Transmission thresholds and pupal/demographic surveys in Yogyakarta, Indonesia for developing a dengue control strategy based on targeting epidemiologically significant types of water-holding containers Dana A. Focksa , Michael J. Bangsb, Cole Churchc, Mohammad Juffried and Sustriayu Nalime a b
Infectious Disease Analysis, P .O. Box 12852, Gainesville, FL, USA
Freeport Indonesia, Public Health and Malaria Control-Kuala Kencana, P .O. Box 616, Cairns 4870 Australia c d
Ouachita Parish Mosquito Abatement District, Monroe, LA, USA
Department of Pediatrics, Gadjah Mada University, Yogyakarta, Indonesia
e
Tahija Dengue Project, Jl. Pandega Sakti 159 Kaliurang Street KM 6, 2, Yogyakarta 55283, Indonesia
Abstract All water-holding containers (ca. 3000) associated with approximately 320 residences in Yogyakarta, Indonesia, were examined for the presence of Aedes aegypti (L.), Aedes albopictus Skuse, and Culex quinquefasciatus Say pupae in four replicate surveys conducted during two dry seasons (1996 and 1998) and two wet seasons (1997 and 1999). Less than 6% of these receptacles had pupae. Ae. aegypti pupae collected were ten times more than Ae. albopictus (ca. 1600 vs. 160 respectively); Cx. quinquefasciatus was rarely encountered. From a dengue perspective, the epidemiological significance of a particular type of container is a function of the number of Ae. aegypti pupae per person – calculated simply as the ratio of total number of Ae. aegypti pupae recovered in that type and the number of residents, ca. 2800. Overall, there was an average of 0.57 Ae. aegypti pupae per person. Assuming an overall herd immunity of 33% and an average temperature of 29 °C, we estimate the transmission threshold in Yogyakarta to be approximately 0.43 Ae. aegypti pupae per person. By eliminating mosquito production in two common household containers – wells and used tyres the number of Ae. aegypti pupae would be reduced from 0.57 to 0.29 per person, below our estimate of the transmission threshold. An assessment of the effectiveness of this strategy is currently being conducted in a multi-year study in Yogyakarta using an insect growth regulator (IGR) for mosquito control. Keywords: Dengue; Epidemiology; Prevention and control; Risk assessment; Targeted source reduction and control; Sustainable; Community-based; Transmission threshold.
E-mail: DAFocks@ID-Analysis.com; Fax: +1-352-372-1838 83
Dengue Bulletin – Volume 31, 2007
Targeted source reduction/control strategy for dengue control
Introduction Dengue in Indonesia Dengue and dengue haemorrhagic fever (DF/ DHF) was first observed in Indonesia in 1968 in Surabaya and Jakarta, the two largest metropolitan cities on the island of Java.[1] While this initial epidemic involved less than 60 cases, the case-fatality rate exceeded 40%. Since then the incidence of DF/DHF has increased dramatically in Indonesia and has spread geographically to all regions of the country. The DHF incidence fluctuates monthly and typically reaches its peak in December and January every year, except in large cities such as Jakarta, Bandung and Surabaya, where the highest incidence is reported in April and May. Currently, dengue is the eighth leading cause of hospitalization among Indonesian children.[2] Similar trends of progressively larger epidemics interspersed with quieter, inter-epidemic years in the neighbouring countries of Cambodia, Myanmar, Laos, Thailand and Viet Nam reflect waxing and waning human population “herd” immunity, urbanization, the movement of people, and the influence of weather anomalies associated with El Niño/Southern Oscillation (ENSO) events. In the face of the most significant ENSO event of the century, 1998 witnessed the largest epidemic on record in Indonesia with 72 133 reported cases and 1414 deaths; the case-fatality rate (CFR) in this epidemic was 2.0%, reflecting several decades of improved clinical management.[2] In 2001, the total number of cases (DF and DHF) and deaths reported were almost 20 000 and 180 respectively.[3] In early 2004, DF/DHF made a dramatic rebound with over 58 000 cases and 658 deaths reported in the first four months.
conducted an external review of the dengue/ dengue haemorrhagic fever prevention and control programme of Indonesia in June 2000.[2] The following brief history of the Indonesian dengue prevention and control programme reflects this report. The Indonesian Ministry of Health has considerably modified its strategies to control dengue over the past three decades. Initially, adult control using perifocal space spraying of insecticides with portable and vehicle-mounted thermal fogging and ultra low volume (ULV) machines was the government’s recommended method and response for most areas. The protocol specified treatment within a 100-metre radius of reported DHF cases. In the 1980s, the strategy changed to include the addition of extensive larviciding using temephos (1% sand granules). The policy was to treat all breeding sites in dengue-endemic urban areas a single time each year, timed ideally to precede the onset of the transmission season. The Ministry of Health subsequently modified this strategy to target only those urban areas reporting DHF for three consecutive years, wherein retreatments were scheduled with a frequency of three months. This selective larviciding programme was implemented between 1986 and 1991. Beginning in 1992 and continuing until the present, the national strategic emphasis has been larval control involving community efforts, health education and intersectoral coordination. Currently, national efforts have focused on organizing working groups at the village level under the general guidance of local health centre personnel. This programme, called Bulan Gerakan (or 3M), emphasizes intensive health education using mass media, women’s groups and schoolchildren, community-based breeding source reduction, and door-to-door house inspections to monitor for larvae, and to clean containers and apply temephos as necessary. An important member of the 3M programme is the Family Welfare Education Women’s Movement (Pendidikan Kesejahteraan Dengue Bulletin – Volume 31, 2007
Past dengue control efforts The World Health Organization’s Regional Office for South-East Asia (WHO/SEARO) 84
Targeted source reduction/control strategy for dengue control
Keluarga or PKK). The role of the PKK is education of the house-owner about larval inspections, methods to store water safely, and the elimination or cleaning of breeding containers. An additional role of the PKK involves community-based group education and monitoring programme results. Finally, authorities developed health education programmes for elementary schoolchildren and for use with the mass media. Unfortunately, with only a few notable exceptions, these efforts have not been successful in controlling dengue.[2] However, a pilot project in the city of Purwokerto, aided with funding from Rotary International has shown promise. The project organized a strategy of community partnership based on dasa wisma (ten houses) in several villages in the area. The results were encouraging to the extent that Rotary and others funded an extension to this effort that targeted 11 major urban areas in Indonesia. Patterned after the Purwokerto project, it included educational programmes for the public and medical personnel. The extended project had the endorsement and support of the Indonesian government, WHO, the US Public Health Service, and importantly, mayors’ offices. While funding was available and key public officials remained prominently associated with the project publically, e.g., the mayor’s wife, the effort did reduce dengue cases (Sustriayu Nalim, personal communication).
