Background
There are renewed calls for malaria eradication with a focus on Africa [
1,
2]. Malaria mortality rates are decreasing in many populations, with global incidence having fallen by approximately 37 % since 2000 to 214 million new cases in 2015 [
3,
4]. Despite these gains, malaria remains a major global disease burden, with approximately 438,000 deaths annually [
3,
5].
The elimination of malaria from many western nations has been attributed primarily to social and economic development allowing for screening of windows and doors, destruction of vector breeding sites, and rapid diagnosis and treatment [
6]. The feasibility of
Plasmodium falciparum malaria elimination in most of sub-Saharan Africa is low, with Uganda being among the countries with lowest feasibility [
7]. Sub-Saharan African countries are disproportionately affected by malaria [
4] due to the presence of highly competent mosquito vectors, widespread poverty, limited infrastructure, and overburdened health systems [
8,
9]. Those living in extreme poverty are most vulnerable to infectious diseases, yet within-country disparities are often ignored [
10,
11]. African indigenous populations in particular have consistently poorer health outcomes than their non-indigenous counterparts [
12]. Social determinants of indigenous health include, but are not limited to, poverty, discrimination, limited access to health care, and loss of traditional lands [
13]. Indigenous and ethnic minority populations outside of Uganda experience higher rates of malaria, which have been attributed to relative impoverishment, marginalization, and geographic remoteness [
14‐
18]. In some cases, genetic variations have been identified as drivers of ethnic differences in malaria parasitaemia and immunological response [
19].
Risk factors for malaria can be conceptualized as non-modifiable and modifiable. Non-modifiable factors such as age and sex have been inconsistently associated with higher malaria risk. In endemic areas, children under 5 years of age have high risk of malaria due to their immunological naïveté [
10,
20,
21]. Beyond this, the relationship between age and malaria in adults is less clear [
22]. It is known that regular exposure to malaria results in a functional immunity to the disease which quickly wanes in the absence of exposure. Excluding immunologically-suppressed pregnant women, sex-related variations in malaria risk are generally linked to gender-role and occupational exposures [
23,
24]. Women in roles as household water-collectors may spend more time near mosquito breeding sites. Men with forest-related jobs may spend their time in mosquito-dense areas [
10,
23].
Modifiable risk factors include environmental conditions and human behaviour. Local vector ecology, including the locations of swamps, forests, rice paddies where vectors breed, and the proximity of these sites to human habitations may bring humans into frequent contact with mosquitoes [
25‐
27]. Housing conditions may also affect transmission. Open eaves and windows, for example, may permit mosquito entry into sleeping quarters [
25,
28,
29]. Education and wealth are known to be protective against malaria. Understanding malaria transmission and prevention may result in behaviours such as staying indoors during peak vector activity, and having access to preventive strategies that may reduce exposure of humans to vectors [
27,
30]. Many of these risk factors are mediated by poverty whereby access to building materials and bed nets may be dependent upon income [
31‐
33]. Poverty has been described as a key modifiable determinant of malaria burden [
6].
Livestock are routinely included in analyses of risk factors for malaria infection, yet their role in malaria transmission is not well understood. Livestock are dead-end hosts for human-infectious
Plasmodium parasites and may reduce human malaria risk by drawing vectors away from humans (zooprophylaxis) [
34‐
36]. In some cases, however, livestock act as additional blood meal sources, and they may alter vector longevity and population density to increase human malaria risk (zoopotentiation) [
37,
38]. Empirical research exploring the association between livestock and malaria risk has been complicated by the confounding role of wealth [
39]. Livestock symbolize social standing, provide food and services, and act as an asset to be sold in times of financial need [
40‐
44]. Both livestock and wealth are generally associated with lower malaria prevalence. Given efforts to include zooprophylaxis in integrated vector control programs, there have been calls for further research to understand the impact of livestock on malaria transmission [
39].
The Ugandan Batwa are an indigenous population with life expectancy and child mortality rates significantly worse than national averages [
45]. There remain considerable gaps regarding our understanding of Batwa health; to our knowledge only two peer-reviewed studies currently report the prevalence and risk factors of health outcomes for Batwa in Uganda [
23,
46]. While literature on Batwa livelihoods is currently limited, it is thought that Batwa engage in livestock livelihoods at a much lower rate than non-Batwa due to financial restrictions, in addition to their historical hunter-gatherer culture. Further, Namanya identified consistent differences across risk factors for malaria, including housing and education between Batwa and non-Batwa [
47].
