Interview of Professor Bob Snow, who tells us about his paper published in Nature in October 2017.
VIDEO TRANSCRIPT: 115 YEARS OF MALARIA IN AFRICA
My name is Professor Bob Snow, I work for the Kenya Medical Research Institute Wellcome Trust Programme in Kenya and the University of Oxford in the UK, and I am here today to talk about a paper published this week in Nature (11th October 2017).
The work represents 21 years of research effort, which began in 1996 and focuses on the malaria parasite prevalence rate – that’s a measure of the quantity of malaria, often done through surveys of communities to work out how many people have the parasite in their blood. This is a good indicator of the amount of malaria in any given area, because it’s really hard to measure how many people die of malaria or how many people get sick of malaria in Africa, largely because they share symptoms and signs with many other causes, but actually knowing whether or not you have the parasite in your blood is pretty unambiguous.
We’ve focused on trying to find survey data that spanned over 100 years in Africa, bearing in mind that these surveys were first done in the early 1900s by researchers working in Sierra Leone, Ghana, Nigeria and other areas of Africa, and it’s been a huge effort. Largely we’ve had to go to archives across the continent, some of them incredibly dusty and dirty. We’ve had to dig down into the Ministry of Health’s basements, pull out boxes covered in woodlice and rat droppings to salvage some of these really old survey data. It’s been a fantastic experience, a bit like the Indiana Jones side of my job, trying to find and track down this survey data. We have had some fantastic experiences of finding people who have retired from malaria working in Southern Africa, identify where they live, go to their houses, spend a couple of days with them in their garage finding all the survey records from Botswana and Namibia in the 1950s and 1960s. We’ve been to libraries and archives of the old eradication programmes in Senegal, Ghana, Nigeria, Uganda, Tanzania, Sudan. Some of these have been great successes and little joys finding treasures of original survey data. For example, in Uganda we found the original survey records, cards, with everyone’s name, village and so on, when they had their blood sample taken, results of microscopy in Entebbe. We were told that the eradication programme did surveys in 1965 and 1967, and those records were housed in Jinja not in Entebbe. We found the man who had the key to the old store; we tracked him down, it took us about two days to find him and another day to find the key, only to discover that they weren’t there and that they had been moved to another part of the hospital in Jinja, it took us another two days to find that person with that key; we get there, we open the door and there’s nothing. Then he points out to us that two years beforehand, they were in fact all burnt, just to make space.
Part of this project has been to preserve these original records and find them before they completely disappear. Other examples are in Sudan where we tried desperately to find all the survey data done in the big surveys in the early 60s, where they sampled thousands of villages across Sudan – combined Sudan in those days, North and South – but only to discover, having tried for weeks in all possible locations, that they’re probably somewhere in Europe. So, this is an example a bit like the Elgin Marbles: there are original survey data but we just can’t track them down.
It’s been a long effort, and of course it’s been hugely helped by the very large number of malaria scientists working today across the continent. Over 900 malariologists working in the whole of Africa have been incredibly generous and shared their data with us. This is raw data, it’s often published but in an aggregate form and they’ve disaggregated it by village to share it with us a part of our project.
Once we’ve got access to this original material or published material, we then extract from that report where the village is. This takes a lot of effort because names change over time and we have to provide a precise longitude and latitude for each of the villages, how many people were surveyed, the month and year of the survey and how many were positive for Plasmodium falciparum which is the main parasite that causes the majority of disease in Africa. This then goes into a large database and we attach to that every single PDF, electronic digital copy of the original materials. These have now been archived and we have shared them with the World Health Organisation’s Global Malaria Programme recently, so that there is a legacy of these data and there is an institutionally legitimate partner to house them. They are now with the WHO in Geneva.
Once we have all the data – which took 21 years to assemble – we need to try and understand “what does it mean?” Our main objective, for a very long time now, has been “what has been the long term history of malaria in Africa?” Many people who have published lots of work recently have focused on 2000 to 2015 only, and many of them have shown that malaria has indeed dropped. But actually, that doesn’t give us that lens of what happened before 2000; what did malaria look like in Africa before then? This big assembly of 115 years’ worth of data now adds to over 50,000 survey points, geolocated, in nearly 30,000 villages, gives us a unique opportunity to try and understand what has happened over 100 years. To do this, we didn’t want to over-model it. One of the temptations is to create very complicated models which include lots of layers of information on how climate has changed, how interventions have changed, and you build it all together and you try and use what they call covariates to predict what malaria is in any given location. Rather than do that, we’ve decided to actually just let the data themselves tell the story.
We haven’t used any covariates, we’ve taken each of the individual survey points and displayed them over 520 administrative units across Africa – these are largely administrative units, quite large ones – and Madagascar, the largest of the off-shore islands. We haven’t included the other smaller islands, so it’s just mainland Sub-Saharan Africa and Madagascar divided into 520 units. Then we have tried a simple statistical model, which smooths the data in space and in time, borrowing information from each of the points in each of those administrative areas, to give us a prediction of malaria based on the data in each of those units for one of 16 time points, starting in 1900 and ending in 2010, 2015. So, we’ve segmented time and we’ve segmented space. Then we summarise all that data, just by having the median estimate for the whole of Africa and Madagascar over those 16 time periods.
