Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Do socio-demographic factors modify the effect of weather on malaria in Kanungu District, Uganda?

Do socio-demographic factors modify the effect of weather on malaria in Kanungu District, Uganda? Background: There is concern in the international community regarding the influence of climate change on weather variables and seasonality that, in part, determine the rates of malaria. This study examined the role of sociodemo‑ graphic variables in modifying the association between temperature and malaria in Kanungu District (Southwest Uganda). Methods: Hospital admissions data from Bwindi Community Hospital were combined with meteorological satellite data from 2011 to 2014. Descriptive statistics were used to describe the distribution of malaria admissions by age, sex, and ethnicity (i.e. Bakiga and Indigenous Batwa). To examine how sociodemographic variables modified the associa‑ tion between temperature and malaria admissions, this study used negative binomial regression stratified by age, sex, and ethnicity, and negative binomial regression models that examined interactions between temperature and age, sex, and ethnicity. Results: Malaria admission incidence was 1.99 times greater among Batwa than Bakiga in hot temperature quartiles compared to cooler temperature quartiles, and that 6–12 year old children had a higher magnitude of association of malaria admissions with temperature compared to the reference category of 0–5 years old (IRR = 2.07 (1.40, 3.07)). Discussion: Results indicate that socio‑ demographic variables may modify the association between temperature and malaria. In some cases, such as age, the weather‑malaria association in sub ‑populations with the highest inci‑ dence of malaria in standard models differed from those most sensitive to temperature as found in these stratified models. Conclusion: The effect modification approach used herein can be used to improve understanding of how changes in weather resulting from climate change might shift social gradients in health. Keywords: Malaria, Climate change, Weather, Meteorological, Sex, Age, Indigenous, Batwa, Bakiga, Sociodemographic modifiers, Uganda Background Malaria continues to pose a threat to human health worldwide. Approximately 92% of all malaria cases *Correspondence: ostk91@gmail.com in 2017 occurred in the World Health Organization School of Epidemiology and Public Health, University of Ottawa, Ottawa, (WHO) African Region. Five of these countries, pri- Canada marily in sub-Saharan Africa, accounted for half of the Full list of author information is available at the end of the article © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Ost et al. Malaria Journal (2022) 21:98 Page 2 of 13 malaria cases worldwide [1]. Uganda accounted for 4% Methods of all cases of malaria in 2017, and is among the coun- Kanungu district, Uganda tries facing significant challenges to Plasmodium fal - This study was conducted in Kanungu District, located ciparum malaria elimination due to the presence of in southwestern Uganda, near Bwindi Impenetrable highly competent mosquito vectors and lack of infra- National Park (Fig.  1). Over the last 50 years, this region structure systems in place to support elimination [1, of eastern Africa experienced an increase in seasonal 2]. Climate change threatens progress made towards mean temperature [16, 17]. Warming trends are likely to malaria elimination in many areas of the world includ- continue, with an increase in mean temperature of up to ing Uganda. The Intergovernmental Panel on Climate 2.0 °C projected by 2030, and an increase in regional dry- Change (IPCC) concluded with medium to high con- ing [16]. fidence that climate change could alter the geographic The region is primarily inhabited by Bakiga and indig - range of the Anopheles vector, creating the potential for enous Batwa people, both of whom face relatively high longer transmission seasons and increasing the number ill-health burdens when compared with the national of people at risk, and noted that this projection varies average. Both Bakiga and Batwa people are highly vul- regionally [3]. nerable to health impacts of a warming climate, and Among the most vulnerable to the effects of climate have identified malaria, food insecurity, and gastrointes - change are indigenous people who already face dispro- tinal illnesses as climate-sensitive health concerns [18, portionate burdens of health and social inequity [4]. 19]. The indigenous Batwa experience higher prevalence Sub-Saharan African indigenous people, in particular, of malaria compared to the Bakiga, 9.4% versus 4.5%, often have more health challenges than non-indige- respectively [2]. This difference in malaria prevalence is nous people living in the same geographic areas [5, 6]. paralleled by a range of health and socio-economic dis- This inequity in health outcomes is frequently rooted parities between the two populations (Table  1), includ- in colonization and the social determinants of health, ing reduced life expectancy [19]. The Batwa were forcibly including discrimination, loss of traditional lands, mar- removed from their ancestral lands with the creation of ginalization, and limited access to healthcare services the Bwindi Impenetrable Park in 1991 (Fig.  1), where [7, 8]. they relied on subsistence hunting and gathering; evic- In recent years, a number of countries have moved to tion from the park forced them into settlement in agrar- establish nation-wide policies regarding climate change ian communities outside the park boundaries (19). There adaptation activities, including in Uganda [9–11]. How- are currently approximately 6700 Batwa individuals living ever, nation-wide policies risk aggregation of diverse at- in southwestern Uganda [20], 900 of whom live within risk populations, masking important trends at a more Kanungu District. There are no notable ecological or granular level. Localized research in vulnerable popula- geographic differences in the areas where the Batwa or tions can underpin resource distribution and identifica - Bakiga live that would increase risk of malaria in either tion of focal vulnerabilities [12]. The districts in Uganda population (Fig. 1). vary greatly in their geography, demographic makeup, The Bwindi Community Hospital (BCH) was founded primary health concerns, and in the ways in which they in 2003 as a clinic to primarily serve Batwa [23]. Since experience climate change impacts. Factors such as soci- its founding, BCH has expanded into a large facility odemographic makeup of a district will be important in that includes six in-patient wards, including a pediatric, determining the most effective course of adaptive action adult, maternity, and immunodeficiency hospital wards, within specific communities. While many studies have as well as an out-patient ward, and several satellite clinics quantified the association between temperature and pre - for remote settlements [23]. BCH operates on a fee-for cipitation for malaria transmission [13–15], these studies service model, and donations help to subsidize an insur- typically do not consider how socio-demographic char- ance scheme for residents who qualify [18, 24]. All Batwa acteristics modify these associations. This study aims to residents are covered under this insurance plan (eQuality address this gap through evaluation of the role of soci- health insurance), which has enrolled 34% of its catch- odemographic factors—age, sex, and ethnicity—in modi- ment area in the plan as of 2020 [23, 25]. fying the association between temperature and malaria Kanungu District is a rural area of rolling hills located in Kanungu District Uganda from 2011 to 2014 using at an elevation of 1,310 m above sea level [26]. There are data from the Bwindi Community Hospital. This study four species of malaria parasite that affect humans in explored effect modification by examining how socio- Uganda, the most virulent being P. falciparum [27, 28]. demographic variables interact and by stratifying models, Plasmodium falciparum is the primary endemic malaria which has important implications for public health policy parasite found in the Kanungu region and is most often under a changing climate. carried by the Anopheles gambiae mosquito species [29]. Ost  et al. Malaria Journal (2022) 21:98 Page 3 of 13 Fig. 1 Map of Kanungu district and location of Bwindi community hospital Table 1 Socio‑ economic and health differences between Bakiga and Indigenous Batwa populations (adapted from MacVicar et al. 2017a [21]) Health and socio-economic measure Batwa (proportion of the population (%)) Bakiga (proportion of the population (%)) Malaria prevalence among adult 6.45 4.46 Moderate acute malnutrition among adult women 45.86 0.42 Household mosquito net use (did not have nets) 70.99 53.56 Access to handwashing facilities (did not have access to handwashing) 73.85 56.40 Access to soap (did not have access to soap) 73.85 56.40 Prevalence of positive malaria antigen detection test in both July 2013 and April 2014—survey of all Batwa adults, sample of Bakiga adults[2] Classified as moderately malnourished according to the Uganda ministry of health integrated management of acute malnutrition guidelines)[22] [2] d,e Only asked of people that had access to hand washing facility, for example for the Batwa, 32 or 94% of the households that had access to handwashing had access to soap[2] Data collection (RDT) or blood slide in conjunction with symptoms. Hospital data Individual inpatient records from the hospital were Electronic records of patients with a malaria diagnosis merged with insurance coverage data based on patient from 1-Jan-11 through 21-Dec-14 were obtained through ID to provide additional data on sex, age, and ethnicity partnership with BCH. Malaria diagnosis was defined [24]. Data were de-identified prior to analysis to ensure as any case with a positive rapid antigen diagnostic test the confidentiality of patients. In total, there were data Ost et al. Malaria Journal (2022) 21:98 Page 4 of 13 age, and ethnicity, and were retained for this study; cases with incomplete information were excluded. Approxi- mately 51% of data were missing information on ethnic- ity, which was one effect modifier of interest. Excluded cases had a similar demographic distribution to the final sample used in the analysis based on the initial testing (Additional file 1: Table S1). Meteorological data Meteorological data were estimated from the European Centre for Medium Range Weather Forecasts Re-analy- sis (ERA)-Interim Climate Database that combined data from multiple sources. The ERA-Interim climate data - bases have a spatial resolution of 0.75°  by 0.75°  . Daily values for total precipitation (i.e. rainfall (mm)) as well as maximum, minimum, and average temperature (°C) Fig. 2 Conceptual model illustrating social modification of were obtained for all dates matching the extracted medi- malaria‑ weather relationship: boxes and horizontal arrows represent main pathways of interest in climate–malaria relationship, while the cal records (i.e. 1 January, 2011 to 31 December, 2014) black box represents our sociodemographic variables of interest. Bold [24]. Meteorological data were merged with the BCH vertical arrows indicate theorized effect modification data based on date of admission; lags were then created to account for the assumed time between mosquito/para- site development, point of infection, and finally the day of for 39,287 admissions (all diagnoses) at BCH and 6602 admission. malaria admissions were reported during the years 2011– The research team focused on the extent to which non- 2014. Of these, 18,846 (48%) of the admissions and 3440 meteorological variables modified the effect of tempera - (52%) of malaria admissions, could be matched on sex, ture on malaria hospital admission incidence. As such, IRRs for Stratified Models, hot vs. cool quartiles 0.3 Demographic Variables Fig. 3 Logarithmic scale; incidence rate ratios by demographic category for stratified models, gridline marks IRR of 1, stratified models for season do not include interaction term for season and temperature Incidence Rate Ratio (IRR) with Robust Standard Error Log Scale Baseline Batwa Bakiga Female Male Age 0-5 Age 6-12 Age 13-18 Age 19-55 Age 55+ Season (Dry) Season (Wet) Ost  et al. Malaria Journal (2022) 21:98 Page 5 of 13 models did not aim to maximize precision in the specifi - with malaria hospital admission incidence (Fig. 2). Exam- cation of the temperature-malaria association, but rather ining effect