Am. J. Trop. Med. Hyg., 72(2), 2005, pp. 201-208
Copyright © 2005 by The American Society of Tropical Medicine and Hygiene
SPATIAL PATTERNS OF AND RISK FACTORS FOR SEROPOSITIVITY FOR DENGUE INFECTION
BIRGIT H. B. VAN BENTHEM,
SOPHIE O. VANWAMBEKE,
NARDLADA KHANTIKUL,
CHANTAL BURGHOORN-MAAS,
KAMOLWAN PANART,
LINDA OSKAM,
ERIC F. LAMBIN, AND
PRADYA SOMBOON
Koninklijk Insituut voor de Tropen/Royal Tropical Institute, Biomedical Research, Amsterdam, The Netherlands; Department of Geography, Université Catholique de Louvain, Louvain, Belgium; Office of Vector Borne Disease Control No. 2, Muang District, Chiang Mai, Thailand; Institute of Virology, Erasmus University, Rotterdam, The Netherlands; Department of Parasitology, Faculty of Medicine, Chiang Mai University, Chiang Mai, Thailand
 |
ABSTRACT
|
Spatial patterns of and risk factors for seropositivity of dengue infection were studied in three sites in northern Thailand. A survey was conducted in 2001 among 1,750 persons. Potential risk factors for dengue infection were measured by questionnaire and IgM antibodies against dengue were detected by an enzyme-linked immunosorbent assay. The role of landscape as a risk factor was studied using land cover maps and a geographic information system. Logistic regression identified risk factors for dengue seropositivity. Spatial patterns of seropositive cases were determined by cluster analyses. Six percent of the study population was seropositive. Risk factors for dengue seropositivity differed per site, demonstrating variation in local infection patterns. In the periurban site, seropositivity depended on human behavior and factors related to housing quality rather than environmental factors. In both rural sites, older persons had a higher risk of seropositivity and persons living in houses surrounded by natural and agricultural land covers had a lower risk of seropositivity.
|
INTRODUCTION
|
Dengue fever is a mosquito-borne infection and one of the most important emerging infectious diseases. Today, in several Asian countries, dengue hemorrhagic fever is a leading cause of pediatric hospitalization and death. In Thailand, periodic outbreaks of dengue have been reported throughout the country1, with a large outbreak in 1987 causing more than 1,000 deaths2 and another in 1998 causing 424 deaths.3 Dengue virus infection can manifest itself as clinically unapparent, an undifferentiated febrile illness, classic dengue fever, or dengue hemorrhagic fever. Seropositivity can be used as a marker for dengue infection and assessed by an enzyme-linked immunosorbent assay (ELISA) measuring specific IgM or IgG antibodies against dengue. In a prospective cohort study in Thailand, 87% of the dengue virus infections were subclinical.4 Since passive surveillance is used in the Thai health reporting system, many cases of infection are missed because of this large proportion of asymptomatic infections.
Dengue fever is transmitted by mosquito vectors Aedes aegypti and Ae. albopictus. Aedes aegypti is an urban vector that became more widespread following uncontrolled urbanization in the second half of the 20th century.5 It breeds mostly in artificial containers,5,6 but has been reported in natural containers as well.7 Sparse vegetation, low altitude, good transportation routes, and urban development favor Ae. Aegypti over Ae. albopictus, although it could be found in places where the last criterion was not satisfied.8 Aedes albopictus can be found mainly in rural and periurban areas,5 and it also breeds in artificial as well as natural containers.9 It has been reported in forested areas in Malaysia10 and could be bridging sylvatic and urban cycles.5 Its role in transmission is controversial,11 but Ae. albopictus is considered a vector of dengue in Asia.5 In Thailand, dengue transmission occurs during the rainy season from May through October. Since there is no dengue vaccine available to date, the focus is on control activities such as vector elimination: insecticide spraying to prevent and interrupt outbreaks and community participation to eliminate breeding places.3
The heterogeneity in the incidence of dengue fever observed over time and space reflects the complexity of risk factors involved in disease transmission. Population growth, rural-urban migration, inadequacy of basic urban infrastructure, and exponential growth of consumerism are responsible for conditions that are highly favorable for transmission.12 Dengue fever used to be confined to large cities in Thailand, but recent incidence rates were reported to be higher in rural areas than in urban areas.13 Although the development of transport infrastructure and the increase of traveling and commuting have been mentioned as a cause for the spread of dengue outside urban areas,14,15 another reason for the change in incidence over time and space could be changes in land use. In Thailand, for example, great areas of forest have been cleared to cultivate cash crops and orchards, rice fields have been converted into housing in periurban areas, and irrigation projects have greatly improved the water supply and therefore the length of the growing season for some crops. Only a few attempts have been made at linking land cover or spatial features to dengue infection since it was generally accepted that dengue transmission was restricted to urban areas rather than natural or agricultural environments.15,16
The present comprehensive study was undertaken to investigate personal, household, and environmental risk factors for infection, as measured by its proxy parameter (dengue IgM seropositivity). Accurate knowledge of the distribution of dengue infection is useful for planning and evaluating prevention and control programs. Here, we made use of a geographic information system (GIS) to investigate whether seropositivity for dengue is clustered within the study area and whether it is influenced by landscape features.
