We conducted a retrospective descriptive study of EVD acquisition among HCWs in Sierra Leone from May to December 2014. We obtained the study data from three main sources: 1) the national EVD database, which is an Epi Info application called Viral Hemorrhagic Fever (VHF), developed by the Centers for Disease Control and Prevention (CDC) [10]; 2) a cross-sectional survey conducted through administration of a structured questionnaire to infected HCWs; 3) key informant interviews of select health stakeholders in the affected districts.
We extracted case investigation data for all HCWs listed in the national VHF database from May through December 2014. A HCW is defined as any person involved in the promotion, protection, or improvement of the health of the population [11]. Based on this definition, we included all categories of workers who are directly or indirectly involved in EVD health services delivery. These include doctors, clinical officers, nurses, community health officers, surveillance officers, ambulance drivers, and support staff (hygienists, porters, and others). All HCWs in the database were listed for study. We descriptively analysed the HCW data using Epi Info 7. Preliminary analysis revealed underestimation of HCW infections in several districts and missing data, such as location, mode, and type of exposure, which are required to better characterise EVD transmission among HCWs.
To collect additional data, a structured questionnaire was developed, pre-tested, and administered to all HCWs in the national database, as well as to other infected HCWs identified by the District Health Management Teams (DHMTs). The questionnaire contained 42 questions and sub-questions that were categorised into four main sections, i.e., HCW identity, exposure, and knowledge and practice of IPC. Six data collectors were identified (three from the WHO data management team, who are co-authors of this paper, and three external candidates) and trained on the study protocol and questionnaire administration, to ensure that the questionnaire was administered in a uniform manner. The data collectors then pre-tested the questionnaire and administered the final version to respondents.
Nine districts that reported HCW infections during the study period were covered. All surviving HCWs were interviewed in person. In cases of HCWs who had died, next of kin or close associates were interviewed. A relative or close associate was defined as a spouse, parent, brother or sister who was close to or lived with the deceased HCW. To ensure completeness of data, colleagues working in the same health facility as deceased HCWs were also interviewed, in order to obtain workplace-related information that could not be provided by relatives or associates.
Qualitative data were collected through key informant interviews of District Medical Officers (DMOs) and Matrons at district hospitals in selected districts. A purposive sampling method was used to identify four high transmission districts, namely, Kenema, Port Loko, Bombali and Tonkolili. A fifth district, Bonthe, which had not reported any HCW infections at the time of our study, was also selected to better understand the challenges of EVD prevention in non-transmission districts. A key informant interview guide was developed by the research team and used to interview DMOs and Matrons in these districts. The key informant interview guide had five main questions that explored the availability and implementation of IPC policies in the districts, the perceived causes of HCW infection, the challenges associated with ensuring HCW safety, and what could be done to prevent HCW infection in the future. The key informant interviews were conducted by two members of the research team; six key informant interviews were conducted in total.
Survey data were entered into a Microsoft Excel database, cleaned, and then exported into Epi Info 7. In the first stage of analysis, we conducted univariate analyses on all variables in the database. In the second stage, we performed detailed descriptive analyses on the key variables. The descriptive analyses included the distribution of cases and deaths for person, location, time, IPC knowledge, disease outcome, location and type of exposure. We obtained the total number of HCWs in Sierra Leone from the National Health Sector Strategic Plan 2010–2015 [12] and used this to calculate the rate of EVD infection among HCWs. The key informant interview data were transcribed, and responses were coded and collated [13]. Some respondents did not respond to all the questions, which explains the differences in denominators in some results.
This study was part of extended epidemiological investigations to provide scientific evidence for initiating tailored interventions to control the EVD outbreak in Sierra Leone. The study was approved by the MOHS. All data presented herein are anonymous. In collaboration with the MOHS and DHMTs, the data collectors identified the affected HCWs, explained the study purpose, and obtained their verbal consent to participate in the study.