<.001	−0.080 (−0.111, −0.049)	<.001
3-6 months spline	0.001 (−0.012, 0.013)	.888	−0.003 (−0.018, 0.012)	.669
6-12 months spline	−0.001 (−0.008, 0.006)	.786	−0.001 (−0.009, 0.007)	.818
12-8 months spline	−0.002 (−0.009, 0.005)	.492	0.000 (−0.008, 0.009)	.963
18-24 months spline	0.002 (−0.005, 0.008)	.600	−0.003 (−0.010, 0.005)	.523
Rape Exposed: No Early Counselling
0-3 months spline	−0.342 (−0.366, −0.319)	<.001	−0.337 (−0.365, −0.308)	<.001
3-6 months spline	−0.012 (−0.035, 0.011)	.304	−0.022 (−0.050, 0.006)	.119
6-12 months spline	−0.005 (−0.017, 0.008)	.456	−0.006 (−0.021, 0.009)	.404
12-18 months spline	0.003 (−0.011, 0.016)	.700	0.000 (−0.016, 0.016)	.986
18-24 months spline	0.000 (−0.012, 0.012)	.952	−0.007 (−0.022, 0.007)	.342
Rape Exposed: Early Counselling
0-3 months spline	−0.341 (−0.368, −0.314)	<.001	−0.325 (−0.358, −0.293)	<.001
3-6 months spline	−0.017 (−0.041, 0.006)	.148	−0.041 (−0.070, −0.013)	.005
6-12 months spline	−0.002 (−0.014, 0.011)	.800	−0.017 (−0.032, −0.002)	.025
12-18 months spline	−0.002 (−0.016, 0.012)	.755	0.003 (−0.013, 0.020)	.711
18-24 months spline	−0.010 (−0.023, 0.002)	.109	−0.011 (−0.026, 0.004)	.166
Unexposed: Covariates
Age & 0-3 months spline	0.001 (0.001-0.0020	<.001	0.000 (−0.001, 0.001)	.438
Education & 0-3 months spline	−0.016 (−0.025, −0.007)	.001	−0.006 (−0.017, 0.004)	.239
Childhood trauma & 0-3 months spline	0.007 (0.006, 0.009)	<.001	0.005 (0.003, 0.007)	<.001
Rape exposed: Covariates
Exposed-Age & 0-3 months spline	0.002 (0.001, 0.003)	<.001	0.002 (0.001, 0.003)	<.001
Exposed-Education & 0-3 months spline	−0.005 (−0.016, 0.005)	.314	0.015 (0.002, 0.027)	.021
Exposed-Childhood trauma & 0-3 months spline	0.008 (0.007, 0.009)	<.001	0.008 (0.006, 0.009)	<.001
Non partner sexual violence-rape exposed group	1.985 (1.611, 2.359)	<.001	3.106 (2.654, 3.557)	<.001

^aCoefficients rounded off to three decimal places.

Table 3 presents the predicted mean scores from the final fitted model. Despite a decline in scores for both depression and PTSD among women who attended counselling over the follow-up period, no significant difference in mean depression and PTSD scores were found in comparison with women who did not attend early counselling (see Table 3 and in Figure 5).

Figure 5. Mean estimates from the final fitted model of depression and posttraumatic stress symptom (PTSS) over time over 2 years.

Table 3. Final model predicted mean estimates for Depression and Posttraumatic Stress Disorder (PTSD) and 95% CI.