|
METHODS
|
Study design.
RISKMODEL is a research project that is investigating the relationship between changes in land use and the occurrence of malaria and dengue in northern Thailand. For the prospective dengue study, three study sites with changes in land cover between 1989 and 2000 were selected based on Landsat (U.S. Geological Survey, EROS Data Center, Sioux Falls, SD) images. 1) Ban Pa Nai is a rural area in Chiang Mai province with two villages at an altitude of 450 meters above sea level. The main land use change observed is a shift from one to two rice harvests per year facilitated by the building of a dam in 1996 (Figure 1
). 2) Ban Pang is a rural site in Lamphun province at an altitude of 380 meters above sea level. Surrounding a narrow, irrigated valley, large areas on the hill-slope have been cleared for planting longan trees (fruits used as cash crop) (Figure 2). 3) Mae Hia is situated in the suburbs of Chiang Mai and is composed of two villages at an altitude of 320 meters above sea level (Figure 3). Following land speculation and development, large areas of former rice fields were converted to housing projects or left unused and uncultivated following the Asian financial crisis of 1997.
The landscape features and organization vary among sites and offer a variety of typical northern Thailand lowland environments. Land cover maps were derived from a March 2000 Landsat image with a spatial resolution of 30 meters using maximum likelihood classification methods, with a global accuracy ranging from 81% in the heterogeneous periurban Mae Hia to 86% in Ban Pang and 87% in Ban Pa Nai. All study villages had a history of dengue infection.
Details of the study methodology have been previously described.17 The Medical Ethical Committee of Chiang Mai University reviewed and approved the study, and local permission and collaboration were obtained. The staff consisted of 10 interview teams, each consisting of two interviewers from vector-borne disease units and one local public health volunteer. Each team was responsible for a number of households in the village. The latitude and longitude of main points in the village, including street corners, were registered using a hand-held global positioning system. Households were located on the map along these geo-referenced streets using preliminary hand-drawn maps.
Study population.
All inhabitants of the study sites were asked to voluntarily participate in the study. After explaining the purpose of the study, written informed consent was obtained. A total of 1,928 persons were included in May 2001 at the start of the dengue transmission season. Of these 1,928 persons, 1,750 (91%) were followed-up in September 2001 at the end of the transmission season; the 178 persons not followed-up in September were younger, less often farmers, and more often living in Mae Hia compared with the 1,750 persons under follow-up. Between May and September 2001 study participants completed daily a calendar registering where they spent most of their day (village, fields, orchards, or forest) and if they had a fever that day. People reported fever subjectively. The compliance rate for completing the calendar was almost 100% (missing for two persons) because health volunteers checked them every week.
Blood collection.
Finger prick blood was collected in May and September after informed consent was obtained. Blood was stored on filter paper (903TM Paper; S&S Company, Dassel, Germany) and air-dried in the shade. Within one or two days after the fieldwork, all filter papers were stored in the refrigerator (4°C) until antibody detection after a few months. After reconstitution of the filter papers in phosphate-buffered saline, antibodies were detected by means of a dengue fever virus IgM capture ELISA (Focus Technologies, Herndon, VA).18 A ratio 
1.3 compared with the reference was considered a seropositive result.