modification can be achieved through two assess the extent to which this generalized association methods: interaction and stratification. As such, three is sensitive to effect modification. Lags were created for types of models were used: (1) a baseline model that did both ambient temperature and precipitation, both impor- not include interaction variables or stratify data; (2) mod- tant predictors of malaria risk [3], out to six months prior els with weather- and age, sex, and ethnicity interaction to admission date under the a priori assumption that variables; and (3) models stratified by age, sex, and eth - a biologically plausible time lag for malaria would not nicity. The team compared model results among these extend past four months, and would not be less than one types of models, comparing the magnitude and direc- month [29–34]. A combined 12 and 13  week lag—the tion of the associations between temperature and malaria time between admission date and temperature preceding admissions. Effect modification methods were informed that date by 77–91  days—in mean weekly temperature by [24, 36]. was identified as having the most significant and strong - est association with malaria hospital admission rates Baseline model and was thus chosen for further analysis (Table  2). The The research team first constructed a model with all team converted the variable (temperature) into a binary meteorological and socio-demographic variables to variable reflecting the highest quartile (versus the low - measure the extent to which temperature is associated est 3 quartiles combined) of mean weekly temperature with malaria hospital admissions without accounting for 12 and 13  weeks prior to admission in order to evaluate any effect modification. We retained sociodemographic ‘hot’ weeks versus milder or ‘cooler’ weeks. Precipitation variables age, sex, and ethnicity in the model as fixed was not found to be significant in bivariate analysis, so a effects along with year and the meteorological variable binary variable for season was created based on date of season. admission; rainy seasons were defined as March-June and September–November, and dry seasons were defined Interaction model as December-February and July–August [35]. Season The team used interaction variables to estimate the mar - was retained in all models to account for the dependent ginal effects of temperature and age, sex, and ethnicity. nature of temperature and precipitation in the mosquito- This interaction approach took into account interaction weather relationship [13]. Because of the importance of and confounding between variables in the model. While precipitation in this relationship the research team per- theoretically more analytically robust, the coefficients of formed an additional sensitivity analysis using rainfall as interaction variables are sensitive to sample size and can the predictive variable in the models, and assessed the be less intuitive to interpret. effect on the size, direction, and significance of the other model coefficients. Stratified models Stratification of models involved fitting separate models Data analysis for each level of strata. Using stratified models provided Conceptual approach a more interpretable estimates of the effects of weather Effect modification occurs when a variable differentially on malaria in different age, sex, and ethnicity strata. This modifies the observed effect of a risk factor on disease approach is less sensitive to small sample sizes, which status. This study examined the ways in which age, sex, was particularly pertinent given the sample size for Batwa and ethnicity changed the association of temperature in the dataset was small. The team compared incidence rate ratios (IRRs) between demographic variables using a ratio of ratios approach [37]. Table 2 Incidence Rate Ratio (IRR) for malaria hospital admission incidence and temperature, by weekly lag Model variables Lag by week IRR p-value 95% CI A summary of variables used in the models is pro- vided in Table  3. The count of weekly malaria hospi - 10 1.02 0.63 (0.94, 1.11) tal admissions was the dependent (outcome) variable 11 0.99 0.82 (0.91, 1.08) of interest. Temperature was the independent (expo- 12 1.09 0.04 (1.00, 1.19) sure) variable of interest: specifically, the highest quar - 13 1.12 0.01 (1.03, 1.22) tile (versus the lowest 3 quartiles combined) of mean 14 1.06 0.15 (0.98, 1.16) weekly temperature 12 and 13  weeks prior to admission 15 1.07 0.09 (0.98, 1.18) was examined. Age, sex, and ethnicity were examined as a,b Bold text indicates time lags between admission date and temperature potential effect modifiers. Age was categorized into the preceding that date selected for use in final models Ost et al. Malaria Journal (2022) 21:98 Page 6 of 13 Stratified models following categories: < 5  years, 6–12  years, 13–18  years, The team ran the baseline models stratified by age, sex, 19–55  years, and > 55  years of age. Additionally, the ethnicity. Stratified models also contained all control research team conducted a sensitivity analysis on age to and interaction variables as fixed effects to minimize the verify that results were not sensitive to different age cut- effect of confounders in the analysis. offs. Ethnicity was divided into the two main groups: the Indigenous Batwa, and all other ethnic groups, primarily Ethics consisting of ethnic Bakiga. Year and season were identi- This study approved by ethics boards at McGill Univer - fied a priori as confounding variables. sity, the University of Guelph, the University of Alberta, and the University of Washington, as well as Bwindi Negative binomial multivariable regression models Community Hospital. All personal identifiers were Researchers used a negative binomial multivariable removed from the dataset before analysis. This research regression model with the count of total weekly malaria is conducted within the broader IHACC project, in cases as the dependent (outcome) variable. For the pop- partnership with Makerere University, and in collabora- ulation at risk (offset) variable, we used the weekly total tion with Batwa health and development programmes in admissions to BCH for any diagnosis. Year was controlled Kampala and in the region. Communities have already for in all models as a fixed effect. All analyses were con - consented to ongoing collaboration with the IHACC ducted in STATA v.15.1 (Stata Corp., USA). research project and team. At the time of data collection, the UNSCT was not accepting applications and was not Baseline model granting ethics approvals. We first built an unstratified model with no interactions that included sociodemographic variables as fixed effects Limitations and meteorological exposures: Data represent a short period of time (less than 4  years) and were, therefore, insufficient to infer rela- ln weekly malaria counts = β + β mean weekly T L12 − 13 0 1 tionships between malaria and climate change for long + β (sex) + β age + β ethinicity 2 3 4 timeframes. Additionally, though wet and dry season + β (season) + β (Year) 5 6 were included in the models to represent the complex + ln (population at risk) relationship between meteorological variables like temperature and precipitation, and malaria, there are limitations to excluding precipitation/rainfall as a vari- Interaction model able. Future work should examine these meteorologi- The team examined interaction between socio-demo - cal variables in this context in more detail. The Batwa graphic variables (i.e. age, sex, ethnicity, and season) and sample in the dataset was small, reducing statistical temperature, and their association with malaria hospital power in our analysis. Despite this, the team chose to admissions. To illustrate the size, direction, and confi - examine ethnicity, given its indicative role as a poten- dence interval of interactions, researchers evaluated lin- tially important driver of vulnerability to malaria in ear combinations of estimates for weather with season, the region and given the Batwa rank malaria as a top sex, age, and ethnicity. The team included all interaction climate-sensitive health outcome [2, 19, 38]. In doing variables in a single model. The final model equation so, the research team sought to avoid exclusion of used for analyses was: small, but marginalized and high-risk indigenous peo- ple from climate-health analytic research [19]. Under- ln weekly malaria counts = β + β mean weekly T L12 − 13 0 1 standing malaria impacts in view of weather and + β (sex) + β age + β ethinicity 2 3 4 climate in small vulnerable groups is vital in informing + β season + β Year 5( ) 6( ) health policy. + β mean weekly T L12 − 13 ∗ season + β mean weekly T L12 − 13 ∗ sex 8 Results Descriptive statistics + β mean weekly T L12 − 13 ∗ age ◦ Of the hospital admissions data with complete sociode- + β mean weekly T L12 − 13 ∗ ethinicity mographic records over the study period, the research + β mean weekly T L12 − 13 ∗ Year + ln (population at risk) Ost  et al. Malaria Journal (2022) 21:98 Page 7 of 13 Table 3 Description of dependent, independent, effect modification, and control variables used in effect modification analysis Variable (units) Description Dependent (outcome) variable Weekly malaria cases Total case count per 7 day period Independent (exposure) variable Mean weekly temperature ( ̊C) with a 12–13 weekly average tem‑ Binary: lower (cooler) quartiles 1–3 (referent); top (hottest) quartile perature lag Eec ff t modification variables Ethnicity Binary: Bakiga and other (referent); Batwa Sex Binary: male (referent); female Age Categorical: < 5 years (referent); 6‑12 years; 13‑18 years; 19‑55 years; > 55 years Confounding (control) variables Season Binary: dry (referent); wet Modelled with an interaction term with the independent variable Year Categorical: 2011 (referent), 2012, 2013, 2014 team found that 56.7 percent of malaria cases in the distributed, with 53.1 percent of malaria cases being sample occurred during the wet season from March– female, comprising 56.1 percent of the total admissions June or September–November. Sex was relatively evenly data. A majority (36.9%) of the total hospital admissions Table 4 Descriptive statistics of variables included in final models data from Bwindi Community Hospital, Uganda (2011–2014). Demographics No. of (x) demographic Proportion of (x) No. of (x) demographic out Proportion of (x) out of all admissions demographic out of all of all malaria admissions demographic out of all admissions malaria admissions Female 10,565 56.10% 1826 53.10% Male 8281 43.90% 1614 46.90% 0–5 years 5687 30.20% 1143 33.20% 6–12 years 2514 13.30% 896 26.00% 13–18 years 1950 10.30% 414 12.00% 19–55 years 6957 36.90% 895 26.00% 55 + years 1738 9.20% 92 2.70% Ethnicity (Bakiga) 18,608 98.70% 3,386 98.40% Ethnicity (Batwa) 238 1.30% 54 1.60% Season (wet) 10,957 58.14% 1948 56.63% Season (dry) 7889 41.86% 1492 43.37% Total number* 18,846 3,440 Meteorological variables Descriptive temperature statis‑ Mean Min Max tics throughout (x) year (celsius) 2011 18.91 13.13 27.51 2012 19.07 12.22 28.67 2013 19.54 12.96 28.98 2014 19.55 13.32 29.19 Average daily and yearly total Average daily Yearly total rainfall (mm) 2011 3.55 1296 2012 3.55 1300 2013 3.22 1174 2014 3.07 1197 Ost et al. Malaria Journal (2022) 21:98 Page 8 of 13 Table 5 Results of baseline model measuring the effect of sample) were recorded as being Batwa; 22.7% of Batwa temperature on malaria with sociodemographic variables as hospital admissions were malaria cases during the study fixed‑ effects period, and 18.2% percent of the Bakiga hospital admis- sions were malaria cases (Table 4). Model variables Baseline model IRR (95% CI) p-value Temperature ‘Cool’ quartiles Referent Referent Baseline model Temperature ‘Hot’ quartile 1.27 (0.90, 1.80) 0.18 The weekly incidence rate of malaria hospital admissions Bakiga Referent Referent was 1.27 (0.90, 1.80) times higher during weeks with hot Batwa 1.08 (0.76, 1.55) 0.66 weather (highest quartile 29.30–29.42  °C) compared to Male Referent Referent weeks with cooler weather (three lower quartiles 29.08– Female 0.91 (0.84, 0.98) 0.02 29.30  °C) (Table  5). The weekly malaria hospital admis - 0–5 years old Referent Referent sions incidence rate among the indigenous Batwa was 6–12 years old 1.64 (1.48, 1.82) < 0.001 1.08 (0.76, 1.55) times higher than for Bakiga. The weekly 13–18 years old 1.08 (0.94, 1.24) 0.28 incidence rate of malaria hospital admissions for females 19–55 years old 0.65 (0.59, 0.71) < 0.001 was 0.91 (0.84, 0.98) times the rate of males. Compared 