Our study is a population based epidemiological study driven by a clear objective of site preparation for a future dengue vaccine field trial. The Namuang sub-district of Muang district, Ratchaburi province, was chosen for the study site because it had high reported rates of dengue virus transmission. The incidence of dengue disease was determined from a prospective, long- term active fever surveillance of study participants in a well-defined cohort of school children throughout a 48-month period. Daily tracking of school absenteeism during school days and telephone contact with parents twice a week during school holidays made a broader capture of dengue possible. Most febrile illness cases detected in the cohort surveillance presented at Ratchaburi hospital were examined by pediatricians and had samples collected for dengue diagnostic testing.
An important aspect of this study was that our surveillance system captured children with febrile illness who sought medical care in both the public and private sectors. Prior to the beginning of the study, we met with a private hospital director and pediatricians who worked on the private hospital and clinics in Muang district to inform them of the study project. They were asked to inform us whenever participants attended their hospital/clinics. As a result, our staff visited all participants admitted to the private hospital and obtained acute serum samples from febrile patients. Convalescent samples were subsequently collected at RH. In addition, by end of the first year, we had strengthened parent-staff communications and education by use of telephone explanation and written documents that explained provision of study care with no charge fast track service and telephone consultation service at RH. Our efforts to educate parents were evidenced by increased in parent initiated hospital visits for febrile illness to 87% in the years 2007–2009.
Our rates of dengue incidences over the four-year period, 1.77, 3.58, 5.74 and 3.29 per 100 person-years in 2006–2009 reflect a parallel pattern of dengue incidence reported from the national surveillance data base in 2006–9 for Muang district, Ratchaburi Province: 159, 200, 361 and 155 per 100,000, respectively. Our prospectively determined dengue incidence in Na-muang sub-district is 11 to 21 (average 16.5) fold higher than those derived from the national surveillance database in Muang district. One difference between our incidence rates and the national rates is that we prospectively determined DHF, DF and UF cases among febrile illnesses of primary school children, whereas, the national data presents mainly DHF reported from hospitalized patients in all age groups.
Although there were some methodological differences in how active surveillance was conducted in our study and previously reported studies in childhood cohorts from Kampang Phet, Thailand 
, Kampong Cham, Cambodia 
and Managua, Nicaragua 
, the incidences of laboratory confirmed dengue found by active surveillance for acute febrile illness were much higher than that of the corresponding national reporting data. Using a calculated expansion factor to estimate differences between national reporting and laboratory based surveillance for dengue incidence, Wichmann et al. 2011 found the average under-recognition of dengue across three cohort studies (Kampang Phet, Ratchaburi and Kampong Cham) of dengue incidence to be more than 8-fold 
. The Nicaraguan study found under-reporting of dengue cases in relation to national surveillance systems to be 21.3 fold 
. The national under-reporting of dengue incidence of cases hinders accurate knowledge of disease burden.
All four DENV serotypes were involved in all levels of disease severity observed over the study period. Although there was strong seasonal variation, laboratory confirmed dengue virus infections were observed in every month of the four-year observation period in all 7 schools indicating that there is continuous DENV transmission in Ratchaburi. DENV-1 was the most commonly virus isolated (43% of total), followed by DENV-2 (29%) and DENV-3 (20%). New introduction in 2007 of DENV-2 and DENV-3 after a period of relatively low frequency in occurrence was followed by a severe outbreak in 2008 wherein DEN2 frequency was 30-fold increased and DENV-3 8-fold increased. This observation is consistent with a previous report that in Thailand DENV-2 was a dominant isolate during moderately severe dengue outbreak years and DENV-3 was associated with subsequent severe outbreaks after a time of relatively low frequency 
Dengue virus isolation methods differed between 2006 and the subsequent years in that PCR was used from 2007–2009. A lower DENV isolation rate was observed when culture in C6/36 cells and mosquito inoculation (67%) were used compared to detection of virus genome using RT-PCR (>83%) in this study is likely due to lower sensitivity of mosquito methods compared to PCR molecular methods 
. The overall viral identification rate using combined mosquito inoculation and PCR tests of 84.5% reported in this study is higher than the 65% in the previous report which also used both methods 
Secondary-type DENV infection described in this study was based on the ratio of anti-DENV IgM: IgG<1
. It has been observed from serology test results that there is cross reaction between JEV IgG and DENV IgG in S1 as well as in S2 with resulting inability to distinguish JEV IgG from DENV IgG. This observation had also been reported previously 
. As the Thai National Immunization Program has included JE vaccination since 2000, JEV IgG from vaccination might have had some influence on the observed DENV IgG titers and may have confounded the interpretation of secondary infection data reported here. However, the majority (84.5%) of the patients with secondary DENV infection was accompanied with positive dengue virus detection and had low JEV IgG titer in their acute (S1) specimens suggesting a minimal impact on the validity of anti-DENV IgM/IgG ratio used to estimate secondary DENV infection.
