Demographics

The demographics of women presenting to Panzi Hospital between 2004 and 2008 requesting postsexual violence care were disaggregated according to type of sexual violence (Table 1). Both the mean and median age was 35 years, with an age range of 3.5 to 80 years. In total, 6% of survivors were younger than 16 years of age and 10% were 65 years of age or older. Fifty-three percent of women were married, 22% were widowed, and 9% reported that their spouses had abandoned them. The mean number of children per woman was 3.6, with a median of 3. The number of children per woman ranged from 0 to 12. Also, 59% of women were illiterate, and only 9% had attended secondary or postsecondary school. The majority of women in this data set reported farming as their source of livelihood (74%), and another 10% reported that they were unemployed. The majority of women self-identified with the Bashi tribe (65%).

TABLE 1

Demographics of Sexual Violence Survivors Presenting to Panzi Hospital From 2004-2008 by Type of Sexual Violence

For the purpose of regression analysis, sexual assaults were divided into 2 categories: rape NOS versus all other types (including gang rape, sexual slavery, combined gang rape and sexual slavery, and other types of sexual violence). This division was based on earlier evidence that rape NOS is more likely to be perpetrated by civilians and carries attack characteristics that are distinct from both gang rape and sexual slavery.Reference Bartels, VanRooyen, Leaning, Scott and Kelly12 The complete regression analysis is shown in Table 2.

TABLE 2

Odds Ratios (OR) for Demographic Predictors of Rape NOS and Presentation to Panzi Hospital After Sexual Assault^a

In summary, survivors who had completed secondary education were more likely to report rape NOS than other types of sexual assault in comparison to survivors who were illiterate (Odds Ratio [OR] = 1.417, 95% CI = 1.123-1.788). Survivors who were students (OR = 3.832, 95% CI = 2.774-5.294) and survivors who reported “other” occupations (OR = 5.081, 95% CI = 3.759-6.869) were more likely to report rape NOS than other types of sexual assault in comparison with survivors who reported being farmers. Furthermore, survivors who self-identified with the Barega (OR = 0.691, 95% CI = 0.540-0.882) and Bahavu (OR = 0.476, 95% CI = 0.332-0.681) tribes were less likely to report rape NOS than other types of rape in comparison to survivors who self-identified with the Bashi tribe. Survivors who were 15 years of age or younger were more likely to report rape NOS than other types of sexual assault in comparison to survivors in all other age groups (Table 2). Finally, survivors who reported being single were more likely to report rape NOS than other types of sexual assault in comparison to women who reported being married, widowed, or abandoned (Table 2).

Presentation to Panzi Hospital

Many sexual violence survivors who presented to Panzi Hospital came alone (36%). Other women were accompanied by aid workers (13%), family members (7%), or other patients traveling to Panzi Hospital (7%). Less than 1% of women were accompanied by their husbands.

In total, 73% of women reported “vehicle” as their mode of transportation to Panzi Hospital, without differentiating private vehicle, taxi, bus, ambulance, or other. Another 18% walked to the hospital to receive care. Less than 1% of women took a flight to reach Panzi Hospital, and less than 1% traveled by boat.

Most sexual violence survivors (92%) reported that they had originally accessed the services of a nongovernmental organization (NGO) or community-based organization (CBO) and were then referred to Panzi Hospital for care. For 86.29% of those women, the service rendered by the NGO/CBO was referral and transportation to Panzi Hospital. For another 13.62%, the service received was not specified, and only 4 survivors (0.09%) indicated that they had received medical care from the NGO/CBO.

A third of the records did not include the month that the sexual assault occurred. In many instances in which the month of attack was missing, the assault had occurred several years before arrival at Panzi Hospital, thus representing cases in which significant delays occurred between sexual violence and presentation to medical care. To calculate the time intervals between sexual assault and seeking medical care, the month of attack was assumed to be June for any record in which the actual month was not provided.

The Figure provides the time intervals between sexual assault and presentation to Panzi Hospital requesting postsexual assault services. The mean time interval was 10.4 months and the median time interval was 7 months. Only 12% of women presented within the first month following the sexual assault. Many women came within the first year following the attack or had a much-delayed presentation, not seeking help for more than 3 years following the assault.

FIGURE

Time Interval Between Sexual Assault and Presentation to Panzi Hospital Requesting Postsexual Assault Services.

Data labels represent percentage of total women in the data set.