that are particularly important by some recent developments outlined below. This strategy targets only the most epidemiologically important types of breeding containers. We measure the epidemiological importance of each type of container in the environment using the statistic the total standing crop of Ae. aegypti (L.) pupae per hectare or per person associated with each particular type.
Results from prior research Transmission models Recently, there has been a movement in the epidemiological community to recognize the pervasive influence of the environment and climate on various vector-borne diseases. The efforts of Martens et al.[4] and Patz et al.,[5] for example, have documented substantial ties of disease activity to environmental features and climate trends for dengue, schistosomiasis and malaria. The work of Bouma et al.[6] establishing statistical relationships between weather anomalies associated with El Niño and malaria in Colombia is especially encouraging in the context of developing early warning/mitigation systems for weather-driven infectious diseases. Recognizing these ties, mathematical epidemiologists and public health specialists are beginning to construct disease models incorporating environment and climate parameters. A number of researchers have recently been involved in these types of studies on the dengue system and have developed simulation models and estimates of transmission thresholds.[7,8,9,10,11] These results have had a degree of success and are being evaluated by the public health community. The algorithms of the dengue models[7,9] take into account key factors known to influence dengue epidemiology; the result is a software tool used by researchers and public health practitioners that is orientated toward site-specific simulation. 85
Proposed targeted source reduction and control efforts The goal of the present work is to build on the foundation of the 3M and PKK programmes by reducing the number of types of breeding containers that must be controlled or eliminated to only a select few that are responsible for most of the adult vector production. We believe it is possible to estimate the degree of reduction required and to identify the types of containers Dengue Bulletin – Volume 31, 2007
Targeted source reduction/control strategy for dengue control
Validation studies compared model output with field and laboratory observations at sites in Asia and the Americas.[8] Funding from WHO and the US National Institutes of Health (NIH) has led to ongoing evaluations of transmission thresholds derived from the models in Viet Nam and Peru respectively. The hope is that the published thresholds[11] can be used in tropical locations to predict disease vulnerability, assess control measures and provide guidance in targeting the especially important classes of breeding containers. The recent development of the pupal/demographic survey, coupled with estimates of transmission thresholds, make the results of simulation studies with the dengue models available to operational control programmes in the developing world.[10,11,12] The models are currently being re-written and extended through funding from the Tahija Family Foundation (Jakarta) and the Innovative Vector Control Consortium (IVCC) funded by the Bill and Melinda Gates Foundation (the citation for IVCC/dengue is: Hemingway J, Beaty BJ, Rowland M, Scott TW, Sharp BL. The Innovative Vector Control Consortium: Improved Control of mosquito-borne diseases. Trends in Parasitology 2006 (22): 308-312.) Transmission thresholds and the pupal/demographic survey The expense and ineffectiveness of drift-based insecticide aerosols to either prevent or control dengue epidemics has led to suppression strategies based on eliminating larval breeding sites.[13] With the notable, albeit short-lived, exceptions of Cuba and Singapore, these sourcereduction efforts have met with little documented success. Public health workers attribute failure to two factors: inadequate participation of the communities, and a strategy that entailed destruction or treatment of virtually every breeding container in the environment. The transmission thresholds for dengue based on the standing crop of Ae. aegypti pupae per 86
person[11] were developed for use in the assessment of risk of transmission and to provide targets for the actual degree of suppression by type of breeding container required to prevent or eliminate transmission in source-reduction programmes. When coupled with field observations from pupal/demographic surveys (as reported herein for Yogyakarta), it is possible for the first time for control specialists to know how important the various types of containers in the environment are in terms of contributing to the transmission threshold.[10,11] This strategy of concentrating on only the types of containers most responsible for the majority of adult vector production and hence transmission, e.g. outdoor drums and tyres vs typically low-producing indoor vases and domestic containers, was recently evaluated in a 9-country study in the Americas and SE Asia with WHO/TDR funding (citation is: Focks DA, Alexander N. Multicountry study of Aedes aegypti pupal productivity survey methodology. Findings and recommendations. 2006. TDR/IRM/Den/06.1)). The WHO’s Tropical Diseases Research (TDR) programme has commissioned a review article on the current state of the science for entomological surveying for dengue risk assessment and control.[14] Central in this document are the concepts of the pupal/demographic survey, transmission thresholds and targeted source reduction and control of especially productive containers. There is a growing recognition that adherence to the current strategy of attempting to eliminate or control all containers, irrespective of productivity or time of the year, is doomed to continued failure.[10] Dengue early warning system (EWS) In the last several years, there has been a call by the directors of national anti-dengue programmes in Indonesia, Thailand and Viet Nam for the operational need of an early warning system (EWS) that would provide sufficient lead time (1 to 3 months) to permit mobilization of Dengue Bulletin – Volume 31, 2007
