Box 1.3 Counting deaths differently

Different methods of estimating mortality yield different results, making it difficult to know which methods are preferable. Because actual mortality data are hard to come by for developing countries, the World Health Organization (WHO) has historically estimated these global numbers using a range of techniques, including modeling and extrapolation from local studies. Using these well-accepted techniques, WHO estimates that for 1993, infectious diseases accounted for 40 percent of the 51 million deaths, while noncommunicable diseases such as heart disease and cancer accounted for 36 percent, and injuries another 8 percent. WHO attributed fully 16 percent of all deaths in 1993 to unknown causes (1).

Using a different approach, other scientists within WHO and Harvard University have recently generated quite different estimates, published in a 1996 book The Global Burden of Disease (2). In this study, the researchers strived to avoid double-counting deaths. (In other words, they ensured that the sum of all childhood deaths of specific diseases did not exceed the number of childhood deaths.) The Global Burden of Disease (GBD) team also developed a technique to assign deaths from unknown causes to a specific cause, distributing those that occurred among children younger than age 5 to infectious diseases, and those that occurred to children older than age 5 to chronic diseases. The result? Noncommunicable diseases, rather than infectious diseases, top the list as the world’s largest killer, accounting for 56 percent of all deaths. The infectious disease toll is 34 percent, and injuries account for 10 percent. In all, this study provides a very different snapshot of the world, showing developing countries to be much farther along in the demographic transition than previously expected. The table below shows global estimates for specific diseases from these two different studies, one for 1993 and the other for 1990. (Because disease trends vary only slightly from year to year, the fact that these two studies took place in different years is not believed to have a bearing on results.)

Each estimate has its own, often vociferous, advocates and detractors. Alan Lopez of WHO, one of the GBD study coauthors, along with Christopher Murray of Harvard, concedes that “there is large room for error in global mortality estimates” – both theirs and WHO’s. “We might well be off by a factor of 2, but I doubt that we are off by a factor of 10” (3). These differences, and the debate over them, underscore the tentative nature of all global mortality estimates, which should be seen as approximate ranges, not absolutes. It also reinforces the need for better data collection.

The controversy over mortality estimates pales in comparison to that engendered by a new indicator the same team developed to measure both death and disability attributable to a particular disease