The effect of long-lasting insecticidal water container covers on field populations of Aedes aegypti (L.) mosquitoes in Cambodia

Transcription

1 Vol. 33, no. 2 Journal of Vector Ecology 333 The effect of long-lasting insecticidal water container covers on field populations of Aedes aegypti (L.) mosquitoes in Cambodia Chang Moh Seng 1, To Setha 2, Joshua Nealon 1, Ngan Chantha 2, Doung Socheat 2, and Michael B. Nathan 3 1 World Health Organization Cambodia. # , Pasteur Street (51), Phnom Penh, Cambodia 2 National Centre of Parasitology, Entomology and Malaria Control, Ministry of Health, Phnom Penh, Cambodia 3 Department of Control of Neglected Tropical Diseases, World Health Organization, Geneva, Switzerland Received 2 April 2008; Accepted 31 July 2008 ABSTRACT: Dengue in Cambodia is mainly transmitted by Aedes aegypti (L.) mosquitoes that primarily breed in large, concrete jars ( 200 liters) used for the storage of water for domestic use. Following a preliminary risk assessment, longlasting insecticidal netting (LN) treated with deltamethrin was incorporated into the design of the covers for these jars. Their effect on immature and adult female populations of Ae. aegypti in six villages in a peri-urban area of Cambodia were compared with populations in six nearby control villages before and for 22 weeks after distribution of the jar covers. There were significantly fewer pupae per house in intervention villages than in control villages (6.6 and 31.9, respectively, p<0.01). Fewer pupae were recovered from intervention houses than from control houses at every post-intervention assessment. Two weeks after the intervention, the average number of indoor resting female Ae. aegypti per house in the intervention villages had declined approximately three-fold, whereas in the controls there was only a slight reduction (16%). The magnitude of the difference between the two areas diminished over time, which contact bioassays confirmed was likely due to a gradual reduction of insecticidal effect of the jar covers. In the study area, insecticide-treated covers for large concrete water storage jars were efficacious for controlling Ae. aegypti in the protected water jars and with a demonstrable effect on adult densities and survival. Further studies of this targeted container strategy in Cambodia, and elsewhere, are recommended. However, improvements in technology that would extend the duration of insecticidal effectiveness of LN materials may be needed for the development of cost-effective public health applications. Journal of Vector Ecology 33 (2): Keyword Index: Mosquito, jar covers, vector control. INTRODUCTION Dengue is of major public health importance in Cambodia, and during the main transmission season is the disease responsible for most pediatric hospitalizations (DeRoeck et al. 2003). All four virus serotypes circulate in both urban and rural areas and children under 15 are disproportionately affected. An unprecedented nationwide epidemic occurred in 2007 with a reported total of 39,850 hospitalized cases and 407 deaths. Previous studies in Cambodia have shown that concrete, household water storage jars constitute over 80% of Aedes aegypti larval habitats in many areas of the country (Socheat et al. 2004). These jars, mostly liters in capacity, have consequently become a target for dengue vector control activities in an attempt to reduce dengue transmission. Since 2001, large scale, seasonal application of larvicides to these container habitats has been carried out to reduce the risk of infection in a population of approximately 3 million urban inhabitants (Suaya et al. 2007), some 25% of the population of the country. This strategy, however, is viewed as an interim measure for high population density urban areas. Alternative, long-term, cost-effective interventions for Ae. aegypti control are being sought for incorporation into the national dengue prevention and control program as part of an integrated vector management strategy. One intervention under investigation is an insecticidal cover that aims to prevent mosquito breeding in domestic water storage jars. Jar covers of various types are commonly used by householders in Cambodia for reasons that seldom include mosquito control. The development and distribution of one which would prevent mosquito breeding is a logical extension to current cultural practices. There has been only limited success in suppressing vector populations in some endemic countries through the use of covers as physical barriers to prevent access to ovipositing female mosquitoes (Kittayapong and Strickman 1993). However the incorporation of insecticides into covers may confer two important additional benefits, namely the killing of mosquitoes emerging from the water in the jars; and reducing survival rates of female mosquitoes that are seeking oviposition sites in the community, including in water jars. Technological advances in the manufacture of long-lasting insecticidal mosquito nets (LNs), mainly for malaria control, have recently enabled insecticidal jar covers and other treated materials to be explored for their potential in controlling vectors of other diseases, including those which transmit dengue viruses (Kroeger et al. 2006, Lenhart et al. 2008). Through a process of participatory research and field

