Article Text
Abstract
Background Despite worldwide efforts to eradicate malaria over the past century, the disease remains a significant challenge in the Democratic Republic of the Congo (DRC) today. Climate change is even anticipated to worsen the situation in areas with higher altitudes and vulnerable populations. This study in Haut-Katanga, a highland region, aims to evaluate the effectiveness of past control measures and to explore the impact of climate change on the region’s distinct seasonal malaria pattern throughout the last century.
Methods We integrated colonial medical records (1917–1983) from two major mining companies (Union Minière du Haut-Katanga and the Générale des Carrières et des Mines) with contemporary data (2003–2020) from Lubumbashi. Concurrently, we combined colonial climate records (1912–1946) with recent data from satellite images and weather stations (1940–2023). We used Generalised Additive Models to link the two data sources and to test for changing seasonal patterns in transmission.
Results Malaria transmission in Haut-Katanga has fluctuated significantly over the past century, influenced by evolving control strategies, political conditions and a changing climate. A notable decrease in cases followed the introduction of dichlorodiphenyltrichloroethane (DDT), while a surge occurred after the civil wars ended at the beginning of the new millennium. Recently, the malaria season began 1–2 months earlier than historically observed, likely due to a 2–5°C increase in mean minimum temperatures, which facilitates the sporogonic cycle of the parasite.
Conclusion Despite contemporary control efforts, malaria incidence in Haut-Katanga is similar to levels observed in the 1930s, possibly influenced by climate change creating optimal conditions for malaria transmission. Our historical data shows that the lowest malaria incidence occurred during periods of intensive DDT use and indoor residual spraying. Consequently, we recommend the systematic reduction of vector populations as a key component of malaria control strategies in highland regions of sub-Saharan Africa.
- Epidemiology
- Public Health
- Malaria
Data availability statement
Data are available upon reasonable request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Malaria has been a persistent health issue in the Democratic Republic of Congo for centuries, with fluctuating transmission rates influenced by climate change, control strategies and political conditions.
WHAT THIS STUDY ADDS
This study reveals that minimum temperatures in Haut-Katanga have increased significantly over the past century, leading to an earlier malaria season by 1–2 months. It also highlights that despite modern control efforts, malaria incidence has returned to pre-DDT levels, emphasising the critical role of historical interventions that reduced mosquito densities.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The study’s findings underscore the need for sustained and intensified malaria control strategies, particularly in high-altitude regions impacted by climate change. It advocates for renewed use of vector control measures, such as indoor residual spraying.
Introduction
Infectious diseases have likely plagued Central African communities since ancient times, but it was only in the late 19th century that their spread and impact were systematically documented by soldiers, explorers and missionaries.1 2 While the next centuries witnessed important advancements in identifying and controlling such diseases (often exemplified by the successful eradication of smallpox), some remain a difficult challenge to the health communities even today.3 Malaria unequivocally falls into this latter category, persisting as a leading cause of mortality in the region, where it still accounts for approximately 200 000 deaths annually.4 5
Malaria is caused by parasitic protozoans of the genus Plasmodium that are transmitted by female Anopheles mosquitoes.1 The disease exhibits a clear spatiotemporal distribution influenced by climate and land use factors, as well as the local capacity to manage it.6 7 While rainfall and land use primarily impact the abundance of the vector, temperature plays a dual role in affecting both the vector abundance and the duration of sporogony, that is, the time it takes for the parasite to develop within a mosquito and become transmissible.8 9 For example, under optimal temperature conditions (20°–25°C), the sporogony of human Plasmodium typically takes 8–14 days to complete.10 Consequently, if adult Anopheles mosquitoes perish before day 12, they are unlikely to contribute much to parasite transmission.11 12 Similarly, temperatures below a critical threshold (usually 16°C for human Plasmodium species) impede parasite development in the mosquito, resulting in the absence of malaria transmission in the area despite the presence of the vector.13 This interplay between local climate conditions and transmission dynamics often manifests as a distinct seasonal malaria pattern and a decrease in transmission risk over an elevation gradient (usually absent over 1850 metres above sea level (m.a.s.l)).14