Throughout the data collection period, the delay between sexual violence and time to presentation at Panzi Hospital did not shorten. For instance, the average time delay in 2005 was 9.4 months versus 13.7 months in 2008. The analysis for 2004 was difficult because so many of the sexual assault records for 2004 did not include the month of the sexual assault.

We examined whether any demographic factors predicted delay to seeking medical care after sexual assault. For the purposes of this analysis, the time interval between sexual assault and presentation to Panzi Hospital was divided into less than 1 month versus 1 month and longer, as 1 month was the shortest time interval that could be consistently identified in this retrospective analysis. The full regression analysis is provided in Table 2. Survivors who had completed primary education were more likely to present to medical care in the first month than those who were illiterate (OR = 1.636, 95% CI = 1.282-2.088). In addition, survivors who reported being students (OR = 3.973, 95% CI = 2.700-5.868) and survivors who reported other occupations (OR = 2.401, 95% CI = 1.590-3.627) were more likely to present to Panzi Hospital in the first month in comparison to survivors who reported being farmers. Survivors who self-identified with the Barega tribe (OR = 0.546, 95% CI = 0.336-0.888) and Bafulero tribe (OR = 0.413, 95% CI = 0.191-0.899) were less likely to present to Panzi Hospital in the first month as compared with survivors who self-identified with the Bashi tribe. Survivors who were 15 years of age and younger were more likely to present to medical attention in the first month in comparison to survivors in all other age groups (Table 2). Finally, survivors who reported being single were more likely to present to Panzi Hospital within a month in comparison to survivors who were married, widowed, or abandoned (Table 2).

Type of sexual violence and type of perpetrator were also predictive of delayed presentation to Panzi Hospital. Survivors who reported being assaulted by civilians were more likely to present to Panzi Hospital in the first month than were those who reported being assaulted by unidentified soldiers (OR = 5.813, 95% CI = 3.968-8.547), soldiers belonging to an identified armed group (OR = 4.739, 95% CI = 3.289-6.849), and non-specified perpetrators (OR = 3.968, 95% CI = 2.865-5.464). Survivors who experienced gang rape were less likely to present to Panzi Hospital within a month in comparison to survivors who experienced rape NOS (OR = 0.541, 95% CI = 0.422-0.695).

Qualitative Analysis

The qualitative analysis provided insight into why many sexual violence survivors delayed seeking medical attention. For some, presentation to a hospital seemed to be prompted by physical symptoms that worsened over a period of time, finally becoming severe enough that the survivor decided to seek medical care. This was the case with the following woman:

Other sexual violence survivors did not seek early medical attention because they did not have the means to access care, as was the case in the following instance:

In some instances, the spouses and families of survivors were not aware of the sexual assault, and the survivors waited for an opportunity to present to medical attention privately.

For some survivors of sexual slavery, delayed medical care was simply a function of having been abducted for a prolonged period of time. This was the case for the following woman:

Finally, stigma surrounding sexual violence appears to have prevented some women from seeking prompt medical attention. This was the case for the following woman:

Demographics

Analysis of this data set, consisting of 4311 female survivors presenting to Panzi Hospital seeking postsexual violence care between 2004 and 2008, illustrates sexual violence that is pervasive in the communities of South Kivu. It affects young girls and elderly women who are single, married, divorced, and widowed. It affects the educated and the uneducated, as well as women of different occupations and ethnicities. In general, the educational status,2728 occupation,29 and ethnicity28 of women in this data set reflect the general demographics of women in South Kivu Province, implying that no one is spared. Furthermore, our finding that sexual violence is widespread in Eastern DRC, and affecting women and girls of all demographic backgrounds, is consistent with other reports.Reference Steiner, Benner, Sondorp, Schmitz, Mesmer and Rosenberger9Reference Johnson, Scott and Rughita1628Reference Peterman, Palermo and Bredenkamp30

