Assessing the future threat from vivax malaria in the United Kingdom using two markedly different modelling approaches
© Lindsay et al; licensee BioMed Central Ltd. 2010
Received: 24 December 2009
Accepted: 5 March 2010
Published: 5 March 2010
The world is facing an increased threat from new and emerging diseases, and there is concern that climate change will expand areas suitable for transmission of vector borne diseases. The likelihood of vivax malaria returning to the UK was explored using two markedly different modelling approaches. First, a simple temperature-dependent, process-based model of malaria growth transmitted by Anopheles atroparvus, the historical vector of malaria in the UK. Second, a statistical model using logistic-regression was used to predict historical malaria incidence between 1917 and 1918 in the UK, based on environmental and demographic data. Using findings from these models and saltmarsh distributions, future risk maps for malaria in the UK were produced based on UKCIP02 climate change scenarios.
The process-based model of climate suitability showed good correspondence with historical records of malaria cases. An analysis of the statistical models showed that mean temperature of the warmest month of the year was the major factor explaining the distribution of malaria, further supporting the use of the temperature-driven processed-based model. The risk maps indicate that large areas of central and southern England could support malaria transmission today and could increase in extent in the future. Confidence in these predictions is increased by the concordance between the processed-based and statistical models.
Although the future climate in the UK is favourable for the transmission of vivax malaria, the future risk of locally transmitted malaria is considered low because of low vector biting rates and the low probability of vectors feeding on a malaria-infected person.
Over the last 30 years there has been a rapid increase in emerging and re-emerging infectious diseases in the human population [1–3], with over 175 species of pathogens now classified as emerging or re-emerging . Emergence of these diseases are frequently associated with ecological changes  and many are transmitted by insect or tick vectors [1, 4, 5]. Because of the sensitivity of malaria to a range of environmental factors there is concern that the disease may spread to new regions of the world [6, 7], previously disease free, like the UK.
Malaria was once common in marshland communities in central and southern England between 1500 and 1800, before finally disappearing in the early 1900s . Those areas most badly affected included the Fens, Thames estuary, the Kent coast, the Somerset levels, the Severn Estuary and the Holderness of Yorkshire; all substantial areas of marshland. The disease declined progressively from the 1850s onwards, as living conditions in the marshes improved and quinine, an effective anti-malarial, became more affordable and was used more frequently. The last significant epidemic of malaria in England and Wales occurred in 1917 and 1918, with the main focus on the Isle of Sheppey, near the mouth of the Thames Estuary [9, 10]. During this epidemic, there were 330 cases of locally-transmitted vivax malaria when infected servicemen returning from Macedonia were billeted near salt marshes. Indeed, all reported cases of locally transmitted malaria in the 1900s were vivax malaria, except for an unusual case of falciparum malaria in Liverpool . Whilst vivax malaria is a serious and debilitating infection, it is rarely a killer like falciparum malaria . Thus although historically mortality rates in the marshlands of southern England were several orders higher than inland areas, it is unlikely that the elevated mortality in the marshes was due to vivax malaria.
There are six species of Anophelines in Britain capable of transmitting both temperate and tropical strains of vivax malaria: Anopheles atroparvus, Anopheles algeriensis, Anopheles claviger, Anopheles daciae, Anopheles messeae and Anopheles plumbeus[13, 14]. Anopheles atroparvus is considered the most important vector of vivax malaria in the UK, since its distribution coincides best with the historical distribution of the disease . It also rests inside homes and will feed on people .
Modelling the threat from malaria is a necessary first step in estimating the populations at risk and where they live [17–19], helping to inform disease surveillance programmes across the world. To date there have been two broad approaches to modelling the distribution of malaria: process-based models and statistical ones. Process-based models of malaria spread are considered favourable since they are based on an understanding of the biology of malaria. They are largely driven by the manner in which temperature influences mosquito survival, frequency of blood-feeding and the development of parasites within mosquitoes [20, 21]. The major concern with this approach is that the models are solely temperature driven and no allowance is made for two potentially important environmental parameters known to affect vector populations; rainfall and relative humidity. Rainwater, groundwater and diluted seawater provide mosquito breeding habitats, along with a moderately humid environment which is conducive to vector survival . Statistical models capture the climate envelope of where the disease is present and produce relatively robust models of the disease's current distribution [18, 19]. However, they lack the clarity and understanding provided by process-driven models. Rarely are the two different modelling approaches used together. A simple process-driven model of malaria was compared with a statistical model to examine the risk of malaria transmission in the UK today and in the future under projected climate change.