55 + years old 0.28 (0.20, 0.39) < 0.001 to children 0–5 years old, youth 6–12 years had the high- Season (wet) Referent Referent est weekly incidence rate of malaria hospital admissions, Season (dry) 1.30 (1.01, 1.68) 0.04 followed by youths aged 13–18  years. Weekly malaria *Bold indicates a p-value of < 0.05 hospital admission incidence rates were 1.30 (1.01, 1.68) times higher during the dry season compared to the wet season. Furthermore, the sensitivity analysis using rain- fall as the predictive variable in the models suggested were 19–55 years old, and 35.6 percent of 6–12 year old very little difference in the models, so it was removed children who were admitted to the hospital had a malaria from the analysis. infection. Only 238 hospital admissions (1.2% of the Table 6 Incidence rate ratio (IRR) interaction model results, including sociodemographic variables as effect modifiers Interaction model, results by temperature quartile Quartile 1–3 (cool) Quartile 4 (hot) IRR hot/IRR cool within strata IRR (95% CI) IRR (95% CI) of ethnicity, sex, age, and p-value p-value season Ratio of Ratios (ROR) Bakiga Referent Referent Referent Batwa 0.82 (0.34, 1.99) 1.63 (0.64, 4.16) 1.99 0.67 0.31 Male Referent Referent Referent Female 1.02 (0.86, 1.22) 2.02 (1.03, 3.09) 1.98 0.81 0.001 0–5 years old Referent Referent Referent 6–12 years old 0.96 (0.74, 1.24) 1.90 (1.31, 2.74) 1.98 0.75 0.001 13–18 years old 0.92 (0.64, 1.33) 1.82 (1.19, 2.77) 1.98 0.65 0.01 19–55 years old 0.96 (0.72, 1.29) 1.90 (1.32, 2.73) 1.98 0.78 0.001 55 + years old 1.43 (0.57, 3.59) 2.83 (1.05, 7.67) 1.98 0.44 0.04 Season (wet) Referent Referent Referent Season (dry) 0.23 (0.10, 0.51) 0.45 (0.20, 0.99) 1.96 < 0.001 0.047 Interpretation for ethnicity: the Batwa weekly malaria hospital admission incidence rate was 1.63 times the rate of admission for Bakiga during the lagged hot temperatures. The ratio of ratios for Batwa vs Bakiga in the hot quartile over the cool quartiles was 1.99 *Bold indicates a p-value of < 0.05 Ost  et al. Malaria Journal (2022) 21:98 Page 9 of 13 Table 7 Incidence rate ratio (IRR) stratification model results with sociodemographic variables as effect modifiers Models stratified by social factor; IRR (95% CI) Ethnicity Sex Age (years) Season Bakiga Batwa Male Female 0–5 6–12 13–18 19–55 55 + Season (wet) Season (dry) Temperature Referent Referent Referent Referent Referent Referent Referent Referent Referent Referent Referent quartile 1–3 Temperature 2.09 (1.49, 0.71 (0.10, 2.07 (1.45, 1.82 (1.25, 2.04 (1.36, 2.07 (1.40, 1.29 (0.84, 1.61 (1.08, 1.01 (0.48, 1.87 (1.34, 0.40 (0.17, quartile 4 2.94) 4.81) 2.96) 2.65) 3.06) 3.07) 1.98) 2.40) 2.12) 2.62) 0.96) p value < 0.001 0.72 < 0.001 0.002 0.001 < 0.001 0.25 0.02 0.99 < 0.001 0.04 IRR strata1/ IRR Referent 0.34 Referent 0.88 Referent 1.01 0.63 0.79 0.50 Referent 0.21 strata0 Interpretation for ethnicity: the Bakiga weekly malaria hospital admission incidence was 2.09 times greater during weeks in the hottest temperature quartile than in the coolest quartiles, compared to Batwa, who had 0.71 times greater incidence in weeks in the hottest quartile. The ratio of ratios (ROR) for Batwa vs. Bakiga in the hot season only was 0.34, indicating that the indicative ‘effect’ of the hottest quartile on malaria incidence was 0.34 times the rate in the Batwa than Bakiga, or that Bakiga incidence was more sensitive to temperature compared to Batwa incidence *Bold indicates a p-value of < 0.05, **Stratified model for season does not include season-temperature interaction term Ost et al. Malaria Journal (2022) 21:98 Page 10 of 13 Evidence of effect modification of the temperature -malaria Discussion association This study investigated whether social factors, such as Interaction model age, sex, and ethnicity, modified the association between The association of temperature with malaria differed by temperature and malaria hospital admission incidence. age, sex, and ethnicity (Table  6). Women experienced a Results indicated that the social variables examined in higher weekly incidence rate of malaria hospital admis- our models do modify this association, although this sions compared to men, with this difference substan - modification was not significant or lacked sufficient sta - tially higher during hotter lagged weeks (top quartile of tistical power to achieve statistical significance in all mean temperature) compared to cooler lagged weeks. cases. Although subject to wide confidence intervals, the During weeks prior to admission in the three combined findings point to the strongest associations between tem - cooler quartiles of temperature, the weekly incidence perature and malaria incidence among 6–12  year olds, rate of malaria hospital admission between men and the elderly, and indigenous Batwa. women were similar (IRR = 1.02 (0.86, 1.22)). During The results for ethnicity differed between the baseline the hottest weeks before admission, however, the weekly and interaction model results with regards to the magni- incidence rate of malaria hospital admissions among tude of the association, both of which showing that the women was significantly higher (IRR = 2.02 (1.03, 3.09)) Indigenous Batwa have a higher incidence of malaria than the rate among men. Increases in the weekly inci- than the Bakiga. Both the baseline and interaction results dence rate of malaria hospital admissions were higher differed in the direction of the association between the in the wet season than the dry season, and higher in stratified model which suggested that the indigenous the hottest lagged temperature quartile (IRR = 0.45 Batwa had a lower incidence of malaria in the hot, com- (0.20, 0.99)) than the cooler lagged quartiles of the dry pared to cold, temperature quartiles, whereas the Bakiga season (IRR = 0.23 (0.10, 0.51)). Compared to children had a greater incidence of malaria in the hot, compared 0–5  years old, 13–18  year olds had the lowest inci- to cold temperature, quartile. Furthermore, descrip- dence rate ratio, and the highest being among 55 + year tive statistics indicated that 18.2% of all Bakiga hospi- olds. The association between temperature and malaria tal admissions were for malaria, and 22.7% of all Batwa was higher among the Batwa than the Bakiga, and was admissions were for malaria. The baseline and interac - more than 50% greater for the Bakiga during the hottest tion model results were also consistent with community lagged temperature quartile (IRR = 1.63 (0.64, 4.16)). survey research conducted by Donnelly et  al. [2], who All of these estimates, however, had wide confidence found the burden of malaria to be substantially higher intervals, which could be due to the small Batwa sam- among the Batwa than Bakiga. These findings, therefore, ple size that limits statistical power to detect significant illustrate the importance of evaluating data for effect differences. modification to capture how weather impacts malaria differentially for Batwa and Bakiga. Results for age also differed between the baseline and Stratified model effect modification models. The baseline and stratified Results differed slightly between interaction and strati - results indicated that the highest incidence of malaria fied models (Table  7). The Bakiga had a higher associa - hospital admissions was among 6–12  year old children, tion between temperature (hot versus cold quartiles) and which is similar to the established literature on malaria, malaria incidence compared to Batwa (Batwa IRR = 0.71 who found that 0–5 year olds have the highest burden on (0.10, 4.81) versus Bakiga IRR = 2.09 (1.49, 2.94)). This a global scale [1]; however, in the study interaction model translates to a ratio of ratios of 0.34, indicating that the results, the team found that individuals over the age of increase in weekly malarial hospital incidence rates 55 had a higher incidence rate of malaria hospital admis- during the hottest quartile weeks was 0.34 times lower sions in the highest temperature quartile when compared than the rate of the Bakiga. Similar to the interaction with the referent category of 0–5 year olds. The variation models, the results indicated that 6–12  year old chil- between the established literature on highest risk malaria dren and males had a higher magnitude of association age groups and the baseline and stratification results of weekly incidence rate of malaria hospital admissions in this study could possibly be explained by findings with temperature compared to other age categories and from local community surveys, which found that while females, respectively. There was an overall increase in 0–5  year old children had higher rates of malaria, this weekly malaria hospital admission incidence rates dur- age group was more likely to sleep under an ITN at night ing the wet season during times of high (4th quartile) than any other age group, suggesting that while malaria temperatures (Fig. 3). Ost  et al. Malaria Journal (2022) 21:98 Page 11 of 13 rates are high overall for those < 5years, seasonal fluctua - Uganda has several national level calls for stronger cli- tions in infection may be moderated by ITN protection mate policy including: the Lake Victoria Basin Report [39]. Children 6–12 years old, in contrast, were less likely 2018, Uganda’s National Adaptation Program of Action to have ITN protection [39], and may thus experience (NAPA) 2007, The Uganda National Climate Change wider fluctuations in infection risk associated with tem - Policy 2015, and a National Policy for Disaster Prepar- perature, which is consistent with our interaction model edness and Management 2010 [9, 10, 17]. Most of these results. policies have broad goals that address national level con- Our results cannot be directly used to make con- cerns such as water, agriculture, economic, and prepar- clusions about climate change and malaria due to the edness adaptation. In Uganda’s more remote districts, short study period. Climate-health projections cannot such as Kanungu District, interaction results suggest that be inferred from weather and temperature associa- while the entire population is more susceptible to malaria tions. These study results indicated that the association compared to the national average, some, like the Batwa between temperature and malaria was stronger among and youth, experience higher rates of malaria hospital particular social-demographic strata in the Kanungu admissions during periods of high temperatures, and may District region. Notably, this interaction and/or strati- need additional planning and resource allocation, such as fication approach implied that sub-populations with assistance with the removal of mosquito breeding sites the highest incidence rates of malaria will not neces- around the home, or distribution of mosquito bed nets sarily be the same as those with the strongest associa- to achieve more equitable adaptation. Currently Uganda tions between meteorological variability and malaria policy prioritizes the distribution of mosquito nets to incidence. The insights from understanding the causal pregnant women or households with children under the reasons for these differences can point to how malaria age of 5 [39]. risks might shift differentially across sub-populations under climate change. For example, if climate change Conclusion acts to magnify and/or extend the number of hot The effect modification approach used herein can be used weeks, researchers could speculate that these changes to improve understanding of how changes in weather could increase malaria incidence more rapidly among resulting from climate change might shift social gra- age categories who are unprotected by ITNs at night. dients in health. These study findings suggest that local A traditional approach to projecting climate risk would level policy may be beneficial in addressing some of the assume that since children < 5  years have the high- more ‘micro’ level concerns that Ugandan Districts will est current rates of malaria according to the literature, face, such as differential risk of malaria infection among emphasis on that population is the highest priority. sub- populations. Local policy could expand to include This effect modification approach suggests that while the Batwa population, youth, and the elderly in their high protecting children < 5 years remains a priority, malaria priority prevention efforts, and prioritize follow-up and incidence among children 6–12  years and the elderly retention programming among Batwa. may be more sensitive to warming, meriting interven- tion to prevent increased