2 334 Journal of Vector Ecology December 2008 testing of a range of prototypes, a suitable jar cover design for local use was identified prior to the study (Socheat et al. 2004). The subsequent choice and incorporation of LN technology into the design (deltamethrin 50 mg/m 2 ) included preliminary contact bioassay studies and a risk assessment to ensure that the dosages and rates of insecticide release from the fibers were such that the covers would not pose undue risk for persons who drank the protected water. Contact bioassay results indicated that under laboratory conditions of permanent soaking or a standardized weekly flushing protocol using water poured from above, the bioefficacy of deltamethrin-impregnated LN netting on the susceptible Rockefeller strain of Ae. aegypti persisted for at least 26 weeks. When subjected to outdoor weathering, the insecticidal activity was reduced by several weeks. Field strains of Ae. aegypti originating from Kampong Cham were found to be similarly susceptible (Chang unpublished data). For the risk assessment, under simulated field conditions, concentrations of deltamethrin in the water samples taken from water jars with LN covers were consistently below the limit of detection of the analytical method (0.05 μg/liter), which is considerably lower than the acceptable daily intake of mg/kg (WHO 1990). It was concluded that drinking water from jars protected by deltamethrin-impregnated covers does not pose an undue hazard. The decision to utilize LN covers was based on issues of operational simplicity, projected efficacy, and marginal additional expense over using untreated covers. It was postulated that the lethal insecticidal effect on gravid females and emerging adults would provide a considerable advantage over non-insecticidal covers. This field study of the effectiveness of the LN jar cover design for Ae. aegypti control was undertaken between May and November 2006, in Ampil commune, approximately 120 km northeast of Phnom Penh, the capital city of Cambodia. MATERIALS AND METHODS Description of the study area Ampil commune, Kampong Cham Province, is close to the provincial capital, Kampong Cham. It is comprised of 12 villages with a population of approximately 13,000 in 2,500 households. The area has a history of dengue outbreaks. The villages are typical Cambodian communities. Many houses are built on stilts and are of wooden construction, with a variety of livestock living in close proximity. Rice cultivation provides the main source of income. There is no piped water supply or sanitation and electrical supply is uncommon. Water for domestic use is stored in concrete water jars, mostly with capacities of or liters. The jars are located outdoors where roof catchment rainwater is harvested and stored for domestic use during periods of rainfall. At other times, the jars are filled from alternative sources, e.g., rivers and open wells. Rainfall data was obtained from the government weather station, Kampong Cham Province, approximately 2 km from the study site. Water jar cover design The fabric and netting of the water jar cover was attached to a circular, painted steel hoop. Each pre-sewn cover was comprised of a central circle of white polyester netting (PermaNet II with 150 denier, 156 mesh, incorporating deltamethrin at a dosage of 55 mg/m 2 and an ultra-violet protectant) through which water could pass (thereby allowing for in situ rainwater harvesting), and an outer ring of blue polyester fabric, also treated with deltamethrin and ultra-violet protectant at the same dosage. The blue polyester fabric was incorporated in the cover design to increase resistance from wear and tear during regular household use caused by abrasion of the cover against the rim of the concrete jar. The blue fabric was sewn with a hem into which a drawstring was inserted. The cover, slightly larger than the diameter of the hoop, was fitted by placing it over the hoop and tying the drawstring tightly. To best fit the two main jar sizes, two cover sizes were used, one with a diameter of 75 cm and a central, circular netting portion of 60 cm diameter, and the other with a diameter of 85 cm and a central, circular netting portion with a diameter of 70 cm. Study design A cluster of six villages (with a total of 1,307 households) was assigned to the intervention arm of the study and a further six (with a total of 1,174 households) to the control arm, on a non-randomized basis. The closest distance between the two study arms was >200 m. After discussion with and agreement of the village elders to carry out the study, the informed consent