Climate change, malaria and neglected tropical diseases: a scoping review

Introduction

Human activity is driving incremental changes in climate patterns globally. The burning of coal, oil and natural gas; clearing land by cutting down forests; and intensified agriculture all release greenhouse gases, primarily CO₂, methane and N₂O. Deforestation also reduces capacity for CO₂ absorption. Increases in atmospheric concentrations of greenhouse gases are driving mean global temperatures upwards, with consequent effects including rising sea levels, changes in rainfall and increases in the frequency and intensity of extreme weather events.¹

Climate change will perturb human health in profound and long-lasting ways, both directly and indirectly.² Multiple direct and indirect mechanisms will contribute. Human physiology can be affected directly by changes in air temperature, humidity, stress and diet. Human behaviour can be altered by changing weather, changing economic circumstances, migration, natural disasters and access to or quality of healthcare. Environmental conditions also influence pathogen transmission, reproduction, development and genetics; the reproduction, development, genetics, behaviour, range, longevity and predation of vectors, intermediate hosts and reservoir hosts; and the feasibility and effectiveness of interventions. In fact, there is already empirical evidence of climate change having amplified more than one-half of all known human infectious diseases.³

The impact of climate change is likely to be disproportionately borne by the poorest people, who are also disproportionately affected by malaria and neglected tropical diseases (NTDs; Table 1). In part because of this association with poverty, malaria and NTDs are often co-endemic. Many of these diseases are currently suitable for coordinated control via integrated programmes.^4–7

Developing an understanding of the likely effects of climate change on the epidemiology of malaria and NTDs is critical to minimising their health implications. This scoping review explores current predictions of the effects of historical and future climate change on malaria and NTDs, and the potential amelioration of these effects through climate change mitigation and adaptation strategies.

Methods

A comprehensive scoping review was conducted. To complete the review efficiently but with high fidelity, we used automation and artificial intelligence-based tools to help define search terms, translate them across different platforms, de-duplicate search results and conduct title-and-abstract screening.⁸ We did not otherwise use artificial intelligence assistance. This report adheres to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews guidance⁹ (Supplementary Table 1).

Results

Database searches yielded 31 560 records. Forward and backward citation searching yielded an additional 11 133 (of which 8587 were within the specified publication range), producing a combined total of 42 693 records (40 147 within the specified range). After removal of 14 879 duplicates, 27 814 records were available for title-and-abstract screening. The stopping rules for title-and-abstract screening were met after 9013 records had been examined; of these 9013, 7442 (83%) were excluded on the basis that their titles and abstracts indicated a failure to meet the inclusion criteria. Full-text papers were therefore sought for 1571 records (17% of 9013), of which 1543 (98%) were able to be retrieved. Following full-text review, 511 papers (33% of 1543) were included for data extraction; these are summarised in Supplementary Table 2. Extracted data are presented in full in Supplementary Table 3. A flow diagram is included as Figure 2.

Papers that met the inclusion criteria considered outcomes relevant to malaria (185 papers), dengue and chikungunya (181), the leishmaniases (53), schistosomiasis (29), Chagas disease (19), foodborne trematodiases (17), lymphatic filariasis (14), snakebite envenoming (11), rabies (9), human African trypanosomiasis (8), Buruli ulcer (6), echinococcosis (4), onchocerciasis (2), leprosy (1), scabies (1) and soil-transmitted helminthiases (1). No papers meeting the inclusion criteria considered outcomes relevant to dracunculiasis; mycetoma, chromoblastomycosis or other deep mycoses; taeniasis/cysticercosis; trachoma; or yaws (Figure 3A).

The vast majority of papers used modelling to study the effect of climate change (435 papers). There were 72 field study papers, 54 laboratory research reports, 28 ecological case reports, 10 cohort studies and seven other types of study (Figure 3B; multiple study types were possible in a single paper). There was little variation in the numbers of papers from year to year (Figure 3A and B, left plots).