incidence in those age groups. Supplementary Information Similarly, while research previously highlighted higher The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12936‑ 022‑ 04118‑5. incidence of malaria among the Indigenous Batwa com- pared to their Bakiga neighbours, these interaction Additional file1: Table S1. Descriptive statistics of variables from the results suggest that Batwa face the additional burden original (full dataset) including variables with missing demographic infor‑ of higher sensitivity to temperature when compared mation from Bwindi Community Hospital, Uganda (2011–2014). to the Bakiga. Research by Clark et al. [39] highlighted very low retention of ITNs following free distribution, Acknowledgements Thank you to Carlee Wright, the project manager of IHACC for creating the indicating that Batwa may lack ITN protection during map in this manuscript, and to the Thomas Francis Jr. Fellowship. Finally, thank peak infection conditions, and also be less likely to ben- you to Samuel Des Rochers‑ Jette for his coding support. efit from ITN-distribution interventions. These results Authors’ contributions point to the particularly high vulnerability of Batwa in a KO was responsible for the data analysis and manuscript writing with changing climate, with existing high burdens of malaria supervision from, LBF, SLH, ABK, and KE. KBW and MC guided and informed compounded by higher weather sensitivity and lower the methods of this paper. SL, DBN, YH, Bwindi Community Hospital, and the IHACC Research Team were responsible for primary data collection for this uptake of interventions compared to neighbouring project. All authors were involved in the design of this project. All authors read Bakiga. and approved the final manuscript. Ost et al. Malaria Journal (2022) 21:98 Page 12 of 13 Funding 7. Gracey M, King M. Indigenous health part 1: determinants and disease No funding was provided for this project. patterns. Lancet. 2009;374:65–75. 8. King M, Smith A, Gracey M. Indigenous health part 2: the underlying Availability of data and materials causes of the health gap. Lancet. 2009;374:76–85. Due to the nature of this research, participants of this study did not agree for 9. USAID. Lake victoria basin climate change adaptation strategy and action their data to be shared publicly, so supporting data is not available. plan 2018–2023. 2018. https:// www. clima telin ks. org/ resou rces/ lake vic‑to ria‑ basin‑ clima te‑ change‑ adapt ation‑ strat egy‑ and‑ action‑ plan‑ 2018‑ 2023. Accessed 15 Apr 2019. Declarations 10. Nyasimi M, Radeny M, Mungai C, Kamini C. Uganda’s National Adaptation Programme of Action: Implementation, Challenges and Emerging Les‑ Ethics approval and consent to participate sons. 2016. https:// ccafs. cgiar. org/ resou rces/ publi catio ns/ ugand as‑ natio This study approved by ethics boards at McGill University, the University of nal‑ adapt ation‑ progr amme‑ action‑ imple menta tion‑ chall enges . Guelph, the University of Alberta, and the University of Washington, as well as 11. The Republic of Uganda Ministry of Water and Environment. National Cli‑ Bwindi Community Hospital. All personal identifiers were removed from the mate Change Policy: Transformation through Climate Change Mitigation dataset before analysis. This research is conducted within the broader IHACC and Adaptation. Uganda: The Republic of Uganda Ministry of Water and project, in partnership with Makerere University, and in collaboration with Environment. 2015. p. 1–67. https:// www. mwe. go. ug/ sites/ defau lt/ files/ Batwa health and development programs in Kampala and in the region. Com‑ libra ry/ Natio nal% 20Cli mate% 20Cha nge% 20Pol icy% 20Apr il% 202015% munities have already consented to ongoing collaboration with the IHACC 20fin al. pdf . research project and team. At the time of data collection, the UNSCT was not 12. Adger WN, Arnell NW, Tompkins EL. Successful adaptation to climate accepting applications and was not granting ethics approvals. change across scales. Glob Environ Change. 2005;15:77–86. Not Applicable; secondary data analysis. 13. Beck‑ Johnson LM, Nelson WA, Paaijmans KP, Read AF, Thomas MB, Bjørn‑ stad ON. The importance of temperature fluctuations in understand‑ Consent for publication ing mosquito population dynamics and malaria risk. R Soc Open Sci. All authors have seen and approved of the final version of this manuscript and 2017;4:160969. consent to the publication of this article. 14. Okuneye K, Gumel AB. Analysis of a temperature—and rainfall‑ dependent model for malaria transmission dynamics. Math Biosci. Competing interests 2017;287:72–92. Authors have no competing interest to declare. 15. Caminade C, Kovats S, Rocklov J, Tompkins AM, Morse AP, Colón‑ González FJ, et al. Impact of climate change on global malaria distribution. Proc Author details Natl Acad Sci USA. 2014;111:3286–91. School of Epidemiology and Public Health, University of Ottawa, Ottawa, 16. Intergovernmental Panel on Climate Change. Africa. Climate Change Canada. Priestley International Centre for Climate, University of Leeds, Leeds, 2014 – Impacts, Adaptation and Vulnerability: Part B: Regional Aspects: UK. School of Interdisciplinary Science, McMaster University, Hamilton, Working Group II Contribution to the IPCC Fifth Assessment Report. Canada. Department of Geography, McGill University, Montreal, Canada. Cambridge: Cambridge University Press; 2014. p. 1199–266. 5 6 School of Public Health, University of Alberta, Edmonton, Canada. Depar t‑ 17. USAID. Climate Risk Profile: East Africa Regional. USAID; 2020. p. 14. ment of Geography, Geo‑Informatics and Climatic Sciences, School of Forestry, https:// www. clima telin ks. org/ resou rces/ clima te‑ risk‑ profi le‑ east‑ africa‑ Environmental and Geographical Sciences, College of Agricultural and Envi‑ regio nal. ronmental Sciences, Makerere University, Kampala, Uganda. I ndigenous 18. Labbé J, Ford JD, Berrang‑Ford L, Donnelly B, Lwasa S, Namanya DB, et al. Health Adaptation To Climate Change, Research Team, Edmonton, Canada. Vulnerability to the health effects of climate variability in rural southwest ‑ 8 9 Uganda Martyrs University, Kampala, Uganda. Faculty of Health Sciences, ern Uganda. Mitig Adapt Strateg Glob Change. 2016;21:931–53. Uganda Martyrs University, Kampala, Uganda. Department of Atmospheric 19. Berrang‑Ford L, Dingle K, Ford JD, Lee C, Lwasa S, Namanya DB, et al. and Ocean Sciences, McGill University, Montreal, Canada. Depar tment Vulnerability of indigenous health to climate change: a case study of of Health Services, University of Washington, Seattle, USA. Bwindi Commu‑ Uganda’s Batwa Pygmies. Soc Sci Med. 2012;75:1067–77. nity Hospital, Kanungu, Uganda. Center for Health and the Global Environ‑ 20. Kulkarni MA, Garrod G, Berrang‑Ford L, Ssewanyana I, Harper SL, Bara‑ ment, University of Washington, Seattle, USA. heberwa N, et al. Examination of antibody responses as a measure of exposure to malaria in the indigenous batwa and their non‑indigenous Received: 2 July 2021 Accepted: 5 March 2022 neighbors in Southwestern Uganda. Am J Trop Med Hyg. 2017;96:330–4. 21. MacVicar S, Berrang‑Ford L, Harper S, Steele V, Lwasa S, Bambaiha DN, et al. How seasonality and weather affect perinatal health: comparing the experiences of indigenous and non‑indigenous mothers in Kanungu District. Uganda Soc Sci Med. 2017;187:39–48. References 22. Sauer J, Berrang‑Ford L, Patterson K, Donnelly B, Lwasa S, Namanya D, 1. 1. World Health Organization. World malaria report 2018 [Internet]. et al. An analysis of the nutrition status of neighboring indigenous and Geneva: World Health Organization; 2018. https:// apps. who. int/ iris/ non‑indigenous populations in Kanungu District, southwestern Uganda: handle/ 10665/ 275867. close proximity, distant health realities. Soc Sci Med. 2018;217:55–64. 2. Donnelly B, Berrang‑Ford L, Labbé J, Twesigomwe S, Lwasa S, Namanya 23. Bwindi community hospital. Patient care—hospital services—Bwindi DB, et al. Plasmodium falciparum malaria parasitaemia among indigenous community hospital. http:// www. bwind ihosp ital. com/ index. php/ hospi Batwa and non‑indigenous communities of Kanungu district. Uganda tal‑ servi ces/ patie nt‑ care . Accessed 10 Apr 2019. Malar J. 2016;15:254. 24. Bishop‑ Williams K, Berrang‑Ford L, Sargeant J, Pearl D, Lwasa S, Namanya 3. Fischer H, Fraedrich K, Fuss S, Hoegh‑ Guldberg O, Jacob D, Taylor M, et al. D, et al. Understanding weather and hospital admissions patterns to Impacts of 1.5°C global warming on natural and human systems. In: inform climate change adaptation strategies in the healthcare sector in Global Warming of 1.5°C. Intergovernmental Panel on Climate Change. Uganda. Int J Environ Res Public Health. 2018;15:2402. 2018. https:// www. ipcc. ch/ sr15/. 25. Bwindi community hospital. Bwindi community hospital; eQuality insur‑ 4. Ford JD. Indigenous health and climate change. Am J Public Health. ance plan. https:// www. bwind ihosp ital. com/ equal ity. php. Accessed 4 Jan 2012;102:1260–6. 5. Anderson I, Robson B, Connolly M, Al‑ Yaman F, Bjertness E, King A, et al. 26. Kilama M, Smith DL, Hutchinson R, Kigozi R, Yeka A, Lavoy G, et al. Estimat‑ Indigenous and tribal peoples’ health ( The Lancet‑Lowitja institute global ing the annual entomological inoculation rate for Plasmodium falciparum collaboration): a population study. Lancet. 2016;388:131–57. transmitted by Anopheles gambiae s.l. using three sampling methods in 6. Ohenjo N, Willis R, Jackson D, Nettleton C, Good K, Mugarura B. Health of three sites in Uganda. Malar J. 2014;13:111. indigenous people in Africa. Lancet. 2006;367:1937–46. Ost  et al. Malaria Journal (2022) 21:98 Page 13 of 13 27. Breman J. The ears of the hippopotamus: manifestations, determi‑ nants, and estimates of the malaria burden. Am J Trop Med Hyg. 2001;64(1_suppl):1–11. 28. Yeka A, Gasasira A, Mpimbaza A, Achan J, Nankabirwa J, Nsobya S, et al. Malaria in Uganda: challenges to control on the long road to elimination: I. Epidemiology and current control efforts. Acta Trop. 2012;121:184–95. 29. Okello PE, Van Bortel W, Byaruhanga AM, Correwyn A, Roelants P, Talisuna A, et al. Variation in malaria transmission intensity in seven sites through‑ out Uganda. Am J Trop Med Hyg. 2006;75:219–25. 30. Githeko AK, Ndegwa W. Predicting malaria epidemics in the Kenyan Highlands using climate data: a tool for decision makers. Glob Change Hum Health. 2001;2:54–63. 31. Huang F, Zhou S, Zhang S, Wang H, Tang L. Temporal correlation analysis between malaria and meteorological factors in Motuo County. Tibet Malar J. 2011;10:54. 32. Sewe M, Rocklöv J, Williamson J, Hamel M, Nyaguara A, Odhiambo F, et al. The association of weather variability and under five malaria mortality in KEMRI/CDC HDSS in Western Kenya 2003 to 2008: a time series analysis. Int J Environ Res Public Health. 2015;12:1983–97. 33. Nanvyat N, Mulambalah CS, Barshep Y, Ajiji JA, Dakul DA, Tsingalia HM. Malaria transmission trends and its lagged association with climatic fac‑ tors in the highlands of Plateau State. Nigeria Trop Parasitol. 2018;8:18–23. 34. Okiring J, Routledge I, Epstein A, Namuganga JF, Kamya EV, Obeng‑ Amoako GO, et al. Associations between environmental covariates and temporal changes in malaria incidence in high transmission settings of Uganda: a distributed lag nonlinear analysis. BMC Public Health. 2021;21:1962. 35. McSweeney C, New M, Lizcano G, Lu X, McSweeney C, New M, et al. The UNDP climate change country profiles. Bull Am Meteorol Soc. 2010;91:157–66. 36. Charette M, Berrang‑Ford L, Coomes O, Llanos‑ Cuentas EA, Cárcamo C, Kulkarni M, et al. Dengue incidence and sociodemographic conditions in Pucallpa, Peruvian Amazon: what role for modification of the dengue ‑ temperature relationship? Am J Trop Med Hyg. 2020;102:180–90. 37. Knol MJ, VanderWeele TJ. Recommendations for presenting analyses of effect modification and interaction. Int J Epidemiol. 2012;41:514–20. 38. Donnelly B, Berrang‑Ford L, Ross NA, Michel P. A systematic, realist review of zooprophylaxis for malaria control. Malar J. 2015;14:313. 39. Clark S, Berrang‑Ford L, Lwasa S, Namanya D, Twesigomwe S, Kulkarni M, et al. A longitudinal analysis of mosquito net ownership and use in an indigenous Batwa population after a targeted distribution. PLoS ONE. 2016;11:e0154808. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations. Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Malaria Journal Springer Journals

Do socio-demographic factors modify the effect of weather on malaria in Kanungu District, Uganda?

Loading next page...