There appears to have been a shift in modal age of dengue incidence over the past four decades in Thailand. A retrospective hospital based study of laboratory confirmed dengue in 15,376 patients from Bangkok, Thailand reported a modal age of 5 years during 1973–79 which increased to 8 years during 1990–99 
. We found a modal age of 11 years in the present study. This finding might reflect a somewhat older group presenting with less severe disease (UF) in outpatient settings, but it is consistent with the increasing modal age previously reported 
. Hospitalization rates for acute symptomatic dengue infection could be considered a measurement of dengue disease severity. The hospitalization rate for DHF, DF, and UF were 100- 84- and 15- percent, respectively. Over the four years of our study, the provision of fast track service without payment, led to progressively better compliance of febrile cases from the cohort reporting directly to hospital OPD for evaluation, minimizing the possibility that cases may have been missed, especially in years 2007–9. However it is possible that the study missed some mild febrile cases that did not attend RH in 2006.
Because of the sensitivity of our surveillance system only 184 (46%) of laboratory confirmed dengue cases (DHF 42 cases and DF 142 cases) met the 1997 WHO case definition 
. Clinical manifestations of 210 (52.5%) of laboratory confirmed UF cases did not meet the WHO case definition for dengue 
. UF was the most common clinical manifestation of children infected with dengue virus, whether primary or secondary DENV infections, and can be difficult to distinguish from other childhood febrile illnesses 
. Our findings that a high proportion of dengue cases manifested as UF has an important impact on present vector control practices. In Ratchaburi province, mosquito control campaigns against dengue occur broadly over the province, four times per year. In addition, implementation of mosquito control measures including source reduction, application of larvicides, and spraying residual insecticide, are implemented in the area of 100 meters in diameter around all the reported houses of patients with dengue infection at all times of year. Due to the current national surveillance system only DHF/DF cases observed in clinical facilities are reported. Mosquito control measures are not implemented around houses of UF patients. This may contribute to inability to adequately control dengue and result in the long persistence and wide spread of dengue in the province as well as in Thailand in general.
We performed an extensive dengue epidemiology study and report accurate background data on dengue incidence in a cohort of individuals at high risk of dengue, children ages 3–15 years living in Namuang subdistrict of Muang district of Ratchaburi province, Thailand. We found sufficiently high incidence of four DENV serotypes for four consecutive years to make vaccine efficacy studies possible. Our study methods and findings fulfilled the epidemiological criteria recommended by WHO for dengue vaccine trial site selection 
. Following the reports of safety and immunogenicity of phase I studies of live-attenuated tetravalent dengue vaccine in dengue–naïve and dengue-endemic populations 
, the site was selected for the first dengue vaccine efficacy and safety trial, phase 2b. This trial uses a live-attenuated yellow fever vaccine virus based chimeric tetravalent dengue vaccine in a 3-dose vaccine administration of doses spaced at six-month intervals. This trial was launched in early 2009 and dose administration in approximately 4000 participants was completed in March 2011 [ClinicalTrials.gov Identifier: NCT00842530). Safety of the vaccine within the trial enrollment and vaccine administration period has been established. An analysis for vaccine efficacy is planned in 2012. If the vaccine is sufficiently immunogenic and efficacious after long term follow up, it is anticipated that the vaccine might be available for use in 2015.