Data analyses.
Differences in seropositivity in September 2001 by study site and other risk factors were calculated by chi-square tests and a P value < 0.05 was considered statistically significant. Logistic regression was used to identify risk factors for seropositivity. Odds ratios (ORs) and their 95% confidence intervals (95% CIs) were calculated. Factors associated (P < 0.15) with seropositivity in univariate analyses were selected for multivariate analyses. In multivariate analyses, we tested statistically significant (P < 0.05) interactions between determinants in the final model and confounding. Since analyses were performed separately by site, numbers were small and therefore factors with a P value < 0.15 were considered significant. All statistical analyses were performed using SPSS software (SPSS Inc., Chicago, IL).
Potential individual risk factors were asked by questionnaire and included sex, age, profession, site, place of birth, knowledge of dengue, location of spending day time, location of spending the evening, number of days spent in the village, fields, orchards, or forest between May and September 2001, housing construction, screening of windows, availability of mosquito nets, and number of days with fever between May and September 2001.
We linked the occurrence of seropositivity within a household with the landscape attributes of its surroundings. The dependent variable was the existence of one or more seropositive persons in a household. Based on land cover maps two series of landscape factors were calculated for each household using ArcView 3.2 (Environmental Science Research Institute, Redlands, CA). The first series was the percentage of each land cover class in a 200-meter buffer (circle with a 200-meter radius) around each house. The second series was the distance between each house and the nearest 2,700 m2 (i.e., four Landsat pixels) patch of each land cover class (except village zones classes). This yielded 22 land cover factors, which were organized in categories per site. The distance from a house to the edge of the village was also calculated.
On the land cover maps, villages could not be distinguished from orchards due to dense fruit tree vegetation in both land use categories. Villages were therefore delineated on the land cover map by digitizing housing from topographic maps (1/50,000; Royal Thai Survey Department, Bangkok, Thailand). The topographic maps were printed in 1991 for Ban Pang and Mae Hia and 1985 for Ban Pang. Topographic shadows and burned areas were not taken into account since they are only temporary effects. The legend of the land cover maps was composed of 1) mixed deciduous forest, 2) dry deciduous forest, 3) bush or sparse forest, 4) irrigated fields (wet): irrigated in March, 5) irrigated fields (dry): not irrigated in March, 6) old orchards (tree cover > 60%), 7) water bodies and wide rivers, 8) upland fields/young orchards (tree cover < 60%), 9) sparsely vegetated area related to various human activities with no buildings or agriculture (e.g., wasteland, grassy area), 10) densely built areas, 11) village zones with dense vegetation, and 12) village zones with sparse vegetation.
Furthermore, to investigate whether seropositive persons were clustered in a certain geographic area within a study site, cluster analyses were performed using the Kulldorff spatial scan statistic.19 Clustering occurs when the probability of seropositivity is not randomly distributed, but concentrated at certain parts of the study area. A circular window, of which the radius changed continuously, was moving across the map of the village. For each circle, a spatial scan statistic tested if there is an increased risk of seropositivity inside the circle compared with the area outside the circle using the likelihood ratio test. The P value was obtained from a likelihood ratio test based on Monte Carlo simulation with 9,999 replicates. The maximum size of the circle never exceeded 50% of the total scanned area. The cluster analyses were performed per village using seropositive persons as cases and seronegative persons as controls. Houses were used as census areas.
|
RESULTS
|
Of the 1,750 persons included in the study, nine reported having had dengue (seven cases were confirmed by a medical doctor) between May and September 2001 and only three of them were seropositive. Overall, 113 persons (6.5%, 95% CI = 5.4–7.7) were seropositive in September 2001. Only 3 of the 113 seropositive persons were diagnosed with dengue fever, which implies that 97% of all infections were undiagnosed. Fifty-four persons seroconverted during the study period, giving a incidence rate of 3.1% (95% CI = 2.3–4.0). The seroprevalence and seroincidence did not significantly differ between study sites, whereas sex, age, and profession did (P < 0.0001) (Table 1). The land cover classification also differed by site (Table 2).