of all householders was obtained for their participation. All houses in the study area were assigned a unique identification number. For the first survey in each village, a systematic sample of 10% of the houses was selected based on random selection of the first house from among the numbers For successive surveys, the 10% random selection process was based on the remaining numbers. When all numbers from 1-10 had been used once, the process was started anew. In May 2006, at the beginning of the rainy season, two baseline surveys were carried out, respectively four weeks and one week prior to the distribution of jar covers. After jar cover distribution, ten post-intervention surveys were carried out at approximately two-week intervals. Volunteers distributed free jar covers to all households in the intervention villages. One cover was provided for each water jar and their proper use was explained to householders. A total of 4,052 covers were distributed to 1,307 households in the intervention area, an average of 3.1 per house. Entomological methods Every two weeks, all water storage jars in the sampled houses were examined with the aid of torchlight for the presence of mosquito pupae. All pupae from each infested water jar were removed using a pipette and white plastic

3 Vol. 33, no. 2 Journal of Vector Ecology 335 Table 1. House, container, and Breteau indices before distribution of jar covers (baseline). Intervention area Control area 1 st baseline survey 2 nd baseline survey Combined baseline surveys 1 st baseline survey 2 nd baseline survey Combined baseline surveys No. of houses surveyed No. of houses with Aedes larvae/pupae House Index No. of containers with water No. of wet containers with Aedes larvae/pupae Container Index Breteau Index (no. of containers with Aedes larvae/pupae per 100 houses) bowl and transported live to the laboratory in vials labelled with house- and jar-specific identification details. Pupae were killed in hot water (~70 C), identified to species using the taxonomic keys of Mattingly (1971), and counted. During two baseline surveys, and again 13 weeks and 22 weeks after jar cover distribution, the surveys for pupae in water jars were extended to other larval habitats. On these occasions, each sampled house and surrounding land was systematically searched and all containers capable of holding water were identified, categorized by type, and counted. All pupae were collected, identified to species in the laboratory, and counted to provide container-specific information on the standing crop of pupae and comparison between intervention and control villages. Population and household data were obtained from the Provincial Health Department and used to estimate pupae per person indices for each village in the study. During the baseline surveys only, mosquito larvae (up to a maximum of ten per container) were also collected from all infested containers and identified to species in order to derive house, container, and Breteau indices. The covered, partially covered, or uncovered status of each surveyed water jar was recorded. Adult mosquitoes resting in the bedrooms of sampled houses were collected using a battery-powered knapsack aspirator (modified after Clark et al. 1994). A systematic 10-min procedure was followed. A separate, labelled holding cup was used for each house. After collection and transportation to the field laboratory, the species and sex of mosquitoes were determined and the numbers collected from each house recorded. All unfed or recently bloodfed female Ae. aegypti were dissected to determine parous rates using the tracheolar examination method of Detinova (1962). Only female Ae. aegypti were used for analysis of the relative abundance of adult Aedes over the study period. Contact bioassays were performed on the netting portion of jar covers taken from the field after 4, 8, 12, 16, and 20 weeks of use by householders. A local strain of second-generation Ae. aegypti originating from Kampong Cham was colonized, and 15 sugar-fed, two-to-three dayold females were exposed to pieces of the netting (10 cm diameter) for 10 min in standard WHO bioassay cones. Four replicates were performed on each cover and between one and three covers were used at each interval, depending on the availability of sufficient numbers of reared mosquitoes. Knockdown after 10 min (KD 10 ) and 60 min (KD 60 ) and 24 h mortalities were recorded. The sampled covers were replaced with new ones. All data were analyzed using Stata 6.0 (Stata Corp., Texas, U.S.A.). Confidence intervals of means were calculated from the standard error. The Wilcoxon rank-sum test was used to compare differences in quantitative variables between the intervention and control areas at each assessment point and as a whole. Comparisons between parous rate data from the intervention and control areas and the proportion of jars and their infested/non-infested status were examined using Pearson s chi-square test.