While climate change is anticipated to increase the risk of extreme temperature and rainfall events, its impact on malaria transmission remains equivocal.15 16 Modelling studies predict that its most pronounced effects will be observed in fringe areas, where the disease is inherently unstable due to factors such as higher altitudes (resulting in temperature fluctuations around the sporogony threshold) and vulnerable populations with limited immunity, which is particularly the case in the highlands in Africa.17–21 Conversely, in non-fringe regions, the influence of climate change on malaria is expected to be lower given the overriding impact of other critical socioeconomic factors.22 In these lower regions, increasing temperatures could potentially even reduce transmission in areas where the thermal optimum of 25°C is exceeded.23 Demonstrating a clear link between climate change and malaria risk is thus challenging and would require well-documented long-term time series data that span potential intervention periods.7 22 These time series must also be contextualised within the observed global decline in malaria during the 20th century. Indeed, in the context of sub-Saharan Africa, the disease’s prevalence decreased from 40% from 1900 to 1929 to 24% from 2010 to 2015.24 However, this overall trend has experienced interruptions characterised by sudden transmission increase or decrease phases, primarily contingent on the implementation or cessation of various control interventions (eg, dichlorodiphenyltrichloroethane (DDT) use or distribution of mosquito nets).25
In the Democratic Republic of Congo (DRC), malaria continues to pose a significant public health challenge today, ranking globally second with 13% of all cases.26 27 During the colonial period of the country (1885-1960), the study of malaria has been integral to the activities of the colonial medical department.1 28 Annual reports meticulously detailed malaria infection rates in the communities and the ecology of the vector species (mainly in the bigger cities and around mining camps). For example, D’Haenens and colleagues29 used this data to produce the first comprehensive map of malaria in the DRC, dividing the country into ecological zones relevant to malaria stratification. More recently, the National Strategic Plan for Malaria Control created a map of malaria zones for the country based on temperature and rainfall thresholds.30 This map underscores the persistent nature of malaria transmission in the DRC, with year-round transmission occurring throughout most of the country, except in the Eastern and Southern high-elevation areas that are characterised by seasonal malaria (1200–1800 m.a.s.l) or its absence (>1800 m.a.s.l).
This study focuses on the province of Haut-Katanga, one of these high-elevation areas in the DRC that exhibits a distinct seasonal malaria pattern.28 31 32 It is situated in the extreme south of the country and encompasses the cities of Lubumbashi and Likasi. The region’s rich mineral resources have attracted both national and international mining companies throughout the past century. The most important company in this region was the Union Minière du Haut-Katanga (UMHK),33 later nationalised as Générale des Carrières et des Mines (Gécamines).34 35 Recognising malaria as a substantial impediment to the economic development of the mine, UMHK’s managers established a hygienic service for infectious disease control shortly after the mine’s opening in 1917.36 The data collected by this service offered a unique opportunity to (1) describe the impact of historical malaria control interventions in the region of Haut-Katanga over the past century and (2) investigate the influence of climate change on the seasonal malaria pattern in the region.
Methods
Historical context Union Minière du Haut-Katanga and Gécamines
At the start of the 20th century, European colonists established settlements in the sparsely inhabited yet mineral-rich Katanga region, giving rise to the mining cities of Elisabethstad (now Lubumbashi) and Jadotstad (Likasi).33 The UMHK was founded in 1906 with funding from Belgian and British investors and experienced rapid growth, producing 1000 tons of copper in its inaugural year and scaling to 135 000 tons by 1929. This expansion resulted in an increase in African employers, initially recruited from Zambia and later from the Congolese provinces, Rwanda and Burundi. Consequently, workers’ numbers surged from 8500 to 17 200 between 1919 and 1929. The global economic crisis in the 1930s led to a production decline of more than half, pushing the company to the brink of bankruptcy. The UMHK’s revival occurred during World War II when it became a significant copper, tin and uranium supplier for the Allied war effort. Post-war, the company continued to flourish, reaching its peak in the 1950s, contributing around 10% of global copper production and dominating the international cobalt market with approximately 60% of global production. At this peak, the UMHK employed roughly 100 000 people.