The aims of this paper were (1) to estimate malaria prevalence in indigenous Batwa of Kanungu District and compare this with their non-indigenous neighbours, (2) explore modifiable risk factors for malaria parasitaemia in order to identify potential entry points for intervention to reduce malaria prevalence among Batwa and non-Batwa and, (3) to test the hypothesis of zooprophylaxis in a sub-set of livestock owners from both Batwa and non-Batwa populations.
Discussion
The period prevalence of
P. falciparum parasitaemia for July 2013 and April 2014 for adult Batwa was 9.35 %, which is higher than the 4.1 % rate previously reported for Batwa over the age of 5 years [
23]. A bed net distribution was carried out in November 2012, which could explain the low prevalence rate found previously in this population in January 2013 [
59]. These findings are consistent with parasite prevalence estimates for Kanungu District as being relatively low (10–40 % prevalence when including children) compared to other parts of Uganda [
8,
60].
Ethnicity was associated with malaria parasitaemia, with Batwa at higher risk. This is an important result given that wealth and other theorized risk factors for parasitaemia were controlled for; there remains a residual effect of ethnicity beyond these factors. There are marked differences in livelihoods between the two populations that may be driving these results. Batwa are currently undergoing a transition as they adapt to life outside of the forest. This transition is reflected in the low levels of education and livestock ownership relative to non-Batwa. There may also be unmeasured genetic differences between these ethnic groups that influence malaria prevalence [
61‐
64]. For instance, ethnic differences in immune responses [
19,
65] and lower susceptibility to malaria infection among ethnic minority Fulani tribes relative to sympatric ethnic groups in West Africa have been attributed to genetic differences at various loci [
66,
67]. It might be expected, given their higher prevalence of infection, that Batwa mount a weaker immune response relative to non-Batwa in the face of exposure. Immunological studies would be required to test this hypothesis, but it is unclear how the results of such studies would lead to intervention strategies [
6].
These results provide tentative indication that there may be important modifiable risks for malaria infection among Batwa and non-Batwa. Iron sheet roofing may play a role in malaria risk. House construction is an important risk factor for malaria infection [
68‐
70] and thatched houses have previously been related to higher malaria prevalence of inhabitants in other parts of sub-Saharan Africa [
28] due to the propensity of mosquitoes to rest indoors [
25]. However, a systematic review and meta-analysis on the effect of house construction on malaria found that modern roof materials (iron sheets, tiles) were not consistently associated with decreased odds of infection [
71]. Thatched or wooden roofs may confer some protection to individuals relative to corrugated iron sheets. It is possible that the open eaves of iron sheet roofs facilitate mosquito house entry and thereby increased malaria risk. Open eaves and gaps in housing materials are frequently associated with higher rates of parasitaemia [
24,
29,
71]. However, this result may be confounded by wealth. Among non-Batwa, iron sheet roofs are a reflection of wealth since they must be purchased at a high cost compared to thatch or banana fibre, which may be obtained at no cost from the surrounding environment. In contrast, iron sheet roofing is purchased for Batwa by a local NGO with priority given to the most impoverished families as identified by the community. A Ugandan study carried out in part in Kanungu District found that closed eaves and modern house construction were associated with significantly decreased human biting rates and incidence rates of malaria in children when controlling for age, sex, and household wealth [
72]. Further evaluation of the impact of house construction on malaria risk among Batwa may help to inform and improve these community development strategies.
The analysis of livestock owners suggested that when controlling for wealth and bed net use, keeping animals inside at night reduced the odds of malaria infection. Previous studies suggest that keeping animals indoors at night increases malaria risk where mosquitoes are zoophilic [
39]. Entomological studies in the region suggest that
Anopheles funestus and
Anopheles gambiae sensu lato are the predominant vector species [
60].
Anopheles funestus tends towards anthropophily; however,
An. gambiae s.l. consists of seven morphologically indistinguishable species [
73] of which the most important vectors may be more zoophilic, as with
Anopheles arabiensis, or anthropophilic, as with
An. gambiae sensu stricto. This finding might suggest that the predominant local species are anthropophilic and, in turn, that keeping animals indoors at night results in reduced malaria risk among livestock owners, however further entomological evaluation of local vector populations is required. The mechanism through which livestock infer protection against malaria remains poorly understood and their role in malaria transmission has been much contested [
32,
35,
37,
38,
74,
75]. Livestock may draw mosquitoes away from humans, reducing their exposure to malaria [
76,
77], or may provide an abundance of blood-meals, increasing vector density and longevity [
32,
38,
74]. The zooprophylatic effect is, however, complicated by the relationship between livestock and wealth [
32,
39]. It is well recognized that livestock contribute significantly to household economies for the rural poor, including the most marginalized [
40,
41,
43]. Wealthy community members in Kanungu are those with the greatest livestock and land holdings. The findings of this study are consistent with others showing that poverty is positively associated with malaria prevalence [
31,
78,
79].