View this table:
[in this window]
[in a new window]
|
TABLE 2 Characteristics of the study sites: percentage of each land cover class in a one-kilometer buffer around the village edges
|
|
Individual factors.
Risk factors for seropositivity in September 2001 differed between the study sites (Table 3). In both univariate and multivariate analyses for Ban Pa Nai, sex and age were significantly related with seropositivity. Females had more than half the risk of seropositivity compared with males (adjusted OR = 0.45, 95% CI = 0.21–0.94). Persons 30 years old had more than five times a higher risk of being seropositive compared with persons < 30 years old. Comparable with Ban Pa Nai, the prevalence of antibodies in Ban Pang was significantly related to age (P = 0.002). For Mae Hia, the periurban site, only knowledge of dengue was related with seropositivity. Persons ignorant about dengue had a risk twice as high of being seropositive compared with persons with knowledge of dengue (OR = 2.05, 95% CI = 0.99–4.28).
View this table:
[in this window]
[in a new window]
|
TABLE 3 Logistic regression to identify personal risk factors for dengue seropositivity in northern Thailand by site*
|
|
Household and environmental factors.
As in the individual level analysis, risk factors varied between sites. Ban Pa Nai has a very homogenous landscape and both villages are surrounded by irrigated rice fields, an inappropriate environment for Aedes development. Significant factors in univariate analysis included type of housing, distance to irrigated fields (dry), proportion of orchards within 200 meters, and distance to the edge of the village (Table 4). In the multivariate model, adjusted for the existence of two villages, households within the village had a 1.94 times higher risk compared with households at the edge of the village (95% CI = 0.85–4.44) and households with a higher proportion of orchards within 200 meters had half the risk (95% CI = 0.22–1.14) for seropositivity.
View this table:
[in this window]
[in a new window]
|
TABLE 4 Logistic regression to identify household and environmental risk factors for dengue seropositivity in Ban Pa Nai, Thailand*
|
|
No housing factors were significant in the analysis for Ban Pang (Table 5). Significant environmental factors in univariate analysis were proportion of village area with dense vegetation within 200 meters, distance to orchards, distance to dry forest, and proportion of upland fields or young orchards within 200 meters. Only the latter two variables remained significant in the multivariate analysis. The larger the distance to dry forest, the higher the risk for seropositivity. Houses with upland fields or young orchards within 200 meters had half the risk (95% CI = 0.23–1.10) compared with houses with no upland fields or orchards within 200 meters.
View this table:
[in this window]
[in a new window]
|
TABLE 5 Logistic regression to identify household and environmental risk factors for dengue seropositivity in Ban Pang, Thailand*
|
|
In periurban Mae Hia, all the housing factors were significantly related to dengue seropositivity in univariate analyses (Table 6), but only the presence of mosquito nets remained significant in the multivariate analysis. This factor is a proxy for the housing quality since mostly people in non-concrete houses with no window screening used mosquito nets. Although many environmental factors were significant in the univariate analysis, only the proportion of densely built areas within 200 meters remained significant in the adjusted model. This factor was correlated with dengue knowledge and housing factors and is therefore probably a proxy for other factors such as housing quality (i.e., newer houses).
View this table:
[in this window]
[in a new window]
|
TABLE 6 Logistic regression to identify household and environmental risk factors for dengue seropositivity in Mae Hia, Thailand*
|
|
Repeating the analyses for one person per household, we could test the significance of environmental factors adjusted for individual risk factors. In Ban Pang and Mae Hia, the same adjusted models were obtained, except for the use of mosquito nets in Mae Hia being replaced by window screening. No significant environmental risk factors were identified for Ban Pa Nai.
Cluster analyses.