The process-based approach
where f2 is a thermal sum, measured in degree days, representing the accumulation of temperature units over time to complete the development (105 degree days), g2 is a development threshold below which development ceases (14.5°C), and T is ambient temperature . Generally, as conditions warm the rate of parasite development increases [26, 27]. However, there is uncertainty about the minimum threshold for parasite development, with figures ranging from 14.5-16.0°C [25, 26, 28]. b is the proportion of female mosquitoes developing parasites after taking an infective blood meal (0.19) . r is the rate at humans with malaria recover from the infection. It is usually considered that the duration of each infection is therefore 1/r days. It was assumed that an average infection would be patent for 60 days, giving a value for r of 0.0167/day .
In the model, these formulae were combined with the 1961-90 UKCIP climate data at a 5 km resolution for present-day climate. The model outputs the number of months of the year when R0 is >1, indicating potential disease spread. Under conditions when R0 is <1 for a considerable proportion of the year, the disease probably cannot persist without continuous introduction from elsewhere, or possibly as quiescent stages within apparently recovered people. Future risk was modelled for 2015 and 2030 using the UKCIP02 Medium-high climate change scenario for the UK. This scenario uses the Hadley Centre global climate model (HADCM3) for a medium-high climate change scenario (SRES A2), which is used, via a further stage, to drive a regional version of the model that has a resolution of 50 km over Europe. Results from this regional model over the UK then form the UKCIP02 scenario. The distribution of saltmarsh was derived from the Land Cover Map of Great Britain 1990  which classifies land use in the UK into 25 classes at a 25 m resolution from satellite information. 1 km cells were classified as containing saltmarsh if they contained more than 0.5% saltmarsh according to the Landcover map. To validate this model, climate suitability maps showing the potential distribution of malaria in 1961-1990 were compared with distribution records of malaria recorded in 1859 and 1864 . To demonstrate whether temperatures were comparable between these periods, annual summer temperatures (June to September) from the Central England Temperature record  were compared for 1831-1860 and 1961-1990 using a t-test with SPSS software (PASW Statistics 18).
The statistical approach
where θ M is the set of non-zero parameters that define model M, and x is the set of factors associated with cell i. Here, it was assumed that the 1,618 cells, for which we have data, are independent. Given the scarcity of locally-contracted malaria (only 23 recorded cases), more complex spatially-explicit models that incorporate non-independence was not considered in the analysis, as there was unlikely to be sufficient information in the data to support such complexity.
where M denotes the model and θ is the set of K model parameters that are estimated from the data using maximum likelihood. Once AIC-values had been calculated for all models a Δ-value was calculated for each model, which is the difference between the AIC value of the model and the lowest AIC value calculated among all models. Thus, the model with the lowest AIC value was associated with a Δ-value of zero. Models considered to be parsimonious with the data were those that had a Δ-value < 6 . In addition, to avoid selecting overly complex models, we also disregarded models if they were more complex versions of models having a lower Δ-value .
Model selection results from the AIC analysis investigating the importance of environmental and population data on the prevalence of new malaria cases between 1917 and 1918.
MTW + APET
MTW + APET + LnP
The simple temperature-driven processed-based model demonstrated that present temperatures in the south of England are able to support transmission of vivax malaria for a few months each year. For most areas, this is just for two months, although the heat island effect of London means this city could support transmission for three months. If the climate becomes warmer, conditions for transmission become more favourable and last for longer. Under the UKCIP02 medium-high scenario used here, the risk of transmission is predicted to increase in the south of England, spreading northwards towards the Scottish border. The areas at greatest risk include the Thames Estuary, the Suffolk coast, the Fens, Romney marshes, the Southampton coastline, the Severn Estuary and parts of the South-West coastline. Interestingly, by 2030, the areas where the climate is predicted to be suitable for malaria for three to four months are those that once supported malaria in the past: coastal and inland marshes of southern England. It is important to appreciate that this analysis reflects average monthly values, and that in reality there will be large variations in climate from year to year, making some years considerably more suitable for malaria transmission than others.