Final model mean estimates
Depression
	Rape Exposed and early counselling	Rape Exposed and no early counselling	Unexposed
Time	Mean estimate (95% CI)	Mean estimate (95% CI)	Mean estimate (95% CI)
Baseline	34.5 (33.1, 35.9)	33.6 (32.8, 34.4)	13.5 (12.8, 14.2)
Month 3	19.5 (18.1, 20.9)	18.6 (17.3, 19.9)	14.0 (13.3, 14.8)
Month 5	17.7 (16.4, 19.0)	17.1 (15.8, 18.4)	14.3 (13.6, 15.0)
Month 9	17.3 (16.3, 18.3)	16.9 (15.8, 17.9)	14.2 (13.6, 14.7)
Month 12	17.2 (15.7, 18.6)	16.1 (14.7, 17.5)	14.1 (13.3, 14.9)
Month 18	16.7 (14.8, 18.5)	16.6 (14.8, 18.4)	13.7 (12.7, 14.7)
Month 24	14.8 (12.6, 17.0)	17.0 (14.9, 19.1)	13.9 (12.8, 15.1)
PTSD
	Rape Exposed and early counselling	Rape Exposed and no early counselling	Unexposed
Time	Mean estimate (95% CI)	Mean estimate (95% CI)	Mean estimate (95% CI)
Baseline	36.7 (35.0, 38.4)	34.6 (33.7, 35.6)	7.1 (6.3, 7.9)
Month 3	21.5 (19.8, 23.2)	18.4 (16.9, 20.0)	7.5 (6.6, 8.4)
Month 6	17.6 (16.0, 19.1)	15.9 (14.3, 17.5)	7.4 (6.6, 8.3)
Month 9	15.6 (14.4, 16.9)	15.4 (14.2, 16.7)	7.3 (6.6, 7.9)
Month 12	14.3 (12.6, 16.1)	14.5 (12.8, 16.2)	7.3 (6.3, 8.3)
Month 18	14.4 (12.2, 16.6)	14.5 (12.3, 16.6)	7.2 (6.0, 8.4)
Month 24	12.6 (10.0, 15.2)	13.6 (11.0, 16.1)	6.7 (5.3, 8.1)

5. Discussion

This is the first study that followed a cohort of 1520 rape survivors for two years to describe the trajectory of PTSD and depression over time. Our study showed two years post rape, 45.2% of the women met the cut-off for depression and 32.7% for PTSD which was significantly higher than levels found among the women not exposed to rape (33.1% for depression and 12.8% for PTSD). This study confirms the long-term mental health burden post-rape. To date, there have been few longitudinal, naturalistic studies tracking the mental health outcomes of rape survivors in the immediate aftermath of rape. As far as we know, this is the first study in a global south setting to show the sustained mental health burden post-rape as compared to a comparison group. The depression and PTSD burden among survivors in the RICE study is considered high with a third (32.7%) of survivors having PTSD and nearly one in two (45.2%) having depression at 24 months post rape, We have found no comparative studies that measured PTSD and depression 2 years after a rape experience. Although not directly comparable, a recent meta-analysis of psychological disorders associated with sexual assault (Dworkin, 2020) reported a pooled 12 months prevalence of 26% PTSD and 24% for depression (Dworkin, 2020). Although there is no evidence for a differential mental health impact of different forms of sexual abuse, rape has consistently been shown to be the most pathogenic trauma for PTSD compared to other traumas (Yehuda et al., 2015).

The trajectory of both depression and PTSD over the two years of follow-up is in keeping with a clinical presentation of the natural recovery from trauma (Hughes, 2012), with a decline in average scores seen from 3 months but then remaining at the same levels over the full follow-up period. The profile of depression is most revealing as the average score remained above the cut off value of 16 for up to 24 months. The persistent mental health associations place in perspective the immense problem of rape in South Africa where under-reporting to services are well known.

Global estimates of the disease burden show that mental health problems contribute the largest burden to disability and account for a third (32.4%) of years lived with disability (YLD) (Vigo et al., 2016). Mental health problems are also a risk factor for multiple other adverse health outcomes (Vos et al., 2020). More specifically the 2019 estimates show a growing burden, with depressive and anxiety disorders among the top six causes contributing to this burden (Vos et al., 2020). This is of particular relevance in a country such as South Africa with high level of co-occurring HIV, chronic non-communicable diseases and mental disorders (Mayosi et al., 2009). The recent evidence of how poor mental health mediates HIV acquisition in the aftermath of rape also points to the urgency to effectively address mental health post rape (Abrahams et al., 2021).