In September 2001, one cluster of seropositive cases was found in Mae Hia (P = 0.04) that included 10 households with 22 persons. There were 6 cases, whereas only 0.98 cases were expected if cases were randomly distributed over study site. When taking into account persons who seroconverted during follow-up from May to September 2001, one significant cluster was found in Mae Hia (P = 0.04) that included 17 households with 25 persons. There were 6 cases present instead of the 1.42 cases expected.
|
DISCUSSION
|
Risk factors for dengue seropositivity differed substantially per site, which demonstrates the variation in local infection patterns. In an urban environment, there are plenty of suitable habitats for Aedes mosquitoes and the presence of breeding sites for the vector is thus not the limiting factor for dengue infection. Rather, infection is likely to be dependent on the quality of housing and use of prevention measures by exposed people since knowledge of dengue resulted in a higher use of preventive measures in our study population.17 Environmental risk factors were important in the two rural sites as was suggested by others.20 In both sites, persons living in houses surrounded by natural and agricultural land covers had a lower risk of infection. For some land covers, such as forest, the explanation is straightforward since a forest is not a favorite breeding place for Aedes mosquitoes. Living near orchards did not increase the risk of infection. This is of particular importance since orchard development is among the most important land cover changes observed in Thailand nowadays and its impact in terms of vector-borne diseases was not yet determined. Thus, infection in rural sites mainly occurred among persons living in the center of the village and among the elderly. To study whether the surrounding natural or agricultural environment is a less favorable location for dengue transmission, integration of data on ecology, presence of Aedes mosquitoes, and the respective role of Ae. albopictus and Ae. aegypti is necessary. Only one cluster of seropositive cases could be identified underscoring the variation in the rate of infection over small distances. This probably corresponds with the variation in the abundance of Aedes larvae and dengue transmission over such small distances.6 Again, to study factors related to dengue transmission rather than infection, data on vector abundance are required.
Other landscape factors, such as land cover complexity or spatial structure, could be taken into account. However, they are mostly relevant to the landscape as a whole and these factors cannot easily be linked with household level data, notably because houses in a village often belong to the same landscape patch or unit. Conversely, there is little information on the impact of these factors on mosquito breeding and therefore on transmission risk.
Several studies showed an association between age and IgG dengue seropositivity,21,22 which is consistent with the fact that levels of IgG antibody to dengue remain high once a person is infected. How long antibodies remain present is subject to large interpersonal variation, depending on the immune status of the person. Our decision to test IgM seropositivity was based on the knowledge that IgM antibodies specific for dengue normally disappear within three months after infection. One can argue that presence of IgM antibodies for dengue is also a marker for lifetime exposure, since age is an important risk factor for dengue seropositivity in the rural sites. However, if that is the case, one would expect an age trend in all sites, and the percentage of seropositive persons did not increase with age in Mae Hia.
Dengue seropositivity and environmental factors were linked through location of households. Since not all people stay at home during the day and Aedes mosquitoes bite mainly during daytime, we cannot be sure that transmission takes place in or around the house. Information from the questionnaire showed that there was no significant difference in the risk of seropositivity for persons spending most of their day in or around their house or elsewhere. However, it could be that this information is incomplete. The calendar gave only information on staying in areas with buildings rather than staying in their own house during daytime. However, it likely that transmission occurred in or around the house since older persons were at higher risk and they tend to stay at home
In our study, 97% of the infections were undiagnosed, which is comparable with another Thai study.4 In a recent study in Kamphaeng Phet, Thailand, the ratio between unap parent and symptomatic infections measured by active sur veillance among schoolchildren differed over the years with ratio between 1.8 and 1.0.23 The difference in the ratio be tween unapparent and symptomatic infections between our study and the study in Kamphaeng Phet are likely to be ex plained by the difference in surveillance methods.
Few attempts had been made to link dengue to land cover, although it is known to have spread in rural areas, as shown here and by others.13,24 It is nevertheless particularly relevant to make such links in areas where basic information about disease incidence and prevalence are sparse, such as in many developing countries.25 In this study, using simple GIS and remote sensing techniques, we could show the role of various types of land cover in transmission risk of dengue. Remote sensing has shown its efficiency in surveying large areas whatever their accessibility; standard classification procedures yielded accurate maps even in the fragmented and heterogeneous northern Thailand environment. Furthermore, with the use of a GIS, clusters of seropositivity as an indicator for recent infection could be identified by linking serology to residence of people. This method is a useful tool in public health decision making since it visualizes the spread of a disease or infection and could therefore identify targeted regions or populations to be targeted for interventions.