4 336 Journal of Vector Ecology December 2008 RESULTS Table 2. Number of each container type (No.) and Aedes pupae (Pupae) collected in the intervention and control areas during the combined baseline surveys, and at 13 and 22 weeks after distribution of jar covers. The number of houses surveyed is provided in brackets. Intervention area Control area Baseline (297) 13 weeks (142) 22 weeks (140) Baseline (251) 13 weeks (124) 22 weeks (128) No. Pupae No. Pupae No. Pupae No. Pupae No. Pupae No. Pupae Container Type Drum Concrete water jar 896 9, , , , ,176 Concrete tank Small pot Flower vase Tires Tin can Broken pot Other Total 2,238 11, , ,963 1,474 9, , ,551 The total rainfall for 2006 was 1,516 mm, more than 80% of which fell during the study period (mid-may to early November), which corresponds to the main dengue transmission season. During this period, the month with the highest rainfall was August with a total of mm and the driest was November, with a total of 12.7 mm. In the intervention villages prior to distribution of the jar covers, the combined results of the two baseline surveys yielded house, container, and Breteau indices of 88.2, 31.3, and 235.7, respectively. By comparison, the respective indices in the control villages were 82.5, 37.5, and (Table 1). A total of 20,805 Aedes pupae were collected during the two baseline surveys (Table 2). Of these, 98.4% (20,469) were Ae. aegypti and 1.6% (336) were Ae. albopictus. The great majority of pupae (83.2%) were collected from water jars. A further 6.0% of the total was collected from concrete tanks and 3.7% from small pots. The distribution of pupae among containers in the intervention and control areas was similar at baseline. In the intervention villages, the standing crop of Aedes pupae declined from 35.3/house (95% CI ) at baseline to 11.0/house (95% CI ) after 13 weeks, later increasing to 14.2/house (95% CI ) after 22 weeks (Table 3). The corresponding rates in the control area were 37.6/house (95% CI ), 33.0/house (95% CI ) and 20.1/house (95% CI ) respectively. For water storage jars, the targeted habitat in the intervention villages, the rates were 29.6 pupae/house (95% CI ) at baseline, 6.6/house (95% CI ) after 13 weeks and 10.9/house (95% CI ) after 22 weeks, whereas in the control villages the rates were 31.0 pupae/house (95% CI ), 22.7/house (95% CI ) and 17.2/house (95% CI ), respectively. Over the course of the follow-up, a cumulative total of 3,698 water storage jars were examined and 3,247 (88.0%) were fully covered at the time of survey. At each assessment point, over 80% of water jars were covered (range %, Table 4). On examination, 536 (16.5%) of the fully covered jars were infested, in comparison with 115 of the 262 (43.9%) uncovered jars (p<0.001). In the intervention villages, the number of pupae per person decreased from 6.9 at baseline to 2.1 after 13 weeks, and increased to 2.8 after 22 weeks. The corresponding rates in the control area were 7.3, 6.4, and 3.9, respectively (Table 5). Of 11,254 adult mosquitoes captured indoors during the two baseline surveys, 6,458 (57.4%) were Ae. aegypti. Only one (<0.01%) Ae. albopictus was collected. The remainder were either Anopheles spp. (2,657; 26.3%) or Culex spp. (2,138; 19.0%). Of the Ae. aegypti population, 52.5% were females. Hereafter, the results presented relate only to adult female Ae. aegypti. In the intervention area, an average of 5.2 and 5.9 females were collected per house during the first and second baseline assessments, respectively, compared to 6.4 and 7.2 per house in the control area.