Targeted source reduction/control strategy for dengue control
control operations. In response to this call, preliminary EWSs for Yogyakarta and Bangkok were developed; they are based on logistic regression analysis.[15] The predictor variables are sea surface temperature (SST) anomalies over the tropical Pacific and monthly cases of dengue in each city. The predicted variable is the probability of an epidemic year forecast 1 to 3 months before the peak transmission season. The Java EWS was sufficiently accurate to be operational. The Yogyakarta EWS gave perfect 1- and 2-month forecasts; the 3-month forecast incorrectly classified one year in the 14-year period of record. Biological control agents A very promising recent measure is the application of biological agents like Mesocyclops, Micronecta and larvivorous fish. In Viet Nam, Mesocyclops, a tiny copepod crustacean, have been found to be good predators of Ae. aegypti larvae. Since February 1993, Mesocyclops have been released into water containers of 400 houses of a hamlet in My Van district, Hai Hung province. In March 1994, this measure was supplemented by a campaign to eliminate discarded containers (which are too small for Mesocyclops to survive), and education about how to maintain adequate populations of Mesocyclops in domestic water containers. After 17 months, Ae. aegypti had been completely eliminated and this result has been sustained until today. Since July 1995, mobilization of the community for peridomestic hygiene and use of Mesocyclops to control Aedes larvae was implemented for 1600 houses in one commune in Thuong Tin district, Ha Tay province. Working together with the network of health collaborators, primary-school pupils, local authorities and health staff, and using the system of local communication, health education campaigns were organized in order to improve people’s knowledge and to mobilize every member of the community. Almost all big water Dengue Bulletin – Volume 31, 2007
containers in the commune received Mesocyclops or larvivorous fish; it is now very difficult to find discarded containers, and mosquito indices have been reduced almost to zero. From 1996, this model has been extended to three provinces, Nam Ha, Hai Hung and Hai Phong in northern Viet Nam.[16]
Goals of the present work The purpose of this report is to provide an analysis of four annual pupal/demographic surveys conducted in Yogyakarta between 19961999 highlighting the epidemiological significance of the twenty-some types of Ae. aegypti-breeding containers in the environment. From this analysis, and using the estimates of transmission thresholds for dengue, we propose to develop a targeted source-reduction/control strategy for Yogyakarta that will require substantially less effort than the traditional community-based efforts without targeting where the goal is to control or eliminate all containers irrespective of their contribution to the adult population of Ae. aegypti.
Methods Study site Yogyakarta, a city of over 520 000 people, is the provincial capital of Yogyakarta, located in central Java. The province is divided into administrative districts called kabupatens with each district divided into progressively smaller units beginning with sub-districts called kecamatans, and these, in turn, divided into kelurahans, and further divided into rukun warga (RW), and finally into rukun tetangga (RT), the smallest administrative unit composed of approximately 50-80 families each. The study site was located in kecamatan Gondokusuman, within Yogyakarta city. The actual study area covered approximately 6.34 ha, from which 87
Targeted source reduction/control strategy for dengue control
323 houses were selected and subsequently sampled over four time periods. Disease and demographic data were derived from a recent study conducted in the same area.[17] An average of five houses were chosen from each RW where a dengue case had occurred and included 64 RWs distributed among five kelurahans (Kota Baru = 4 RWs; Terban = 11 RWs; Baciro = 21 RWs; Klitren = 16 RWs, and Demangan = 12 RWs). The location of each sampled house was provided coordinates in 1999 using a hand-held geographical positioning device (Magellan GPS 300tm, San Dimas, CA). Demography data were taken from each house, including number of permanent residents, house size and land area. Climatological data (daily rainfall and maximum/ minimum ambient temperatures) were obtained from the local Meteorology and Geophysics Agency (Station Bulaksumar, University of Gadjah Mada).
abundance by type of container. With the aid of a flashlight, hand-held fine-meshed netting devices and pipettes were used to remove all pupae from the container with captured specimens placed in white trays for easy observation. While in the field, all pupae would be first immobilized using hot water supplied from a thermos and immediately placed in labelled plastic bags containing 70% ethyl alcohol and sealed. Bag labels included house number, container type, location (indoor or outdoor) and number of pupae collected. Additionally, all information, including lot and house size, was hand-recorded in a field logbook and later transcribed into a computerized database. The only water-holding containers not surveyed in this study were residential wells, which are known to harbour Ae. aegypti and Culex quinquefasciatus Say larvae.[18]
Specimen identification Preserved pupae were returned to the laboratory for identification. Using an illustrated key developed for this purpose, pupae were examined using a stereomicroscope and easily identified for species and sex.[19] For the purposes of this study, pupae were identified as either Ae. aegypti, Aedes albopictus Skuse, or Cx. quinquefasciatus, with all other pupae identified only to genera.