This study also investigated in a naturalistic design the role of early counselling on mental health outcomes. We found no evidence of the role of early counselling on depression and PTSD levels over time. International treatment guidelines for PTSD following trauma (all forms of trauma) recommend both early (within 3 months) and late interventions (International Society for Traumatic Stress Studies, 2018). A recent randomised control trial (RCT) assessed two sessions of Eye Movement Desensitisation and Processing (EMDR), one of the recommended treatments of PTSD, within 14–28 days post rape as an early intervention to reduce PTSD symptoms post rape. The sample of 57 rape survivors was randomised to either the intervention or to a control group, the latter receiving usual care (watchful waiting) and at three months post-rape the analysis showed that the EMDR was not more effective in reducing PTSS than usual care (Covers et al., 2021). In contrast a systematic review on early interventions to prevent PTSD after sexual assault found that they could lead to lasting effects, but due to the limited availability of longitudinal data (only 4 studies) and high risk of bias in many studies, they could not draw definitive conclusions on the long-term safety and efficacy of early interventions (Oosterbaan et al., 2019). Clearly, more research is needed to understand what works best for sexual assault survivors in the high prevalence and limited resource settings such as South Africa. In our study, we recruited women from post-rape services, and all women received initial counselling as part of their acute care. We have no information on the counselling women received in the 6-month period after the rape beyond attendance of counselling at the post-rape care centre. We can safely assume that all participants received the same counselling modality as it was offered by the same NGO and was very likely trauma focused. However, we do not have information on the number, duration and timing of these sessions which may all have contributed to the quality of counselling received. The full post-rape package of services is well described in an evaluation report of the TCCs, with mental health care identified as not a priority (Jordaan et al., 2016). Rape survivors constitute a particularly vulnerable group with multiple comorbid traumas such as childhood adversity, social disadvantage, and revictimisation experiences from service providers, which intersect with rape stigma experiences (Jewkes et al., 2021). All these factors influence a survivor’s pathway to counselling and help-seeking. Closing the treatment gap for mental health care in low- and middle-income countries has been a global concern for more than a decade (Petersen et al., 2019). A recent review of the efficacy of secondary prevention for PTSD after a sexual assault shown much more work is still needed (Short et al., 2020). Current discussion on addressing the mental health needs of those in resource-limited settings focuses on advocating for the implementation of low-threshold and structured (i.e. manualised) evidence-based interventions that can be adapted for the local context and delivered by non-specialist providers (Holmes et al., 2018). Additionally, we have to better understand rape survivors’ perspectives on counselling in our setting as this may influence uptake. Other important variables should be taken into account when providing care such age and past traumas since the model showed that older women, childhood trauma experiences (either raped or not) and non-partner sexual violence in the group who experienced rape, had negative associations on the depression and PTSD symptoms, significantly so in the first three months.

6. Strengths and limitations

A key strength of our study is the prospective cohort design with rape-exposed women recruited soon after the rape event and the inclusion of a non-rape-exposed comparison group. Both groups were also health care seekers. Our large sample size also allowed us to have statistical power to examine both outcomes (depression and PTSD). In addition, the long follow-up period of 2 years is substantially longer than most other longitudinal studies of rape survivors.

Our study focussed on rape survivors who sought care immediately and we cannot extrapolate to other rape survivors who do not seek care or seek care long after the rape. Doing follow-up research with rape survivors is not easy and the decrease in retention over time in our exposed group is unsurprising. Anecdotally, some women told us they did not want to return for follow-up visits because they ‘want to move on’. We do not know if this was because the study served as a reminder of the rape and exiting the study was therefore a coping strategy. We have described our experiences to improve retention elsewhere (Abrahams et al., 2017). We assumed a more conservative missing data mechanism (MAR assumption) given that the women who experienced a rape were more likely to dropout due to the multiple vulnerabilities associated with rape experience. Further analysis of the dropout patterns and reasons for dropout is required to address this limitation.