Even in the initial years of Congolese independence (1960–1965), there was a sustained increase in mining exports from UMHK.35 President Mobutu’s nationalisation of UMHK’s assets in 1967, followed by their transfer to the state-owned Gécamines in 1972, marked the establishment of Gécamines as the exclusive mining entity nationwide. Despite reaching a peak copper production of approximately 500 000 tons annually in the 1980s, foreign shareholders withdrew from Gécamines that encountered financial challenges attributed to mismanagement, corruption and civil unrest. This resulted in a significant decline in exports around the 1990s. In 2002, the Congolese government officially terminated the Gécamines' monopoly, leading to the increased transfer of many mining sites to foreign ownership in the region.
The forced labour policies, during the beginning of the UMHK through agencies like the Bourse du Travail du Katanga, meant that many men were driven into dangerous labour, often leaving their families and villages to work in mining camps under harsh conditions.37 Wages were low, and working circumstances were unsanitary, leading to high rates of disease and death. Moreover, traditional social structures were disrupted, as the colonial administration systematically dismantled chieftaincies and imposed new forms of taxation, forcing people into wage labour to survive.38 After the First World War, rather for economic reasons, the UMHK started to implement a social health policy to enhance the viability of industrial camps and stabilise the workforce.33 34 36 This policy favoured mass-oriented colonial medicine over individualised care to reduce costs. Only after the Second World War did UMHK invest in more professional training and medical, social and moral education for both European and African employers.31 During this period, they also actively promoted family life, assuming married workers would settle faster, cause fewer problems and exhibit lower disease and death rates. This comprehensive approach resulted in a significant reduction (86%) in overall mortality between 1920 and 1970, reflecting advancements in effective health prevention, treatment and care.34
Malaria control was one of the first initiatives undertaken by the UMHKs medical branch. Following a severe malaria outbreak at the campsite in 1922, which disrupted normal operations, the company’s initial response involved terminating the contracts of workers with severe malaria symptoms and alcoholics.36 In the years that followed, the focus of malaria control expanded to include the broader population, incorporating the use of mosquito nets, quinine treatment (both therapeutic and prophylactic) and measures targeting both larval and adult mosquito vectors (eg, DDT). To evaluate the effectiveness of these strategies, UMHK meticulously tracked malaria cases over time, data that we used in this article.
Malaria data
Historical data
We extensively searched the African Archives of the Belgian Ministry of Foreign Affairs (Joseph Cuvelier Depot) to acquire quantitative data on malaria cases in Katanga. The examined reports, all composed in French, were entitled ‘Union Minière du Haut-Katanga: rapports annuels du département adressés au secrétariat de la direction de Bruxelles’36 or Rapports annuels du département médical de la Génerale Congolaise des Carrières et des Mines.34 These annual reports were authored by members of the ‘Service hygienic de UMHK’ and encompassed epidemiological data on a yearly (1917–1983) or monthly basis (1930–1946), focusing on the most prevalent diseases the company faced. The reports also discussed annual trends, specific control interventions and reflections on how to enhance disease control. Malaria data included the number of treated symptomatic cases confirmed through microscopy. Notably, this information was disaggregated for European and Congolese employees, as well as for men, women and children. The data was generally aggregated for all the headquarters of the mine concessions in Katanga (±40 000 km2). The most important mining cities were: Jadotstad (Likasi, 1318 m.a.s.l), Elisabethville (Lubumbashi, 1276 m.a.s.l), Kipushi (1329 m.a.s.l), Kambove (1450 m.a.s.l) and Kolwezi (1448 m.a.s.l). Given the total number of individuals associated with the company, we could compute the malaria incidence per 1000 employees, differentiating between Congolese and European individuals and further stratifying by gender and age (adult vs child). The analysis was conducted over varying time units, yearly or monthly.
