Research Article: Attacks on healthcare facilities as an indicator of violence against civilians in Syria: An exploratory analysis of open-source data

Date Published: June 10, 2019

Publisher: Public Library of Science

Author(s): Sayaka Ri, Alden H. Blair, Chang Jun Kim, Rohini J. Haar, Mary C. Smith Fawzi.


Grasping the human cost of war requires comprehensive evaluation of multiple dimensions of conflict. While the number of civilian casualties is a frequently used indicator to evaluate intensity of violence in conflict, the inclusion of other indicators may provide a more complete understanding of how war impacts people and their communities. The Syrian conflict has been specifically marked by attacks against healthcare facilities, and the advancement of technology has provided an avenue for remote data analysis of conflict trends. This study aims to determine the feasibility of using publicly available, online data of attacks on healthcare facilities to better describe population-level violence in the Syrian Civil War.

This study utilized publicly available datasets from the Violations Documentation Center (VDC) and Physicians for Human Rights (PHR) to compare trends in attacks on healthcare facilities and civilian casualties from March 2011 to November 2017 in the Syrian Civil War. We used descriptive statistics, bivariate tests and a multivariable hypothesis testing model to measure the association between the two indicators while adjusting for confounding variables.

We examined for associations between attacks on healthcare facilities and overall civilian casualties. In the adjusted regression model, each attack on a healthcare facility in the Syrian conflict corresponded to an estimated 260 reported civilian casualties in the same month (95% CI: 227 to 294). This model adjusted for population displacement (using number of registered refugees as a proxy). The May 2014 interaction term, used a transition point of early/late war based on political events during that time, illustrated that each healthcare facility attack after May 2014 corresponded to a statistically significant decrease of 228 civilian deaths. This suggests that although attacks on healthcare facilities continued to contribute to overall civilian deaths, the scale that this was happening was lower after May 2014.

In the Syrian Civil War, our findings suggest that the inclusion of other humanitarian indicators, such as attacks on hospitals, may add granularity to traditional indicators of violence (e.g. such as civilian casualties) to develop a more nuanced understanding of the warring tactics used and violence against civilians in the Syrian conflict. This exploratory case study represents a novel approach to utilizing open-source data along with statistical analysis to interpret violence against civilians. Future research could benefit from analyzing attacks on healthcare facilities and other civilian infrastructure concurrently with civilian casualty data for further data-driven utilization of open-source data.

Partial Text

Understanding violent trends against civilians is important in informing relief efforts, advocating for foreign policy, and holding stakeholders accountable to violations of international humanitarian law (IHL). Despite its importance, measuring the scope of violence towards civilians remains challenging [1]. Although many indicators have been applied to measure conflict-associated violence, the total number of civilian casualties is commonly used as the standard index to evaluate the severity of war [2–4]. However, the methods used to count casualties are often unreliable in an ongoing conflict [5]. The breakdown of civil registries and absence of a centralized death registration system can result in an under-reporting of total casualties. It may also result in an over-reporting of identified casualties as databases may contain duplicate counts, or may include reports of non-conflict related fatalities [6]. Since accurate documentation of civilian deaths is difficult, challenges may arise from using casualties as the sole indicator of violence against civilians.

Several non-governmental and not-for-profit organizations maintain publicly available, online databases of statistics from the Syrian Civil War. Databases were identified through a broad, web-based search and were chosen based on the following inclusion criteria: (1) open-source and available publicly, (2) availability of monthly aggregated data, (3) a published and transparent methodology for how data were collected and presented, and (4) a multi-tier verification procedure before publishing data on incidents. Given the wide variability in civilian death counts, the source for the civilian casualty database was selected based on additional requirements: (1) differentiation of conflict related deaths by civilian and combatant status, (2) data reported from March 2011 to November 2017, and (3) a database that was updated at least monthly.

Six online, publicly available databases that record information on the Syrian conflict were evaluated. These databases include Médecins Sans Frontières (MSF) and Physicians for Human Rights (PHR) for data on healthcare facility attacks, and the Syrian Center for Statistics and Research (CSR-SY), Syrian Observatory for Human Rights (SOHR), Syrian Network for Human Rights (SNHR), and the Violations Documentation Center (VDC) for civilian casualty data [20–25]. Based on the inclusion criteria discussed above, databases from PHR and the VDC were selected from this list (S1 Table). PHR provides the most comprehensive open-source dataset on healthcare facility attacks as the only organization that started documenting attacks from the start of the Syrian conflict in March 2011. VDC was chosen as the only organization to provide a database to distinguish between civilian or combatant casualties with a published, multi-tier methodology for users to review online. The UNHCR’s online database was also sourced to extract cumulative totals of registered Syrian refugees as a confounding variable (S1 Table). Table 1 outlines the sources, verification methodology, and descriptions of the online sources.

To our knowledge, this is the first study to utilize publicly available, online data to explore the relationship between attacks on healthcare facilities and civilian casualties during an active conflict. The descriptive analysis highlights that documented attacks on healthcare facilities can be useful to understand the extent of violence against civilians in conflict. From Fig 1, there were upward trends in civilian casualties and attacks on healthcare facilities early in the conflict that diverged later in the conflict. As the war progressed, specifically from May 2014 onwards, attacks on healthcare facilities continued to increase erratically while the number of civilian casualties per month plateaued. The two indicators of violence, civilian casualties and attacks on healthcare facilities, provide different narratives on the level of violence committed against civilians in the latter period of the Syrian Civil War. Using civilian casualties as the only indicator to assess violence against civilians would suggest that Syria is becoming less dangerous for civilians, which according to attacks on healthcare facilities, might be misleading.

This study highlights the added value of including indicators of violence other than civilian casualties to provide a more nuanced narrative around the human costs of war. In conflict settings, damage to social infrastructure can provide insight to the warring tactics of armed forces and violence perpetrated against civilians. With the introduction of recent technological advances, investigators are offered an opportunity for a more comprehensive understanding of violence from data on healthcare facilities in conflict that may not be illustrated in civilian casualty reports.