The pupal/demographic survey Each premise was sampled consecutively for immature stages of container-breeding mosquitoes during four different time periods between 1996 and 1999, with two sample surveys during wet seasons and two during the dry periods of the year.# A team of three or four people would visit each house and carefully inspect the inside and around the outside perimeter all natural and artificial containers for preimaginal stages of mosquitoes. The number and type of containers present at each house were recorded as well as the number of permanent residents. The presence or absence of mosquito larvae in each container was recorded without regard to the number present. Collections concentrated on quantifying pupal Collection dates: Dry season: 8–22 May 1996, Wet: 23 January–7 February 1997, Dry: 15–30 September 1998, and Wet: 30 March–19 April 1999. #
Results With very few exceptions, the four pupal/ demographic surveys conducted in the kecamatan Gondokusuman returned to the same ca. 316 houses each year. The residents associated with these houses numbered approximately 2800 (Table 1). Each exhaustive survey collected pupae from the ca. 3000 water-filled containers associated with the study houses. Approximately 5.5% of the containers were positive for one or more pupae of Ae. Dengue Bulletin – Volume 31, 2007
88
Targeted source reduction/control strategy for dengue control
Table 1: Summary statistics for pupal/demographic surveys conducted in Yogyakarta, Indonesia, between 1996 and 1999
a
Total precipitation (mm) two months prior to survey and average temperature during survey (°C)
Figure 1: Plot of 30- and 60-day rainfall accumulations before surveys and the average of daily average temperature (°C) in Yogyakarta, Indonesia [Approximate dates of surveys are indicated by vertical, downward-pointing arrows. Total precipitation during the two months prior to each survey and the average temperature during survey are presented in Table 1] 30.0 29.5 29.0 Avg. monthly temperature 30-day accumulation 60-day accumulation
1,400 1,200 1,000 800 600 400 200 0
28.5 28.0 27.5 27.0 26.5 26.0 25.5 25.0
Jan-96
Jan-97
Jan-98
Jan-99
aegypti, Ae. albopictus or Cx. quinquefasciatus. Approximately ten times more Ae. aegypti pupae were recovered than Ae. albopictus (ca. 1600 vs 160); Cx. quinquefasciatus was rarely encountered in the sites being examined. The average temperatures during the dry- and wetDengue Bulletin – Volume 31, 2007
seasons surveys were not substantially different, averaging 28.0 °C and 27.2 °C respectively (Figure 1). However, the accumulated rainfall for the 60 days prior to each dry-season survey was only about 10% of the wet-season accumulations (ca. 84 vs. 794 mm). 89
Rainfall totals for 1 and 2 months
Avg. temperature
Targeted source reduction/control strategy for dengue control
Types and numbers of water-filled containers as a function of season and location A total of 71 different types of water-filled containers were observed during the four
surveys; many of these types, however, were only seen once or at most only a few times during the surveys. The 3, 5 and 11 most common types of containers accounted for >60%, >75% and 90% of all water-filled containers (Table 2). Some types were
Table 2: Most frequently encountered types of containers during surveys conducted during the dry and wet seasons in Yogyakarta, Indonesia S. No. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. a b
Type Bird watering dish Bucket Storage in water closet Water container (large) Plastic water container Water container (large) Clay water container Refrigerator water pan Padasan Flower pot Flower vase Bottle Plant axil Tin can Tyre Drinking glass Pool, pond, tank Bowl Drum Fish pond Pan Cover or lid Plate, dish Clay water pot (small) a b
Dry 670 549 538 251 196 112 91 58 41 19 45 34 2 17 10 4 22 3 11 16 7 4 2 6
Wet 681 619 549 220 232 152 90 58 56 78 41 44 77 53 47 51 22 30 16 4 8 11 12 8
Mean 675 584 543 235 214 132 91 58 48 48 43 39 39 35 28 27 22 17 14 10 8 7 7 7
Proportion of mean 0.23 0.20 0.18 0.08 0.07 0.04 0.03 0.02 0.02 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00
Accumulation 0.23 0.42 0.61 0.68 0.76 0.80 0.83 0.85 0.87 0.88 0.90 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.97 0.97 0.98 0.98 0.98 0.98
Indonesian: Bak mandi Indonesian: Bak air Dengue Bulletin – Volume 31, 2007
90
Targeted source reduction/control strategy for dengue control
exclusively found either indoors or outdoors and others could be found in both locations. By location, 34 container types were observed indoors and 66 types outdoors. Rainfall accumulations preceding the surveys influenced the number of types of containers in the environment (Table 2). During the two dryseason surveys, a total of 26 and 40 different types of containers were observed indoors and outdoors respectively; the corresponding numbers for the two wet-season surveys were 29 and 60 different types respectively. However, somewhat surprisingly, the average number of water-filled containers was largely independent of the season of the surveys (Table 1); the average number of water-filled containers in the environment was 2732 and 2770 for the dryand wet-season surveys respectively. Numerically, the most common types of containers observed were (in descending
frequency) bird watering dishes, buckets (ember), water storage container in water closet (bak mandi), large water tanks (bak air), plastic water containers (tempayan) and large water storage containers. These particular containers accounted for ca. 80% of all water-filled containers. Thirtyfive of the 71 types observed were never found positive for the pupae of any species in any of the surveys. Table 2 provides a list of the 24 most common container types in descending order of abundance. Of the nine most common types, accounting for 87% of all containers, there were no significant changes in abundance as a function of season, suggesting that the most common types of containers are not rain-filled but filled manually. Those container types listed in Table 2 that are more commonly found in the wet season are also those that are located 0primarily outside of the residence, e.g. flowerpots, plant axils, tin cans and tyres.
Figure 2: Frequency of being positive for larvae in the four surveys conducted between 1996 and 1999 in the 18 most common types of water containers
0.50 0.45 0.40 0.35
Frequency
0.30 0.25 0.20 0.15 0.10 0.05 0.00
Dengue Bulletin – Volume 31, 2007
T ow yre e o rp Cl ay rag ot e w in at e W W rc at on C er co tain nt er ai ne r( lg) B ow Fl ow l er va s Ti e n ca Pa n Re D d a r fri ge inki san ng ra to gla r Po wat ss e ol ,p rp on an d, Co tank nt ai ne Bu r ck e Bi Plan t rd ta w at xil er (sm ) W at Bot er t ta le nk (lg ) St Fl
91
Targeted source reduction/control strategy for dengue control
Prevalence of larvae by season and container type The average proportion of containers with larvae in the dry and wet seasons was 0.128 and 0.173 respectively. The prevalence of larvae in the 18 most common types of containers is presented in Figure 2. Recent evaluations of the utility of the traditional Stegomyia indices (the House, Container and Breteau indices, and various related derivations) concluded that: (i) they are of only limited operational value
for measuring the entomological impact of larval control interventions; (ii) that they are not proxies for adult vector abundance; and (iii) are not useful in the development of targeted control strategies.[14] Neither are they useful for assessing transmission risk because they do not take into consideration the epidemiologically important variables, including adult vector and human abundance, temperature and herd immunity in the human population. For these reasons, we will confine our analysis to the pupal data.