Recent data
For 2003–2011, we used monthly malaria incidence rates per 1000 inhabitants.32 39 Data were sourced from both primary healthcare facilities situated in the 104 health areas across nine health districts within Lubumbashi. Malaria diagnosis was assumed for children under 5 years in the presence of fever. In adults, a suspected malaria diagnosis was confirmed through microscopy testing. Weekly data validation was performed by the National Malaria Control Programme (PNLP) and other members of the Malaria Task Force (UNICEF, WHO and health development partners in the Health Zones). To achieve maximum completeness (100%), active data collection was conducted at the end of the month when weekly epidemiological reports were delayed. For this study, we used aggregated data at the city level. For 2012–2020, we downloaded modelled annual malaria incidence rates (cases per thousand inhabitants) for Haut-Katanga from the Malaria Atlas Project (MAP) database.40 We did not find any data for the period 1984–2002, which corresponds to the period of political unrest and wars.
Patient and public involvement
This study is based exclusively on the analysis of published and/or publicly available data. No new data were collected, and no human subjects or sensitive information were involved in this research
Climate data
We extracted climate data (cumulative rainfall, mean minimum and mean maximum temperature) from various sources from 1912 to 2023. Historical data from Lubumbashi (1912–1946) were obtained from the State Archives of Belgium, specifically from the archive of the ‘Institut National pour l'Etude Agronomique du Congo Belge (INEAC)’, which houses all climatological records from the Belgian Congo. At the time, these climatological records were noted on carbon copy paper at the different meteorological institutes of the country. Despite the availability of data from various stations in Katanga, our focus was on the data from the observatory of Lubumbashi (27° 28’ 16.639”, −11° 39’ 17.564; elevation 1230 m.a.s.l, located in the city centre) due to its possession of the longest undisrupted time series from the region. Recent data from Lubumbashi (2003–2023) were sourced from the meteorological station at METTELSAT de Luano.41 This institute is currently operating under the ‘Régie des Vois Aeriennes’ and is situated near the Lubumbashi airport (27° 28’ 59.88’’, −11° 40’ 0.12”; elevation 1276 m.a.s.l), 10 km from the original observatory of Lubumbashi. Additionally, climate data from Lubumbashi (27° 28’ 16.639”, −11° 39’ 17.564) spanning the years 1940–2023 were extracted from ERA5 (European Centre for Medium-Range Weather Forecasts Reanalysis V.5).42 This georeferenced data results from reanalysing models with global observations, providing a complete and consistent data set using the principles of physics.
Data analyses
Malaria incidence: long-term time effect
Due to variations in the diagnostic criteria and data aggregation by the official entities during the different periods, we did not pursue statistical investigations into the fluctuations of malaria incidence over the entire period. Instead, we focused on describing the general trend in malaria incidence for each data source and sought to link it with the malaria control efforts outlined in the relevant documents in the discussion.
Malaria incidence: seasonal effect
While investigating the direct impact of climate change on long-term malaria incidence was impossible due to differences in diagnostics over time and confounding effects (eg, different control efforts), we could test its potential influence on the seasonal transmission pattern in the region. Indeed, despite the differences in case definitions and data aggregation between the historical and recent data sets, we assumed consistency in case definitions and control efforts within the respective periods. To enable a relative comparison of seasonal patterns, we compared monthly aggregated data from the European and Congolese communities at UMHK (1930–1946) to recent data from Lubumbashi (2003–2011). Monthly data from UMHK for 1947–1954 were excluded due to a tremendous decrease in malaria incidence attributed to DDT interventions. Similarly, we restricted the use of recent data to the period 2003–2011, preceding the mass distributions of long-lasting insecticidal nets in Lubumbashi (limited distribution in 2009) and because data from after 2011 was from MAP only (that includes climate variables for its malaria predictions, making it unsuitable for our study). Data from before 1930 and 1954–1983 could also not be used, as we only found annually aggregated data in the historical documents for these periods.
To investigate the timing of the seasonal malaria peak, we constructed Generalised Additive Models (GAM) where malaria incidence served as the response variable (h(malaria incidence)), and continuous time (f1(continuous time)) and seasonal time (f2(seasonal time)) were employed as explanatory variables. Consistent with the model formulation proposed by Wood,43 our model can be expressed as follows:
E(h(malaria incidence)) = f1(continuous time, k=2) + f2(seasonal time, k=12).
