D.R. Congo’s war has killed roughly 5.4 million people due to death and disease, according to a report released by International Rescue Committee on Tuesday. But should we care about this number? Or maybe a better way to ask this question: is this the kind of research that humanitarian agencies should be doing?

According to IRC’s survey, there are an estimated 45 000 excess deaths per month, almost all due to hunger and disease—a rate of death that has not declined since combat activities began to subside several years ago. Death tolls are in fact higher in areas of central DRC, which has not seen war for several years. This is clearly an issue we should care about.

I’d discuss some of the problems with IRC’s figures, but to my surprise the New York Times devoted half an article to unpacking their statistical credibility. Turns out that the 5.4 million number is correct–plus or minus two million.

I love the Times, but it does not usually get high marks for (1) statistical sensibility or (2) nuanced coverage of African events. The credit goes to reporter Lydia Polgreen, who offers some of the consistently best African reporting by a Westerner.

Does it matter if the number is off by two million or more? Maybe not. The best use of such data, I imagine, is to prompt the New York Times (and maybe even lowly academic bloggers) to write an article on the subject–something that just might put the crisis on the agenda in the G8 or the UN security council (that would be the Times, not me). The IRC team interviewed 14,000 households across 700 remote locations. I’m going to hazard a wild guess and say that the research effort probably cost $350,000. Not a bad deal for coverage in several of the world’s leading newspapers.

Even so, I can imagine research that produces the same advocacy impact but has much more operational use. By which I mean any operational use. One of the researchers is quoted saying that, “These data can help us understand the scale of the problem and target our solutions to save lives.” This statement sounds much less impressive if you add “plus or minus two million people” to the end. No humanitarian program manager I know bases decisions on such data.

Wouldn‘t it be better if the survey and research design were more attuned to measuring the determinants of mortality than the levels? The influence of government programs? The effectiveness of alternative interventions? The relationship between mortality and local health care, or the difference when doctors and nurses show up to work?

A national mortality rate could easily be a byproduct of such an effort. It would be more poorly measured, of course, but nobody except me (and Lydia Polgreen!) is probably going to notice.

P.S. For everyone out there who is infuriated at this post, don’t worry. I’m having dinner with IRC’s research director, a close friend, on Friday night and she’ll let me have it.

P.P.S. I should also mention that IRC and others are already doing the kind of research I suggest, even in eastern DRC. The trick is to think more critically and creatively about the standard epidemiological surveys that are still a major focus of many humanitarian disasters.