Table 3: Summary of numbers of Ae. aegypti and Ae. albopictus pupae collected and containers positive for the same by location (indoors or outdoors) based on survey year and season
92
Dengue Bulletin – Volume 31, 2007
Targeted source reduction/control strategy for dengue control
Table 4: Average total numbers of pupae collected as a function of season and location
Figure 3: Total number of Ae. aegypti pupae recovered during the four surveys in Yogyakarta by type of container and location (indoors or outdoors) [Containers are distributed across the horizontal axis in descending order of total standing crop in that type. The container types shown are the 18 most productive classes of containers] 3500 In 3000 Out
2500
2000
1500
1000
500
0 St W ora g at er e in co nt WC ai ne Cl r( ay lg) w at er Ty r c Bi ont e rd ai n w at er er (sm Bu ) Fl cke ow t er p o Tr as t h ca Ti n n c Co an nt Fl ain ow er Re er fri va ge F se i ra sh to p r o Po wat nd er ol ,p p on an d, ta nk Bo w l D D ru rin m k W ing at gla er ta ss nk (lg )
Standing crop of pupae as a function of the number of containers, container type and season The average proportion of containers with pupae in the dry and wet seasons was 0.047 and 0.061 Dengue Bulletin – Volume 31, 2007
respectively. Of the pupae collected in all surveys, independent of season, Ae. aegypti accounted for 89.8%, Ae. albopictus 9.0% and Cx. quinquefasciatus 1.2% of the total (Table 1). A summary of pupal collection as a function of survey, container location, species of mosquito 93
Targeted source reduction/control strategy for dengue control
and season (Table 3) indicates that Ae. aegypti can be found both in indoor and outdoor containers, and that the number of pupae coming from outdoor containers increases during the rainy season; the standing crop of Ae. aegypti indoors is remarkably constant and independent of season. Outdoor breeding accounts for virtually all additional production during the rainy season. In contrast, Ae. albopictus pupae are essentially found only outdoors and in rain-filled containers. It is therefore not surprising that the average standing crop of Ae. aegypti is somewhat less a function of rainfall (dry season average– 1312 vs wet season–1862, an increase of ca. 42%) than Ae. albopictus (dry season average – 38 vs 281 in wet season, an increase of 640%). Table 4 provides an average of total pupal collections for the two seasons by mosquito and location. While the numbers collected are low and preclude confident statements, Cx. quinquefasciatus immatures in these surveys were only found outdoors and their abundance seems to be independent of rainfall. Aedes albopictus immature abundance is a strong function of rainfall and they are only found outdoors. The total numbers of Ae. aegypti pupae recovered in the four surveys are combined to provide the best estimate of production by container type (Figure 3). The total numbers of Ae. aegypti pupae in each of the types of containers highlights an important point: the epidemiological importance of a class of containers is not simply a function of the abundance of the containers, but rather the product of the containers’ abundance and average standing crop of pupae (Table 5). The water storage containers located in water closets (bak mandi) account for 22% of all containers but 50% of all Ae. aegypti pupae. The classes “large water container” (bak air) and “tyre” account for 6% and 1% of all containers, yet they are responsible for 13% and 6% of all pupae respectively. The large water tank (12% of all containers) contributes essentially nothing to Ae. aegypti production. 94 a b
Table 5: The frequency of containers by type and the proportion of all Ae. aegypti pupae associated with that type [Container types are sorted in descending order of Ae. aegypti pupae]
Indonesian: Bak mandi Indonesian: Bak air
The total number of Ae. albopictus pupae recovered in each survey was correlated with the number of water-holding containers present (0.84). The likely explanation is that Ae. albopictus largely breeds in outdoor containers, which are substantially filled by rainwater; there is seasonality in abundance as a function of the number of containers (Table 4). In contrast, the total number of Ae. aegypti pupae recovered in each survey was independent of the number of water-holding containers present (correlation: –0.10). This is a bit unexpected in so far as Ae. aegypti production in outdoor containers increases substantially in the rainy season (cf. Tables 3 and 4). The total numbers of Ae. aegypti and Ae. albopictus pupae caught in each survey were only slightly correlated (0.32). Dengue Bulletin – Volume 31, 2007
Targeted source reduction/control strategy for dengue control
Figure 4: Frequency histogram of the numbers of Ae. aegypti pupae per house [Note that the frequency of the 0 to 5interval class (0.80) is off scale] 0.06 0.05 0.04 0.03 0.02 0.01 0.00 0 20 40 60 80 100 120 140 160 180 200
Ae. aegypti pupae per house
positive houses were more than three times more likely to remain positive over time than a negative premises that was to become positive over the course of a year.[21,22] These persistently productive premises were given the name key premises. Their notion was that, in government suppression programmes with limited resources, it might improve effectiveness to focus on former key premises for subsequent visits rather than use a systematic approach of visiting every house. It should be noted that the effectiveness of such a strategy would be dependent not only on the existence of a clumped distribution of productivity at the house level, but also on the persistence of productive houses between years. With this in mind, we looked into the nature of the distribution of pupae per household. The results presented in Figure 4 clearly indicate a non-linear distribution associated with the existence of key premises. The second question – the persistence or stability of such households between years – is addressed in Table 6. Here, the number of Ae. aegypti pupae recovered at each premises in each of our sequential surveys is summed; and the premises are then sorted in descending order by this sum. Table 6 presents the 28 most productive households, our key premises. While they account for only ca. 8% of all homes surveyed, they accounted for 51% of all Ae. aegypti pupae recovered. The important conclusion to be drawn from this, however, is that unusually high (or low) counts of pupae in a particular survey are not highly correlated with subsequent or previous counts; this is further corroborated by the low correlation of Ae. aegypti pupae per house (Table 7) – average correlations for surveys separated by 1, 2 and 3 years were 0.11, 0.00 and 0.13 respectively. This lack of inter-year correlation is also observed with the number of Ae. albopictus pupae per house – average correlations for surveys separated by 1, 2 and 3 years were 0.08, 0.14 and 0.00 respectively (Table 8).
