improvements should not distract from data accuracy, which would have been significantly improved with investments in health infrastructure. Contact tracing was far more effective when done by healthworkers and volunteers than through cell phone data, for example. NGOs filled much of the healthcare gap in the three most affected countries, but that made coordination and collection of data difficult during 25 the emergency. Rural regions, like the area where Ebola first emerged, have the lowest levels of both health care practitioners and facilities.68 g. Data sharing arrangements made before public health emergencies: For some relationships, particularly those involving large private sector and for-profit actors, template agreements on data ownership, conditions of subsequent and secondary uses, and distribution of rights may facilitate data sharing. Similarly, it appears that agreements between public health or aid agencies and contractors may be informed by pre-existing terms so that program officers will have guidance when data sharing questions are posed from contractors. Agreements with national, public funders may also require data sharing and enhance coordination. For example, trials of two Ebola vaccine candidates (ChAd3-ZEBOV and rVSVZEBOV) benefited greatly from an open collaboration between investigators and institutions in Africa, Europe, and North America. These teams, coordinated by the WHO, were able to generate and exchange critical data for the development of urgently needed, novel vaccines along faster timelines than have ever before been achieved.61 The consensus solution in the secondary literature, confirmed by interviewees, was to enhance data management capacity and analytic expertise in under-resourced settings and to establish data transfer agreement templates now in order to set conditions in the future for the proper use of data and assignment of credit.61 5. Conclusions: In the context of the Ebola outbreak, where there was a known pathogen without a licensed intervention, data sharing was important not only to the initial response including treatment of those infected and tracing their contact with others, it was also critical to the research response involving diagnostics, therapeutics, and vaccines. Epidemiological and surveillance data sharing was impeded by the unwillingness or inability of contact tracing organizations to do so because of contractual or funding interests. Once the data was shared, it was widely distributed on open platforms although its underlying accuracy was problematic because data entry was not standardized. Health facility data sharing similarly suffered in the early days of the epidemic because health facilities were being spontaneously constructed. After the dedication of a centralized response authority and a specific means to communicate that information (e.g. 117 in Sierra Leone), that data became more widely and effectively shared. Pathogen genome data was not widely shared. Indeed, it appears that thousands of samples sent out of Guinea and Sierra Leone remain unaccounted for, potentially subject to study, while few results were released. Genomic sequencing data bottlenecked with one or two 26 academic authorities who researchers trusted to verify results as part of the scientific process, but ultimately for publication, not for response. For research on biomedical interventions, data sharing was at its most robust when it was centrally coordinated and funded, for example, by the National Institute for Allergy and Infectious Diseases, the Wellcome Trust, Canadian Institute for Health Research, the World Health Organization, and similar large charitable and governmental stakeholders. This data sharing was less robust for negative or inconclusive results, which were either not released or delayed. In general, community engagement during the West Africa Ebola outbreak involved meetings with community leaders and public information campaigns. While many of these efforts were successful, the implementation of each exposed the need for preventative, ongoing community engagement to prepare for future outbreaks. Finally, political pressures negatively affected data sharing by creating an incentive for the Guinea and Sierra Leone governments specifically to view skeptically data pointing to a severe public health emergency, data which was in turn made questionable by the lack of health infrastructure available to gather and report it in a standardized way. Once the extent of the emergency became clear and decision-making infrastructure was put in place at the national level, data sharing became more robust. Each of these data sharing barriers may be addressed, at least partially and in some cases wholly, through planning and targeted investments. For some barriers, enablers emerged over the course of the Ebola public health emergency that may serve as models for future efforts at data generation, collection, sharing and analysis. For others, focused policy measures and monetary investments will be necessary to fill in gaps and remedy weaknesses, especially where centralized public health infrastructure is underdeveloped. 27 Bibliography: 1. Roberts, M. First Ebola boy likley infected by playing in bat tree. BBC, Dec. 30, 2014. https://www.bbc.com/news/health-30632453. Accessed July 24, 2018. 2. World Health