A retrospective study of environmental predictors of dengue in Delhi from 2015 to 2018 using the generalized linear model

Even though several studies were performed earlier1,7,8,910, the changes in climatic conditions make it imperative to revisit and re-examine the impact of climate change on the occurrence of dengue cases. The alteration in climate could change both the spatial and temporal dynamics of dengue ecology by increasing vector ranges, broadening the duration of vector activity, and snowballing the mosquito’s infectious period14.

The seasonal pattern of dengue cases was similar every year (2015–2018) in Delhi which was also reported by others10. Our findings reveal that the dengue cases progressed from July to August, hit the highest point in September to October and declined by December. Dengue cases were recorded high during the post-monsoon period10.

Analysis of dengue cases with monthly average max and min temperature revealed that the number of reported cases was the highest from August to October across the study period (2015 to 2018) which was a few months after the highest average max and minimum temperature.

The average maximum temperature (lag 0–3 months) was significantly correlated with dengue cases and it was high at lag 2 and 3 months (0.695 and 0.694; p<0.01) as reported in another study10however, another study reported at lag 013. However, a non-significant effect was also reported in another study15. This is because the temperature is an important determinant of egg and immature mosquito development, biting rate, the development time of virus in the mosquito (extrinsic incubation period), and survival at all stages of the mosquito life cycle1.

The effect of minimum temperature (lag 0–3) was also significant, but a strong correlation at lag 2 was observed (0.887; p<0.001) as reported by others13. The temperature at peak of dengue was between 25 and 27 ° C similar to another study that has shown temperatures in the range of 20 to 31.7 ° C have provided a suitable environment for breeding and abundance of Aedes mosquito species and thereby increasing the risk of dengue cases16.

The correlation between average temperature and dengue cases was significant and high at lag 2 (0.795, p<0.01), while 0–3 months lag was reported in another study17. However, a moderate positive correlation was indicated between average monthly temperature and dengue cases15 while contrasting results were reported in a study by Su18.

A significant correlation was also seen between the difference in monthly average minimum and maximum temperature at lag 1 and 2 with dengue cases (- 0.675 and – 0.663, p<0.01). The negative correlation indicates reverse relation between them. However, the diurnal temperature range (DTR) was reported to be associated with the dengue epidemic19.20. High mean temperatures with narrow daily temperature variation, are important for dengue transmission as it influences the biology and vectorial capacity of Ae. Aegypti21. It is highly probable that as the number of cold days and nights decreases and the number of warm days and nights increases on the global scale (IPCC), it would impact dengue incidence.

The temporal trend across seasons revealed a rise in the occurrence of dengue fever in the monsoon seasons and post-monsoon seasons. The dengue cases reached the peak following the months with the highest rainfall, post-monsoon (Fig. 1) also reported by another study in Delhi10 which may be related to inherent delays between weather conditions and their impact on mosquito populations, virus replication with their subsequent influence on transmission patterns3. However, rainfall was not associated with dengue incidence in the Chitwan district of Nepal13.

The total and cumulative total rainfall (lag 1–3 months) were significantly correlated and it was high for total rainfall at lag 2 (0 0.738, p<0.01) and cumulative rainfall lag 0 (0.795, p<0.01) as reported by others3,10,22. Rainfall was significantly related to dengue as precipitation provides habitats for the aquatic stages of the mosquito life cycle and strongly influences vector distribution14however, extreme rainfall decreases in dengue risk due to adverse impact on vector habitat23.A low correlation between humidity and dengue cases was observed as compared to rainfall as reported by others3. The humidity (lag 0) was found significantly correlated with dengue (0.299, p<0.05), however, it was reported at lag 0 and 29.13.

The Generalized linear model (GLM) using negative binomial regression was fitted to know the combined effect of environmental factors. The maximum temperature and cumulative rainfall had a significant positive impact on dengue incidence while the difference in maximum and minimum temperature, and relative humidity had a reverse effect on dengue incidence. Similar results were reported in a study conducted in Delhi and rainfall, temperature, and humidity at lag 2 were the significant predictors of dengue10. A study in Nepal indicated that minimum temperature at lag 2, the maximum temperature at lag 0, the maximum temperature at lag 3, and relative humidity at lag0 were significant predictors in the model13. In a study in Cambodia, the model reflected average temperature, maximum temperature, minimum temperature and rainfall as significant predictors of dengue17while another study in Dhaka city revealed the best fit for maximum temperature, rainfall and humidity at lag 2 months8. With such a prediction model it is possible to have better control measures and preparedness for better case management to avoid the epidemic. However, the best fit model of this study is shown to overestimate the number of dengue cases during peak season for the years 2017 and 2018. The wide gap between actual and predicted may be attributed to various other underlying factors such as the presence of susceptible in the population, adopted effective control measures, better case management, case-reporting, better awareness, etc. which have not been considered in the model.

The correlation between dengue incidence and weather factors also seemingly varies by locality, suggesting that a future dengue early warning system would likely be best applied at a local / regional scale, rather than at a nationwide level. The present study is consistent with findings of other studies24,2526that a persistent peak in dengue cases each year following the highest rainfall and temperature, indicated the influence of the preceding month’s climatic factors, which was comparable to other countries17.27.28, a time during which the mosquito can develop and contaminate the population. In a study on the dengue situation in India, the high transmission potential was also reported throughout the monsoon period29. Other studies in Taiwan, Thailand, Brazil, Singapore, etc. also show the association between dengue incidence and seasonal patterns in temperature, relative humidity, and rainfall1,7,8,910. The time lag can be also explained by the influence of weather conditions on the biological development of the mosquito vector, including prolonged egg hatching periods and the propensity of Aedes eggs to survive without water for many months27.

The limitation remains that the dengue data is for the entire Delhi and not area-specific hence dengue cases have been taken for the entire Delhi. Therefore, accessing climate data and dengue cases for Municipalities wise was not possible. More years of data need to be studied further to validate the best fit model. The study needs to be extended to socio-demographic components such as population growth, travel or migration rate, water storage habit, etc. This work has been limited to Delhi only, hence future studies encompassing diverse geographic regions should also be included.

Leave a Comment

Your email address will not be published.