In conclusion, risk factors for dengue seropositivity differed between rural and urban sites. In rural sites, spatial variation exists, with persons living further from agricultural or natural land covers (i.e., in the center of the village) having the highest risk of dengue seropositivity. In the periurban area, dengue seropositivity was clustered in certain neighborhoods and mainly related to factors such as knowledge of dengue, use of preventive measures, and housing quality rather than natural and agricultural land covers. Whether these results will vary between seasons and over time remains a subject for future investigations.
Received March 23, 2004.
Accepted for publication September 8, 2004.
Acknowledgments: We thank all participants in the epidemiologic dengue survey for their participation, and the staff of the Vector Borne Disease Control (VBDC) units and VBDC Office No. 2, as well as the local public health volunteers for their collaboration. We also thank our partners of RISKMODEL for their collaboration, and Paul Klatser and Mirjam Bakker for critically reading the manuscript.
Financial support: This study was supported by European Union grant QLRT-1999-31787, provided within the Quality of Life and Management of Living Resources Program (1998–2002).
Authors’ addresses: Birgit H. B. van Benthem and Linda Oskam, Koninklijk Insituut voor de Tropen/Royal Tropical Institute, Biomedical Research, Meibergdreef 39, 1105 AZ Amsterdam, The Netherlands, Telephone: 31-20-566-5450, Fax: 31-20-697-1841, E-mails: b.v.benthem{at}kit.nl and l.oskam{at}kit.nl. Sophie O. Vanwambeke and Eric F. Lambin, Department of Geography, Université Catholique de Louvain, Louvain, Belgium, E-mails: vanwambeke{at}gorg.ucl.ac.be and lambin{at}geog.ucl.ac.be. Nardlada Khantikul and Kamolwan Panart, Office of Vector Borne Disease Control No. 2, 18 Boonruangrit Road, Muang District, Chiang Mai 50200, Thailand, E-mails: ornardlada{at}hotmail.com and malar{at}chmai.loxinfo.co.th. Chantal Burghoorn-Maas, Institute of Virology, Erasmus University, Rotterdam, The Netherlands, E-mail: c.maas{at}erasmusmc.nl. Pradya Somboon, Department of Parasitology, Faculty of Medicine, Chiang Mai University, Chiang Mai 50200, Thailand, E-mail: psomboon{at}mail.med.cmu.ac.th.
|
REFERENCES
|
- Cummings DAT, Irizarry RA, Huang NE, Endy TP, Nisalak A, Ungchusak K, Burke DS, 2004. Travelling waves in the occurrence of dengue haemorrhagic fever in Thailand. Nature 427: 344–347.[Medline]
- Ministry of Public Health, Division of Epidemiology, 1989. Annual Epidemiological Surveillance Report. Bangkok: Ministry of Public Health, 67–74.
- Ministry of Public Health, Department of Communicable Disease Control, 2000. Communicable Disease Control in Thailand. Bangkok: Ministry of Public Health.
- Burke DS, Nisalak A, Johnson DE, Scott RM, 1988. A prospective study of dengue infections in Bangkok. Am J Trop Med Hyg 38: 172–180.
- Vazeille M, Rosen L, Mousson L, Failloux AB, 2003. Low oral receptivity for dengue type 2 viruses of Aedes albopictus from southeast Asia compared with that of Aedes aegypti. Am J Trop Med Hyg 68: 203–208.[Abstract/Free Full Text]
- Strickman D, Kittayapong P, 2002. Dengue and its vectors in Thailand: introduction to the study and seasonal distribution of Aedes larvae. Am J Trop Med Hyg 67: 247–259.[Abstract]
- Zeller HG, 1998. Dengue, arbovirus et migrations dans l’Océan Indien. Bull Soc Pathol Exot 91: 56–60.[Medline]
- Kittayapong P, Strickman D, 1993. Distribution of container-inhabiting Aedes larvae (Diptera: Culicidae) at a dengue focus in Thailand. J Med Entomol 30: 601–606.[ISI][Medline]
- Lounibos LP, O’Meara GF, Nishimura N, Escher RL, 2003. Interactions with native mosquito larvae regulate the production of Aedes albopictus from bromeliads in Florida. Ecol Entomol 28: 551–558.