 
/lp/springer-journals/do-socio-demographic-factors-modify-the-effect-of-weather-on-malaria-3zJnlaaD5e

References (39)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2022
eISSN
1475-2875
DOI
10.1186/s12936-022-04118-5
Publisher site
See Article on Publisher Site

Abstract

Background: There is concern in the international community regarding the influence of climate change on weather variables and seasonality that, in part, determine the rates of malaria. This study examined the role of sociodemo‑ graphic variables in modifying the association between temperature and malaria in Kanungu District (Southwest Uganda). Methods: Hospital admissions data from Bwindi Community Hospital were combined with meteorological satellite data from 2011 to 2014. Descriptive statistics were used to describe the distribution of malaria admissions by age, sex, and ethnicity (i.e. Bakiga and Indigenous Batwa). To examine how sociodemographic variables modified the associa‑ tion between temperature and malaria admissions, this study used negative binomial regression stratified by age, sex, and ethnicity, and negative binomial regression models that examined interactions between temperature and age, sex, and ethnicity. Results: Malaria admission incidence was 1.99 times greater among Batwa than Bakiga in hot temperature quartiles compared to cooler temperature quartiles, and that 6–12 year old children had a higher magnitude of association of malaria admissions with temperature compared to the reference category of 0–5 years old (IRR = 2.07 (1.40, 3.07)). Discussion: Results indicate that socio‑ demographic variables may modify the association between temperature and malaria. In some cases, such as age, the weather‑malaria association in sub ‑populations with the highest inci‑ dence of malaria in standard models differed from those most sensitive to temperature as found in these stratified models. Conclusion: The effect modification approach used herein can be used to improve understanding of how changes in weather resulting from climate change might shift social gradients in health. Keywords: Malaria, Climate change, Weather, Meteorological, Sex, Age, Indigenous, Batwa, Bakiga, Sociodemographic modifiers, Uganda Background Malaria continues to pose a threat to human health worldwide. Approximately 92% of all malaria cases *Correspondence: ostk91@gmail.com in 2017 occurred in the World Health Organization School of Epidemiology and Public Health, University of Ottawa, Ottawa, (WHO) African Region. Five of these countries, pri- Canada marily in sub-Saharan Africa, accounted for half of the Full list of author information is available at the end of the article © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Ost et al. Malaria Journal (2022) 21:98 Page 2 of 13 malaria cases worldwide [1]. Uganda accounted for 4% Methods of all cases of malaria in 2017, and is among the coun- Kanungu district, Uganda tries facing significant challenges to Plasmodium fal - This study was conducted in Kanungu District, located ciparum malaria elimination due to the presence of in southwestern Uganda, near Bwindi Impenetrable highly competent mosquito vectors and lack of infra- National Park (Fig.  1). Over the last 50 years, this region structure systems in place to support elimination [1, of eastern Africa experienced an increase in seasonal 2]. Climate change threatens progress made towards mean temperature [16, 17]. Warming trends are likely to malaria elimination in many areas of the world includ- continue, with an increase in mean temperature of up to ing Uganda. The Intergovernmental Panel on Climate 2.0 °C projected by 2030, and an increase in regional dry- Change (IPCC) concluded with medium to high con- ing [16]. fidence that climate change could alter the geographic The region is primarily inhabited by Bakiga and indig - range of the Anopheles vector, creating the potential for enous Batwa people, both of whom face relatively high longer transmission seasons and increasing the number ill-health burdens when compared with the national of people at risk, and noted that this projection varies average. Both Bakiga and Batwa people are highly vul- regionally [3]. nerable to health impacts of a warming climate, and Among the most vulnerable to the effects of climate have identified malaria, food insecurity, and gastrointes - change are indigenous people who already face dispro- tinal illnesses as climate-sensitive health concerns [18, portionate burdens of health and social inequity [4]. 19]. The indigenous Batwa experience higher prevalence Sub-Saharan African indigenous people, in particular, of malaria compared to the Bakiga, 9.4% versus 4.5%, often have more health challenges than non-indige- respectively [2]. This difference in malaria prevalence is nous people living in the same geographic areas [5, 6]. paralleled by a range of health and socio-economic dis- This inequity in health outcomes is frequently rooted parities between the two populations (Table  1), includ- in colonization and the social determinants of health, ing reduced life expectancy [19]. The Batwa were forcibly including discrimination, loss of traditional lands, mar- removed from their ancestral lands with the creation of ginalization, and limited access to healthcare services the Bwindi Impenetrable Park in 1991 (Fig.  1), where [7, 8]. they relied on subsistence hunting and gathering; evic- In recent years, a number of countries have moved to tion from the park forced them into settlement in agrar- establish nation-wide policies regarding climate change ian communities outside the park boundaries (19). There adaptation activities, including in Uganda [9–11]. How- are currently approximately 6700 Batwa individuals living ever, nation-wide policies risk aggregation of diverse at- in southwestern Uganda [20], 900 of whom live within risk populations, masking important trends at a more Kanungu District. There are no notable ecological or granular level. Localized research in vulnerable popula- geographic differences in the areas where the Batwa or tions can underpin resource distribution and identifica - Bakiga live that would increase risk of malaria in either tion of focal vulnerabilities [12]. The districts in Uganda population (Fig. 1). vary greatly in their geography, demographic makeup, The Bwindi Community Hospital (BCH) was founded primary health concerns, and in the ways in which they in 2003 as a clinic to primarily serve Batwa [23]. Since experience climate change impacts. Factors such as soci- its founding, BCH has expanded into a large facility odemographic makeup of a district will be important in that includes six in-patient wards, including a pediatric, determining the most effective course of adaptive action adult, maternity, and immunodeficiency hospital wards, within specific communities. While many studies have as well as an out-patient ward, and several satellite clinics quantified the association between temperature and pre - for remote settlements [23]. BCH operates on a fee-for cipitation for malaria transmission [13–15], these studies service model, and donations help to subsidize an insur- typically do not consider how socio-demographic char- ance scheme for residents who qualify [18, 24]. All Batwa acteristics modify these associations. This study aims to residents are covered under this insurance plan (eQuality address this gap through evaluation of the role of soci- health insurance), which has enrolled 34% of its catch- odemographic factors—age, sex, and ethnicity—in modi- ment area in the plan as of 2020 [23, 25]. fying the association between temperature and malaria Kanungu District is a rural area of rolling hills located in Kanungu District Uganda from 2011 to 2014 using at an elevation of 1,310 m above sea level [26]. There are data from the Bwindi Community Hospital. This study four species of malaria parasite that affect humans in explored effect modification by examining how socio- Uganda, the most virulent being P. falciparum [27, 28]. demographic variables interact and by stratifying models, Plasmodium falciparum is the primary endemic malaria which has important implications for public health policy parasite found in the Kanungu region and is most often under a changing climate. carried by the Anopheles gambiae mosquito species [29]. Ost  et al. Malaria Journal (2022) 21:98 Page 3 of 13 Fig. 1 Map of Kanungu district and location of Bwindi community hospital Table 1 Socio‑ economic and health differences between Bakiga and Indigenous Batwa populations (adapted from MacVicar et al. 2017a [21]) Health and socio-economic measure Batwa (proportion of the population (%)) Bakiga (proportion of the population (%)) Malaria prevalence among adult 6.45 4.46 Moderate acute malnutrition among adult women 45.86 0.42 Household mosquito net use (did not have nets) 70.99 53.56 Access to handwashing facilities (did not have access to handwashing) 73.85 56.40 Access to soap (did not have access to soap) 73.85 56.40 Prevalence of positive malaria antigen detection test in both July 2013 and April 2014—survey of all Batwa adults, sample of Bakiga adults[2] Classified as moderately malnourished according to the Uganda ministry of health integrated management of acute malnutrition guidelines)[22] [2] d,e Only asked of people that had access to hand washing facility, for example for the Batwa, 32 or 94% of the households that had access to handwashing had access to soap[2] Data collection (RDT) or blood slide in conjunction with symptoms. Hospital data Individual inpatient records from the hospital were Electronic records of patients with a malaria diagnosis merged with insurance coverage data based on patient from 1-Jan-11 through 21-Dec-14 were obtained through ID to provide additional data on sex, age, and ethnicity partnership with BCH. Malaria diagnosis was defined [24]. Data were de-identified prior to analysis to ensure as any case with a positive rapid antigen diagnostic test the confidentiality of patients. In total, there were data Ost et al. Malaria Journal (2022) 21:98 Page 4 of 13 age, and ethnicity, and were retained for this study; cases with incomplete information were excluded. Approxi- mately 51% of data were missing information on ethnic- ity, which was one effect modifier of interest. Excluded cases had a similar demographic distribution to the final sample used in the analysis based on the initial testing (Additional file 1: Table S1). Meteorological data Meteorological data were estimated from the European Centre for Medium Range Weather Forecasts Re-analy- sis (ERA)-Interim Climate Database that combined data from multiple sources. The ERA-Interim climate data - bases have a spatial resolution of 0.75°  by 0.75°  . Daily values for total precipitation (i.e. rainfall (mm)) as well as maximum, minimum, and average temperature (°C) Fig. 2 Conceptual model illustrating social modification of were obtained for all dates matching the extracted medi- malaria‑ weather relationship: boxes and horizontal arrows represent main pathways of interest in climate–malaria relationship, while the cal records (i.e. 1 January, 2011 to 31 December, 2014) black box represents our sociodemographic variables of interest. Bold [24]. Meteorological data were merged with the BCH vertical arrows indicate theorized effect modification data based on date of admission; lags were then created to account for the assumed time between mosquito/para- site development, point of infection, and finally the day of for 39,287 admissions (all diagnoses) at BCH and 6602 admission. malaria admissions were reported during the years 2011– The research team focused on the extent to which non- 2014. Of these, 18,846 (48%) of the admissions and 3440 meteorological variables modified the effect of tempera - (52%) of malaria admissions, could be matched on sex, ture on malaria hospital admission incidence. As such, IRRs for Stratified Models, hot vs. cool quartiles 0.3 Demographic Variables Fig. 3 Logarithmic scale; incidence rate ratios by demographic category for stratified models, gridline marks IRR of 1, stratified models for season do not include interaction term for season and temperature Incidence Rate Ratio (IRR) with Robust Standard Error Log Scale Baseline Batwa Bakiga Female Male Age 0-5 Age 6-12 Age 13-18 Age 19-55 Age 55+ Season (Dry) Season (Wet) Ost  et al. Malaria Journal (2022) 21:98 Page 5 of 13 models did not aim to maximize precision in the specifi - with malaria hospital admission incidence (Fig. 2). Exam- cation of the temperature-malaria association, but rather ining effect