Umeå universitets logga

Dengue is the major public health burden in Sri Lanka. Kalutara is one of the highly affected districts. Understanding the drivers of dengue is vital in controlling and preventing the disease spread. This study focuses on quantifying the influence of weather variability on dengue incidence over 10 Medical Officer of Health (MOH) divisions of Kalutara district. Weekly weather variables and data on dengue notifications, measured at 10 MOH divisions in Kalutara from 2009 to 2013, were retrieved and analysed. Distributed lag non-linear model and hierarchical-analysis was used to estimate division specific and overall relationships between weather and dengue. We incorporated lag times up to 12 weeks and evaluated models based on the Akaike Information Criterion. Consistent exposure-response patterns between different geographical locations were observed for rainfall, showing increasing relative risk of dengue with increasing rainfall from 50 mm per week. The strongest association with dengue risk centred around 6 to 10 weeks following rainfalls of more than 300 mm per week. With increasing temperature, the overall relative risk of dengue increased steadily starting from a lag of 4 weeks. We found similarly a strong link between the Oceanic Niño Index to weather patterns in the district in Sri Lanka and to dengue at a longer latency time confirming these relationships. Part of the influences of rainfall and temperature can be seen as mediator in the causal pathway of the Ocean Niño Index, which may allow a longer lead time for early warning signals. Our findings describe a strong association between weather, El Niño-Southern Oscillation and dengue in Sri Lanka.

BACKGROUND: Dengue is one of the major public health problems in Sri Lanka. Its outbreak pattern depends on a multitude of drivers, including human mobility. Here we evaluate the impact of COVID-19 related mobility restriction (lockdown) on the risk of dengue in Sri Lanka.

METHODOLOGY: Two-stage hierarchical models were fitted using an interrupted time-series design based on the notified dengue cases, January 2015 to July 2020. In the first stage model, the district level impact was estimated using quasi-Poisson regression models while accounting for temporal trends. Estimates were pooled at zonal and national levels in the second stage model using meta-analysis. The influence of the extended period of school closure on dengue in children in the western province was compared to adults.

FINDINGS: Statistically significant and homogeneous reduction of dengue risk was observed at all levels during the lockdown. Overall an 88% reduction in risk (RR 0.12; 95% CI from 0.08 to 0.17) was observed at the national level. The highest impact was observed among children aged less than 19 years showing a 92% reduction (RR 0.8; 95% CI from 0.03 to 0.25). We observed higher impact in the dry zone having 91% reduction (RR 0.09; 95% CI from 0.05 to 0.15) compared to wet zone showing 83% reduction (RR 0.17; 95% CI from 0.09 to 0.30). There was no indication that the overall health-seeking behaviour for dengue had a substantial influence on these estimates.

SIGNIFICANCE: This study offers a broad understanding of the change in risk of dengue during the COVID-19 pandemic and associated mobility restrictions in Sri Lanka. The analysis using the mobility restrictions as a natural experiment suggests mobility patterns to be a very important driver of dengue transmission.

Background: Dengue has become a major public health problem in Sri Lanka with a considerable economic burden. As a response, in June, 2014, the Ministry of Health initiated a proactive vector control programme in partnership with military and police forces, known as the Civil-Military Cooperation (CIMIC) programme, that was targeted at high-risk Medical Officer of Health (MOH) divisions in the country. Evaluating the effectiveness and cost-effectiveness of population-level interventions is essential to guide public health planning and resource allocation decisions, particularly in resource-limited health-care settings.

Methods: Using an interrupted time series design with a non-linear extension, we evaluated the impact of vector control interventions from June 22, 2014, to Dec 29, 2016, in Panadura, a high-risk MOH division in Western Province, Sri Lanka. We used dengue notification and larval survey data to estimate the reduction in Breteau index and dengue incidence before and after the intervention using two separate models, adjusting for time-varying confounding variables (ie, rainfall, temperature, and the Oceanic Niño Index). We also assessed the cost and cost-effectiveness of the CIMIC programme from the perspective of the National Dengue Control Unit under the scenarios of different levels of hospitalisation of dengue cases (low [25%], medium [50%], and high [75%]) in terms of cost per disability-adjusted life-year averted (DALY).

Findings: Vector control interventions had a significant impact on combined Breteau index (relative risk reduction 0·43, 95% CI 0·26 to 0·70) and on dengue incidence (0·43, 0·28 to 0·67), the latter becoming prominent 2 months after the intervention onset. The mean number of averted dengue cases was estimated at 2192 (95% CI 1741 to 2643), and the total cost of the CIMIC programme at 2016 US$271 615. Personnel costs accounted for about 89% of the total cost. In the base-case scenario of moderate level of hospitalisation, the CIMIC programme was cost-saving with a probability of 70% under both the lowest ($453) and highest ($1686) cost-effectiveness thresholds, resulting in a net saving of $20 247 (95% CI −57 266 to 97 790) and averting 176 DALYs (133 to 226), leading to a cost of −$98 (−497 to 395) per DALY averted. This was also the case for the scenario with high hospitalisation levels (cost per DALY averted −$512, 95% CI −872 to −115) but with a higher probability of 99%. In the scenario with low hospitalisation levels (cost per DALY averted $690, 143 to 1379), although the CIMIC programme was cost-ineffective at the lowest threshold with a probability of 77%, it was cost-effective at the highest threshold with a probability of 99%.

Interpretation: This study suggests that communities affected by dengue can benefit from investments in vector control if interventions are implemented rigorously and coordinated well across sectors. By doing so, it is possible to reduce the disease and economic burden of dengue in endemic settings.

Funding: None.