- Chan YC, Ho BC, Chan KL, 1971. Aedes aegypti (L.) and Aedes albopictus (Skuse) in Singapore City. Observations in relation to dengue haemorrhagic fever. Bull World Health Organ 44: 651–657.[ISI][Medline]
- Gubler DJ, 2003. Aedes albopictus in Africa. Lancet Infect Dis 3: 751–752.[ISI][Medline]
- World Health Organization, 1997. Dengue Haemorrhagic Fever: Diagnosis, Treatment, Prevention and Control. Geneva: World Health Organization.
- Chareonsook O, Foy HM, Teeraratkul A, Silarug N, 1999. Changing epidemiology of dengue hemorrhagic fever in Thailand. Epidemiol Infect 122: 161–166.[Medline]
- Wellmer H, 1983. Dengue Haemorrhagic Fever in Thailand: Geomedical Observations on Developments over the Period 1970–1979. Berlin: Springer-Verlag, 1–40.
- Strickman D, Sithiprasasna R, Kittayapong P, Innis BL, 2000. Distribution of dengue and Japanese encephalitis among children in rural and suburban Thai villages. Am J Trop Med Hyg 63: 27–35.[Abstract]
- Muttitanon W, Kongthong P, Kongkanon C, Yoksan S, Gonzales JP, Barbazan P, Spatial and temporal dynamics of dengue hemorrhagic fever epidemics (Nakhon Pathom province, Thailand, 1997–2001). Available from www.gisdevelopment.net/application/health/planning/healthp0010pf.htm.
- van Benthem BHB, Khantikul N, Panart K, Kessels PJ, Somboon P, Oskam L, 2002. Knowledge and use of prevention measures related to dengue in northern Thailand. Trop Med Int Health 7: 993–1000.[ISI][Medline]
- Groen J, Koraka P, Velzing J, Copra C, Osterhaus ADME, 2000. Evaluation of six immunoassays for detection of dengue virus-specific immunoglobulin M and G antibodies. Clin Diagn Lab Immunol 7: 867–871.[Abstract/Free Full Text]
- Kuldorff M, Nagarwalla N, 1995. Spatial disease clusters: detection and inference. Stat Med 14: 799–810.[ISI][Medline]
- Endy TP, Nisalak A, Chunsuttiwat S, Libraty DH, Green S, Rothman AL, Vaughn DW, Ennis FA, 2002. Spatial and temporal circulation of dengue virus serotypes: a prospective study of primary school children in Kamphaeng Phet, Thailand. Am J Epidemiol 156: 52–59.[Abstract/Free Full Text]
- Reiskind MH, Baisley KJ, Calampa C, Sharp TW, Watts DM, Wilson ML, 2001. Epidemiological and ecological characteristics of past dengue virus infection in Santa Clara, Peru. Trop Med Int Health 6: 212–218.[ISI][Medline]
- Bartley LM, Carabin H, Vinh Chau N, Ho V, Luxemburger C, Hien TT, Garnett GP, Farrar J, 2002. Assesment of factors associated with flavivirus seroprevalence in a population in southern Vietnam. Epidemiol Infect 128: 213–220.[Medline]
- Endy TP, Chunsuttiwat S, Nisalak A, Libraty DH, Green S, Rothman AL, Vaughn DW, Ennis FA, 2002. Epidemiology of inapparant and symptomatic acute dengue virus infection: a prospective study of primary school children in Kamphaeng Phet, Thailand. Am J Epidemiol 156: 40–51.[Abstract/Free Full Text]
- DeRoeck D, Deen J, Clemens JD, 2003. Policymakers’view on dengue fever/dengue haemorrhagic fever and the need for dengue vaccines in four southeast Asian countries. Vaccine 22: 121–129.[ISI][Medline]
- Hay SI, Packer MJ, Rogers DJ, 1997. The impact of remote sensing on the study and control of invertebrate intermediate hosts and vectors for diseases. Int J Remote Sensing 18: 2899–2930.
TOP
ABSTRACT
INTRODUCTION