Year-to-year variability in production at the household level Especially productive households – key premises In an effort to facilitate the location of positive premises and containers in Queensland, Australia, Tun-Lin et al. used various forms of statistical analyses to develop the Premise Condition Index (PCI). In essence, they were looking for proxies or surrogates to detect the presence of high-level outliers among containers and premises.[20] They found that the condition of the house, the degree of shade and tidiness of the yard, both observable without entering the house or yard, were strongly correlated with both the proportion of positive premises and the numbers of infested containers. If only premises with the highest PCI scores were surveyed, they found that the probability of finding a positive home or container was increased approximately fourfold. These particular houses represented <10% of all sites, yet they accounted for 35% of all positive containers. An important observation was that
Dengue Bulletin – Volume 31, 2007
Frequency
95
Targeted source reduction/control strategy for dengue control
Table 6: Listing of the households responsible for the highest production of Ae. aegypti pupae during all four surveys [By way of explanation, the first row of data are the results for house TE179. A total of 247 Ae. aegypti pupae were recovered at this household during the surveys conducted between 1996 and 1999; this total represents 0.039 of all production. The table indicates that about 8% of the houses were responsible for ca. 51% of all production observed. The table indicates that unusually high counts of pupae in a particular survey are not highly correlated with subsequent or previous counts]
96
Dengue Bulletin – Volume 31, 2007
Targeted source reduction/control strategy for dengue control
Table 7: Correlations between the total number of Ae. aegypti pupae per house over time [Average correlations for surveys separated by 1, 2 and 3 years are 0.11, 0.00 and 0.13 respectively]
Table 9: Correlations between the number of people per house over time [Average correlations for surveys separated by 1, 2, and 3 years are 0.79, 0.69, and 0.64, respectively]
Table 8: Correlations between the total number of Ae. albopictus pupae per house over time [Average correlations for surveys separated by 1, 2, and 3 years are 0.08, 0.14 and 0.00 respectively]
Number of water-holding containers per house There is significantly more correlation between the numbers of natural and artificial waterholding containers per house between seasons than the number of pupae per household (Table 10). Average correlations for surveys separated by 1, 2 and 3 years are 0.59, 0.57 and 0.49 respectively. Table 10: Correlations between the total number of artificial and natural water-holding containers per house over time [Average correlations for surveys separated by 1, 2, and 3 years are 0.59, 0.57, and 0.49, respectively]
Number of people per household Not surprisingly, the number of residents per household is rather consistent. Correlations of the number of household residents observed during the four surveys declined from an average of 0.79 for surveys conducted within one year of each other to an average of 0.69 for surveys separated by two years; the correlation between the two surveys conducted three years apart, i.e. those of 1996 and 1999, was 0.64 (Table 9).
Dengue Bulletin – Volume 31, 2007
97
Targeted source reduction/control strategy for dengue control
In conclusion, while there are definitely key premises in Yogyakarta (Figure 4), the lack of correlation between years (Table 6) in productivity probably makes any attempt to develop proxies (co-variates) for the rapid identification of especially productive households unlikely to succeed. The source of variability in Ae. aegypti production at the household level between years is not obviously related to rainfall, nor changes in the numbers of residents or wet containers. However, we cannot discount the possibility that the lack of correlation is due to variability in pupation on a daily basis. Because the key premise concept has potential for control, further study using the total pupation over perhaps a week is warranted.
Figure 5: Monthly number of reported cases of DHF in Yogyakarta province during 1985-2001 1,200 1,000
Cases DHF
800 600 400 200 0 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01
Year
Targeted source-reduction strategy for Yogyakarta So far, we have estimated the number of Ae. aegypti pupae per person associated with each type of container in the wet and dry seasons. Our goal in developing a targeted sourcereduction/control strategy is to identify which types of containers, if production in them were eliminated, would result in the area being below the transmission threshold. This will involve deciding which seasonal estimates, wet or dry, to use, what temperature to use, and what value for overall seroprevalence of dengue antibody to use. In all of these deliberations, we will be conservative. We already know that wells will have to be controlled – we do not presently know their production.
period. Given that most transmission occurs during the wet season (Figure 6) and that there are more pupae per person then (0.47 vs 0.67), we will use the wet season survey results.
Temperature to use in estimating the transmission threshold For temperature, we could use the average mean temperature for the first five months of the year, 27.5 °C when ca. 60% of all cases Figure 6: Average monthly rainfall, temperature and number of dengue haemorrhagic fever cases by month for the period 1985–2001 for Yogyakarta province [Almost 60% of cases occur in the 5-month period January through May] 450
Avg. rainfall & no. cases
Total (mm)
Cases
AvgT (°C)
28.5 28.0 27.5 27.0 26.5 26.0 25.5
400 350 300 250 200 150 100 50 0
Wet or dry season estimates of pupae per person Figure 5 includes the annual number of cumulative DHF cases in Yogyakarta province for the period 1985–2001 and Figure 6 includes the cumulative monthly DHF cases for the same 98
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Dengue Bulletin – Volume 31, 2007
Avg. temperature
Targeted source reduction/control strategy for dengue control
are observed. Yet, epidemic transmission is often associated with anomalously high temperatures, almost 29 °C, for the Yogyakarta time series (data not shown). Therefore, to be conservative, we will use a value of 29 °C.