Among papers meeting the inclusion criteria, a total of 174 (35%) considered the possible ameliorating effect of climate change mitigation for any outcome, 24 (5%) considered adaption strategies and two considered both. Sixty-nine papers considered mitigation in relation to malaria outcomes. Nine explicitly considered adaptation strategies for malaria outcomes. Fifty-three papers addressed climate change mitigation and nine considered adaptation strategies in relation to dengue and chikungunya; four of the papers, including mitigation analyses for dengue and chikungunya, also considered malaria outcomes in the light of climate change mitigation. Sixty other papers addressed climate change mitigation or adaptation with respect to NTD outcomes.

There was wide variation in the geographical coverage of papers by disease (Figure 4). Apart from malaria, dengue and chikungunya, and schistosomiasis, there were <10 papers per disease per country for every country, with many endemic countries not featured in any papers for several NTDs. We also observed distinct distribution patterns for malaria, dengue and chikungunya, and schistosomiasis. Most papers examining the impact of climate change on malaria focused on countries in Africa, Brazil, China or India. Dengue and chikungunya papers were focused on Australia, China, India, countries in Europe and the USA, many of which are countries where these diseases may spread in future years. Several schistosomiasis papers focused on China. For leishmaniasis, papers meeting our inclusion criteria considered countries that were widely distributed around the globe, with the exception of countries in East Africa.

The observed variation in geographical coverage of papers across diseases prompted us to further investigate the links between study location, number of papers, disease burden and country vulnerability to climate change. Given the discrepancy in the number of papers between diseases, we grouped these into four categories: (i) malaria; (ii) dengue and chikungunya; (iii) other vector-borne NTDs (Chagas disease, dracunculiasis, human African trypanosomiasis, leishmaniasis, lymphatic filariasis, onchocerciasis, schistosomiasis, trachoma); and (iv) non-vector-borne NTDs (Buruli ulcer, echinococcosis, foodborne trematodiases, leprosy, rabies, scabies/tungiasis, soil-transmitted helminthiases, snakebite envenoming, taeniasis and cysticercosis).

Our analysis showed different patterns across groups of diseases. For malaria there were clear trends towards more papers covering countries with a high malaria DALY burden, low HAQI and high vulnerability to climate change (Figure 5, first column). For dengue and chikungunya, there was a trend towards increasing numbers of papers covering areas with high burden, but at low DALY burden there was an increase in papers due to the relatively large number of studies looking at potential expansion of these diseases into new areas (Figure 5, second column). This also meant that there was a relative abundance of papers studying these diseases in areas where there is good access to healthcare, and where the climate vulnerability score is low. For the remaining vector-borne NTDs (Figure 5, third column), there was a suggestion of an increasing numbers of papers addressing countries with increasing burden and decreasing HAQI, but no suggestion that analyses were more commonly focused on areas with high climate vulnerability. For the non-vector-borne NTD group, papers more frequently considered countries with high DALY burden for that disease, high HAQI and low climate vulnerability (Figure 5, fourth column).

Detailed analysis of the full text of papers meeting the inclusion criteria revealed additional insights. For malaria, the consensus among papers that met the inclusion criteria was that global warming will extend the area in which transmission is possible to some areas where it was previously too cold for vector or parasite development, while conditions in some currently endemic areas may become too severe to maintain transmission. Zones suitable for transmission may shift both poleward and upwards in altitude. Future expansion could be balanced by more frequent droughts, making environmental conditions unsuitable for transmission in some previously endemic areas, including parts of the Sahel.¹⁶ Other papers concluded that malaria transmission seasons will last longer, as more months of the year will have a suitable climate.¹⁷ While these changes could place a greater proportion of the global population at risk,¹⁸ the potential net impact of climate change on the global burden of malaria remains unclear;^19–22 papers that met the inclusion criteria were vastly different and difficult to compare with each other. Making predictions for regional and global populations was previously difficult because of a general lack of high-resolution monthly incidence data.²³ There is likely to be small-scale heterogeneity in effects on the ground.

