In the nineteenth century the world was transformed by the industrial revolution. In the twentieth century it was the turn of the information revolution. The scale and pace of change is truly amazing. In the 1940s, Thomas Watson, the head of IBM, is reported to have forecast that the total world demand for computers would be at most five machines! Today, the vast majority of families in high income countries are connected to the internet.
These technological advances are being used for many purposes. Some are clearly beneficial. It is a great advantage to be able to check one’s bank balance or book an airline ticket whenever you want to. Yet some are more problematic. This week it was revealed that the US Department of Homeland Security has been accumulating comprehensive details of all travel undertaken by American citizens (and presumably others as well). The European Commission is planning a similar system. The United Kingdom, a country where urban areas are already almost entirely covered by surveillance cameras, is proposing to introduce a biometric identity card that will track every encounter that an individual has with an official agency, in the same way that loyalty cards allow supermarkets to monitor individual’s shopping habits. Data protection laws seem simply to be ignored.
Yet, while every move that we make in high income countries is being recorded by someone, in poor countries people are still born, live their often short lives, and die without anyone ever recording anything about them. Worldwide, only about 70 countries have any reasonable data on deaths of its adult citizens.

This will probably come as a surprise to many people, familiar with graphs and tables that purport to show life expectancy in countries such as Liberia or Sierra Leone. In fact, these data are simply guesses, albeit guesses that are informed by some scraps of evidence (or what some people would call “estimates”).
What has happened is that standard life tables have been created, showing what is thought to be the probability of death at different ages in countries exhibiting certain characteristics. Then, data on deaths in infancy and childhood are identified, typically from surveys, and are fed into the life tables to give an overall life expectancy. Obviously, this is critically dependent on us having a good understanding of the relationship between deaths in childhood and deaths in adulthood, which we now realise we don’t have. In other words, we really have no idea about what is happening to adult mortality in much of the world.
This week I was invited to Seattle to join a small group of people to discuss what might be done. The meeting was organised by Chris Murray, who has recently moved from Harvard to the University of Washington, where he has established the Institute for Health Metrics and Evaluation. We had convened within the framework of Grand Challenge 13, funded by the Bill and Melinda Gates Foundation. The challenge is to devise new ways of accurately measuring population health.
We spent the first day looking at the problem of simply capturing data on how many people have died. As Alan Lopez reminded us at the end of our discussions, the gaps in our knowledge are a “scandal of ignorance”. There is little doubt about where we need to be. All countries should have effective systems of vital registration. Yet for many this is still at best a distant prospect, especially those where establishing even the most basic governance functions seems as far away as ever. There are, however, possible intermediate steps, such as sample surveillance, where data are collected from a sample of locations, in the hope that they will be reasonably representative of the overall population. This is what is done in India and China. Then there are the indirect methods, based on data from surveys. Yet none of these are perfect and we still face many unanswered questions about the validity of the methods we are using. A problem in many parts of the world is that many people do not know what age they are. This can lead to what is called heaping, where reported ages are concentrated in numbers ending in 5 or 10. However I was fascinated to learn, although I suppose I should have realised, that in societies where astrological correlates of birth dates are important, people are much better informed. Ken Hill told me that the distribution of ages in the 1953 Chinese census is perfect.
Day two looked at the even more difficult problem of collecting data on cause of death. Here, a degree of realism is needed in what can be achieved. Even in countries with the best possible systems, there will always be considerable uncertainty about the main cause of death in older people who have multiple disorders. Yet it is clear that even here we can do better, in particular by understanding the principles that are used in different countries in assigning a single cause of death where several co-exist.
Where vital registration systems don’t exist, an alternative is to use a “verbal autopsy”, where surviving relatives are asked a series of structured questions about the deceased. Yet here too there are many methodological issues unresolved about how best to allocate a cause of death. Computerised systems are consistent but not always correct. Physicians inspecting the data are less consistent, but may be more often correct. One interesting possibility proposed by Chris Murray was the use of a computerised model that would take the reported signs and symptoms and, based on a validated data set from the same (or a similar) location, allocate a probability to different causes of death. If combined with clinical judgement (i.e. the physician is presented with the probabilities of different causes and, using any additional information available, decides on the most likely cause) this could be a valuable way forward. Clearly the increasing availability of hand held computers offers considerable potential. This would also overcome the problem seen in many existing sentinel surveillance sites of piles of paper forms lying uncoded long after the events they describe took place.
There are other opportunities too. It was pointed out that we are coming up to the next round of censuses in many countries, typically conducted every ten years. It would be possible to include a question asking whether anyone had died in a household in the past year or so and, where this had happened, to follow it up with survey teams applying a shortened verbal autopsy instrument.
Of course, none of this will happen unless the world community begins to take adult mortality seriously, something that it has so far singularly failed to do. What efforts exist have focused on child and maternal mortality. It was even suggested that these efforts have diverted attention away from adult mortality. The problem, as is so often the case, is that we are in a vicious cycle. The priority for international development is the need to reduce child and maternal mortality rates, because these are often the only figures we have on population health in many parts of the world. Yet because these are the priority, no-one (except the Gates Foundation) is willing to invest in the collection of data on anything else.
But maybe there are solutions to the problem of resources. As I came back through Heathrow the iris scanning machine, designed to let frequent travellers pass through immigration a little quicker, was yet again out of order (as it had been last week too). As I noted above, the British government is about to spend billions of pounds (the exact amounts are shrouded in spin and obfuscation, as usual) on a system of biometric identity cards that is doomed to failure (the full account of the failings are in an excellent account by a team at the London School of Economics). If only a fraction of the resources being devoted by the British and American governments could be diverted from the almost entirely pointless and futile attempts to track every move made by their citizens, then maybe we might at least be able to move away from a position where our fellow human beings can live and die without anyone ever recording it. What is more, the much simpler technology required is at least likely to work.