Identification of the targeted classes of containers In Table 11 are listed the most productive classes of containers observed in the wet season surveys. We see that if we controlled the wells and a single type of container, the Storage in WC (Bak mandi), we would achieve a standing crop of Ae. aegypti of 0.37 pupae per person. With the elimination of just these two types of containers, we are already below our transmission threshold. Again, to be conservative, if we add to our list of types of containers to be controlled – Wells, Water container (lg) (Bak air) – a fourth type, Tyre, we can anticipate being at levels of 0.29 and 0.23 pupae per person respectively, if all of the targeted types are completely controlled.
The value of seroprevalence to be used in estimating the transmission threshold This one is a bit more difficult and we will have to make a crude estimate. Graham et al. conducted a serosurvey in the late 1990s in Yogyakarta and estimated that the annual seroconversion rate among children aged 4 to 9 years was, for serotypes 1–4, 0.048, 0.077, 0.042 and 0.034 respectively. [16] In the endemic situation, the rate of seroconversion declines with age as those susceptible to infection become increasingly rare. We estimate the age-specific antibody prevalence rate up through age 80 by multiplying the 4–9year-olds’ rates for each year by one less the prevalence in the previous year. Now, to estimate the average overall prevalences for each of the dengue strains, we need a population estimate of the human age distribution in Yogyakarta. For this we used the WHO demographic data for Indonesia. Taking the average of the product of the age-specific age and prevalence proportions gives us the overall population seroprevalence for each serotype – 0.54, 0.66, 0.51 and 0.45 for serotypes 1–4 respectively (average: 0.54). The transmission thresholds for 0.33 and 0.67 for 29 °C from the table of transmission thresholds given in Focks et al.[11] are 0.43 and 0.96 respectively. Again, to be conservative, we will use the value associated with a seroprevalence of 0.33, i.e., a strategy to bring the area-wide abundance of Ae. aegypti to somewhat below 0.43 Ae. aegypti pupae per person.
Discussion The three targeted types of containers account for 45% of all production, they also account for 65% of all containers. Is this much of a strategy, given the number of containers we are trying to control? We think so for the following reasons: wells, bak mandis and bak airs are in known locations within the house and are typically masonry in construction. This means that they are easy to find, identify and control with several methods currently available in Indonesia, e.g. IGRs (Altocid and Sumilarv) and temephos (Abate). Tyres are similarly easy to locate and identify.
Research topics Most containers are negative for larvae and pupae Why is this so? Is it because they are used domestically and frequently used, cleaned or emptied? Is it because of natural biological
Dengue Bulletin – Volume 31, 2007
99
Targeted source reduction/control strategy for dengue control
control, or perhaps because they do not receive oviposition? Could an understanding of this phenomenon aid in controlling productive containers? Some containers are especially productive and account for the majority of breeding Studies on the mechanisms promoting productivity are needed. A related issue is
the time course of production – are productive containers only episodically so, with intervals of low or no production? Alternatively, are productive containers more or less continuously so? Studies on the mechanisms leading to certain classes of containers being especially productive are also needed. Time series of daily pupation in undisturbed containers in the field would be useful.
Table 11: Types of containers most responsible for the observed standing crop of Ae. aegypti pupae per person observed during the wet season surveys; averages for the two surveys are shown [Number refers to the average number of containers observed by type; the average total number of water-holding containers in the wet season surveys was 3724. Proportion of production is the average proportion of all Ae. aegypti observed in the wet season surveys. Accumulation refers to a summing of the column to the left in a downward direction; the Type “Storage in WC” accounted for 0.456 of all production, that type and the next most productive type, “Water container (lg)” account for 0.571 of all production, etc. Pupae per person refers to the actual contribution of that type of container. Balance if removed is the number of Ae. aegypti pupae per person that would remain in the environment if that Type were controlled and the ones above it. The average number of pupae per person in the environment was 0.672]
a b
Indonesian: Bak mandi Indonesian: Bak air
100
Dengue Bulletin – Volume 31, 2007
Targeted source reduction/control strategy for dengue control
Seasonality in transmission and the utility of an early warning system A recent National Research Council (USA) publication investigating the feasibility of developing practical and sustainable early warning systems (EWSs) for infectious diseases concluded that EWSs would provide significant utility where mitigation methods were available.[23] Their primary value lies in the ability temporally to focus scarce resources for control in those periods when epidemics were likely. Cases of dengue and DHF occur in virtually every province and during every month of the year in Indonesia. Does the seasonality of dengue transmission in Indonesia suggest that there would be merit in gaining the ability to time suppression activities to precede peak transmission periods based on an EWS? Initial attempts to develop an EWS for dengue in Yogyakarta have been promising and will be pursued. Given that the first five months of the year account for an average of 60% of all cases and the remaining months report about ca. 5% References [1] Suroso T, Holani A, Ali I. Dengue haemorrhagic fever outbreaks in Indonesia 1997-1998. Dengue Bulletin. 1998(22): 45-48. [2] World Health Organization, Regional Office for South-East Asia, New Delhi. Report of an external review: dengue/dengue haemorrhagic fever prevention and control programme in Indonesia. New Delhi: WHO-SEARO. 2001. Document SEA-Haem Fever-75, SEA-VBC-79. [3] Kusriastuti R. Internal report. Jakarta, Indonesia: Arbovirus Sub-directorate, Directorate of Vector-Born Disease Control, Ministry of Health 2002. [4] Martens WJM, Jetten TH, Focks DA. Sensitivity of malaria, schistosomiasis and dengue to
each, it seems likely that control would typically be continuous. Perhaps the utility of validated EWSs for Indonesia would be to forecast epidemic years as an aid to the national programme in securing adequate funding for anticipated epidemic years.