One aspect of climate change that is already impacting malaria transmission is the increasing frequency of extreme weather events. Severe flooding in Pakistan in 2022 and cyclones in Mozambique and Madagascar in 2023 were accompanied by local spikes in malaria cases, driven by breeding of Anopheles mosquitos in flood waters.²⁴

For dengue and chikungunya, a significant proportion of papers that met our inclusion criteria described subnational studies with subdecadal time frames. Models with broader geographic and temporal boundaries predict dramatic expansion in the future range of relevant Aedes vector species,^25–27 in line with trends that are already being observed.^28–30 Some local retreat from geographies currently occupied by Aedes spp. is also predicted.³¹ Alongside the overall increase in vector range, considerable increases are projected in the number of future dengue cases under more adverse climate change scenarios (Box 1) in some models.^30,32 Other models predict a plateauing of dengue in highly endemic regions by 2050.²⁸ There is undoubtedly uncertainty about the magnitude and direction of future geographic expansion.

For the leishmaniases, 53 papers that met the inclusion criteria considered relevant outcomes. However, there are many different pathogens and vector species contributing to this complex of diseases (Table 1), making a reliable, complete picture very difficult to draw. Several large-scale models predict changes in the range of relevant sandfly species around the Mediterranean and in the Americas, with transmission in the latter extending as far north as southern Canada in some future climate scenarios.^33–37 The number of people in North America living in areas in which leishmaniasis is transmitted could double from 2010 to 2080 under SRES B2.³⁷ The paucity of information on the potential impact of climate change on leishmaniasis in Africa (Figure 4) was striking.

For schistosomiasis, one model predicted increased prevalence and intensity of infection in some areas of East Africa over the next few decades, particularly in Rwanda, Burundi, south-west Kenya and eastern Zambia, with concurrent substantial decreases in risk in parts of Kenya, southern South Sudan and eastern Democratic Republic of the Congo.³⁸ Decreases in transmission are predicted for China,³⁹ although the possibility was also invoked that disease will re-emerge in parts of mainland China where it was previously eliminated.⁴⁰ This concern may account for the relative concentration of publications that focused on China (Figure 4).

There were few papers on the remaining NTDs, with a bias towards vector-borne diseases, possibly due to the existence of established methodologies for examining vector suitability and its links to climate. Climate impacts on these other NTDs, whether through environmental changes or societal ones, such as changes in time to diagnosis due to changes in access to health systems,⁴¹ remain largely unexplored.

Discussion

For thousands of years, societies have been shaped and reshaped by epidemics, with Ebola virus disease, COVID-19 and mpox presenting only the most recent striking examples. A distinct feature of the climate crisis is the pace of change in underlying global ecosystems. This generates uncertainty about the future epidemiology of multiple diseases: not only those that have historically manifested as epidemics, but also those formerly considered as stable and endemic, and those being driven towards elimination or eradication. Climate change will simultaneously reshape the epidemiology of many non-infectious diseases, threaten health infrastructure, affect the health workforce and alter other foundational determinants of human health. These parallel effects will exacerbate the challenge presented by the evolving epidemiology of infectious diseases.^42,43

Malaria and many NTDs have relatively complex life cycles, involving overlapping webs of interactions between humans and vertebrate and invertebrate animals. Multiple points of exposure to ecological, biological and social systems increase the probability that climate change will alter disease incidence or prevalence.^43,44 This environmental sensitivity makes prediction of future scenarios both difficult and important. The perceived difficulty is borne out in the literature identified here: for most of the outcomes in scope, projections of the effects of future climate change at large scale are scarce. Projections that do exist incorporate considerable uncertainty.