It is likely that our predictions are robust for two reasons. Firstly, 93 of the 96 sites where malaria was recorded in the early-nineteenth century occurred within the climate suitability map for malaria in 1961-90, as reported previously . Although the cases and climate surfaces were from different periods, the summer temperatures, when conditions are best for transmission, were similar. Secondly the climate suitability map generated using a statistical model, fitted to an independent data set, was very similar to that generated using the process-based model.
An analysis of the statistical models demonstrated that the most important parameter to include was the mean temperature of the warmest month, providing support for using the temperature-driven process-based model in the analysis. Risk of malaria also increased with higher human population densities and drier conditions. Whilst it is logical to appreciate that the risk of an infective mosquito feeding on a person will be greater when there are more people, it is perhaps surprising that malaria was predicted to be more likely in drier conditions. This unexpected result was largely due to five cases of malaria occurring around the Suffolk coast that could not be accounted for by warm temperatures or high population density. However, since this region is close to saltmarsh, the favoured habitat of the historical malaria vector, An. atroparvus, it may simply reflect that the cases lived close to suitable vector habitats. This raises a general problem with our mapping approach. And that is, although climate governs the distribution of vectors and the pathogens they transmit at a broad scale, the actual risk depends on local conditions (e.g. habitat type) and the complex interactions between pathogen, vector and people.
Within the temperature-suitability zone we are uncertain where the vectors occur. The actual distribution of potential vectors in the UK is poorly defined and is based on data collected over 100 years ago or by entomologists in an unsystematic fashion [31, 38–40]. The concern is that these distributions may not reflect the true distribution of vector species. As a proxy measure to identify sites likely to have An. atroparvus, a landcover map of saltmarsh areas was used as a filter over the temperature-suitability map, since this mosquito is a coastal species, tolerant of brackish water .
At first sight these maps may give the impression that there is a significant risk of malaria returning to the UK. However, the field studies cast doubt on this conclusion and demonstrate the importance of carrying out fieldwork to validate the model predictions. In 2003, a survey of adult mosquitoes was carried out on the saltmarsh on the Isle of Sheppey , the last place in England to experience an epidemic of malaria . Using two carbon dioxide baited CDC light traps and two Mosquito Magnet® traps mosquitoes were sampled continuously from June to September, the period when most mosquitoes are searching for a host. At 1240 ha this is the largest area of saltmarsh in the UK/England, yet only 40 An. maculipennis (probably An. atroparvus) were collected. Of 83 blood-fed An. maculipennis collected from a horse stable, a derelict school and a derelict concrete pill box on the Isle of Sheppey (RH unpublished data), and analysed using a precipitin test , just two had fed on people indicating that around 2% of the population fed on people. The risk of being bitten by a potential malaria mosquito is, therefore, extremely low, even though there is a general problem with nuisance biting by other mosquito species on the Island .
How likely is it that vivax malaria could be transmitted in the UK? Malaria transmission requires that a potential vector should feed on someone carrying gametocytes, the stage of the parasite that is capable of maturing and becoming infective within a mosquito. Between 9-24 days after taking an infectious blood meal of gametocytes, depending on temperature  a vector mosquito will be potentially able to transmit the infection to anyone that it bites. In 2003 there were 206 cases of imported vivax malaria and approximately 66% of these were contracted by visitors to the Indian subcontinent . It is likely that this will under-represent the true number of people with circulating gametocytes since even in areas of low transmission many infectious semi-immune patients will not appear sick . Nonetheless, overall there are likely to be relatively few people carrying vivax gametocytes in the country, and since most of these are amongst people of Asian decent who tend to live in major urban areas, away from coastal marshes, the possibility of An. atroparvus, biting an infectious patient are remote. There remains the possibility that there are other vectors of vivax malaria in the UK since not all indigenous cases of malaria recorded in the 1917/18 outbreak lived near saltmarsh. In such cases, An. plumbeus or An. claviger might have transmitted the infection.