The cross-sectional nature of this study limits the ability to infer causal relationships between the risk factors and malaria infection. Unemployed people who were available to be surveyed during data collection events were overrepresented in the sample. This may have reduced variation within the population and perhaps led to an underestimation of the importance of employment for malaria infection. Similarly, men were more likely to be working away, leading to an overrepresentation of women. Given that the role of sex in malaria risk remains unclear, it is difficult to predict the direction of this effect. The small sample size and number of cases limited statistical power to detect effect sizes. Sample size also prevented the stratification of uni- and multi-variable analyses of parasitemia based on ethnicity, although sensitivity analysis suggested results were consistent between ethnic groups. Similarly, sample size prevented ethnicity-based stratification for the zooprophylaxis analysis. As a result, the identification of risk factors, and recommendations for malaria control, apply to the survey population as a whole. In highly vulnerable remote indigenous populations, sample sizes are typically small and demographics frequently differ from non-indigenous populations; lack of statistical power must be balanced with the importance of prioritizing research within vulnerable and at-risk sub-populations.
Authors’ contributions
LBF, SL, DBN, SH and IHACC Research Team conceptualized, designed, and obtained funding for, the Indigenous longitudinal health survey tool Health. BD, LBF, NAR, and PM conceptualized and obtained funding for the non-indigenous sampling design and data collection. BD, ST, JL, SH, and MK carried out data collection. BD carried out statistical analysis and drafted the manuscript. LBF, NAR, and PM supervised statistical analysis. All authors contributed to manuscript editing. All authors read and approved the final manuscript.
Acknowledgements
We sincerely thank the ten Batwa communities in Kanungu District: Buhoma, Byumba, Bikuuto, Karehe, Kitahurira, Kihembe, Kebiremu, Mukongoro, Rurangara, and Kitariro, for their engagement with the project. We also thank the Local Council 1 chairpersons of Buhoma, Byumba, Bikuuto, Mukono, Mpungu, Kihembe, Kebiremu, Mukongoro, Kishanda, and Kitariro and the non-Batwa community members for their participation. We deeply appreciate the IHACC survey administrators Aine Tom, Ainembabazi Brian, Amanya Prince, Asaasira Grace, Asinath Kyomuhangi, City Tyson Magezi, Collins, Cranmar Magezi, Dan Mugume, Ivan Arineitwe, Kaheru John, Kenneth Kirihariwe, Kesande Annet, Kesande Charity, Kokunda Sylvia, Komuhangi Rovance, Matukunda Judith, Ninsiima Evas, Orikiriza Mathias, Orishaba Rabecca, Tom Kabategyire, Tumuhimbise Ishmael and Turyaeyira Bright. We also acknowledge the work of IHACC research team members Emmanuel Eloku, Hubert Nkabura, Jamen Kasumba, Fortunate Twembabaze, Christine Nantogo, Martin Kigozi, and Allan Gordon. We would like to thank Sierra Clark and Margot Charette for help with data entry. BD would like to thank Kaitlin Patterson for her assistance with statistical analysis. BD would like to acknowledge funding from an IDRC Doctoral Research Award.
This research is part of an international project entitled the “Indigenous Health Adaptation to Climate Change” (IHACC) project (
http://www.ihacc.ca), with parallel field study sites in the Canadian Arctic and Peru. We thank our local partners; Ugandan Ministry of Health, Kanungu District Administration, Bwindi Community Hospital, Batwa Development Program. Funding was provided by CIHR/NSERC/SSHRC and IDRC Tri-Council Initiative on Adaptation to Climate Change, Indigenous Health Adaptation to Climate Change (IHACC), IDRC File N. 106372-003, 004, 005; CIHR Open Operating Grant, Adaptation to the health effects of climate change among Indigenous peoples in the global south (IP-ADAPT), Application N. 298312. We would also like to thank the anonymous reviewer for their thoughtful feedback.
IHACC Research Team: James Ford, Cesar Carcamo, Alejandro Llanos, and Victoria Edge.