‘Seasonal time’ was the primary variable of interest and represents the months per year. This variable was smoothed non-linearly using cyclic cubic regression splines (k=12), limiting discontinuity between the end and the beginning of a new year. The variable ‘continuous time’ was introduced to account for potential long-term effects on malaria incidence spanning multiple years, such as those arising from control interventions. Its degrees of freedom (df) were constrained (k=2) to prevent confounding with the other time component. Given that the response data were count-based, we used a model with a Poisson distribution. Besides an analysis of the compiled data, we explored whether a comparable seasonal pattern was discernible in the disaggregated historical data, conducting separate analyses for men, women and children among both European and Congolese employees.
Trends in climate variables
To assess the temporal evolution of climate conditions in Lubumbashi, we developed three GAMs with cumulative rainfall, mean maximum temperature and mean minimum temperature per month as the response variables. In all GAMs, seasonal components (cyclic cubic regression splines, k=12) and a continuous time component (k=2) were integrated as explanatory variables. This analysis was conducted both on the combined data set of the INEAC and ERA5 data spanning 1912–2023 and on ERA5 data alone covering 1940–2023. To ensure comprehensiveness, we also computed mean monthly climate variables using METTELSAT data from the Lubumbashi airport separately.
Malaria incidence: seasonal effect and climate change
To further examine the relationship between climate variables and malaria incidence, we assessed potential time delays between climate variables (monthly rainfall and mean minimum temperature) and malaria incidence. This investigation involved analysing cross-correlations at various time lags (monthly) using the ccf function in the software R. Subsequently, we developed GAMs with malaria incidence as the response variable, incorporating distinct lags of rainfall and mean minimum temperature as explanatory variables. These variables were also subjected to non-linear smoothing using cyclic cubic regression splines (k=12). Models incorporating time lags (the delay in months between climate conditions and malaria incidence) with lower Akaike information criterion (AIC) values would indicate stronger correlations between the climate and malaria variables and would be selected based on the parsimony principle.
For all GAMs, we determined the optimal level of smoothing for the time variables through cross‐validation using the built‐in function of the R‐package MGCV.44 The importance of the explanatory variables was assessed based on χ2 tests (p values) or differences in the AIC (ΔAIC), achieved by comparing the full model to models with each respective variable removed. The difference in effective df, which estimates the fit’s complexity, are reported where applicable. Evaluation of the final model fit included an examination of the diagnostic plots. To interpret the direction of the response variable (increase or decrease) based on explanatory variables, we generated graphs illustrating the fit and its confidence bounds.
Results
Long-term malaria incidence
Figure 1 shows that changes in malaria incidence coincide with different control strategies implemented and worldwide malaria eradication programmes (online supplemental table 1). The malaria incidence in Haut-Katanga exhibited substantial variations within and between data sets used in this study, as indicated by the significant continuous time effect in the GAMs (table 1). The incidence was the highest at the beginning of the 20th century in the population of European UMHK employees with a peak incidence of 1000 cases per 1000 employees annually (almost everybody became infected). Monthly cases declined to approximately 300 in the 1930s with the introduction of the first control measures like quinine prophylaxis and mosquito net use. After a surge in cases during World War II (±400 cases per 1000 employees per year), the incidence rapidly declines (<50 cases per 1000 employees per year) during the Global Malaria Eradication Programme (GMEP) era, attributed to DDT use. The incidence started to increase again during the middle of the 1970s due to relaxed control measures. At the beginning of the 21st century, Lubumbashi experienced a resurgence, with malaria incidence increasing from ±100 to ±200 cases per 1000 inhabitants annually. Current projections from the MAP suggest that malaria in Haut-Katanga returned to pre-GMEP era levels, reaching ±300 cases per 1000 inhabitants per year.
Supplemental material
Climate variables
All climate variables (cumulative rainfall, mean minimum/maximum temperature per month) exhibited a distinct seasonal pattern across data sets, as denoted by the significant seasonal time component in the GAMs (table 2). The annual cycle delineated a hot-rainy season from October to April and a cold-dry season from May to September, characterised by minimal precipitation (figure 2). While this seasonal pattern itself persisted over time, a significant increase in mean minimum temperature was observed consistent with climate change (table 2), particularly pronounced during the cold season with a temperature rise of 2–5°C over the entire study period (figure 2, online supplemental figure 1). Interestingly, the minimum temperature also increased by 2–3°C during the hot months (online supplemental figure 1), surpassing the threshold for malaria sporogony (from 16°C to 18–19°C). We highlight that the extent of the observed temperature increase may vary depending on the dataset used (ERA5 or ERA5+INEAC), illustrated by a GAM for September in online supplemental figure 2.