modification can be achieved through two assess the extent to which this generalized association methods: interaction and stratification. As such, three is sensitive to effect modification. Lags were created for types of models were used: (1) a baseline model that did both ambient temperature and precipitation, both impor- not include interaction variables or stratify data; (2) mod- tant predictors of malaria risk [3], out to six months prior els with weather- and age, sex, and ethnicity interaction to admission date under the a priori assumption that variables; and (3) models stratified by age, sex, and eth - a biologically plausible time lag for malaria would not nicity. The team compared model results among these extend past four months, and would not be less than one types of models, comparing the magnitude and direc- month [29–34]. A combined 12 and 13  week lag—the tion of the associations between temperature and malaria time between admission date and temperature preceding admissions. Effect modification methods were informed that date by 77–91  days—in mean weekly temperature by [24, 36]. was identified as having the most significant and strong - est association with malaria hospital admission rates Baseline model and was thus chosen for further analysis (Table  2). The The research team first constructed a model with all team converted the variable (temperature) into a binary meteorological and socio-demographic variables to variable reflecting the highest quartile (versus the low - measure the extent to which temperature is associated est 3 quartiles combined) of mean weekly temperature with malaria hospital admissions without accounting for 12 and 13  weeks prior to admission in order to evaluate any effect modification. We retained sociodemographic ‘hot’ weeks versus milder or ‘cooler’ weeks. Precipitation variables age, sex, and ethnicity in the model as fixed was not found to be significant in bivariate analysis, so a effects along with year and the meteorological variable binary variable for season was created based on date of season. admission; rainy seasons were defined as March-June and September–November, and dry seasons were defined Interaction model as December-February and July–August [35]. Season The team used interaction variables to estimate the mar - was retained in all models to account for the dependent ginal effects of temperature and age, sex, and ethnicity. nature of temperature and precipitation in the mosquito- This interaction approach took into account interaction weather relationship [13]. Because of the importance of and confounding between variables in the model. While precipitation in this relationship the research team per- theoretically more analytically robust, the coefficients of formed an additional sensitivity analysis using rainfall as interaction variables are sensitive to sample size and can the predictive variable in the models, and assessed the be less intuitive to interpret. effect on the size, direction, and significance of the other model coefficients. Stratified models Stratification of models involved fitting separate models Data analysis for each level of strata. Using stratified models provided Conceptual approach a more interpretable estimates of the effects of weather Effect modification occurs when a variable differentially on malaria in different age, sex, and ethnicity strata. This modifies the observed effect of a risk factor on disease approach is less sensitive to small sample sizes, which status. This study examined the ways in which age, sex, was particularly pertinent given the sample size for Batwa and ethnicity changed the association of temperature in the dataset was small. The team compared incidence rate ratios (IRRs) between demographic variables using a ratio of ratios approach [37]. Table 2 Incidence Rate Ratio (IRR) for malaria hospital admission incidence and temperature, by weekly lag Model variables Lag by week IRR p-value 95% CI A summary of variables used in the models is pro- vided in Table  3. The count of weekly malaria hospi - 10 1.02 0.63 (0.94, 1.11) tal admissions was the dependent (outcome) variable 11 0.99 0.82 (0.91, 1.08) of interest. Temperature was the independent (expo- 12 1.09 0.04 (1.00, 1.19) sure) variable of interest: specifically, the highest quar - 13 1.12 0.01 (1.03, 1.22) tile (versus the lowest 3 quartiles combined) of mean 14 1.06 0.15 (0.98, 1.16) weekly temperature 12 and 13  weeks prior to admission 15 1.07 0.09 (0.98, 1.18) was examined. Age, sex, and ethnicity were examined as a,b Bold text indicates time lags between admission date and temperature potential effect modifiers. Age was categorized into the preceding that date selected for use in final models Ost et al. Malaria Journal (2022) 21:98 Page 6 of 13 Stratified models following categories: < 5  years, 6–12  years, 13–18  years, The team ran the baseline models stratified by age, sex, 19–55  years, and > 55  years of age. Additionally, the ethnicity. Stratified models also contained all control research team conducted a sensitivity analysis on age to and interaction variables as fixed effects to minimize the verify that results were not sensitive to different age cut- effect of confounders in the analysis. offs. Ethnicity was divided into the two main groups: the Indigenous Batwa, and all other ethnic groups, primarily Ethics consisting of ethnic Bakiga. Year and season were identi- This study approved by ethics boards at McGill Univer - fied a priori as confounding variables. sity, the University of Guelph, the University of Alberta, and the University of Washington, as well as Bwindi Negative binomial multivariable regression models Community Hospital. All personal identifiers were Researchers used a negative binomial multivariable removed from the dataset before analysis. This research regression model with the count of total weekly malaria is conducted within the broader IHACC project, in cases as the dependent (outcome) variable. For the pop- partnership with Makerere University, and in collabora- ulation at risk (offset) variable, we used the weekly total tion with Batwa health and development programmes in admissions to BCH for any diagnosis. Year was controlled Kampala and in the region. Communities have already for in all models as a fixed effect. All analyses were con - consented to ongoing collaboration with the IHACC ducted in STATA v.15.1 (Stata Corp., USA). research project and team. At the time of data collection, the UNSCT was not accepting applications and was not Baseline model granting ethics approvals. We first built an unstratified model with no interactions that included sociodemographic variables as fixed effects Limitations and meteorological exposures: Data represent a short period of time (less than 4  years) and were, therefore, insufficient to infer rela- ln weekly malaria counts = β + β mean weekly T L12 − 13 0 1 tionships between malaria and climate change for long + β (sex) + β age + β ethinicity 2 3 4 timeframes. Additionally, though wet and dry season + β (season) + β (Year) 5 6 were included in the models to represent the complex + ln (population at risk) relationship between meteorological variables like temperature and precipitation, and malaria, there are limitations to excluding precipitation/rainfall as a vari- Interaction model able. Future work should examine these meteorologi- The team examined interaction between socio-demo - cal variables in this context in more detail. The Batwa graphic variables (i.e. age, sex, ethnicity, and season) and sample in the dataset was small, reducing statistical temperature, and their association with malaria hospital power in our analysis. Despite this, the team chose to admissions. To illustrate the size, direction, and confi - examine ethnicity, given its indicative role as a poten- dence interval of interactions, researchers evaluated lin- tially important driver of vulnerability to malaria in ear combinations of estimates for weather with season, the region and given the Batwa rank malaria as a top sex, age, and ethnicity. The team included all interaction climate-sensitive health outcome [2, 19, 38]. In doing variables in a single model. The final model equation so, the research team sought to avoid exclusion of used for analyses was: small, but marginalized and high-risk indigenous peo- ple from climate-health analytic research [19]. Under- ln weekly malaria counts = β + β mean weekly T L12 − 13 0 1 standing malaria impacts in view of weather and + β (sex) + β age + β ethinicity 2 3 4 climate in small vulnerable groups is vital in informing + β season + β Year 5( ) 6( ) health policy. + β mean weekly T L12 − 13 ∗ season + β mean weekly T L12 − 13 ∗ sex 8 Results Descriptive statistics + β mean weekly T L12 − 13 ∗ age ◦ Of the hospital admissions data with complete sociode- + β mean weekly T L12 − 13 ∗ ethinicity mographic records over the study period, the research + β mean weekly T L12 − 13 ∗ Year + ln (population at risk) Ost  et al. Malaria Journal (2022) 21:98 Page 7 of 13 Table 3 Description of dependent, independent, effect modification, and control variables used in effect modification analysis Variable (units) Description Dependent (outcome) variable Weekly malaria cases Total case count per 7 day period Independent (exposure) variable Mean weekly temperature ( ̊C) with a 12–13 weekly average tem‑ Binary: lower (cooler) quartiles 1–3 (referent); top (hottest) quartile perature lag Eec ff t modification variables Ethnicity Binary: Bakiga and other (referent); Batwa Sex Binary: male (referent); female Age Categorical: < 5 years (referent); 6‑12 years; 13‑18 years; 19‑55 years; > 55 years Confounding (control) variables Season Binary: dry (referent); wet Modelled with an interaction term with the independent variable Year Categorical: 2011 (referent), 2012, 2013, 2014 team found that 56.7 percent of malaria cases in the distributed, with 53.1 percent of malaria cases being sample occurred during the wet season from March– female, comprising 56.1 percent of the total admissions June or September–November. Sex was relatively evenly data. A majority (36.9%) of the total hospital admissions Table 4 Descriptive statistics of variables included in final models data from Bwindi Community Hospital, Uganda (2011–2014). Demographics No. of (x) demographic Proportion of (x) No. of (x) demographic out Proportion of (x) out of all admissions demographic out of all of all malaria admissions demographic out of all admissions malaria admissions Female 10,565 56.10% 1826 53.10% Male 8281 43.90% 1614 46.90% 0–5 years 5687 30.20% 1143 33.20% 6–12 years 2514 13.30% 896 26.00% 13–18 years 1950 10.30% 414 12.00% 19–55 years 6957 36.90% 895 26.00% 55 + years 1738 9.20% 92 2.70% Ethnicity (Bakiga) 18,608 98.70% 3,386 98.40% Ethnicity (Batwa) 238 1.30% 54 1.60% Season (wet) 10,957 58.14% 1948 56.63% Season (dry) 7889 41.86% 1492 43.37% Total number* 18,846 3,440 Meteorological variables Descriptive temperature statis‑ Mean Min Max tics throughout (x) year (celsius) 2011 18.91 13.13 27.51 2012 19.07 12.22 28.67 2013 19.54 12.96 28.98 2014 19.55 13.32 29.19 Average daily and yearly total Average daily Yearly total rainfall (mm) 2011 3.55 1296 2012 3.55 1300 2013 3.22 1174 2014 3.07 1197 Ost et al. Malaria Journal (2022) 21:98 Page 8 of 13 Table 5 Results of baseline model measuring the effect of sample) were recorded as being Batwa; 22.7% of Batwa temperature on malaria with sociodemographic variables as hospital admissions were malaria cases during the study fixed‑ effects period, and 18.2% percent of the Bakiga hospital admis- sions were malaria cases (Table 4). Model variables Baseline model IRR (95% CI) p-value Temperature ‘Cool’ quartiles Referent Referent Baseline model Temperature ‘Hot’ quartile 1.27 (0.90, 1.80) 0.18 The weekly incidence rate of malaria hospital admissions Bakiga Referent Referent was 1.27 (0.90, 1.80) times higher during weeks with hot Batwa 1.08 (0.76, 1.55) 0.66 weather (highest quartile 29.30–29.42  °C) compared to Male Referent Referent weeks with cooler weather (three lower quartiles 29.08– Female 0.91 (0.84, 0.98) 0.02 29.30  °C) (Table  5). The weekly malaria hospital admis - 0–5 years old Referent Referent sions incidence rate among the indigenous Batwa was 6–12 years old 1.64 (1.48, 1.82) < 0.001 1.08 (0.76, 1.55) times higher than for Bakiga. The weekly 13–18 years old 1.08 (0.94, 1.24) 0.28 incidence rate of malaria hospital admissions for females 19–55 years old 0.65 (0.59, 0.71) < 0.001 was 0.91 (0.84, 0.98) times the rate of males. Compared 55 + years old 0.28 (0.20, 