This article was originally published on The Conversation website.

The bombing of a hospital in the Syrian city of Aleppo once again brings to the fore the precarious situation of humanitarian workers in war zones. Dozens were killed. The hospital was supported by the humanitarian organisation Médecins sans Frontières (MSF). Another MSF hospital was targeted in Kunduz, Afghanistan in 2015. There, 42 doctors, nurses and patients were killed.

Since the 1990s and the rise of conflicts in West Africa, Somalia, Chechnya, the former Yugoslavia and Africa's Great Lakes region, humanitarian organisations have been warning of greater insecurity for their staff. These observations are bolstered by surveys aimed at objectively quantifying violence against humanitarian workers.

The first survey of deaths among humanitarian workers was published in July 2000. But the importance of statistics in discussions on humanitarian security was established by the creation of the Aid Workers Security Database (AWSD). Its purpose is to catalogue every attack reported by the press or aid operators since 1997 that has resulted in the death, serious injury or kidnapping of aid workers.Unfortunately, these surveys have significant methodological weaknesses.

Methodological problems

There is no consensus on what defines humanitarian work. It is often very hard to know on what basis the employee of a humanitarian organisation was targeted by violence. Was it as a private individual, as someone from a particular country, as a member of a relief organisation, or something else?

The AWSD catalogues all attacks committed against a very wide range of people. These include representatives of foreign governments and funding bodies, UN officials and employees, representatives of the Red Cross and Red Crescent, volunteers, employees of international and local NGOs, members of religious or community solidarity networks, or employees of transport and private security companies.

And the meaning of such indicators is ambiguous. A decline in the number of attacks could, for example, reflect an improvement in the security conditions. But it could also reflect the withdrawal of humanitarian workers from areas that have become too dangerous.

Security databases count attacks and victims from reports in the administrative records of aid organisations or the media. This method cannot tell us whether what is changing is the phenomenon being studied (the violence), the reporting of it (documentation by of humanitarian organisations or the media), or some combination of the two. We don't know if there are more attacks, that more attacks are being reported, or both.

Studies that attempt to calculate attack and victim rates (the number of violent incidents and victims, respectively, per 10,000 workers per year) are hampered by the lack of a reliable denominator. It's a rare organisation that can provide accurate figures on its workforce by country and by year.

Despite such reservations, many studies and quantitative analyses conclude that the burden of insecurity on humanitarian workers is growing.

The AWSD data show that in absolute terms, the average annual number of victims has quadrupled over the past fifteen years. It was 78 in 1997-2001 and reached 376 in 2012-2013.

Compared with the rise in the number of humanitarian workers deployed in crisis areas, however, the number of attacks has been stable. The rate of workers killed, injured and kidnapped ranged between 4 and 6 per 10,000 per year from 1997 to 2012.

In other words, according to AWDS' own data, the number of victims has increased in proportion to the number of aid workers. In that sense, humanitarian action is no more dangerous than it was in 1997.

Is increasing humanitarian insecurity a myth?

In the final analysis, the quantitative studies on humanitarian insecurity are based on ambiguous indicators interpreted to confirm the preconceived idea that the danger is getting worse. Does this mean that no conclusions can be drawn from these studies, and that increasing humanitarian insecurity is a myth? Not necessarily.

There have been deliberate attacks against aid workers throughout the history of humanitarian action. Such violence is neither new, nor widespread, nor increasing in relative terms.

But the concerns expressed by aid workers are not irrational. The unfortunate reality is that in the past 15 years, the threat of kidnapping associated with insurrectionary Salafism has expanded geographically. Many contemporary conflicts are marked by extreme violence (in Syria and South Sudan, for example) from which aid workers are not exempt.

Context is everything

The problem is that quantitative studies such as the AWSD divert practitioners' and specialists' attention away from the real challenge - that of analysing each major security incident as a unique event, placing it in its local context and the historical context of the relief operation.

While it is urgent to rigorously identify attacks, it is even more urgent to discuss them based on detailed accounts that help us understand exactly what happened, and why.

Discussing the practices that contributed to (or helped mitigate) an attack can help an organisation understand whether the risks involved were worth taking.