Although sexual violence affected women from all demographic backgrounds, disaggregation of the data revealed that certain demographic factors were associated with the type of sexual violence experienced. For instance, girls younger than 15 years of age and single women were more likely to report rape NOS than other types of sexual violence. Prior work has suggested that young girls are more likely to be sexually assaulted by civilians,Reference Nelson, Collins, VanRooyen, Joyce, Mukwege and Bartels31 and that civilian perpetrators are more likely to perpetrate rape NOS.Reference Bartels, Scott, Mukwege, Lipton, Vanrooyen and Leaning15 Consistent with our earlier findings, Duroch et al reported that minors were more likely to be sexually assaulted by unarmed civilians.Reference Duroch, McRae and Grais26 Level of education and occupation also appeared to impact the type of sexual violence. Survivors who had completed secondary education, survivors who reported that they were students, and survivors who reported other occupations (including teaching, cooking, housekeeping, health care, and dressmaking) were all more likely to experience rape NOS than other types of sexual violence. It remained unclear why educated women, students, and women in other professions would have been more likely to experience rape NOS rather than gang rape, sexual slavery, or a combination of gang rape and sexual slavery. In South Kivu, urban areas tended to have a more educated population, the population tended to engage in different occupations as opposed to rural areas, and there was less military/militia presence in comparison to rural regions. Therefore, we speculated that it may have been because educated women and women in other occupations live in more urban areas and because a greater proportion of urban sexual assaults was perpetrated by civilians (due to lesser presence of armed combatants) that this trend was observed. Prior work suggested that civilian perpetrators are more likely to commit rape NOS than other types of sexual assault.Reference Bartels, Scott, Mukwege, Lipton, Vanrooyen and Leaning15

Presentation to Panzi Hospital

In this study, there were notable delays before women presented to Panzi Hospital for medical attention. Similar delays were reported by Malteser International,Reference Steiner, Benner, Sondorp, Schmitz, Mesmer and Rosenberger9 by Casey et al,Reference Casey, Gallagher, Makanda, Meyers, Vinas and Austin32 and Duroch et alReference Duroch, McRae and Grais26 at other medical facilities in DRC. In Sierra Leone, lengthy delays before presentation to medical care have also been reported (5 months on average).Reference Amowitz, Reis and Lyons33 These delays have important implications for physical, reproductive, and mental health. Many women sought medical care only when their physical symptoms persisted or became very severe. In an ideal setting, postsexual violence care is recommended within the first 72 hours such that HIV PEP,Reference Workowski and Berman22Reference Smith, Grohskopf, Bkack, JD, Veronese and Struble24 prophylactic antibiotics for STIs,Reference Workowski and Berman22 and emergency contraceptives can be given.Reference Bates and Marx23Reference Linden25 However, with a mean delay of 10.4 months from sexual assault to seeking care, most sexual violence survivors in this study presented to Panzi Hospital too late to benefit from these preventative medical therapies.

Women who reported being sexual assaulted by civilian perpetrators were more likely to present to Panzi Hospital within the first month after the assault than were women who were sexually assaulted by armed combatants or nonspecified perpetrators. This finding contradicted our hypothesis that women who experienced brutal sexual assaults perpetrated by armed militias would suffer more injuries and therefore be more likely to seek prompt medical attention. We believe that civilian-perpetrated sexual assault is more common in larger towns and cities such as Bukavu, and women in these towns and cities can more easily access care at Panzi Hospital. Therefore, this finding may simply reflect sampling bias. Women who reported being gang raped were less likely to present to Panzi Hospital in the first month after the assault; again, this may represent sampling bias, as gang rape is more likely to be perpetrated by armed combatantsReference Bartels, Scott, Mukwege, Lipton, Vanrooyen and Leaning15 in rural villages, and women living in rural villages probably have more difficulty accessing medical care in Bukavu.

Girls younger than 15 years of age and single survivors were also more likely to present to Panzi Hospital within the first month after sexual assault. In the narratives of many of these young girls, it was apparent that they were being brought for medical care by their parents after their parents were informed of the sexual assault.

Qualitative analysis identified multiple reasons why sexual violence survivors had a delayed presentation to Panzi Hospital. Some women presented late because they had been abducted and forced into sexual slavery for a prolonged period, sometimes for months or years. Since the date of the assault was entered as the day that the woman was abducted, prolonged periods of sexual slavery undoubtedly contributed to the prolonged intervals between assault and seeking care (average of 10.4 months). However, other reasons were also noted for delayed presentations. Some women specifically mentioned that they had no means to come to Panzi Hospital earlier. Although our study was not designed to examine this, we speculated that women who lived in rural villages further from Panzi Hospital had more difficulty finding the means to come to the hospital in comparison to women who lived in or close to Bukavu.

