Background Humanitarian settings often present unique scientific challenges and conditions that distinguish them from standard research settings. While a number of these challenges are faced in both standard settings and humanitarian settings, factors unique to humanitarian settings such as inaccessibility and time sensitivities further exacerbate the effects of these challenges. This analysis focuses on experiences in post-disaster contexts such as Indonesia and India following the 2004 Indian Ocean Tsunami, the Philippines following Typhoon Haiyan in 2013, and Nepal following the 2015 earthquake. Discussion Particular issues that we faced in undertaking research in post-disaster settings include challenges with uncharted ethical and cultural considerations, non-standardised administrative methods for record keeping, data sharing and dissemination. While these issues are not unique to post-disaster humanitarian settings, the time-sensitive nature of our work exacerbated the effects of these concerns. Relying on local partners and making quick decisions to tackle issues is imperative for navigating both foreseen and unforeseen challenges. While pre-emptive action to address these concerns is the most efficient means to expedite research protocols, adaptability and contingency planning are key components of practical research implementation in dynamic situations. Conclusions Research is not always a priority in humanitarian settings, so innovative methods are necessary to conduct meaningful and situationally appropriate research in these venues. By understanding available resources, local culture, and political considerations and working efficiently and decisively, we can begin to jump hurdles associated with epidemiologic research in humanitarian settings.

Conflict in Yemen has displaced millions and destroyed health infrastructure, resulting in the world’s largest humanitarian disaster. The objective of this paper is to examine mortality in Yemen to determine whether it has increased significantly since the conflict began in 2015 compared to the preceding period. We analysed 91 household surveys using the Standardized Monitoring and Assessment of Relief and Transitions methodology, covering 2,864 clusters undertaken from 2012–2019, and deaths from Armed Conflict Location & Event Data Project database covering the conflict period 2015–2019. We used a Poisson-Gamma model to estimate pre-conflict (μp, baseline value) and conflict period (μc) mean death rates using household survey data from 2012–2019. To analyse changes in the distribution of deaths and estimate nationwide excess deaths, we applied pre- and post-conflict death rates to total population numbers. Further, we tested for association between excess death and security levels by governorate. The national estimated crude death rate/10,000 in the conflict period was 0.20 (95% CI: 0.17, 0.24), which is meaningfully higher than the estimated baseline rate of 0.19 (95% CI: 0.17, 0.22). Applying the conflict period rate to the Yemeni population, we estimated 168,212 excess deaths that occurred between 2015 and 2019. There was an 17.8% increase in overall deaths above the baseline during the conflict period. A large share (67.2%) of the excess deaths were due to combat-related violence. At the governorate level, posterior crude death rate varied across the country, ranging from 0.03 to 0.63 per 10,000 per day. Hajjah, Ibb, and Al Jawf governorates presented the highest total excess deaths. Insecurity level was not statistically associated with excess deaths. The health situation in Yemen was poor before the crisis in 2015. During the conflict, intentional violence from air and ground strikes were responsible for more deaths than indirect or non-violent causes. The provision of humanitarian aid by foreign agencies may have helped contain increases in indirect deaths from the conflict.

No abstract