Primary outcomes were assessed with self-reported questionnaires and self-reporting bias may have influenced measurement of these outcomes. We only included frequency PTSS/PTSD scores in our analysis as severity data were not available for everyone (a range between 53% at baseline to 1% at 24 months were missing). We did, however, find a strong positive correlation between frequency and severity scores from baseline to 24 months for participants with both scores (p ≤ .001). A study limitation is that we did not anchor the DTS symptoms to the same trauma at each visit. We are however very confident that the women who experienced a rape were referring to this event when they answered the PTSD assessments at each follow-up visit. To demonstrate the vast and complex trauma experiences in the sample, we did an analysis of baseline trauma exposures reported (LEC, IPV, NPSV and childhood trauma) and assessed the proportion of each trauma type among those who met the PTSD cuff-off at each return visit. This is presented in Table S8 and we show the most common trauma events reported among those with PTSD at each visit were childhood trauma, IPV as well as having experienced being robbed/highjacked at gunpoint or knife point. This confirms the wide range and extent of trauma experienced by many South Africans (Mayosi et al., 2009) and the complexity of identifying a single trauma to the PTSD in our setting. This is different to studies reported from the global north.

Another unmeasured confounder was pharmacological management. We had no information on the use of antidepressants or other psychotropic medications. Use of these medications may have influenced the symptom trajectories observed. We recorded medication use in our nursing notes (i.e. source documentation during clinical assessments) that were not part of the formal RICE dataset. Research nurses generally reported that very few women, if any, reported taking psychiatric medication. We also do not have past psychiatric histories of women included in the study (i.e. pre-rape history of depression, PTSD, other anxiety disorders). In addition, our study did not explore other psychiatric disorders. We chose to focus on the two most common disorders among rape-exposed women – depression and PTSD.

An important limitation of our study was that it was not designed to assess the impact of early counselling and the interpretation of these results must be considered in this light. First, we had few details on the content, format and frequency of the counselling that women received and whether it was evidence-based. Second, we do not have information on whether women attended counselling outside of that which was offered by the rape services. Third, counselling information was self-reported and there may have been social desirability bias. Fourth, the RICE study also had a trauma counsellor as part of the team who offered immediate counselling to participants with high PTSS/PTSD and depression scores or who were at risk for suicide. Few participants accepted the offer to return for counselling with the RICE counsellor. The data on attending counselling at the RICE study was kept confidential and was not included in data analysis in accordance with our ethics protocol. Finally, women may have had counselling after 6-months which may have influenced the trajectory of depression & PTSD. However, the data show very little change in depression and PTSD scores after six months. Despite these limitations, our study design and large cohort of vulnerable participants remains a first in developing settings.

7. Conclusion

Depression and PTSD persist two years post-rape, and this prolonged mental health contributes to the health burden globally. Comprehensive post-rape care, which includes evidence-based prevention strategies, is critical to reducing the risk of PTSD and depression and to managing these mental health conditions when they occur. Enhancing the capacity of providers to deliver basic mental health and psychosocial support to violence survivors is a must.

Acknowledgements

The authors are very grateful to the participants in the RICE study for sharing their time and intimate experiences. We thank all the staff who worked to make the study possible. The authors also thank the National Prosecuting Authority for giving us access to the TCCs and the KwaZulu Natal Department of Health for access to Primary Health Care services.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability

The data used in this paper are available on request to the corresponding author.

Additional information

Funding

This research and the publication are the result of funding provided by the South African Medical Research Council in terms of the SAMRC’s Flagships Awards Project SAMRC-RFA-IFSP-01-2013/ RAPE COHORT’ Grand Challenges Africa Round 6-Data Science approaches to improve Maternal, Neonatal and Child Health in Africa. The analysis in this publication commenced with a research project which formed part of the first author’s (SM) Masters’ degree in Biostatistics and which was supported through the DELTAS Africa Initiative SSACAB [grant number 107754/Z/15/Z] and South African Medical Research Council (SAMRC).