Assessing the effects of air temperature and rainfall on malaria incidence: an epidemiological study across Rwanda and Uganda

Abstract

We investigate the short-term effects of air temperature, rainfall, and socioeconomic indicators on malaria incidence across Rwanda and Uganda from 2002 to 2011. Delayed and nonlinear effects of temperature and rainfall data are estimated using generalised additive mixed models with a distributed lag nonlinear specification. A time series cross-validation algorithm is implemented to select the best subset of socioeconomic predictors and to define the degree of smoothing of the weather variables. Our findings show that trends in malaria incidence agree well with variations in both temperature and rainfall in both countries, although factors other than climate seem to play an important role too. The estimated short-term effects of air temperature and precipitation are nonlinear, in agreement with previous research and the ecology of the disease. These effects are robust to the effects of temporal correlation. The effects of socioeconomic data are difficult to ascertain and require further evaluation with longer time series. Climate-informed models had lower error estimates compared to models with no climatic information in 77 and 60% of the districts in Rwanda and Uganda, respectively. Our results highlight the importance of using climatic information in the analysis of malaria surveillance data, and show potential for the development of climateinformed malaria early warning systems. IntroductionDespite the global contraction in range over the past century , malaria still imposes a significant health and socioeconomic burden to many countries (WHO, 2013). The World Health Organization estimates that about 3.4 billion people are at risk of malaria (WHO, 2013). Approximately 207 million cases and 627,000 deaths occurred in 2012 worldwide (WHO, 2013). About 90% of the total mortality occurs in sub-Saharan Africa, and 77% of that percentage happens in children under 5 years of age (WHO, 2013). Two countries significantly affected by malaria are Rwanda and Uganda. Malaria has long been considered the main cause of morbidity and mortality in both countries (NISR, MOH and ICF International, 2012; UBOS and ICF International, 2012). Over the period 2002 to 2011, more than five million malaria cases were reported in Rwanda to government health facilities. The number was significantly greater in Uganda with about 100 million reports of suspected malaria cases between 2002 and 2010.Trends in malaria incidence could be attributed to the complex interplay of a range of determinants including climatic, environmental, and socioeconomic factors (Kazembe et al., 2006;Lowe et al., 2013;Rulisa et al., 2013). Statistical models are useful tools that allow us: i) to understand how disease outcomes change as a function of variations in their key driver; and ii) to predict disease outcomes based on the dynamics of such drivers (James et al., 2013). This paper aims to investigate the ways in which malaria incidence varies as a function of short-term changes in air temperature and rainfall over the period [2002][2003][2004][2005][...

Using remote sensing environmental data to forecast malaria incidence at a rural district hospital in Western Kenya

“…Studies have shown nonlinear relationships between weather factors and malaria morbidity and mortality 16,19,36–39 . We developed two different general additive models, one using a boosting algorithm to optimize model fit and the other without boosting.…”

Long‐term rainfall regression surfaces for the Kruger National Park, South Africa: a spatio‐temporal review of patterns from 1981 to 2015

“…It is strongly associated with vegetation dynamics and thus ungulate population dynamics (Owen-Smith and Ogutu, 2003;Dunham et al, 2004), predator densities and demography (Celesia et al, 2010) and disease outbreaks (e.g. malaria, Colón-González et al, 2016). Reductions in rainfall can have severe implications for the provision and management of artificial water, which in itself may prove to have far-reaching impacts (Smit, 2013).…”

Assessment of climate-driven variations in malaria incidence in Swaziland: toward malaria elimination

“…Dlamini et al constructed multiple models that correlate mosquito larva abundance with 4 weeks lagged land surface temperature [35]. The non-linear patterns of temperature and precipitation relevant to malaria transmission have been discussed as well [36, 37]. …”