Due to the unavailability of a successful prophylactic or curative treatment thus far until now, vector control is the main driving force in dengue control programmes [
Regardless of different entomological indices that capture various life stages of vector mosquitoes such as larvae [
The current vector control activities within Sri Lanka are mainly driven based on BI and HI [
Setting up a reliable threshold has become a challenging task, due to few major knowledge gaps such as defining an outbreak of dengue, dengue risk situations (in terms of cases), type of mosquito density required to reflect the risk, method to correlate the epidemiological, and entomological cases and spatial scale to be considered [
The study was conducted in two major districts of Sri Lanka; Colombo (6.70° to 6.98°N and 79.83° to 80.22°E) and Kandy (6.93° to 7.50°N and 80.43° to 81.04°E) as indicated in Figure
Location of the studied MOH areas in the districts of Colombo (Dehiwala, Kaduwela, Kolonnawa, Moratuwa, and Piliyandala) and Kandy (Akurana, Gampola, Gangawata Korale, Kandy Municipal Council, and Kundasale) within Sri Lanka.
Routine entomological surveillance activities were conducted on a monthly basis within the selected eight MOH areas from January 2016, to June 2019, by following standard dipping, siphoning, and pipetting methods as recommended by the National Dengue Control Unit, Sri Lanka [
Meanwhile, past entomological surveillance data corresponding to the period of January, 2010, to June 2019, were also collected at the monthly scale from the MOH offices along with the number of reported dengue cases covering the entire study period (January 2010 to June 2019).
House Index [HI] (percentage of houses positive for
Receiver Operating Characteristic (ROC) curves were used to assess the discriminative power of three larval indices (namely, House Index [HI], Breteau Index for
The location of the maximum vertical distance from the line of equality (reference line) or in other words the points on the ROC curve located farthest from the line of equality was selected as threshold values based on the Youden index, to discriminate different phases in the dengue incidence [
Ethical approval was obtained from the Ethics Review Committee of the Faculty of Medicine, University of Kelaniya (P/155/10/2015).
The temporal variations of average annual vector indices, with respect to BI values of both
Temporal variation of reported dengue cases in Colombo and Kandy districts of Sri Lanka.
According to the Pearson correlation analysis,
Pearson’s correlation coefficients for the association between the dengue cases and the larval indices at lag periods of zero, one, and two months in the study areas.
District | MOH area |
|
|
|
|||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
BIagp | BIalb | HI | CI | BIagp | BIalb | HI | CI | BIagp | BIalb | HI | CI | ||
Colombo | Dehiwala | 0.309 | -0.181 | -0.119 | 0.120 | 0.471 |
-0.270 | -0.287 | 0.197 | 0.685 |
-0.287 | -0.387 |
0.158 |
Kaduwela | 0.357 | -0.212 | -0.149 | 0.046 | 0.479 |
-0.294 | -0.304 |
0.102 | 0.742 |
-0.321 | -0.574 |
0.194 | |
Kolonnawa | 0.337 | -0.146 | -0.209 | 0.141 | 0.439 |
-0.242 | -0.357 |
0.189 | 0.724 |
-0.287 | -0.491 |
0.234 | |
Piliyandala | 0.243 | -0.241 | -0.278 | 0.059 | 0.411 |
-0.267 | -0.378 |
0.146 | 0.587 |
-0.367 | -0.472 |
0.159 | |
Moratuwa | 0.342 |
-0.175 | -0.267 | 0.173 | 0.580 |
0.294 | -0.352 |
0.214 | 0.767 |
-0.285 | -0.549 |
0.167 | |
|
|||||||||||||
Kandy | KMC & GK | 0.321 | -0.254 | -0.278 | 0.148 | 0.459 |
-0.347 | -0.480 |
0.281 | 0.764 |
-0.310 | -0.512 |
0.375 |
Akurana | 0.313 | -0.310 | -0.307 | 0.211 | 0.419 |
-0.358 | -0.368 |
0.385 | 0.676 |
-0.362 | -0.387 |
0.349 | |
Gampola | 0.287 | -0.291 | -0.235 | 0.165 | 0.387 | -0.371 | -0.379 |
0.293 | 0.648 |
-0.354 | -0.413 |
0.194 | |
Kundasale | 0.257 | -0.243 | -0.269 | 0.107 | 0.409 |
-0.310 | -0.374 | 0.189 | 0.629 |
-0.297 | 0.398 |
0.204 |
Note: “
Since the lag periods of one and two months resulted in significant associations, the ROC was also conducted for one and two month lag periods. At one-month lag period, only BIagp of Akurana (0.530) and Moratuwa (0.507) had moderately accurate discriminative powers in reflecting dengue epidemics as shown in Table
Area under the Receiver Operating Characteristic curves of different larval indices in terms of dengue epidemic incidence at lag periods of 1 and 2 months.
