Fatal Rift Valley Fever Outbreak Caused By Exposure To Meat From Sick And Dead Livestock: Uganda, July 2018

Angella Musewa (  musewaa@musph.ac.ug ) Republic of Uganda Ministry of Health https://orcid.org/0000-0002-9399-1522 Bernadette B Mirembe Republic of Uganda Ministry of Health Alex R. Ario MOH: Republic of Uganda Ministry of Health Doreen Birungi MOH: Republic of Uganda Ministry of Health Lilian Bulage MOH: Republic of Uganda Ministry of Health Esther Kisaakye Republic of Uganda Ministry of Health Benon Kwesiga Republic of Uganda Ministry of Health Steven N. Kabwama Republic of Uganda Ministry of Health David Muwanguzi Republic of Uganda Ministry of Health Daniel Kadobera Republic of Uganda Ministry of Health Steven Balinandi Uganda Virus Research Institute Deo N. Birungi National Disease Diagnostic and Epidemiology Centre


Introduction
Rift Valley fever (RVF) is a zoonotic viral hemorrhagic fever (VHF) caused by the arthropod-borne RVF virus (1) (2). RVF is transmitted from infected animals to humans through handling of animal tissue, ingesting unpasteurized or uncooked milk of infected animals, and bites from infected Aedes mosquitoes (3) (4). Although disease in humans is often asymptomatic, a subset of patients develop hemorrhagic symptoms, with a case-fatality rate as high as 50% (6) (7). Early supportive treatment, including uid replacement, is associated with improved outcomes and survival (8).
The rst case of RVF was identi ed in 1931, among sheep on a Merino sheep farm in Rift Valley, Kenya during an unusually heavy rainfall season (9). Since that time, RVF cases have been reported in over 30 African countries (10). In Uganda, the rst RVF case was reported in 1968 in livestock (11). In 2016, an outbreak occurred in Kabale District, Western Uganda, which affected both animals and humans (12). Subsequent outbreaks have been reported in Mityana and Kiboga Districts with both animal and human cases (13).

Case de nition and case nding
We de ned a suspected case as acute onset of fever (> 38 o C) with a negative malaria test, and ≥ 2 of the following symptoms: headache, myalgia, and any gastroenteritis symptom (nausea, vomiting, abdominal pain, diarrhoea) in a person from Mbarara, Sheema, Ntungamo, case was a suspected case with thrombocytopenia, low white blood cell counts, or elevated hematocrit plus at least one of the following: unexplained bleeding symptoms, such as bleeding from the nose, vomiting blood, coughing blood (without history of TB), blood in stool, blood in urine, sudden change in vision (e.g. blurred vision, poor vision, seeing oaters, pain in the eye, bilateral orbital pain), jaundice, or any unexplained death in a person with an epidemiologic link to con rmed cases. A con rmed case was a suspected or probable case which tested positive for RVF nucleic acid by reverse-transcriptase polymerase reaction (RT-PCR) (15).
We performed a record review at Mbarara Regional Referral Hospital and other health facilities in the affected districts. The main variables for which data were collected included fever (≥ 37 o C), vomiting, headache, and diarrhoea. We conducted a community case search with the help of community health workers. We conducted case investigations using a standard VHF data collection form, and collected additional information related to the patients' illness (16).
Additionally, for case-patients who were butchers, abattoir workers, and meat dealers, we asked about the sources of meat they purchased.

Descriptive epidemiologic analysis
We conducted a descriptive epidemiologic analysis of the cases by time, person, place, and symptoms. The population data for the descriptive epidemiologic analysis were obtained from each district's headquarters.

Hypothesis generation
We interviewed a convenience sample of 13 case-patients (or, for those that had died, their next-of-kin) who sought care at Mbarara Regional Referral Hospital about their possible exposures. We conducted an initial environmental assessment, where we observed grilled meat being sold on the streets of the eleven affected districts and butchering of dead livestock with little inspection.

Case-control study
We conducted a case-control study to test the hypotheses generated. A control was a person who had no fever from 1 June 2018 to the time of investigation. For each case, we randomly selected ve neighborhood controls, individually matched by sex. Since all cases were adults (≥ 19 years in age), we only selected adult controls.
We administered a structured questionnaire to case-patients/next-of-kin and controls to collect information on demographic characteristics (age, sex, and occupation) and potential exposures during the case-patients' effective exposure periods. We de ned the effective exposure period to be 2-6 days before the case-patients' onset of fever (i.e., between the minimum and maximum incubation period for RVF) (17).