Overall, it is seems extremely unlikely that vivax malaria will be locally transmitted within the UK. Therefore, not more than a few very rare cases of autochthonous malaria in the UK would be expected over the next 50 years. Indeed, one is much more likely to be struck by lightning  than to get malaria from an English mosquito.
The threat posed by malaria in Europe has been assessed previously [47–49]. Processed-based  models and statistical ones  have both been used to explore the global risk of malaria, whilst at the local and countrywide scales processed-based models, of differing levels of complexity, have been used for assessing future risk in Europe (France [51, 52], Germany  and Portugal ). The present study differs from these since it uses both processed-based and statistical models to explore the risk of malaria, and validates both models with independent data sets of historical cases of malaria.
In conclusion, predicting disease risk based solely on climate-driven models is a logical and useful first step in the process of predicting risk. In the case of vivax malaria transmitted by An. atroparvus in the UK, more local factors become important, including the presence of specific habitats for mosquitoes and the distribution and abundance of infectious patients. The important message is that considerably more field-based research should be undertaken to test and improve the predictions made by the relatively simple models shown here.
This research was funded by the Department of Health and the Office of Science and Technologies Foresight Project on The Detection and Identification of Infectious Diseases and the Research and Policy for Infectious Disease Dynamics (RAPIDD) Program of the Science and Technology Directory, Department of Homeland Security, and Fogarty International Center, National Institutes of Health.
- Woolhouse M: Population biology of emerging and re-emerging pathogens. Trends Microbiol. 2002, 10: S3-7. 10.1016/S0966-842X(02)02428-9.View ArticlePubMedGoogle Scholar
- Cleavland S, Laurenson M, Taylor L: Diseases of humans and their domestic mammals: pathogen characteristics, host range and the risk of emergence. Phil Trans Roy Soc Lond B. 2001, 356: 991-999.View ArticleGoogle Scholar
- Morens D, Folkers G, Fauci A: The challenge of emerging and re-emerging infectious diseases. Nature. 2004, 430: 242-249. 10.1038/nature02759.View ArticlePubMedGoogle Scholar
- Taylor L, Latham S, Woolhouse M: Risk factors for human disease emergence. Phil Trans Roy Soc Lond B. 2001, 356: 983-989. 10.1098/rstb.2001.0975.View ArticleGoogle Scholar
- Guebler D: The global emergence/resurgence of arboviral diseases as public health problems. Arch Med Res. 2002, 33: 330-342. 10.1016/S0188-4409(02)00378-8.View ArticleGoogle Scholar
- Confalonieri U, Menne B, Akhtar R, Ebi K, Hauengue M, Kovats R, Revich B, Woodward A: Human health. Climate Change 2007: Impacts, Adaptation and Vulnerability Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Edited by: Parry M, Canziani O, Palutikof J, van der Linden P, Hanson C. 2007, Cambridge: Cambridge University Press, 391-431.Google Scholar
- WHO: Climate change and human health. 1996, Geneva: World Health OrganizationGoogle Scholar
- Dobson MJ: Contours of death and disease in early modern England. 1997, Cambridge: Cambridge University PressView ArticleGoogle Scholar
- James SP: Malaria at home and abroad. 1920, London: John Bale, Sons & DanielssonGoogle Scholar
- Newman G: Reports and papers on malaria contracted in England, 1918. Rep Loc Govern Board Publ Health Medl Subj. 1919, 123: 1-42.Google Scholar