Acknowledgement This work was partially supported by a number of institutions including the Jean and Julius Tahija Family Foundation, the Office of Global Programmes, National Oceanographic and Atmospheric Administration (NOAA), Gadjah Mada University, and the Navy Medical Research Center, Silver Spring, MD. We thank all of the staff who dedicated long hours in the field to carefully collecting mosquito specimens and the data used in this study. We are especially thankful to Lely Sianturi, Saptoro Rusmiarto, Yoyo R. Gionar, Dwiko Susapto, and the health staff of Gondokusuman for their diligence and hard work during the years of investigation. We are also grateful to Iqbal Elyazar for data set preparation.
global warming . Climate Change . 1997 (35):145-156. [5] Patz JA, Martens WJ, Focks DA, Jetten TH. Dengue fever epidemic potential as projected by general circulation models of global climate change. Environ Health Perspect. 1998 Mar; 106(3): 147-53. [6] Bouma MJ, Poveda G, Rojas W, Chavasse D, Quiñones M, Cox J, Patz J. Predicting high-risk years for malaria in Colombia using parameters of El Niño Southern Oscillation. Trop Med Int Health. 1997 Dec; 2(12): 1122-7. [7] Focks DA, Haile DG, Daniels E, Mount GA. Dynamic life table model for Aedes aegypti (Diptera: Culicidae): analysis of the literature
Dengue Bulletin – Volume 31, 2007
101
Targeted source reduction/control strategy for dengue control
and model development. J Med Entomol. 1993 Nov; 30(6): 1003-17. [8] Focks DA, Haile DG, Daniels E, Mount GA. Dynamic life table model for Aedes aegypti (diptera: Culicidae): simulation results and validation. J Med Entomol. 1993 Nov; 30(6): 1018-28. [9] Focks DA, Daniels E, Haile DG, Keesling JE. A simulation model of the epidemiology of urban dengue fever: literature analysis, model development, preliminary validation, and samples of simulation results. Am J Trop Med Hyg. 1995 Nov; 53(5): 489-506. [10] Focks DA, Chadee DD. Pupal survey: an epidemiologically significant surveillance method for Aedes aegypti: an example using data from Trinidad. Am J Trop Med Hyg. 1997 Feb; 56(2): 159-67. [11] Focks DA, Brenner RJ, Hayes J, Daniels E. 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. 2000 Jan; 62(1): 11-8. [12] Focks DA, Brenner RJ, Chadee DD, Trosper J. The use of spatial analysis in the control and risk assessment of vector-borne diseases. Am Entomol. 1998; (45): 173-183. [13] World Health Organization. Key Issues in dengue vector control towards the operationalization of a global strategy: report of consultation. Geneva: WHO; 1995. Document CTD/FIL (Den)/IC.96.1). [14] Focks DA. A review of entomological sampling methods and indicators for dengue vectors. Geneva: World Health Organiation, 2004. Document TDR/IDE/Den/03.1. [15] Focks DA, Lele SR, Juffrie M, Suvannadabba S, Sobel AL, Trahan MW. Early warning systems for dengue in Indonesia and Thailand. Proc Unified Science & Tech Reducing Biological Threats & Countering Terrorism. Albuquerque: New Mexico. 2002. [16] Nam VS, Yen NT, Holynska M, Reid JW, Kay BH. National progress in dengue vector control
in Vietnam: survey for Mesocyclops (Copepoda), Micronecta (Corixidae), and fish as biological control agents. Am J Trop Med Hyg. 2000 Jan; 62(1): 5-10. [17] Graham RR, Juffrie M, Tan R, Hayes CG, Laksono I, Ma’roef C, Erlin, Sutaryo, Porter KR, Halstead SB. A prospective seroepidemiologic study on dengue in children four to nine years of age in Yogyakarta, Indonesia I. studies in 1995-1996. Am J Trop Med Hyg. 1999 Sep; 61(3): 412-9. [18] Gionar YR, Rusmiarto S, Susapto D, Bangs MJ. Use of a funnel trap for collecting immature Aedes aegypti and copepods from deep wells in Yogyakarta, Indonesia. J Am Mosq Control Assoc. 1999 Dec; 15(4): 576-80. [19] Bangs MJ, Focks DA. Abridged pupa identification key to the common containerbreeding mosquitoes in urban Southeast Asia. J Am Mosq Control Assoc, 2006 Sep; 22(3); 565-572. [20] Tun-Lin W, Kay BH, Barnes A. The Premise Condition Index: a tool for streamlining surveys of Aedes aegypti. Am J Trop Med Hyg. 1995 Dec; 53(6): 591-4. [21] Barker-Hudson P, Jones R, Kay BH. Categorization of domestic breeding habitats of Aedes aegypti (Diptera: Culicidae) in Northern Queensland, Australia. J Med Entomol. 1988 May; 25(3): 178-82. [22] Tun-Lin W, Kay BH, Barnes A. Understanding productivity, a key to Aedes aegypti surveillance. Am J Trop Med Hyg. 1995 Dec; 53(6): 595-601. [23] Burke D, Carmichael A, Focks D, Grimes D, Harte J, Lele S, Martens P , Mayer J, Means L, Pulwarty R, Real L, Ropelewski C, Rose J, Shope R, Simpson J, Wilson M. Under the weather: exploring the linkages between climate, ecosystems, infectious disease, and human health. Washington, DC.: National Research Council, National Academy Press, 2001.
102
Dengue Bulletin – Volume 31, 2007