Laboratory investigations
We collected 5 ml of blood from 17 of the 19 case-patients who survived to hospitalization. Samples were taken at the health facilities where the case-patients were being managed and were transported to the Uganda Virus Research Institute for PCR testing. PCR was performed on the RNA extractions to identify RVF virus using the TaqMan assay (18). RVF primers were designed from published GeneBank sequences, and applied using the established methodology (19). Additionally, as part of its livestock RVF surveillance effort, the Uganda National Animal Disease Diagnostic and Epidemiology Centre (NADDEC) collected serum specimens from livestock (goats, cattle, and pigs) in farms where human RVF virus infections were occurring, to test for RVF-speci c IgM antibodies using an Enzyme Linked Immunosorbent Assay (ID Screen® Rift Valley Fever Competition Multi-species ELISA. Montpellier, France) (15).

Environmental assessment
At sites where human cases were identi ed, we assessed for the presence of livestock that had died suddenly during June-August 2018. We interviewed 35 village local leaders and 18 farmer cooperatives to understand more about how livestock meat was transported, the conditions and hygienic practices at local slaughterhouses, and animal movement and trading between different geographic areas.

Data analysis
To account for the matched study design, we used conditional logistic regression to estimate the odds ratios and their associated 95% con dence intervals, with the matching variable being the matched case-control set. At all levels of the analysis, this was applied for both bivariable and multivariable analysis.

Descriptive epidemiology
We identi ed 19 case-patients, including 17 con rmed and 2 probable cases. Two of the probable case-patients died before hospitalization: clinical data were unable to be established for one, who was found dead on the roadside with severe hemorrhaging but who had an epidemiological link to con rmed cases in Ntungamo. This patient was a known butcher in Sanga (Kiruhura district) and reportedly purchased meat regularly from Ntungamo market, where four con rmed cases were working.

Hypothesis generation
In hypothesis generation interviews, nine (69%) of 13 case-patients/next-of-kin interviewed ate meat from sick animals and eight (62%) cut meat from sick/animals. We therefore hypothesized that exposures to meat from sick/dead animals caused the outbreak.

Case-control study ndings
We analyzed associations between various exposures to meat or tissue of livestock that had died of unknown causes (skinning/butchering dead/sick livestock, handling uncooked livestock meat, cutting meat of dead/sick livestock, eating grilled meat of dead/sick livestock, and handling livestock abortus).
The 18 patients for whom we were able to ascertain data were included in the case-control study. Cutting/handling meat from sick/dead livestock eating grilled meat from livestock that died suddenly, and touching uncooked livestock meat were associated with illness (Table  3). * OR = Unadjusted odds ratio from conditional logistic regression; CI = con dence interval. † OR adj = Adjusted odds ratio from conditional logic regression.
N.B 17 case-patients were virologically con rmed for RVF infection, two probable cases died before con rmation but had epidemiological links to the con rmed cases.

Laboratory and environmental investigation ndings
All 17 human blood samples taken tested positive for RVF. Of the samples from 100 livestock taken from thirteen farms where human RVF infections occurred, 72 tested positive.
We conducted assessments at 18 sites where human cases were lived or worked. During interviews with 18 farmer cooperatives in the affected districts, 22 (52%) reported that livestock on their farms had died recently; the carcasses had been butchered and the meat was shared and sold around villages. After the outbreak was reported, police checkpoints were set up in the affected districts to identify and stop any illegal trade of meat from animals that died of unknown causes. However, police reported that some butchers were able to evade the checkpoints by traveling during times when the checkpoints were not functioning, or by hiding meat in packages not typically used to transport meat.