- Blacklock B: Notes on a case of indigenous infection with P. falciparum. Ann Trop Med Parasit. 1921, 15: 59-72.Google Scholar
- Hutchinson RA, Lindsay SW: Malaria and deaths in the English marshes. Lancet. 2006, 367: 1947-1951. 10.1016/S0140-6736(06)68850-8.View ArticlePubMedGoogle Scholar
- Shute PG, Maryon M: Malaria in England past, present and future. Roy Soc Health. 1974, 1: 23-29 & 49Google Scholar
- Linton Y-M, Lee A, Curtis F: Discovery of a third member of the Maculipennis complex in SW England. Europ Mosq Bull. 2005, 19: 5-9.Google Scholar
- James SP: The disappearance of malaria from England. Proc Roy Soc Med. 1929, 23: 71-85.PubMed CentralPubMedGoogle Scholar
- Jetten T, Takken W: Anophelism without malaria in Europe. A review of the ecology and distribution of the genus Anopheles in Europe. 1994, Wageningen: Wageningen Agricultural UniversityGoogle Scholar
- Hay S, Guerra C, Tatem A, Noor A, Snow R: The global distribution and population at risk of malaria: past, present and future. Lancet Infect Dis. 2004, 4: 327-336. 10.1016/S1473-3099(04)01043-6.PubMed CentralView ArticlePubMedGoogle Scholar
- Snow R, Guerra C, Noor A, Myint H, Hay S: The global distribution of clinical episodes of Plasmodium falciparum malaria. Nature. 2005, 434: 214-217. 10.1038/nature03342.PubMed CentralView ArticlePubMedGoogle Scholar
- Rogers DJ, Randolph SE: The global spread of malaria in a future, warmer world. Science. 2000, 289: 1763-1766. 10.1126/science.289.5478.391b.View ArticlePubMedGoogle Scholar
- Martens P: Health and climate change. Modelling the impacts of global warming and ozone depletion. 1998, London: Earthscan PublicationsGoogle Scholar
- Lindsay SW, Thomas CJ: Global warming and risk of vivax malaria in Great Britain. Glob Chan Hum Health. 2001, 2: 80-84. 10.1023/A:1011995115713.View ArticleGoogle Scholar
- Gill CA: The influence of humidity on the life history of mosquitoes and on their power to transmit infection. Trans R Soc Trop Med Hyg. 1921, 14: 77-82. 10.1016/S0035-9203(20)90152-9.View ArticleGoogle Scholar
- Anderson R, May R: Infectious diseases of humans. 1991, Oxford: Oxford University PressGoogle Scholar
- Dietz K: The estimation of the basic reproduction number for infectious diseases. Stat Meth Med Res. 1993, 2: 23-41. 10.1177/096228029300200103.View ArticleGoogle Scholar
- Detinova T: Age-grouping methods in Diptera of medical importance. 1962, Geneva: World Health OrganisationGoogle Scholar
- Boyd MF: Malariology. A comprehensive survey of all apsects of this group of diseases from a global standpoint. 1949, Philadelphia and London: Saunders, W.B.Google Scholar
- Stratman-Thomas WK: The Influence of temperature on Plasmodium vivax. Am J Trop Med. 1940, 20: 703-714.Google Scholar
- Macdonald G: The epidemiology and control of malaria. 1957, London: Oxford University PressGoogle Scholar
- James S: Some general results of a study of induced malaria in England. Trans Roy Soc Trop Med Hyg. 1931, 24: 477-525. 10.1016/S0035-9203(31)90068-0.View ArticleGoogle Scholar
- Fuller R, Groom G, Jones A: The Land Cover Map of Great Britain: an automated classification of Landsat Thematic Mapper data. Photogram Engineer Rem Sens. 1994, 60: 553-562.Google Scholar
- Nuttall G, Cobett L, Strangeways-Pigg T: Studies in relation to malaria. I. The geographical distribution of Anopheles in relation to the former distribution of ague in England. J Hyg. 1901, 1: 4-44. 10.1017/S0022172400000048.PubMed CentralView ArticlePubMedGoogle Scholar
- Parker D, Legg T, Folland C: A new daily Central England Temperature Series, 1772-1991. Int J Climatol. 12: 317-342. 10.1002/joc.3370120402.Google Scholar