In some instances, sexual violence survivors talked about deterioration in their health or about physical symptoms becoming so severe that they finally were prompted to seek medical attention. Thus, there appeared to be a lack of awareness about the benefits of early medical attention following sexual assault, before symptoms begin, such that HIV PEP, prophylactic antibiotics, and emergency contraception can be provided. A study looking at medicolegal services for sexual violence survivors in Kenya also commented on a lack of community awareness about postsexual violence services and about how this might explain delayed care-seeking behavior (although the delays noted in the Kenya study were several days as compared to our reported 10.4 months).Reference Chaudhry, Sangani, Ojwang and Khan34

It may be possible that some symptoms prompting women to seek care were unrelated to the sexual assault. However, to have been referred to the Victims of Sexual Violence Program, these women must have mentioned the sexual assault while they were being triaged and, therefore, they were concerned that the symptoms were related to the assault. In most cases it will be impossible to know whether a symptom such as pelvic pain or vaginal discharge was actually a consequence of sexual violence or was unrelated. Regardless, we identified this as one area of intervention to reduce lengthy delays to medical care—a culturally appropriate, community educational campaign to raise awareness that HIV PEP and emergency contraceptives are most effective if given within the first 72 hours, and that beyond this period they are unlikely to be beneficial. Furthermore, survivors of sexual assault also need to be informed that prophylactic antibiotics can prevent infection with and complications of STIs. Of course, sexual violence survivors who present for early physical care could also benefit from early and culturally sensitive psychological support and mental health services. If raising community awareness improves timely access to medical services, then the availability of medical interventions needs to be ensured, particularly in light of a recent report that none of the 21 public health facilities assessed in Eastern DRC had HIV PEP or emergency contraception available.Reference Casey, Gallagher, Makanda, Meyers, Vinas and Austin32

Some women discussed stigma surrounding sexual assault and about having been too ashamed to seek care earlier. Other women reported that their spouses were not aware of the assault and to prevent their spouses from finding out, they delayed seeking medical care until their spouses were out of town. Here too, we believe that there is an opportunity for intervention to reduce lengthy delays before seeking postsexual assault care. If stigma surrounding sexual violence can be reduced, sexual violence survivors may be more likely to talk about their assaults, more likely to seek the support of family members in accessing postsexual assault services, and more likely to promptly report the sexual violence to health care providers. Addressing stigmatization will require focused sensitization and education at the community level.

The delays documented here were for the period 2004 to 2008. It is noteworthy that since that period, Panzi Hospital and several aid organizations have implemented outreach programs to educate community members about the importance of seeking medical care promptly following sexual assault. The impact of these programs is unknown, and it is possible that delays in seeking care have been reduced since our data were collected.

Limitations

This study has a number of limitations. First, the study's retrospective nature prevented the clarification of data inconsistencies as well as the completion of missing data. The retrospective design also did not allow for verification of the original information or the manner in which it was collected. Second, this study was limited by its sampling methodology. Data represent only those sexual violence survivors presenting to Panzi Hospital for postsexual assault care, thus introducing an inherent selection bias. The sampling within Panzi Hospital was also a limitation. The Victims of Sexual Violence Program was sporadically understaffed, such that an insufficient number of interviewers were available at times to interview all presenting survivors. During these times of understaffing, the available interviewers chose to interview those women whom they perceived, based on the initial triage, to have experienced the most traumatic violence.

Although the potential for selection bias existed, we did not detect any patterns in the data registry gaps. Rather the gaps were sporadic—arising from irregular staffing patterns. It is possible that the sexual violence survivors in this convenience sample have in fact experienced the most severe violence. However, we believe that it would be difficult to determine during the brief triage process which women had suffered the most severe trauma. It is also possible that the sexual violence survivors sampled are different in some way from the general population of women who have experienced sexual violence in Eastern DRC. Other recent retrospective clinic or program-based data share similar sampling biasesReference Steiner, Benner, Sondorp, Schmitz, Mesmer and Rosenberger9Reference Duroch, McRae and Grais26 and in spite of these potential biases, we believe that our large sample size and the collection of data over a 5-year period provides important insights into sexual violence in the region.

In addition, it is possible that some women included in this data set had presented to Panzi Hospital making false claims of sexual assault to access free medical care under the Victims of Sexual Violence Program. However, given the high levels of stigma surrounding sexual violence in DRC and given the potential consequences faced by survivors (for instance, abandonment by husband or family, not being able to marry), we believe that false claims of sexual assault were rare. Furthermore, before being enrolled in the Victims of Sexual Violence Program, staff routinely screens all survivors with a series of questions to verify the history of sexual assault. As a final limitation, several translations were required before the analysis of these data, thus introducing potential errors and cultural differences also have the potential to create errors in interpretation. To limit the later source of error, results were discussed with local Panzi Hospital staff who provided cultural background and context.