Of all potential outcomes framed for this review, those related to malaria, dengue and chikungunya, and the leishmaniases, were the most studied. Yet even for malaria, long-term projections of future transmission scenarios remain inadequate for robust planning. Beyond the expected short-term effects of extreme weather events on local disease incidence,⁴⁵ it is difficult to be definitive⁴⁶–global incidence and attributable deaths may go up, down or stay about the same, depending on multiple factors, including the success or otherwise of nascent vaccination programmes.⁶ The arboviral NTDs, dengue and chikungunya, on the other hand, are generally predicted to continue their current surge.^28,29 Global expansion in the population at risk of the leishmaniases seems likely. The predicted effect of climate change on these diseases suggests that effects on vectors of other vector-borne NTDs will be similarly important to understand.

For many of the NTDs that we grouped as ‘non-vector-borne’, however, limitations in our capacity for prediction stem in part from gaping deficiencies in our understanding of disease transmission at steady state. Data published only after the searches were conducted for this review pin responsibility for Mycobacterium ulcerans transmission in southeastern Australia on the mosquito Aedes notoscriptus,⁴⁷ whereas transmission by mosquitos was previously hypothesised but unproven, and analogous work has not yet been published for Buruli ulcer-endemic areas of the Americas, Japan, Papua New Guinea or West Africa. Postulated mechanisms for the transmission of leprosy are still based on circumstantial evidence.¹³ The nematode worm Dracunculus medinensis, the target of a global eradication programme that began in 1980, was only relatively recently discovered to infect paratenic and definitive hosts other than humans.⁴⁸ These gaps in our knowledge have lingered because funding for research on NTDs is thin⁴⁹ and spread across a very large number of pathogens (Table 1), some of which are probably relatively rare.^50,51

Multiplying that disease-specific uncertainty are uncertainties surrounding future climate scenarios and their secondary impacts (including, for example, on conflict, migration and demography), which are further clouded by our joint hope that individual and collective behaviours will change sufficiently to allow greenhouse gas concentrations to fall and thereby effectively mitigate climate change. This uncertainty for virtually every parameter of NTD transmission models⁵² makes decadal projections of future NTD prevalence or incidence feel ill-advised. Unfortunately, without those projections, NTDs are likely to continue to be given very limited attention in climate-related discussions, including, for example, in the Assessment Reports of the Intergovernmental Panel on Climate Change.⁵³ Better understanding of NTD transmission dynamics, and estimates—even if heavily caveated—of the potential impacts of climate change, are precisely what is needed now.

The current work builds on previously published reviews,^12,43,54 but continues to have some limitations. First, although we undertook screening of titles and abstracts by two independent observers, our use of tags to highlight certain characteristics of papers within the Covidence platform may have unmasked some second screeners to the first screener's decision.

Second, screening for mitigation and adaptation strategies was challenging; the process required considerable judgement. It is possible that relevant sources were overlooked in the >42 000 records that our searches identified.

Third, we did not critically quality-appraise the methodology used for each included paper (e.g. number, size and origin of datasets; appropriateness of climate models, such as use of downscaling; and inclusion of all potentially relevant variables). Many studies had low power, did not consider all potential explanatory variables or confounders, or were otherwise methodologically weak. To be useful as a summary of coverage within the published literature, our visualisations imply that all studies contribute equally to the evidence base, whereas they do not.

Fourth, under-ascertainment is an issue for many diseases, but is particularly problematic for malaria and NTDs; their concentration in impoverished populations means that patients with these diseases have HAQIs that are far from ideal. Under-ascertainment is a specifically identified issue for dengue, in which second infections may be considerably more likely to produce disease that leads to clinical presentation and therefore registration.⁵⁵

Fifth, to be included in this review, a paper had to explicitly juxtapose climate change and relevant outcomes. Some authors may have made tenuous arguments linking weather-related variables (such as temperature or rainfall) to climate change, resulting in inclusion; others may not have been explicit in framing climate change implications when doing so would have been justifiable.

Sixth, it is likely that source data were used more than once in groups of papers with the same or related outcome measures for overlapping geographies and overlapping timespans.