Mean maximum temperature and cumulative rainfall displayed a less clear trend over time, supported by lower ΔAIC and ΔR2 values for the continuous time component in the GAMs (table 2). While the maximum temperature was generally higher in historical (INEAC) data compared with recent data, the ERA5 time series also exhibited an increase from 1940 to 2023 by 2–3°C in most months (online supplemental figure 3). Notably, cumulative monthly rainfall did not reveal a discernible trend (online supplemental figure 4). However, a fluctuating pattern emerged in cumulative annual rainfall (online supplemental figures 5 and 6). Annual rainfall increased from 1920 to 1960, decreased from 1960 to 2000 and showed a subsequent rise from 2000 to 2023. Nevertheless, when comparing the malaria study periods (historical vs recent), a comparable amount of rainfall was noted between these two periods (figure 2).
Seasonal malaria incidence
Following the clear seasonality in climate variables, the GAMs confirmed the presence of a significant seasonal malaria pattern in Haut-Katanga across historical and recent periods (table 1). However, the timing of peak malaria incidence varied between the two periods (figure 2). In the historical period (1930–1946), the peak occurred in March-April and reached its lowest point in August-September. This pattern was consistent across European and Congolese populations and for all age and sex categories at the UMHK (figure 3). Conversely, during the recent period (2003–2011), the malaria incidence in Lubumbashi peaked in January- February and reached its lowest point in July (figure 2). These temporal variations align with the cross-correlation analyses, supporting robust correlations between malaria incidence and minimum temperature 2 months earlier for the historical data and 1 month earlier for the recent data. Similarly, strong correlations were observed between malaria incidence and cumulative rainfall 1 month earlier in the historical data and the same month in the recent data (figure 4). These findings were also consistent with the time-dependent lag GAMs, where the model with the lowest AIC supported a lag of 2 months between malaria incidence and rainfall/temperature for historical data and a lag of 1 month for recent data (figure 5). Specifically, the analysis suggests that malaria incidence peaks 1 month after the peak in temperature and rainfall in the recent data (2003–2011), whereas during the colonial period (1930–1945), malaria incidence peaked three to 4 months after these climatic peaks.
Discussion
We reconstructed the malaria incidence in Haut-Katanga from 1917 to 2020 based on time series from historical and contemporary surveillance sources (figure 1). Despite variations in diagnostic criteria across data sets, our analysis revealed the impact of different control strategies over time. We identified five pivotal periods that signify important milestones in malaria control within Haut-Katanga, following the classification by Snow et al.25
Between 1917 and 1946, malaria control efforts in Haut-Katanga focused on larval control (LC), environmental management and mass drug (quinine) administration campaigns. The peak malaria incidence occurred in 1921–1922, prompting initial control measures by the UMHK, such as terminating contracts for non-compliants, alcoholics and chronically ill employers. Subsequent interventions during the following years included the distribution of mosquito nets, mosquito-proofing of houses and prophylactic quinine for European employees only, resulting in a significant reduction from 1000 to 400 cases per 1000 European employees annually in our time series data (figure 1).36 Since 1928, antilarval practices, including petrolising breeding sites and removing water pools, further decreased the incidence to 200–300 cases in the 1930s.45 This decade witnessed the expansion of anti-larval control around village compounds, with dedicated teams removing vegetation, closing sewage systems and performing adult mosquito control. Despite these efforts, malaria remained a significant issue in the region, reaching a mortality rate of 25% for the European population by the late 1930s (unfortunately, no numbers were given for the African population). Furthermore, we observed a clear increase in malaria cases in our time series at the end of World War II for both the European and Congolese workers (>400 cases per 1000 workers), potentially caused by a surge in newly incoming employees combined with quinine scarcity due to the ongoing war effort.31