District | MOH area |
|
|
||||||
---|---|---|---|---|---|---|---|---|---|
BIagp | BIalb | CI | HI | BIagp | BIalb | CI | HI | ||
Colombo | Dehiwala | 0.409 | 0.289 | 0.210 | 0.293 | 0.661 |
0.347 | 0.319 | 0.508 |
Kaduwela | 0.437 | 0.319 | 0.196 | 0.237 | 0.731 |
0.374 | 0.287 | 0.407 | |
Kolonnawa | 0.497 | 0.217 | 0.176 | 0.336 | 0.698 |
0.288 | 0.269 | 0.568 |
|
Piliyandala | 0.373 | 0.327 | 0.133 | 0.299 | 0.619 |
0.365 | 0.301 | 0.509 |
|
Moratuwa | 0.507 |
0.304 | 0.219 | 0.361 | 0.794 |
0.354 | 0.334 | 0.589 |
|
|
|||||||||
Kandy | KMC & GK | 0.485 | 0.327 | 0.297 | 0.402 | 0.684 |
0.349 | 0.358 | 0.519 |
Akurana | 0.530 |
0.315 | 0.250 | 0.456 | 0.761 |
0.317 | 0.336 | 0.637 |
|
Gampola | 0.421 | 0.276 | 0.184 | 0.393 | 0.627 |
0.321 | 0.317 | 0.573 |
|
Kundasale | 0.389 | 0.269 | 0.201 | 0.243 | 0.655 |
0.309 | 0.309 | 0.549 |
Note: “
Among the considered larval indices, only BIagp and HI emerged as discriminative indicators of dengue epidemics. Therefore, three threshold levels were defined for these as “Low Risk”, “Moderate Risk”, and “High Risk” phases, to facilitate management of dengue vectors based on the association of
Recommended risk threshold values for BIagp and HI based on the Receiver Operating Characteristic curves (ROC).
District | MOH area | BIagp | HI | ||||
---|---|---|---|---|---|---|---|
Low risk | Moderate risk | High risk | Low risk | Moderate risk | High risk | ||
Colombo | Dehiwala | 2.1 | 3.9 | 4.9 | 5.8 | 7.9 | 9.6 |
Kaduwela | 2.4 | 3.7 | 5.3 | 5.6 | 9.2 | 11.2 | |
Kolonnawa | 2.0 | 3.5 | 4.4 | 5.9 | 8.3 | 12.2 | |
Piliyandala | 3.2 | 4.8 | 5.7 | 4.7 | 11.4 | 16.3 | |
Moratuwa | 2.1 | 3.3 | 4.8 | 5.5 | 7.8 | 10.2 | |
Average | 2.4 | 3.8 | 5.0 | 5.5 | 8.9 | 11.9 | |
|
|||||||
Kandy | KMC & GK | 2.2 | 3.5 | 4.7 | 6.6 | 8.5 | 10.6 |
Akurana | 2.1 | 3.9 | 4.8 | 6.2 | 7.7 | 9.9 | |
Gampola | 3.9 | 5.1 | 5.9 | 7.8 | 12.1 | 15.5 | |
Kundasale | 3.6 | 4.1 | 5.6 | 7.1 | 8.1 | 11.1 | |
Average | 2.9 | 4.2 | 5.3 | 6.9 | 9.1 | 11.8 |
Note: “
Receiver Operating Characteristic curve derived for the Akurana MOH area along with maximum vertical distance from line of equality corresponding to (a) Low risk; (b) Moderate risk; and (c) High-risk thresholds of BIagp at a lag period of two months.
