The present study showed that dengue and other febrile illnesses have a substantial financial impact on households in a rural province of Cambodia. Direct medical costs accounted for about 50% of the economic impact; the remaining 50% included non-medical costs of caring for the ill child and actual loss of income due to work loss or the need to pay someone to take care of their rice fields. To pay these costs, two-thirds of households had to borrow money and 25% had to sell assets or use their savings.
This study assessed the cost of a dengue febrile illness irrespective of whether the person was hospitalized or not. This distinction is particularly important to a country because the greater burden of disease is due to persons cared for as outpatients [4
], yet most published dengue cost-of-illness studies have focused on hospitalized patients [6
]. As expected, the total costs to the household increased dramatically if the child was hospitalized, but at US$ 14, the cost of non-hospitalized dengue was still substantial considering that 39% of people in rural Cambodia have a daily income below US$ 0.43 [5
The total cost of a confirmed dengue illness was not different from a non-dengue febrile illness, a finding that differs from a study in northern Thailand [4
]. There the average total cost of a dengue illness (US$ 16.6) was significantly higher than for a non-dengue illness (US$ 9.8) but also the duration of the dengue-related illness was longer than the non-dengue illness and especially the proportion of hospitalization was much higher among dengue patients (33% vs. 1%) [4
]. In Thailand, the average cost of a dengue illness was lower compared to that of Cambodia, which may be due to higher hospitalization rate in our study population in Cambodia (67%). However, the cost per hospitalized dengue case in our study was similar to that of hospitalized dengue hemorrhagic fever patients in Thailand (US$ 40 vs. 39) [4
A study in two villages of Kampong Cham province in 2003 showed the average direct medical and non-medical cost of a hospitalized case of dengue was US$ 34.5, which did not include indirect costs such as loss of income [9
]. In Banteay Meanchey, a rural Cambodian province bordering Thailand, the average household expenditure for dengue treatment at private providers was US$ 103 [6
]. The Cambodian data suggests that differences in costs-of-illness may exist by type of health service utilized (private or public), and that cost-of-illness assessments should be conducted in multiple sites if possible [11
]. Differences might have also occurred due to different methodologies in how data were collected and which -especially indirect- costs were included.
Health equity funds attempt to improve access to health care services for the poorest. A review of four hospital-based health equity funds in Cambodia showed they increased utilization of hospital services by the poorest patients [12
]. In our study population, 60% of households classified as very poor received reimbursements through an equity fund if their child was hospitalized for a febrile illness, which reduced the direct cost to less than 7.5 US$ per hospitalized child. Still, a large proportion of our study population incurred debts to finance the febrile illness. Hospitalization was highly associated with higher average debt, and inversely related to poverty. These associations indicate that dengue and other febrile illnesses continue to exacerbate a family's financial burden, and lack of financial resources probably played a role in the decision whether to manage a sick child in the hospital or at home. In addition, a qualitative study performed in the same province suggested that lack of financial recourses lead to delays in help seeking and inappropriate treatment of children with dengue [9
Traditionally economists have relied on reported income and expenditures as the preferred indicator of poverty [5
], but recent research has shown this kind of data are often subject to measurement error or systematic reporting biases due to differing interpretations of the questions by respondents [14
]. For this reason we created a poverty scale, which combined information on food expenditure with information on the ownership of a selected asset [14
], which in our opinion reflected best the socioeconomic status of a household in rural Cambodia. Information on housing was also used by the Cambodian socio-economic survey in 2004 [15
]. Although our goal was to only classify our study population and not compare it with other populations, nonetheless, 42% of our study households were classified as very poor, which is almost the same proportion (39%) of Cambodians in rural areas that lived below the national poverty line, as determined by the World Bank in 2004 [5
]. Also the observed associations of poverty with the educational level of the father and the magnitude of incurred debts indicated that our poverty score provided a valid assessment tool.
There were several limitations of the study. Its small sample size didn't allow for robust comparisons by type of health care provider once cases were stratified by dengue and non-dengue febrile illness, and hospitalization. The study was conducted in a rural area of a single province, which is more densely populated in comparison to especially the northern provinces but had in 2004 with 35 to 45% an average poverty headcount for rural areas of Cambodia (with the highest in the south-western plain region and the lowest in the coastal provinces) [5
]. Main socio-economic and structural differences in the country exist, however, between rural and urban areas [5
Thus, our study population may be representative of rural Cambodia, where 84% of the total population lives, but not be representative of the entire country. In addition, costs of a dengue episode may vary broadly from one health facility to another depending on how well the health equity funds operate – User-fees exemption for the poor shown in some Cambodian hospitals could worsen inequity [9
]. Anecdotal reports made us believe that the health equity funds in Kampong Cham hospital has helped increase hospital access for the poor. The study design of matching laboratory-confirmed dengue cases with dengue-negative controls lead to some delay in conducting the interviews, which might have introduced some recall problems. However, recall problems should have occurred for both cases and controls and might have not introduced a bias when comparison them. Last, the strength of the association between poverty score and hospitalization for children may be limited because there were no independent measures of disease severity other than fever duration, which may have affected the decision to hospitalize.
The major strength of the study was that hospitalized and non-hospitalized cases and controls were recruited from a prospective community-based cohort study, and included cases that did not seek medical attention. Importantly, the case-control design allowed for robust comparisons of hospitalized and non-hospitalized cases, and the dengue and non-dengue febrile illnesses.