0.39) < 0.001 to children 0–5 years old, youth 6–12 years had the high- Season (wet) Referent Referent est weekly incidence rate of malaria hospital admissions, Season (dry) 1.30 (1.01, 1.68) 0.04 followed by youths aged 13–18  years. Weekly malaria *Bold indicates a p-value of < 0.05 hospital admission incidence rates were 1.30 (1.01, 1.68) times higher during the dry season compared to the wet season. Furthermore, the sensitivity analysis using rain- fall as the predictive variable in the models suggested were 19–55 years old, and 35.6 percent of 6–12 year old very little difference in the models, so it was removed children who were admitted to the hospital had a malaria from the analysis. infection. Only 238 hospital admissions (1.2% of the Table 6 Incidence rate ratio (IRR) interaction model results, including sociodemographic variables as effect modifiers Interaction model, results by temperature quartile Quartile 1–3 (cool) Quartile 4 (hot) IRR hot/IRR cool within strata IRR (95% CI) IRR (95% CI) of ethnicity, sex, age, and p-value p-value season Ratio of Ratios (ROR) Bakiga Referent Referent Referent Batwa 0.82 (0.34, 1.99) 1.63 (0.64, 4.16) 1.99 0.67 0.31 Male Referent Referent Referent Female 1.02 (0.86, 1.22) 2.02 (1.03, 3.09) 1.98 0.81 0.001 0–5 years old Referent Referent Referent 6–12 years old 0.96 (0.74, 1.24) 1.90 (1.31, 2.74) 1.98 0.75 0.001 13–18 years old 0.92 (0.64, 1.33) 1.82 (1.19, 2.77) 1.98 0.65 0.01 19–55 years old 0.96 (0.72, 1.29) 1.90 (1.32, 2.73) 1.98 0.78 0.001 55 + years old 1.43 (0.57, 3.59) 2.83 (1.05, 7.67) 1.98 0.44 0.04 Season (wet) Referent Referent Referent Season (dry) 0.23 (0.10, 0.51) 0.45 (0.20, 0.99) 1.96 < 0.001 0.047 Interpretation for ethnicity: the Batwa weekly malaria hospital admission incidence rate was 1.63 times the rate of admission for Bakiga during the lagged hot temperatures. The ratio of ratios for Batwa vs Bakiga in the hot quartile over the cool quartiles was 1.99 *Bold indicates a p-value of < 0.05 Ost  et al. Malaria Journal (2022) 21:98 Page 9 of 13 Table 7 Incidence rate ratio (IRR) stratification model results with sociodemographic variables as effect modifiers Models stratified by social factor; IRR (95% CI) Ethnicity Sex Age (years) Season Bakiga Batwa Male Female 0–5 6–12 13–18 19–55 55 + Season (wet) Season (dry) Temperature Referent Referent Referent Referent Referent Referent Referent Referent Referent Referent Referent quartile 1–3 Temperature 2.09 (1.49, 0.71 (0.10, 2.07 (1.45, 1.82 (1.25, 2.04 (1.36, 2.07 (1.40, 1.29 (0.84, 1.61 (1.08, 1.01 (0.48, 1.87 (1.34, 0.40 (0.17, quartile 4 2.94) 4.81) 2.96) 2.65) 3.06) 3.07) 1.98) 2.40) 2.12) 2.62) 0.96) p value < 0.001 0.72 < 0.001 0.002 0.001 < 0.001 0.25 0.02 0.99 < 0.001 0.04 IRR strata1/ IRR Referent 0.34 Referent 0.88 Referent 1.01 0.63 0.79 0.50 Referent 0.21 strata0 Interpretation for ethnicity: the Bakiga weekly malaria hospital admission incidence was 2.09 times greater during weeks in the hottest temperature quartile than in the coolest quartiles, compared to Batwa, who had 0.71 times greater incidence in weeks in the hottest quartile. The ratio of ratios (ROR) for Batwa vs. Bakiga in the hot season only was 0.34, indicating that the indicative ‘effect’ of the hottest quartile on malaria incidence was 0.34 times the rate in the Batwa than Bakiga, or that Bakiga incidence was more sensitive to temperature compared to Batwa incidence *Bold indicates a p-value of < 0.05, **Stratified model for season does not include season-temperature interaction term Ost et al. Malaria Journal (2022) 21:98 Page 10 of 13 Evidence of effect modification of the temperature -malaria Discussion association This study investigated whether social factors, such as Interaction model age, sex, and ethnicity, modified the association between The association of temperature with malaria differed by temperature and malaria hospital admission incidence. age, sex, and ethnicity (Table  6). Women experienced a Results indicated that the social variables examined in higher weekly incidence rate of malaria hospital admis- our models do modify this association, although this sions compared to men, with this difference substan - modification was not significant or lacked sufficient sta - tially higher during hotter lagged weeks (top quartile of tistical power to achieve statistical significance in all mean temperature) compared to cooler lagged weeks. cases. Although subject to wide confidence intervals, the During weeks prior to admission in the three combined findings point to the strongest associations between tem - cooler quartiles of temperature, the weekly incidence perature and malaria incidence among 6–12  year olds, rate of malaria hospital admission between men and the elderly, and indigenous Batwa. women were similar (IRR = 1.02 (0.86, 1.22)). During The results for ethnicity differed between the baseline the hottest weeks before admission, however, the weekly and interaction model results with regards to the magni- incidence rate of malaria hospital admissions among tude of the association, both of which showing that the women was significantly higher (IRR = 2.02 (1.03, 3.09)) Indigenous Batwa have a higher incidence of malaria than the rate among men. Increases in the weekly inci- than the Bakiga. Both the baseline and interaction results dence rate of malaria hospital admissions were higher differed in the direction of the association between the in the wet season than the dry season, and higher in stratified model which suggested that the indigenous the hottest lagged temperature quartile (IRR = 0.45 Batwa had a lower incidence of malaria in the hot, com- (0.20, 0.99)) than the cooler lagged quartiles of the dry pared to cold, temperature quartiles, whereas the Bakiga season (IRR = 0.23 (0.10, 0.51)). Compared to children had a greater incidence of malaria in the hot, compared 0–5  years old, 13–18  year olds had the lowest inci- to cold temperature, quartile. Furthermore, descrip- dence rate ratio, and the highest being among 55 + year tive statistics indicated that 18.2% of all Bakiga hospi- olds. The association between temperature and malaria tal admissions were for malaria, and 22.7% of all Batwa was higher among the Batwa than the Bakiga, and was admissions were for malaria. The baseline and interac - more than 50% greater for the Bakiga during the hottest tion model results were also consistent with community lagged temperature quartile (IRR = 1.63 (0.64, 4.16)). survey research conducted by Donnelly et  al. [2], who All of these estimates, however, had wide confidence found the burden of malaria to be substantially higher intervals, which could be due to the small Batwa sam- among the Batwa than Bakiga. These findings, therefore, ple size that limits statistical power to detect significant illustrate the importance of evaluating data for effect differences. modification to capture how weather impacts malaria differentially for Batwa and Bakiga. Results for age also differed between the baseline and Stratified model effect modification models. The baseline and stratified Results differed slightly between interaction and strati - results indicated that the highest incidence of malaria fied models (Table  7). The Bakiga had a higher associa - hospital admissions was among 6–12  year old children, tion between temperature (hot versus cold quartiles) and which is similar to the established literature on malaria, malaria incidence compared to Batwa (Batwa IRR = 0.71 who found that 0–5 year olds have the highest burden on (0.10, 4.81) versus Bakiga IRR = 2.09 (1.49, 2.94)). This a global scale [1]; however, in the study interaction model translates to a ratio of ratios of 0.34, indicating that the results, the team found that individuals over the age of increase in weekly malarial hospital incidence rates 55 had a higher incidence rate of malaria hospital admis- during the hottest quartile weeks was 0.34 times lower sions in the highest temperature quartile when compared than the rate of the Bakiga. Similar to the interaction with the referent category of 0–5 year olds. The variation models, the results indicated that 6–12  year old chil- between the established literature on highest risk malaria dren and males had a higher magnitude of association age groups and the baseline and stratification results of weekly incidence rate of malaria hospital admissions in this study could possibly be explained by findings with temperature compared to other age categories and from local community surveys, which found that while females, respectively. There was an overall increase in 0–5  year old children had higher rates of malaria, this weekly malaria hospital admission incidence rates dur- age group was more likely to sleep under an ITN at night ing the wet season during times of high (4th quartile) than any other age group, suggesting that while malaria temperatures (Fig. 3). Ost  et al. Malaria Journal (2022) 21:98 Page 11 of 13 rates are high overall for those < 5years, seasonal fluctua - Uganda has several national level calls for stronger cli- tions in infection may be moderated by ITN protection mate policy including: the Lake Victoria Basin Report [39]. Children 6–12 years old, in contrast, were less likely 2018, Uganda’s National Adaptation Program of Action to have ITN protection [39], and may thus experience (NAPA) 2007, The Uganda National Climate Change wider fluctuations in infection risk associated with tem - Policy 2015, and a National Policy for Disaster Prepar- perature, which is consistent with our interaction model edness and Management 2010 [9, 10, 17]. Most of these results. policies have broad goals that address national level con- Our results cannot be directly used to make con- cerns such as water, agriculture, economic, and prepar- clusions about climate change and malaria due to the edness adaptation. In Uganda’s more remote districts, short study period. Climate-health projections cannot such as Kanungu District, interaction results suggest that be inferred from weather and temperature associa- while the entire population is more susceptible to malaria tions. These study results indicated that the association compared to the national average, some, like the Batwa between temperature and malaria was stronger among and youth, experience higher rates of malaria hospital particular social-demographic strata in the Kanungu admissions during periods of high temperatures, and may District region. Notably, this interaction and/or strati- need additional planning and resource allocation, such as fication approach implied that sub-populations with assistance with the removal of mosquito breeding sites the highest incidence rates of malaria will not neces- around the home, or distribution of mosquito bed nets sarily be the same as those with the strongest associa- to achieve more equitable adaptation. Currently Uganda tions between meteorological variability and malaria policy prioritizes the distribution of mosquito nets to incidence. The insights from understanding the causal pregnant women or households with children under the reasons for these differences can point to how malaria age of 5 [39]. risks might shift differentially across sub-populations under climate change. For example, if climate change Conclusion acts to magnify and/or extend the number of hot The effect modification approach used herein can be used weeks, researchers could speculate that these changes to improve understanding of how changes in weather could increase malaria incidence more rapidly among resulting from climate change might shift social gra- age categories who are unprotected by ITNs at night. dients in health. These study findings suggest that local A traditional approach to projecting climate risk would level policy may be beneficial in addressing some of the assume that since children < 5  years have the high- more ‘micro’ level concerns that Ugandan Districts will est current rates of malaria according to the literature, face, such as differential risk of malaria infection among emphasis on that population is the highest priority. sub- populations. Local policy could expand to include This effect modification approach suggests that while the Batwa population, youth, and the elderly in their high protecting children < 5 years remains a priority, malaria priority prevention efforts, and prioritize follow-up and incidence among children 6–12  years and the elderly retention programming among Batwa. may be more sensitive to warming, meriting interven- tion to prevent increased incidence in those age groups. Supplementary Information Similarly, while research previously highlighted higher The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12936‑ 022‑ 04118‑5. incidence of malaria among the Indigenous Batwa com- pared to their Bakiga neighbours, these interaction Additional file1: Table S1. Descriptive statistics of variables from the results suggest that Batwa face the additional burden original (full dataset) including variables with missing demographic infor‑ of higher sensitivity to temperature when compared mation from Bwindi Community Hospital, Uganda (2011–2014). to the Bakiga. Research by Clark et al. [39] highlighted very low retention of ITNs following free distribution, Acknowledgements Thank you to Carlee Wright, the project manager of IHACC for creating the indicating that Batwa may lack ITN protection during map in this manuscript, and to the Thomas Francis Jr. Fellowship. Finally, thank peak infection conditions, and also be less likely to ben- you to Samuel Des Rochers‑ Jette for his coding support. efit from ITN-distribution interventions. These results Authors’ contributions point to the particularly high vulnerability of Batwa in a KO was responsible for the data analysis and manuscript writing with changing climate, with existing high burdens of malaria supervision from, LBF, SLH, ABK, and KE. KBW and MC guided and informed compounded by higher weather sensitivity and lower the methods of this paper. SL, DBN, YH, Bwindi Community Hospital, and the IHACC Research Team were responsible for primary data collection for this uptake of interventions compared to neighbouring project. All authors were involved in the design of this project. All authors read Bakiga. and approved the final manuscript. Ost et al. Malaria Journal (2022) 21:98 Page 12 of 13 Funding 7. Gracey M, King M. Indigenous health part 1: determinants and disease No funding was provided for this project. patterns. Lancet. 