World War II marked a turning point in drug discoveries, leading to the development of chloroquine (CQ) and residual insecticides such as DDT. The first trial of DDT in Haut-Katanga occurred in June 1947 in Likasi.36 46 The effect of DDT spraying was clearly visible in our time series data, as malaria incidence decreased from >400 at the end of World War II to <100 during the next decade (figure 1). Because of this success, subsequent reports from 1954 onwards recommended the use of insecticide (DEET and Dieldrin) spraying twice a year in all European and Congolese houses. Chloroquine (Levaquin) was commonly used as prophylaxis for newborns, and larval breeding sites were treated weekly with petrol and DDT. The enormous success of DDT led to its rapid distribution across all campsites at UMHK, prompting a reduction in other larval control measures. Concurrently, between 1955 and 1969, the focus of malaria control shifted towards the widespread application of these novel technologies, including indoor residual spraying (IRS), as suggested by the GMEP.24 Although the high endemicity and weak infrastructure in Haut-Katanga made disease eradication unfeasible, this period coincides with the lowest malaria incidence of the past century in our time series data (<50 cases per 1000 employees per year). Consequently, the disease was no longer perceived as a significant problem by the UMHK staff and all other control measures (eg, vector control) were suspended.36 47
Between 1970 and 1999, malaria programmes in Africa underwent a shift in objectives, moving from elimination to a renewed emphasis on disease control.48 Notably, we observe an increase in malaria cases from the mid-1970s in Haut-Katanga, potentially attributed to a scarcity of primary medication, decreased vector control efforts and increased human activity near rivers for agriculture.34 These relaxed control measures might also have been the result of civil unrest in the region at that time, such as the Shaba invasions in 1977–1978.34 This period furter coincides with the global emergence of mosquito resistance to organochloride insecticides, first identified in the Katanga region in 1986.49 In the late 1990s, a series of international meetings culminated in the Roll Back Malaria initiative, aiming to reduce malaria mortality by half within a decade.48 Consequently, from 2000 to 2010, Haut-Katanga witnessed the implementation of intermittent presumptive treatment of malaria during pregnancy and a transition from chloroquine and sulfadoxine-pyrimethamine to artemisinin-based combination therapy.32 39 Surprisingly, we noticed that the malaria incidence during this period in Lubumbashi was as high as during the pre-GMEP period and even increased during this decade (figure 1). While part of this increase can be attributed to an increase in chloroquine and insecticide resistance, political instability during the civil war period (1997–2003) has likely affected malaria control efforts due to increased human mobility and the deterioration of healthcare infrastructure. Furthermore, after the civil war, road construction led to more stagnant water pools around the city, which was also suggested to have increased the malaria incidence in Haut-Katanga.39 The launch of the Global Technical Strategy in 2016 revitalised the global ambition to eradicate malaria.25 50 In Haut-Katanga, the subsequent period (2010 to the present) is characterised by the widespread distribution of long-lasting insecticide-treated nets (ITNs), the scale-up of rapid diagnostic tests and the implementation of seasonal malaria chemoprevention.28 30 32 39 Despite these extensive control efforts, we noticed that the current malaria incidence in the region appears to be as high as it was in the 1930s (figure 1).
Malaria in the highlands of Haut-Katanga, predominantly attributed to Plasmodium falciparum and transmitted by Anopheles funestus and Anopheles gambiae, exhibits a distinct seasonality.51–54 Historically (1930–1946), the pattern showed a delayed increase in incidence, 2 months following the onset of the rainy season in November, and a delayed decline 1 month after the start of the dry season in May.36 During 2003–2011, we observed that the malaria season began 2 months earlier compared with the historical period (figure 2). While our study’s correlative nature precludes pinpointing a single factor for this trend, it is noteworthy that the mean minimum temperature rose significantly over the past century in Haut-Katanga, nearing the 16°C threshold crucial for malaria sporogony. Even a slight increase above this threshold has been shown to markedly enhance malaria transmission.12 16 By analysing the temperature profile preceding the seasonal incidence surge (September-October), we noted a nearly 5°C rise, from 12°C to 17°C, between historical and recent climatological data, clearly surpassing this sporogony threshold. Consequently, it seems that the constraints imposed by low temperatures on malaria transmission during the start of the rainy season have diminished in recent years, potentially explaining the temporal shift in peak incidence in Haut-Katanga. We expect that further increases in minimum temperature may increase malaria incidence until reaching an optimal temperature boundary, consistent with predictions for many highland regions in sub-Saharan Africa.14 17 Our analyses also suggest that changing trends in rainfall or maximum temperature are less likely contributors to alterations in malaria incidence in the area.