These data show that there was a huge amount of survey data collected in the 1950s and 1960s in preparation for malaria eradication – Africa was part of the global malaria eradication era. There were lots and lots of survey efforts during that period, and there’s a lot of data more recently during the period of the roll back malaria initiative and new investment by the global funds. Our periods of greatest data are the 1950s/1960s and more recently since 2005. Nevertheless, there is quite an awful lot of data available to make some predictions throughout the rest of the time periods.
These data show that malaria probably affected 40% of children in 1900 to 1920, with parasite prevalence rates of 40% to 50% – every other child was carrying the malaria parasite. Over time that has dropped, quite considerably to about 24% in 2010/2015. That significant drop actually doesn’t tell the whole story, because there have been troughs and there have been peaks. These peaks are probably explained by a coincidence of excessive rainfall – the worst El Niño rainfall that hit Africa happened in the late 1990s – and the emergence of chloroquine resistance. You had this congruence of two factors which led to the perfect storm, and malaria prevalence rose quite rapidly to a very high level in the late 1990s early 2000s. There was a slightly less easy to explain lull before that, and that lull was coincidental with a drought in the Sahara: no rainfall no mosquitoes, no mosquitoes no malaria, that’s kind of how one might understand it. But there was also a period when everyone was taking chloroquine and chloroquine was an incredibly effective drug. Every single fever was treated with this drug, and therefore parasite prevalence is likely to have been suppressed by, to all intents and purposes, mass drug administration of chloroquine. Of course that will then lead to the explanation of that massive rise when chloroquine began to fail. There was a big drop before then, and that big drop did coincide with the introduction of new tools for malaria, then after the Second World War that was chloroquine, but it was also DDT used for indoor residual house spray. So, you had these two so called magic bullets that we introduced to Africa in the late 40s early 50s and began to be used more widely across the continent, and you did see that big drop from the high levels in the 1900s, 1920s, 1930s.
Of course, if you now fast forward to the roll back malaria era, we have two other magic bullets: we have insecticide treated bed nets and artemisinin based combination therapy to replace failing chloroquine and sulfadoxine/pyrimethamine. Again, we see from about 2005 to 2010 one of the biggest drops that Africa has ever witnessed. That gives us an understanding of it, but it’s not as straight forward as people might imagine; people like to think that linear rises are associated with linear declines. For example, there is a very large body of research that suggests that global warming has affected malaria, and it has: global warming does affect malaria at the margins, it affects the sea surface temperatures of the Pacific, it actually leads to excessive rainfall. It’s not to deny the climate’s importance, it’s just that the linear association between an increase in land surface temperatures and an increase in malaria can’t be explained by our data on their own. Nor can for that matter an increase in economic development. Africa has witnessed a huge increase in GDP – almost all countries, not universally but most countries have witnessed a very large increase in GDP; there’s been an increase in the percentage of girls going to school; a lot of development has happened, but that can’t explain the epidemic we witnessed in the late 1990s early 2000s.
What this all tells us is that things are complicated, and that doesn’t really help donors or people wanting to project the future of malaria in Africa. But things are complicated and can’t be explained by intervention alone, or climate, or development – they are congruence and composite of these elements, and actually that is one lesson of this paper: we need to understand the situation in its entirety and don’t make too many assumptions.
There are some things though that we might be able to predict for the future. The first is that resistance is an absolute killer for malaria and the cycles of malaria on the continent. If drug resistance – de novo artemisinin drug resistance – was to hit Africa, if pyrethroid resistance was to escalate to the point where insecticide treated bed nets were no longer effective, and we had another El Niño crisis, I think we will see another epidemic similar to the one we saw in 1990s/early 2000s.
The other thing, if you look at the mapped distribution of malaria over time there is one thing that sort of jumps out at you: there is a central belt of Africa running from below The Gambia from Guinea Bissau, through west Africa, Ghana, Nigeria, Burkina Faso, all the way through Central Africa down to the southern part of Mozambique, where P. falciparum malaria infection prevalence hasn’t changed as much as it has at the margins. As one probably might expect, this is an area that’s home to Anopheles gambiae sensu stricto which is the most efficient vector that we know on the planet.
It does beg the question as to whether or not we will achieve global eradication in our lifetimes. This is a quote that is often used because it might be possible to remove those last residual numbers of cases on the Solomon Islands or Saudi Arabia. But focusing your attention on global eradiation and elimination at the margins ignores the fact that you have this belt that has remained pretty much the same for 100 years in Africa. That’s millions and millions of people who continue to suffer from the disease and for which the tools that we have at our disposal now – which is prompt treatment with effective medicines, insecticide treated bed nets, some attempts at indoor residual house spraying – aren’t enough, so we need new tools.
Two things: we can’t ignore that central belt of Africa, and secondly we need some new tools in our toolkit to tackle that burden. In some ways, I am somebody who doesn’t necessarily buy into the whole sort of elimination notion for Africa over the next 30 years. I think we need to change the narrative, change the language to the point where, bearing in mind that for 100 years or so we haven’t made that much of a dent in that middle belt. One thing we can probably promise and can probably guarantee is that no child should die of malaria, nobody should die of malaria. Zero deaths is an achievable ambition over the next 30 to 50 years, but I think trying to shrink the map in Africa, as shown by the paper published this week (11th October 2017) and eliminate malaria transmission in that central belt is almost an impossible dream, certainly within our lifetimes.