Discussion
Our eld investigation revealed the largest RVF outbreak in the past 50 years in Uganda. The outbreak occurred primarily in the "cattle corridor" of Uganda, and was associated with exposure to meat from livestock that died suddenly.
Exposure to meat from infected animals or exposure to abortus has been well-documented as a risk factor for RVF infection among humans in several studies in Africa and the Middle East (20)(21)(22)(23)(24) (25). Men are often more affected than women due to the frequency of their exposures, as was seen during this outbreak (22)(26). Younger adults were also more affected than older adults, likely due to differential participation in butchering and slaughtering activities.
Isingiro district, which borders Tanzania and Rwanda, was more affected than other districts. In addition, the majority of case-patients reported purchasing their meat from Isingiro District, reportedly because it was more economical than purchasing it elsewhere. During the time this outbreak was occurring, Rwanda, which borders ve of the affected districts including Isingiro, was experiencing an epizootic with suspected RVF cases reported (27). While it is unknown what started this outbreak, animal trade and movement between countries may have played a role. Isingiro also has a central abattoir where livestock are slaughtered and meat was distributed to the neighbouring areas, which could potentially have spread the infection, as well as a high cattle density, which has been associated with RVF outbreaks previously (28).
Beyond handling of meat, eating grilled meat from animals that died of unknown causes was also a risk factor in this outbreak, as in previous outbreaks (29) (25). Opportunities to eat meat in Uganda may be rare for poor communities, and depend on price of meat. Although animals that are sick or die of unknown causes are required by Ugandan regulations to be inspected by a quali ed veterinarian before butchering, this does not always happen in practice. Instead, meat from sick livestock is often sold outside of o cial lanes, and more cheaply (~$1 USD per kilogram) than meat from healthy animals (~$4 USD per kilogram). In the area of Uganda where the outbreak occurred, residents typically grill chunks of meat on open re; often the outside is burned, while the inside is still raw, which can facilitate transfer of infection to humans (30). In areas where RVF is a risk, the public should be educated about these risks and understand the need for complete cooking of meat. In addition, enforcement of regulations about slaughter of uninspected livestock need to be strengthened.
The case-fatality rate during this outbreak was high compared to previous outbreaks. High case-fatality rates for RVF have previously been associated with delays in seeking medical care (31)(26) (23). In this outbreak, patients who died were less likely to have sought care than surviving patients, and their average delay in care-seeking was longer than that of surviving patients. Persons working in high-risk occupations and living in areas where RVF occurs should receive education on the signs and symptoms of RVF infection, and the importance of early care-seeking for survival.
Currently, an inactivated RVF vaccine has been developed for human use (32). However, this vaccine is not licensed and is not commercially available. In animals, several live attenuated vaccines have been developed, and are meant for use in RVF-endemic regions (33). Although Kenya currently vaccinates all livestock against RVF, this is not carried out in Uganda, Rwanda, or Tanzania (34). Periodic and reactive vaccination could be employed strategically to effectively reduce the incidence of RVF outbreaks (35). An evaluation of the cost-effectiveness of livestock RVF vaccination in Uganda, either nationwide or in speci c hotspots, would be helpful to inform policy.
Collaborations between animal, human, and environmental health partners (One Health approach) are key in rapid detection, response, and prevention of zoonotic outbreaks. However, optimizing these collaborations remains a work in progress. During this outbreak, early communication from animal health partners to human health partners about livestock sampling results could potentially have allowed intensi ed and focused risk communications to the populations at risk. Collective efforts among different partners in the entomologic and environmental elds as well as human health elds could also facilitate group decision-making on issues such as vector control, potentially reducing other mechanisms of RVF spread.

Conclusion
This RVF outbreak was caused by physical exposure to raw meat or eating meat from infected livestock. We recommended health education to the public on the dangers of handling and eating meat from livestock known to be sick or found dead, and the importance of thorough cooking of meat. We also recommended to adopt a One Health approach for effective control of RVF outbreaks. Speci cally, partners should work under the existing One Health platform, coordinated by a central body, through which data can be shared in real-time and response activities can be discussed and planned together by partners from multiple areas of health expertise (36). This can both improve e ciency of resource use and reduce the number of areas affected. We further recommended that the Ministry of Agriculture, Animal Industry and Fisheries conduct routine surveillance for RVF in livestock and enforce rules and regulations regarding meat inspection.
Based on our ndings, the district health o ces conducted radio talk shows in the affected communities to sensitize them on the dangers of eating meat from animals that are sick or have died of unknown causes. The local governments of the affected districts also strengthened meat inspection by deploying meat inspectors at border sites as well as several meat-selling markets for the next three months.

Availability of data and materials
The data that support the ndings of this investigation belong to the Uganda Public Health Fellowship Program and are not available in publicly available repositories due to ethical concerns. However, the data can be accessed upon reasonable request from the corresponding author and with permission from the Uganda Public Health Fellowship Program.
Author contributions AM, BBM, ARA, DB, LB, EK, BK, SNK, DM, DK, SB, and DNB conceived and designed the study. AM, BBM, DM, EK, BK, DM and SNK contributed to data collection, cleaning, and analysis. AM, BBM and DB implemented the study. AM and DK analyzed the data. AM, BBM, and DNB took lead in writing the manuscript. AM, BBM, ARA, DB, LB, EK, BK, SNK, DM, DK, SB, and DNB participated in manuscript writing to ensure scienti c integrity and intellectual content. All authors contributed to the nal draft of the manuscript. All authors read and approved the nal manuscript.

Ethical clearance and consent to participate
This investigation was in response to a public health emergency (RVF outbreak) and was therefore determined by the US Centers for Disease Control to be non-research and that its primary intent was public health practice or a, epidemic or endemic disease control activity. The MoH Uganda through the o ce of the Director General of Health Services gave the directive and approval to investigate this outbreak. We obtained verbal informed consent from case-persons during this investigation and other interviewed community members that were above 18 years. We ensured con dentiality by conducting interviews in privacy ensuring that no one could follow proceedings of the interview. The questionnaires were kept under lock and key to avoid disclosure of personal information of the respondents to members who were not part of the investigation.

Consent for publication
Not applicable

Competing interests
Authors declare no competing interests