One limitation of our study was the lack of disaggregated medical malaria records for the various mining sites in Haut-Katanga. As a result, we compared historical malaria data from Lubumbashi and Likasi (along with some minor sites) to recent data from Lubumbashi exclusively. Any potential seasonal differences in malaria incidence among these field sites could potentially explain the observed disparities in malaria seasonality between recent and historical data. However, we anticipate that such geographical influence would be minimal given that all these sites are situated at approximately the same altitude as Lubumbashi (between 1250 and 1450 m.a.s.l), currently exhibit similar annual rainfall and temperature patterns and are located relatively close to each other (<150 km). Another limitation of our study is the use of different climatological data sets over time. While discrepancies in minimum temperature between the historical (INEAC) and recent data sets (ERA5 or METTELSAT airport data) might arise from varying measurement techniques (figure 2), we believe the observed increase in mean minimum temperature reflects genuine climatic changes. This belief is supported by several factors: (1) both colonial and recent stations used standardised methods with thermometers accurate to 0.1°C in the shade, (2) the colonial and modern stations were located within 10 km of each other and (3) ERA5 data, though modelled, also show a significant continuous rise in minimum temperature from 1940 to 2022 (see figures 1 and 2). These points collectively reinforce the credibility of our findings despite the inherent limitations of the data sources.
Conclusion
Despite contemporary control efforts, our findings indicate that the current malaria incidence in Haut-Katanga is comparable to levels observed before the GMEP, matching those from the 1930s. The significant increase in minimum temperatures over the past century, which facilitates prolonged optimal conditions for malaria sporogony, suggests that climate change may be a significant driver of this recent rise in incidence. To counter these climate effects, we conclude that sustained and increased investments in malaria control are crucial to limiting transmission in highland settings. The provision of ITNs has been a significant factor in the worldwide decline of malaria.3 24–26 However, in our time series, we did not observe a clear effect of ITNs, first introduced in 2012 in Katanga, on malaria transmission. This could be attributed to the short effective lifespans of these nets. Indeed, similar studies in other African highland settings have also shown that the effective lifespan of these nets is often shorter than the replacement interval (typically 3 years), highlighting the need for more frequent replacement.14 Moreover, the lowest malaria incidence in Haut-Katanga was observed during periods of intensive DDT use and IRS, emphasising the importance of reducing vector populations as a key strategy for controlling and eliminating malaria. Consistent with other studies,55 we recommend the systematic use of IRS as a key component of malaria control strategies in endemic highland regions of sub-Saharan Africa.
Supplemental material
Data availability statement
Data are available upon reasonable request.
Ethics statements
Patient consent for publication
Acknowledgments
We express our gratitude to Dirk Schoonbaert for his assistance in navigating the archives of the Institute of Tropical Medicine in Antwerp, which led to the discovery of these unique malaria time series. We also extend our thanks to professor Kaat Wils and Hannah de Korte (KU Leuven) for directing us to the Gécamines data and discussing our initial findings. We also thank both anonymous referees for their constructive feedback on this manuscript.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Handling editor Emma Veitch
Contributors Conceived the study: JM. Wrote the paper: JM. Provided data: VMT and EM. Performed the statistical analyses: JM. Supervised work: HL and TH. All authors read and approved the final manuscript. A reflexivity statement is provided as a supporting document. JM is the guarantor.
Funding The research was funded by the Belgian Science Policy Office in the framework of the FED-tWIN project RECORDED (Prf-2021-033): Reconstructing disease dynamics in Central Africa using historical museum collections and archives.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Author note The reflexivity statement for this paper is linked as an online supplemental file 2.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.