5 Vol. 33, no. 2 Journal of Vector Ecology Baseline Post-intervention Ae. aegypti females per house Intervention Control Number of weeks before/after jar cover distribution Figure 1. Mean number of indoor resting female Ae. aegypti per house in the intervention and control areas. Two weeks after jar cover distribution, the indoor resting density of mosquitoes in the intervention area had declined approximately three-fold, from the baseline mean of 5.6 females per house to 1.7 females per house and remained below 2 for the remainder of the study (Figure 1). In the control area, a decline was also observed, but it was less pronounced, decreasing from a baseline of 6.8 females per house to 5.7 per house two weeks after the intervention. Results from the control area followed a steady downward trend over the 22 week post-intervention period, but never fell below two females per house. There was a gradual convergence of densities in the two areas from the 4 th post-intervention week onwards. There were statistically significant differences at the 95% level between populations in the two areas at every assessment point (p<0.001 up to week 16, p<0.05 up to week 22). Over the 22-week postintervention period more than twice as many females per house were collected in the control area (3.5, 95% CI ) as in the intervention area (1.4, 95% CI ; overall p<0.001). Parity was not determined during the baseline surveys. Over the post-intervention period, parity was determined in 635 females from the intervention area and 1,492 females from the control area (Table 6). The overall parous rates in the two areas were 32.4% and 42.0%, respectively, and this overall difference was statistically significant at the 95% level (p<0.01). However these differences were not consistent over time, and at only two assessment points were p-values less than The differences were greatest in the period between four and nine weeks after jar cover distribution. Exceptionally, at week 20, the parous rate was higher in the intervention area than in the control area. The results of the contact bioassays on female Ae. aegypti are summarized in Table 7. With a 10-min exposure, the mean KD 60 and 24-h mortality remained above 80% for 12 weeks but declined to lower levels thereafter. DISCUSSION Following jar cover introduction, there was a large and rapid decrease in the standing crop of Aedes pupae in covered water jars. This was most pronounced early in the project. Over the 22-week follow-up period, there was an approximately five-fold reduction in the mean number of pupae per house in jars in the intervention area (6.6 pupae/ house) compared to the baseline (29.6 pupae/house). This contrasts with the control area where there was an early, large increase in the mean number of pupae per house in jars followed by a progressive decline. This decline is thought to have been largely attributable to seasonal changes in Ae. aegypti abundance, but there may have been other influencing factors such as the removal of all pupae from water jars in a 10% sample of houses every two to three weeks, and seasonal changes in water collection and management practices by householders. However, these would also have exerted similar influences on mosquito populations in the intervention villages. Reductions in control area entomological indices have been observed in other recently completed assessments of insecticidal materials for dengue vector control (Kroeger et al. 2006, Lenhart et al. 2008). The mechanism of these decreases deserves investigation. Entomological indices were largely independent of rainfall, possibly because, in the absence of rain, alternative water sources are used by householders, e.g., open wells, rivers. We assume a complex interaction of factors including

6 338 Journal of Vector Ecology December 2008 Table 3. Average number of Aedes pupae per house in intervention and control areas, by container category. Average number of Aedes pupae per house Weeks No. of houses surveyed Water jars Other containers Total Intervention Control Intervention Control Intervention Control Intervention Control Baseline (combined) no data. Table 4. The number of jars examined, the number properly covered, and the percentage properly covered in the intervention area at each assessment point of the trial. No. of weeks post-intervention Total Total no. of jars examined ,689 No. of jars properly covered ,247 % of jars properly covered

7 Vol. 33, no. 2 Journal of Vector Ecology 339 Table 5. Mean household population, mean number of recovered pupae, and derived pupae per person indices in each intervention and control village. Village Mean household population Baseline (combined BL1 and BL2) Pupae/ house PPI Pupae/ house 13 weeks 22 weeks Intervention Ampil Leu Ampil Krom Chuanghouk Vealsbov Romeas Sya Total Control Andoung Chraoh Sralau Chheung Kouk Krala Banteay Thma Roliek Total PPI Pupae/ house PPI Table 6. Parous rates in indoor resting female Ae. aegypti in the intervention and control areas. Weeks post-intervention Total Intervention Area Number dissected Number parous Parous rate (%) Number dissected Control Number parous Area Parous rate (%) Chi squared p-values* 0.60 < <0.01 *Chi-square-derived p-values of the hypothesis that the observed frequencies arose from the same group.