Seventh, we did not specifically look for the impact of climate change mediated through internal displacement and migration of people;⁵⁶ changes in institutional capacity and service provision;^57,58 vector microbiome, genetics or gene expression; or pathogen genetics or gene expression.^59,60 All of these mechanisms may be important.

Eighth, as for any review, our searches had a fixed date range and were not exhaustive within that range. Not all possible intermediate and reservoir hosts (e.g. for rabies) were specifically included. Insect species with postulated but unproven vectorial capacity were excluded; Culex pipiens is a known vector of lymphatic filariasis in Egypt but papers considering it in other contexts were set aside. Studies in press were excluded by design; we were aware of forthcoming work on the impact of climate change on several diseases that had been submitted for peer review but were not yet published when our searches closed. The December 2023 addition of noma to the WHO's list of NTDs⁶¹ occurred too late for noma to be included in our searches.

Ninth, the high degree of heterogeneity (in questions examined, methods used and so on) precluded quantitative synthesis.

Tenth, projecting disease burdens forward over long timescales means that future changes in treatment and control strategies would ideally also be taken into account. This is difficult to do. Authors of studies from the early part of our 2010–2023 publication window may not have foreseen the scale-up in intervention coverage that has occurred for many diseases in the past 10–15 y.

Despite uncertainty around data, some general conclusions and recommendations are proposed here. It can be inferred from existing data, first, that climate change is likely to have profound direct and indirect implications for malaria, dengue and chikungunya, leishmaniasis and at least several other vector-borne NTDs, even if the amplitude and direction of the effects will probably vary by disease and location, be non-linear^22,62–64 and evolve with time. Changes of two kinds will be apparent: diseases will move around, and where endemicity is constant, there will be local increases or decreases in incidence or prevalence. There is a pressing need to safeguard previous global health gains by scaling up proven interventions and achieving impact before future changes render those interventions ineffective.⁶⁵ Second, the lack of predictability, even over relatively short timescales, calls for existing surveillance and intervention systems to be reinforced and regularly reviewed. Integrated surveillance and intervention systems, covering multiple diseases⁶⁶ and taking a One Health approach,⁶⁷ could offer efficiencies. Third, communities should be consulted and involved in these reviews of surveillance and intervention systems, and in research undertaken at the interface of infectious diseases and climate change, to maximise the relevance of such efforts despite changing human populations. Fourth, integrating climate resilience into health systems is critical. This should encompass investing in health infrastructure, fostering cross-sector collaboration, adapting to the needs of displaced populations, improving access to health products and accelerating research and development to fill known gaps.⁶⁸ A particular requirement is access to existing and new countermeasures to limit future expansions in disease burden. Fifth, we do not know enough.

Based on our analysis, we also propose several key recommendations to guide future studies (Box 2). The most important of these is for standardisation and collaboration. Our scope included 21 diseases and disease groups, at least 76 distinct pathogens, 373 venomous snakes and humans everywhere; there is insufficient modelling capacity globally for this to be investigated on a competitive basis, particularly when the appropriate modelling methodologies change with the level of endemicity and results are needed at multiple scales. Indeed, our analysis suggests that existing studies may not be sufficiently focused on areas where the need to plan for adaptation may be greatest. We recommend holistic approaches to risk assessment, incorporating more of the available data and recruiting more of the available brainpower to undertake ensemble analyses with agreed best-practice methodology. Modelling efforts should incorporate consideration of humans, pathogens, vectors, intermediate and reservoir hosts and the effect of relevant interventions, as appropriate, to generate predictions over decadal time frames—and not ignore populations where these diseases are currently endemic. Collectively, these measures should reduce potential duplication and hopefully produce more complete and more accurate estimates of future vulnerability, exposure and impact. Open-source collaborative modelling platforms^17,23 could facilitate contributions from as many relevant stakeholders as are willing to engage, and allow tailoring of consistently high-quality outputs for specific audiences. Collaboration could include involvement of affected communities through citizen science: in vector surveillance, for example. Accessible global databases on disease and vector occurrence^69–75 should be harnessed and adapted to cover additional diseases. Broader use of remotely sensed climate data should be explored, particularly where locally acquired data are unavailable or microclimates are of relevance. Long-term time-series data should be pooled and re-analysed to tease out the relative contributions of deliberate interventions, secular trends, seasonality and climate change. Production of detailed risk and distribution maps should be facilitated to help plan local control and elimination efforts. The fact that many NTDs are targeted for eradication, interruption of transmission or elimination as a public health problem by 2030 should not dissuade us from taking a long-term view of this work;⁴³ global health ambitions are not always realised, and the best possible current understanding of counterfactual scenarios should help decision-makers to target resources and chart the most appropriate course.