- Great Britain Historical Database Online. [http://hds.essex.ac.uk]
- UKBORDERS. [http://0-edina.ac.uk.brum.beds.ac.uk/ukborders/]
- Burnham KP, Anderson DR: Model selection and multimodel inference: a practical-theoretic approach. 2002, London: Springer-Verlag, 2Google Scholar
- Richards SA: Dealing with overdispersed count data in applied ecology. J Appl Ecol. 2008, 45: 218-227. 10.1111/j.1365-2664.2007.01377.x.View ArticleGoogle Scholar
- Lindsay S, Thomas C: Global warming and risk of vivax malaria in Great Britain. Glob Chan Hum Health. 2001, 2: 80-84. 10.1023/A:1011995115713.View ArticleGoogle Scholar
- Lang WD: A map showing the known distribution in England and Wales of the Anopheline mosquitoes, with explanatory text and notes. 1918, London: British Museum (Natural History)Google Scholar
- Rees AT, Snow KR: The distribution of the genus Anopheles in Britain. Dipt Dig. 1990, 6: 7-19.Google Scholar
- Snow KR, Rees AT, Bulbeck SJ: A provisional atlas of the mosquitoes of Britain. 1998, London: University of East LondonGoogle Scholar
- Marshall J: The British Mosquitoes. 1938, London: W. Clowes & SonsGoogle Scholar
- Hutchinson RA, Lindsay SW: Perceived nuisance of mosquitoes on the Isle of Sheppey, Kent, UK. J biosoc Sci. 2006, 38: 701-712. 10.1017/S0021932005000994.Google Scholar
- Tempelis C, Lofy M: A modified precipitin method for identification of mosquito blood-meals. Am J Trop Med Hyg. 1963, 12: 825-831.PubMedGoogle Scholar
- HPA: Malaria imported into the UK, 2003: implications for those advising travellers. 2004, Health Protection AgencyGoogle Scholar
- Roshanravan B, Kari E, Gilman R, Cabrera L, Lee E, Metcalfe J, Calderon M, Lescano A, Montenegro S, Calampa C: Endemic malaria in the Peruvian Amazon region of Iquitos. Am J Trop Med Hyg. 2003, 69: 45-52.PubMedGoogle Scholar
- Elsom D: Deaths and injuries caused by lightning in the United Kingdom: analyses of two databases. Atmosph Res. 2001, 56: 325-334. 10.1016/S0169-8095(00)00083-1.View ArticleGoogle Scholar
- Bruce-Chwatt L, de Zulueta J: The rise and fall of malaria in Europe. 1980, London: Oxford University PressGoogle Scholar
- Takken W, Martens P, Bogers RE: Environmental change and malaria risk. Global and local implications. 2005, Dordrecht: SpringerView ArticleGoogle Scholar
- Becker N: Life strategies of mosquitoes as an adaptation to their habitats. Bull Soc Vector Ecol. 1989, 14: 6-25.Google Scholar
- Martens P, Kovats RS, Nijhof S, de Vries P, Livermore MTJ, Bradley DJ, Cox J, McMichael AJ: Climate change and future populations at risk of malaria. Glob Environ Change. 1999, 9: S89-S107. 10.1016/S0959-3780(99)00020-5.View ArticleGoogle Scholar
- Linard C, Poncon N, Fontenille D, Lambin E: A multi-agent simulation to assess the risk of malaria re-emergence in southern France. Ecol Model. 2009, 220: 160-174. 10.1016/j.ecolmodel.2008.09.001.View ArticleGoogle Scholar
- Linard C, Poncon N, Fontenille D, Lambin E: Risk of malaria reemergence in southern France: Testing scenarios with a Multiagent Simulation Model. Ecohealth. 2009, 6: 1-10.1007/s10393-009-0236-y.View ArticleGoogle Scholar
- Schroder W, Schmidt G: Modelling of temperature-dependent Malaria risk in Lower Saxony, Germany. Parasitol Res. 2008, 103: S152-10.1007/s00436-008-1051-z.View ArticleGoogle Scholar
- Casimiro E, Calheiros J, Santos F, Kovats S: National assessment of human health effects of climate change in Portugal: Approach and key findings. Environ Health Perspec. 2006, 114: 1950-1956.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.