8 340 Journal of Vector Ecology December 2008 Table 7. Bioassays with Ae. aegypti exposed for 10 min to netting material from jar covers after successive 4-week intervals of field use. Weeks of jar cover use Number of jar covers tested Total number of replicates Number of mosquitoes exposed Mean KD 10 (%) Mean KD 60 (%) Mean 24 hour mortality (%) (KD knockdown min after exposure). the number of rainy days, frequency of intermittent rain, temperature, humidity, and availability of blood meals contribute to fluctuations in abundance of Ae. aegypti. In the intervention villages, the leaching of small amounts of deltamethrin into the water of covered jars will undoubtedly have resulted in mortality among larvae, especially early instars that hatched from the crop of viable eggs that were deposited in the jars before they were covered. Such larvicidal effects had been observed during preliminary testing of covers prior to the study (Socheat et al. 2004). While this could be considered an additional benefit in the short term, by exerting selection pressure simultaneously on immature stages and adults, such characteristics of a pyrethroid-formulated LN cover would likely enhance the development of insecticide resistance if used widely as a public health tool. We considered the possibility that covered water jars might divert gravid Aedes mosquitoes to other accessible container habitats for oviposition. However, there was no evidence of this occurring. On the contrary, in the intervention area the standing crop of Aedes pupae per house in containers other than water jars fell 25% from a baseline of 5.7 to 4.3 after 13 weeks, whereas in the control area there was a corresponding increase of 56%, from 6.6 to This disparity between areas may possibly be explained by the lethality of insecticide treated covers on gravid females. If proven, it may represent a significant advantage over untreated covers which (a) allow escape of emerging adult mosquitoes when covers are removed, and (b) may possibly encourage migration to alternative larval habitats. At the end of the study, by which time the insecticidal efficacy of the jar covers had diminished and seasonal rainfall was declining, there were few pupae in containers other than water jars in either area. In both areas there was an overall decline in the density of indoor resting female Ae. aegypti, but as was the case with the standing crop of pupae, the differences between the two areas was largest during the first few weeks following the introduction of the insecticidal covers. The decrease in indoor resting densities of females in the intervention area was 54% lower than in the control area and was largely consistent with trends in the standing crops of pupae in the two areas. However both the intervention and control areas (1.4 and 3.5 females per house, respectively) showed declines when compared to the baseline (5.6 and 6.8 per house, respectively). Nevertheless the clear difference between the two areas after the intervention strongly suggests that the insecticidal jar covers had an effect on female Ae. aegypti densities. Assessment of pupae per person indices was not a primary aim of the study and population data of the area was not recorded. Our utilization of Provincial Health Department population data results in indices that can only be regarded as approximate. Focks et al. (2000) used such data to estimate transmission thresholds at different ambient temperatures and given population antibody seroprevalence. A 2.5-fold reduction from baseline levels to 2.8 pupae per person was observed in the intervention area. This would be insufficient to prevent transmission according to this model except in situations of low temperature/high antibody seroprevalence. After 22 weeks of follow-up, 77.1% of intervention area pupae were recovered from water storage jars. The importance of a single container type from which the majority of adult mosquitoes emerge presents an opportunity for Ae. aegypti suppression below threshold levels if these containers are adequately controlled (WHO 2003). It is similar to pupal distribution observed in other settings (WHO 2006). The observed reductions in entomological indices were comparable with those recently reported following studies of alternative interventions for dengue vector control. These include the addition of guppy fish to water jars in Cambodia with79% less infestation than controls in targeted containers after one year (Chang et al. 2008), the use of insecticidal curtains and container covers in Mexico and Venezuela with 88% and 71% reduction in Breteau index compared to baseline after twelve and nine months, respectively (Kroeger et al. 2006), and the use of insecticide-treated bednets in Haiti with a 58% reduction in Breteau index compared with control after five months (Lenhart et al. 2008). In addition to reducing the immature and adult populations, the results showed a reduction in the parous rate of adult females resting indoors, an indicator of daily survival and an important and sensitive parameter influencing vectorial capacity. In contrast to indoor residual spraying for malaria vector control and the use of insecticidal bednets whereby blood-feeding females are preferentially targeted, this intervention was designed to