The minimum “Low Risk”, “Moderate Risk”, and “High Risk” threshold values for BIagp were recommended for Kolonnawa MOH area as 2.0, 3.5, and 4.4, respectively. On the other hand, the maximum threshold values were set for the Gampola MOH area as 3.9, 5.1, and 6.2 (“Low”, “Moderate”, and “High Risk” thresholds), respectively. In the case of HI, the minimum “Low Risk” cutoff value (4.7) was suggested for Piliyandala MOH area (located in the district of Colombo), while the minimum “Moderate Risk” (7.7) and “High Risk” thresholds were identified for Akurana MOH in Kandy. The Piliyandala MOH area had the maximum “High Risk” threshold of 16.3 for HI, while Gampola had the maximum “Low Risk (7.8)” and “Moderate Risk (12.1)” thresholds (Table
Comparison of entomological indices that reflect the incidence of dengue epidemics and the establishment of reflective thresholds has remained a challenge to many dengue-endemic countries, including Sri Lanka [
A diverse interplay of multiple factors such as humans,
Even though the density of
Irrespective of using both CI and HI in routine entomological surveillance, only HI stated a significant and moderate relationship with dengue cases. In general, the HI is influenced by densities of both
Thresholds developed based on BI and HI have been successfully used in dengue epidemic prediction and management in Brazil, Havana, Thailand, Trinidad, and America [
In a recent study, Ong et al. [
In a previous study, a set of entomologically driven thresholds have been developed for a selected set of MOH areas in Kandy, solely based on the natural dynamics of dengue vectors [
As a routine practice, the VCE in Sri Lanka utilize
Thresholds defined for BIagp and HI denoted a notable degree of spatial variation among the studied MOH areas. As mentioned previously, vector densities are one of the contributing risk factors of dengue incidence, in addition to climatic conditions, land use, population density, degree of herd immunity, etc. [
The current vector controlling approaches heavily rely upon chemical-based methods (systematic use of larvicides, adulticides, and chemical fogging), along with the physical elimination of breeding sites of
The absence of records on serological circulations of dengue viruses, limitations in the availability of long term surveillance data (arising due to limitations in knowledge, trained staff, financial and physical resources, limited commitment and satisfaction of field staff) and limited knowledge on herd immunity, is the major limitations encountered in defining the thresholds in the current study [
In Sri Lanka, VCE are struggling to manage dengue epidemics through a variety of approaches such as vector control, environmental, and patient management [
In light of this, thresholds defined in this study would be helpful for adopting and directing vector control activities, with optimum utilization of limited human and financial resources to control dengue epidemics [
Based on our findings, both HI and BIagp are significantly associated with dengue epidemics at lag periods of one and two months, emerging as representative
Breteau index
Breteau index for
Breteau index for
Container index
Gangawata Korale
House index
Integrated vector management
Kandy municipal council
Medical officer of health
Pearson correlation coefficients
Receiver operating characteristic curve
Vector controlling entities.
The datasets supporting the conclusions of this article are included within the article and its additional files.
Ethical approval was obtained from the Ethics Review Committee of the Faculty of Medicine, University of Kelaniya (P/155/10/2015). The confidentiality of the acquired data was maintained throughout the study.
The written consent of the participants was acquired for publication prior to the conducting of entomological surveillance and all the authors read and approved the manuscript for the publication. Written consents from selected household heads were obtained along with the permission from the relevant MOH offices, prior to the entomological surveillance.
The authors have declared that they have no competing interests.
LU did the conceptualization of the empirical modelling, conducting field surveys, statistical analysis, and writing the manuscript. SA did the planning, guiding, and supervision of the field surveys; NG did the supervision of the research and reviewing the manuscript; MCMI did the supervision of field surveys and reviewing the manuscript; TF conducted field surveys; WA did the overall supervision of the study and reviewing the manuscript; All authors read and approved the final manuscript.
Surveillance activities were funded by the National Research Council Funded Dengue Mega Grant (NRC TO 14-04), Sri Lanka.
Supplementary Figures S1–S4: temporal variations in the average Aedes larval indices for each MOH area in the districts of Colombo and Kandy (2010 to 2019).