Characterizing the effect of temperature fluctuation on the incidence of malaria: an epidemiological study in south-west China using the varying coefficient distributed lag non-linear model

Xing Zhao, Fei Chen, Zijian Feng, Xiaosong Li, Xiao‐Hua ZhouMalaria 2014 · DOI

TL;DR

This study reveals that the impact of average temperature on malaria incidence isn't constant but varies significantly with daily temperature fluctuations (diurnal temperature range, DTR). Specifically, higher DTR can accelerate malaria spread in cooler conditions but slow it down in warmer conditions, suggesting that predicting malaria requires accounting for both average temperature and its daily variability.

What Problem Does This Paper Address?

Malaria transmission is known to be heavily influenced by temperature, affecting both the mosquito vectors and the malaria parasites. While most research has focused on the average temperature, recent theoretical and laboratory studies suggest that daily temperature fluctuations also play a crucial role. However, there's been a lack of epidemiological evidence at the population level to confirm this. This paper aims to bridge that gap by investigating whether the relationship between mean temperature and malaria incidence is modulated by the diurnal temperature range (DTR).

Methods

The researchers employed a Varying Coefficient Distributed Lag Non-linear Model (VC-DLNM). This advanced statistical modeling framework is designed to simultaneously capture:

  • Non-linear relationships: The effect of temperature isn't a simple linear increase or decrease; it can peak or plateau.
  • Lagged effects: Temperature changes don't impact malaria incidence immediately but rather with a delay, as it takes time for mosquitoes to breed and parasites to develop.
  • Varying coefficients: This is the novel aspect. Instead of assuming the temperature-malaria relationship is the same across all conditions, this model allows the relationship to vary depending on the level of daily temperature fluctuation (DTR).

They used a distributed lag non-linear model (DLNM) as a base, which accounts for non-linear and lagged dependencies. The key innovation was modifying this to allow the coefficients for the mean temperature's effect to change based on the DTR, effectively creating a "varying coefficient" model. This was achieved by categorizing DTR into four levels and incorporating these categories into the model.

Datasets

  • Malaria incidence data: Weekly malaria cases from 30 counties in southwest China, collected between 2004 and 2009.
  • Meteorological data: Daily records of maximum, mean, and minimum temperatures, rainfall, and relative humidity for the same period and counties. This data was sourced from the Chinese Meteorological Data Sharing Service System.
  • Population data: County-level population data from 2004 to 2009, obtained from the National Bureau of Statistics of China.

The study selected the top 30 counties based on average annual malaria incidence and the availability of both malaria and meteorological data.

Key Insights and Ideas

  • DTR is not just noise: Daily temperature fluctuations are not merely random variations but actively modulate the impact of average temperatures on malaria transmission.
  • Cooler conditions amplify DTR effects: In cooler environments, larger daily temperature swings (higher DTR) seem to accelerate malaria incidence. This suggests that even a moderate average temperature might lead to outbreaks if accompanied by significant daily temperature shifts.
  • Warmer conditions dampen DTR effects: Conversely, in warmer environments, high daily temperature fluctuations appear to slow down the effect of the mean temperature on malaria incidence.
  • Optimal temperature is context-dependent: The study identified optimal temperature ranges for malaria transmission (around 21-25°C), but these optima are influenced by the DTR.

Key Results

  • The study found distinct non-linear patterns between mean temperatures and malaria incidence across different DTR levels.
  • In cooler temperature conditions, higher DTR was associated with a greater effect of mean temperature on malaria incidence.
  • In warmer conditions, high daily fluctuations led to a slower effect of mean temperature.
  • The optimal temperature for malaria transmission was estimated to be around 24-25°C or 21-23°C, depending on the lag range considered, with these optima being more pronounced in higher DTR groups.
  • The lag effect of temperature on malaria incidence was strongest around the 4th to 6th week, decreasing thereafter.

Limitations and Future Work

  • Confounding factors: As an observational study, it's acknowledged that other confounding factors (e.g., varying preventive measures, behavior habits, city-specific random effects) could influence the results, and these were not fully controlled.
  • Data quality and microclimate: Variations in data quality over the six-year period and the lack of microclimate data (e.g., indoor vs. outdoor temperature, water temperature) are noted limitations.
  • Parasite types: The study did not differentiate between Plasmodium vivax and Plasmodium falciparum malaria, which might have different non-linear patterns.
  • Mosquito vector data: The absence of mosquito vector information means the study couldn't directly assess their impact.

Future work could involve exploring these different parasite types, incorporating more detailed microclimate data, and potentially using more advanced modeling techniques to disentangle the complex interactions between climate, vectors, and parasites. The authors emphasize the need to consider temperature fluctuations alongside mean temperatures for more accurate malaria transmission prediction.

Summary updated

Abstract

The environment is rarely constant, and the result highlights the need to consider temperature fluctuations as well as mean temperatures, when trying to understand or predict malaria transmission. This work may be the first epidemiological study confirming that the effect of the mean temperature depends on temperature fluctuations, resulting in relevant evidence at the population level.
BibTeX
@inproceedings{zhao2014characterizing,
  author = {Xing Zhao and Fei Chen and Zijian Feng and Xiaosong Li and Xiao‐Hua Zhou},
  title = {{Characterizing the effect of temperature fluctuation on the incidence of malaria: an epidemiological study in south-west China using the varying coefficient distributed lag non-linear model}},
  booktitle = {Malaria},
  year = {2014},
  doi = {10.1186/1475-2875-13-192},
}
Figure