9 Vol. 33, no. 2 Journal of Vector Ecology 341 also kill newly-emerged, nulliparous Ae. aegypti. We would therefore expect the effect on parous rate, which is normally a robust indicator of adulticide performance, to be relatively small when compared to the magnitude of decreased parity described elsewhere. The covers were accepted by all households, correct utilization rates were very high (88.0%), and most reported they would be willing to buy covers if they were affordable and available at retail outlets. Further development of this targeted container strategy in Cambodia, and elsewhere, appears warranted. However, improvements in technology that would extend the duration of insecticidal effectiveness of LN netting in the harsh outdoor tropical environment are likely to be needed for the development of cost-effective public health applications. The current cost of $1.20 per cover would likely prove prohibitive to most rural Cambodians and would have to be reduced or subsidized before wide-scale use could be considered. Safe alternatives to pyrethroids as active ingredients should also be sought if incipient insecticide resistance is to be avoided as novel applications of LNs are developed and used. Acknowledgments We thank Vestergaard-Frandsen for their contributions to the design of the jar covers and for providing the covers free of charge for the purposes of the study. The World Bank funded the study under the dengue control project through the Ministry of Health of Cambodia. Without the interest and enthusiastic participation of the residents of Ampil commune, this study would not have taken place, and for this the authors are deeply grateful. The project was approved by the National Ethics Committee for Health Research, Ministry of Health, Cambodia. The views expressed in this article are those of the authors and do not necessarily reflect the views or stated policies of the World Health Organization and the Ministry of Health of Cambodia. REFERENCES CITED Chang, M. S., T. Setha, J. Nealon, D. Socheat, N. Chantha, and M.B. Nathan Community-based use of the larvivorous fish Poecilia reticulata to control the dengue vector, Aedes aegypti in domestic water storage containers in rural Cambodia. J. Vector Ecol. 33: Clark, G., H. Seda, and D.J. Gubler Use of CDC backpack aspirator for surveillance of Aedes aegypti in San Juan Puerto Rico. J. Am. Mosq. Contr. Assoc. 10: DeRoeck, D., J. Deen, and J. D. Clemens Policymakers views on dengue fever/dengue haemorrhagic fever and the need for dengue vaccines in four Southeast Asian countries. Vaccine 22: Detinova, T.S Age-grouping methods in Diptera of medical importance with special reference to some vectors of malaria. Monogr. Ser. Wld. Hlth. Org. 47: Focks, D.A., R.J Brenner, J. Hayes, and E. Daniels Transmission thresholds for dengue in terms of Aedes aegypti pupae per person with discussion of their utility in source reduction efforts. Am. J. Trop. Med. Hyg. 62: Gubler, D.J Dengue and dengue hemorrhagic fever. Clin. Microbiol. Rev. 11: Gubler, D.J Epidemic dengue/dengue hemorrhagic fever as a public health, social and economic problem in the 21 st century. Trends Microbiol. 10: Guha-Sapir, D. and B. Schimmer Dengue fever: new paradigms for a changing epidemiology. Emerg. Themes Epidemiol. 2: 1. Kittayapong, P. and D. Strickman Three simple devices for preventing development of Aedes aegypti larvae in water jars. Am. J. Trop. Med. Hyg. 49: Kroeger, A., A. Lenhart, M. Ochoa, E. Villegas, M. Levy, N. Alexander, and P.J. McCall Effective control of dengue vectors with curtains and water container covers treated with insecticide in Mexico and Venezuela; cluster randomised trials. Brit. Med. J. 332: Lenhart, A, N. Orelus, R. Maskill, N. Alexamder, T. Streit, and P.J. McCall Insecticide-treated bednets to control dengue vectors: preliminary evidence from a controlled trial in Haiti. Trop. Med. Int. Hlth. 13: Mattingly, P.F Contributions to the mosquito fauna of Southeast Asia. XII. Illustrated keys to the genera of mosquitoes (Diptera, Culicidae). Contrib. Am. Entomol. Inst. (Ann Arbor). 7: Socheat, D., N. Chanta, T. Setha, S. Hoyer, M.S. Chang, and M. Nathan The development and testing of water storage jar covers in Cambodia. Dengue Bull. 28: Suaya, J., D. Shepard, M.S. Chang, M. Caram, S. Hoyer, D. Socheat, N. Chantha, and M. Nathan Costeffectiveness of annual targeted larviciding campaigns in Cambodia against the dengue vector Aedes aegypti. Trop. Med. Int. Hlth. 12: World Health Organization International Programme on Chemical Safety: environmental health criteria 97 deltamethrin. Geneva. World Health Organization A review of entomological sampling methods and indicators for dengue vectors. Geneva. World Health Organization Multicountry study of Aedes aegypti pupal productivity survey methodology: findings and recommendations. Geneva.