2009;374:65–75. 8. King M, Smith A, Gracey M. Indigenous health part 2: the underlying Availability of data and materials causes of the health gap. Lancet. 2009;374:76–85. Due to the nature of this research, participants of this study did not agree for 9. USAID. Lake victoria basin climate change adaptation strategy and action their data to be shared publicly, so supporting data is not available. plan 2018–2023. 2018. https:// www. clima telin ks. org/ resou rces/ lake vic‑to ria‑ basin‑ clima te‑ change‑ adapt ation‑ strat egy‑ and‑ action‑ plan‑ 2018‑ 2023. Accessed 15 Apr 2019. Declarations 10. Nyasimi M, Radeny M, Mungai C, Kamini C. Uganda’s National Adaptation Programme of Action: Implementation, Challenges and Emerging Les‑ Ethics approval and consent to participate sons. 2016. https:// ccafs. cgiar. org/ resou rces/ publi catio ns/ ugand as‑ natio This study approved by ethics boards at McGill University, the University of nal‑ adapt ation‑ progr amme‑ action‑ imple menta tion‑ chall enges . Guelph, the University of Alberta, and the University of Washington, as well as 11. The Republic of Uganda Ministry of Water and Environment. National Cli‑ Bwindi Community Hospital. All personal identifiers were removed from the mate Change Policy: Transformation through Climate Change Mitigation dataset before analysis. This research is conducted within the broader IHACC and Adaptation. Uganda: The Republic of Uganda Ministry of Water and project, in partnership with Makerere University, and in collaboration with Environment. 2015. p. 1–67. https:// www. mwe. go. ug/ sites/ defau lt/ files/ Batwa health and development programs in Kampala and in the region. Com‑ libra ry/ Natio nal% 20Cli mate% 20Cha nge% 20Pol icy% 20Apr il% 202015% munities have already consented to ongoing collaboration with the IHACC 20fin al. pdf . research project and team. At the time of data collection, the UNSCT was not 12. Adger WN, Arnell NW, Tompkins EL. Successful adaptation to climate accepting applications and was not granting ethics approvals. change across scales. Glob Environ Change. 2005;15:77–86. Not Applicable; secondary data analysis. 13. Beck‑ Johnson LM, Nelson WA, Paaijmans KP, Read AF, Thomas MB, Bjørn‑ stad ON. The importance of temperature fluctuations in understand‑ Consent for publication ing mosquito population dynamics and malaria risk. R Soc Open Sci. All authors have seen and approved of the final version of this manuscript and 2017;4:160969. consent to the publication of this article. 14. Okuneye K, Gumel AB. Analysis of a temperature—and rainfall‑ dependent model for malaria transmission dynamics. Math Biosci. Competing interests 2017;287:72–92. Authors have no competing interest to declare. 15. Caminade C, Kovats S, Rocklov J, Tompkins AM, Morse AP, Colón‑ González FJ, et al. Impact of climate change on global malaria distribution. Proc Author details Natl Acad Sci USA. 2014;111:3286–91. School of Epidemiology and Public Health, University of Ottawa, Ottawa, 16. Intergovernmental Panel on Climate Change. Africa. Climate Change Canada. Priestley International Centre for Climate, University of Leeds, Leeds, 2014 – Impacts, Adaptation and Vulnerability: Part B: Regional Aspects: UK. School of Interdisciplinary Science, McMaster University, Hamilton, Working Group II Contribution to the IPCC Fifth Assessment Report. Canada. Department of Geography, McGill University, Montreal, Canada. Cambridge: Cambridge University Press; 2014. p. 1199–266. 5 6 School of Public Health, University of Alberta, Edmonton, Canada. Depar t‑ 17. USAID. Climate Risk Profile: East Africa Regional. USAID; 2020. p. 14. ment of Geography, Geo‑Informatics and Climatic Sciences, School of Forestry, https:// www. clima telin ks. org/ resou rces/ clima te‑ risk‑ profi le‑ east‑ africa‑ Environmental and Geographical Sciences, College of Agricultural and Envi‑ regio nal. ronmental Sciences, Makerere University, Kampala, Uganda. I ndigenous 18. Labbé J, Ford JD, Berrang‑Ford L, Donnelly B, Lwasa S, Namanya DB, et al. Health Adaptation To Climate Change, Research Team, Edmonton, Canada. Vulnerability to the health effects of climate variability in rural southwest ‑ 8 9 Uganda Martyrs University, Kampala, Uganda. Faculty of Health Sciences, ern Uganda. Mitig Adapt Strateg Glob Change. 2016;21:931–53. Uganda Martyrs University, Kampala, Uganda. Department of Atmospheric 19. Berrang‑Ford L, Dingle K, Ford JD, Lee C, Lwasa S, Namanya DB, et al. and Ocean Sciences, McGill University, Montreal, Canada. Depar tment Vulnerability of indigenous health to climate change: a case study of of Health Services, University of Washington, Seattle, USA. Bwindi Commu‑ Uganda’s Batwa Pygmies. Soc Sci Med. 2012;75:1067–77. nity Hospital, Kanungu, Uganda. Center for Health and the Global Environ‑ 20. Kulkarni MA, Garrod G, Berrang‑Ford L, Ssewanyana I, Harper SL, Bara‑ ment, University of Washington, Seattle, USA. heberwa N, et al. Examination of antibody responses as a measure of exposure to malaria in the indigenous batwa and their non‑indigenous Received: 2 July 2021 Accepted: 5 March 2022 neighbors in Southwestern Uganda. Am J Trop Med Hyg. 2017;96:330–4. 21. MacVicar S, Berrang‑Ford L, Harper S, Steele V, Lwasa S, Bambaiha DN, et al. How seasonality and weather affect perinatal health: comparing the experiences of indigenous and non‑indigenous mothers in Kanungu District. Uganda Soc Sci Med. 2017;187:39–48. References 22. Sauer J, Berrang‑Ford L, Patterson K, Donnelly B, Lwasa S, Namanya D, 1. 1. World Health Organization. World malaria report 2018 [Internet]. et al. An analysis of the nutrition status of neighboring indigenous and Geneva: World Health Organization; 2018. https:// apps. who. int/ iris/ non‑indigenous populations in Kanungu District, southwestern Uganda: handle/ 10665/ 275867. close proximity, distant health realities. Soc Sci Med. 2018;217:55–64. 2. Donnelly B, Berrang‑Ford L, Labbé J, Twesigomwe S, Lwasa S, Namanya 23. Bwindi community hospital. Patient care—hospital services—Bwindi DB, et al. Plasmodium falciparum malaria parasitaemia among indigenous community hospital. http:// www. bwind ihosp ital. com/ index. php/ hospi Batwa and non‑indigenous communities of Kanungu district. Uganda tal‑ servi ces/ patie nt‑ care . Accessed 10 Apr 2019. Malar J. 2016;15:254. 24. Bishop‑ Williams K, Berrang‑Ford L, Sargeant J, Pearl D, Lwasa S, Namanya 3. Fischer H, Fraedrich K, Fuss S, Hoegh‑ Guldberg O, Jacob D, Taylor M, et al. D, et al. Understanding weather and hospital admissions patterns to Impacts of 1.5°C global warming on natural and human systems. In: inform climate change adaptation strategies in the healthcare sector in Global Warming of 1.5°C. Intergovernmental Panel on Climate Change. Uganda. Int J Environ Res Public Health. 2018;15:2402. 2018. https:// www. ipcc. ch/ sr15/. 25. Bwindi community hospital. Bwindi community hospital; eQuality insur‑ 4. Ford JD. Indigenous health and climate change. Am J Public Health. ance plan. https:// www. bwind ihosp ital. com/ equal ity. php. Accessed 4 Jan 2012;102:1260–6. 5. Anderson I, Robson B, Connolly M, Al‑ Yaman F, Bjertness E, King A, et al. 26. Kilama M, Smith DL, Hutchinson R, Kigozi R, Yeka A, Lavoy G, et al. Estimat‑ Indigenous and tribal peoples’ health ( The Lancet‑Lowitja institute global ing the annual entomological inoculation rate for Plasmodium falciparum collaboration): a population study. Lancet. 2016;388:131–57. transmitted by Anopheles gambiae s.l. using three sampling methods in 6. Ohenjo N, Willis R, Jackson D, Nettleton C, Good K, Mugarura B. Health of three sites in Uganda. Malar J. 2014;13:111. indigenous people in Africa. Lancet. 2006;367:1937–46. Ost  et al. Malaria Journal (2022) 21:98 Page 13 of 13 27. Breman J. The ears of the hippopotamus: manifestations, determi‑ nants, and estimates of the malaria burden. Am J Trop Med Hyg. 2001;64(1_suppl):1–11. 28. Yeka A, Gasasira A, Mpimbaza A, Achan J, Nankabirwa J, Nsobya S, et al. Malaria in Uganda: challenges to control on the long road to elimination: I. Epidemiology and current control efforts. Acta Trop. 2012;121:184–95. 29. Okello PE, Van Bortel W, Byaruhanga AM, Correwyn A, Roelants P, Talisuna A, et al. Variation in malaria transmission intensity in seven sites through‑ out Uganda. Am J Trop Med Hyg. 2006;75:219–25. 30. Githeko AK, Ndegwa W. Predicting malaria epidemics in the Kenyan Highlands using climate data: a tool for decision makers. Glob Change Hum Health. 2001;2:54–63. 31. Huang F, Zhou S, Zhang S, Wang H, Tang L. Temporal correlation analysis between malaria and meteorological factors in Motuo County. Tibet Malar J. 2011;10:54. 32. Sewe M, Rocklöv J, Williamson J, Hamel M, Nyaguara A, Odhiambo F, et al. The association of weather variability and under five malaria mortality in KEMRI/CDC HDSS in Western Kenya 2003 to 2008: a time series analysis. Int J Environ Res Public Health. 2015;12:1983–97. 33. Nanvyat N, Mulambalah CS, Barshep Y, Ajiji JA, Dakul DA, Tsingalia HM. Malaria transmission trends and its lagged association with climatic fac‑ tors in the highlands of Plateau State. Nigeria Trop Parasitol. 2018;8:18–23. 34. Okiring J, Routledge I, Epstein A, Namuganga JF, Kamya EV, Obeng‑ Amoako GO, et al. Associations between environmental covariates and temporal changes in malaria incidence in high transmission settings of Uganda: a distributed lag nonlinear analysis. BMC Public Health. 2021;21:1962. 35. McSweeney C, New M, Lizcano G, Lu X, McSweeney C, New M, et al. The UNDP climate change country profiles. Bull Am Meteorol Soc. 2010;91:157–66. 36. Charette M, Berrang‑Ford L, Coomes O, Llanos‑ Cuentas EA, Cárcamo C, Kulkarni M, et al. Dengue incidence and sociodemographic conditions in Pucallpa, Peruvian Amazon: what role for modification of the dengue ‑ temperature relationship? Am J Trop Med Hyg. 2020;102:180–90. 37. Knol MJ, VanderWeele TJ. Recommendations for presenting analyses of effect modification and interaction. Int J Epidemiol. 2012;41:514–20. 38. Donnelly B, Berrang‑Ford L, Ross NA, Michel P. A systematic, realist review of zooprophylaxis for malaria control. Malar J. 2015;14:313. 39. Clark S, Berrang‑Ford L, Lwasa S, Namanya D, Twesigomwe S, Kulkarni M, et al. A longitudinal analysis of mosquito net ownership and use in an indigenous Batwa population after a targeted distribution. PLoS ONE. 2016;11:e0154808. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations. Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions

Journal

Malaria JournalSpringer Journals

Published: Mar 22, 2022

Keywords: Malaria; Climate change; Weather; Meteorological; Sex; Age; Indigenous; Batwa; Bakiga; Sociodemographic modifiers; Uganda

There are no references for this article.