Supplementary Material

Authors’ contributions

CLD, TDH, AWS and ISF conceived the overall study design, which was elaborated by PK, JLH, CLD, MA, MB, IB, SD, DLM, CMRT, KRvD, TDH, AWS and ISF, with review by all authors; PK, JLH, CLD, MA, MB, IB, SD, CMRT, KRvD and AWS screened titles and abstracts, located full texts and undertook full-text review; PK and JLH identified seed papers and undertook backward and forward citation searches; PK, JLH, CLD, MA, MB, IB, SD, CMRT, KRvD, JA, AD, DSE, JG, MRJ, AMK, JK, WJK, TLL, JSM, JP, NP, AR, LJR, ZMR, GSGS and AWS extracted data; PK, JLH, KRvD, JFM, CNA, HSR, DRMS and TDH conducted quantitative analyses and constructed figures; PK, KRvD, KCR, CNA and AWS prepared the first draft of the manuscript, which was critically revised by all authors. All authors reviewed and approved the final manuscript. PK and AWS are the paper's guarantors.

Funding

This work was supported by Reaching the Last Mile; The Fred Hollows Foundation; Bill & Melinda Gates Foundation and the European Commission. PK, SK, CNA and TDH were supported by funding from the Bill & Melinda Gates Foundation [INV-030046], via the NTD Modelling Consortium. TDH is supported by funding from the Li Ka Shing Foundation. JLH acknowledges funding from USAID through an Interagency Agreement with CDC awarded to DLM. KRvD and RL acknowledge funding from the European Union's Horizon Europe research and innovation programme [grant agreement no. 101057554] (Horizon Europe project IDAlert, https://idalertproject.eu); IDAlert is part of the EU climate change and health cluster (https://climate-health.eu). BA, AFG, JK, VL, AR, KHS, RV, SW, AY, DCL, AWS and ISF are staff members of, and CLD and CMRT are consultants to, the WHO. JA was supported by an appointment to the Research Participation Program at the Centers for Disease Control and Prevention, administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. Department of Energy and the Centers for Disease Control and Prevention. SK, AMK, JSM, MRJ, TLL and GSGS were supported by funding from the Bill & Melinda Gates Foundation [INV-016807]. NMF was supported by funding from Wellcome, the Jameel Institute and the MRC Centre for Global Infectious Disease Analysis. JFM is supported by grant funding from the Bill & Melinda Gates Foundation and Gavi. RSN and KP are staff members of the Pan American Health Organization.

Competing interests

SD is and JC was an employee of The Fred Hollows Foundation; SJB is an employee of the Bill & Melinda Gates Foundation. Their participation in this work was in their personal capacity. All other authors declare no competing interests.

Ethical approval

Not required.

Data availability

All data relevant to the study are available in Supplementary Table 2 and Supplementary Table 3. Code available upon request.

Disclaimers

The authors alone are responsible for the views expressed in this article and they do not necessarily represent the views, decisions or policies of the institutions with which they are affiliated. The boundaries and names shown and the designations used on the maps in this article do not imply the expression of any opinion whatsoever on the part of the authors, or the institutions with which they are affiliated, concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries.