INVESTIGATION OF SEXUAL VIOLENCE IN EASTERN DRC connections. For example, a sexual violence victim with documented symptoms of STIs may have had those symptoms before the sexual violence. Similarly, a woman who reported a pregnancy immediately after being raped may have actually been pregnant from a pre-existing sexual relationship. Second, this study was limited by its sampling methodology. Because the data are representative only of those sexual violence survivors presenting to Panzi Hospital for post-rape care, there is an inherent selection bias. These cases of sexual assault may have been more violent, thus causing women to seek medical assistance. The sexual violence survivors may also have differed from the general population of raped women in that they had the means to access services. On the other hand, those who did not come to the hospital for services may have been more disabled, more oppressed, more fearful, or more vulnerable in other ways. Women who died before seeking care are obviously not included in this database, which introduces a survivor bias of uncertain dimension. An estimate from local health centers in South Kivu suggests that 3% of women who are raped die as a result(67). Because of these sampling biases, the results cannot be generalized to sexual violence survivors throughout Eastern DRC. The sampling within Panzi Hospital was also a limitation of the study. The Victims of Sexual Violence Program was sporadically understaffed, meaning that at times there were an insufficient number of in-take officers to conduct all the necessary interviews. During these times of understaffing, the existing in-take officers chose for interview those women whom they believed to have suffered the most traumatic violence, based on interactions during the initial registration process. Because of this sampling, only 4,311 of the possible 9,709 sexual violence survivors actually registered as presenting to Panzi Hospital between 2004 and 2008 were interviewed. Although it is possible that these registration gaps represented important sampling biases, we did not detect any patterns in the data registry gaps at Panzi Hospital. Rather, the gaps appeared to have arisen sporadically as a result of insufficient staffing. Nonetheless, the hospital registry did not record an intake interview for half of the sexual assault cases that presented to and were treated by Panzi Hospital during this period. As a result, this study is, by force of circumstance, based on a subset of all hospital patients. It is possible that the 4,311 sexual violence survivors presented here do in fact represent the more extreme cases on the spectrum of violence, since they were apparently selected on that basis for interview by the intake officers. However, we believe that it would be challenging to determine at first glance during the registration process which women had suffered the most severe trauma. In fact, a small number of women presumed to have undergone brutal sexual assaults and consequently chosen for interview later admitted in the interview that they had not actually been raped (some women initially claimed that they were sexual violence survivors believing that this would ensure their eligibility for services at Panzi Hospital). This unsystematic but selective intake approach may also have had some effect on skewing the age distribution of women who were interviewed and whose cases were thus included in this analysis. In other reports, as many as one third of sexual violence survivors were children or adolescents.(68) Only 6% were found to be under the age of 16 in the current study. If it was impossible to interview all survivors presenting on any given day, it is possible that younger girls were not selected for interview because it is often more difficult for children to articulate their 47