Date Published: November 20, 2014
Publisher: Public Library of Science
Author(s): Donald S. Shepard, Eduardo A. Undurraga, Miguel Betancourt-Cravioto, María G. Guzmán, Scott B. Halstead, Eva Harris, Rose Nani Mudin, Kristy O. Murray, Roberto Tapia-Conyer, Duane J. Gubler, Bridget Wills. http://doi.org/10.1371/journal.pntd.0003306
Abstract: Dengue presents a formidable and growing global economic and disease burden, with around half the world’s population estimated to be at risk of infection. There is wide variation and substantial uncertainty in current estimates of dengue disease burden and, consequently, on economic burden estimates. Dengue disease varies across time, geography and persons affected. Variations in the transmission of four different viruses and interactions among vector density and host’s immune status, age, pre-existing medical conditions, all contribute to the disease’s complexity. This systematic review aims to identify and examine estimates of dengue disease burden and costs, discuss major sources of uncertainty, and suggest next steps to improve estimates. Economic analysis of dengue is mainly concerned with costs of illness, particularly in estimating total episodes of symptomatic dengue. However, national dengue disease reporting systems show a great diversity in design and implementation, hindering accurate global estimates of dengue episodes and country comparisons. A combination of immediate, short-, and long-term strategies could substantially improve estimates of disease and, consequently, of economic burden of dengue. Suggestions for immediate implementation include refining analysis of currently available data to adjust reported episodes and expanding data collection in empirical studies, such as documenting the number of ambulatory visits before and after hospitalization and including breakdowns by age. Short-term recommendations include merging multiple data sources, such as cohort and surveillance data to evaluate the accuracy of reporting rates (by health sector, treatment, severity, etc.), and using covariates to extrapolate dengue incidence to locations with no or limited reporting. Long-term efforts aim at strengthening capacity to document dengue transmission using serological methods to systematically analyze and relate to epidemiologic data. As promising tools for diagnosis, vaccination, vector control, and treatment are being developed, these recommended steps should improve objective, systematic measures of dengue burden to strengthen health policy decisions.
Partial Text: Dengue presents a formidable global economic and disease burden with around half the world’s population estimated to be at risk of infection , . Dengue transmission has intensified in the past decades, with outbreaks increasing in frequency, magnitude, and countries involved , . Dengue disease varies across time and age of persons affected. This complexity results from the transmission of four different viruses affected by vector density, the host’s immune status, age, pre-existing medical conditions and other factors , . The impact of dengue has been measured in terms of both monetary value and public health metrics, such as disability-adjusted life-years (DALYs) , . Here we use the term “burden of dengue illness” to refer to the amount of clinically apparent disease and mortality imposed by dengue in a population. Economic burden has three main components: (i) costs of illness, estimated from the total symptomatic episodes multiplied by the average costs per episode , , (ii) costs of dengue prevention, surveillance, and control strategies , , and (iii) other impacts of dengue, usually harder to estimate, such as effects of dengue outbreaks on tourism , co-morbidities and complications associated with dengue virus (DENV) infection –, or the effects of the seasonal clustering of dengue on health systems . Accurate estimates of the economic and disease burden of dengue are critical to track health progress, assess program impact and results, and inform decisions about health policy, research, and health service priorities , –. However, estimates of dengue burden have substantial variability due to limitations in the availability, quality, and use of data.
Available data on the economic and disease burden of dengue are limited. We conducted a systematic literature review of articles published or indexed in the Web of Science, MEDLINE, or in WHO’s Dengue Bulletin, combining the keyword “dengue” with the following list of keywords: surveillance, incidence, reporting, sensitivity, capture-recapture, cohort, economics, costs, burden, Aedes aegypti, and control. In addition, we added findings from previous literature reviews on dengue disease and economic burden , , . For relevance to current dengue surveillance and management, we included articles published from 1995 through 2013 in English, Spanish, French, or Portuguese. The inclusion criteria for articles at each step of the review process (i.e. identification, screening, eligibility, and inclusion) are shown in the PRISMA flow diagram  (Figure 1). The review process left us with 88 articles. Our goal was not to obtain numerical findings from the individual studies, but rather to summarize the main strategies and data used to estimate the economic and disease burden of dengue and the sources of variability in the burden estimates.
Multiple factors contribute to the variability in estimates of dengue burden, making it challenging to obtain accurate estimates. We recommend a series of strategies for improving dengue-burden estimates; however, some of them may be costly and therefore harder to achieve, and strategies themselves may need to be evaluated for their cost-effectiveness. Possibly the most important limitation has to do with limited availability, quality, and use of dengue surveillance data in many countries. New prospective studies to ascertain dengue burden better are needed, particularly in areas where reporting is least complete (or nonexistent), such as Africa or South Asia. However, several improvements in economic and disease burden estimates may be achieved with available data. Reported surveillance data should include a narrative about the system’s main characteristics, including whether it includes the private sector, ambulatory episodes, cases of all ages, and type of lab confirmation, if any, of DENV infections reported. Most importantly, reporting to national surveillance systems should record each dengue episode as either hospitalized or ambulatory (i.e., never hospitalized). The use of covariates to estimate the burden of dengue can adjust for underreporting and/or to extrapolate to areas where there is no reporting at all , , . It would be important to characterize the context for epidemiological dengue studies to describe why these studies were conducted at the specific time and place, and how those settings compare to others in the country or region. Understanding how specific variables affect the burden of dengue will help researchers improve burden estimates. The greatest source of uncertainty in existing burden of dengue studies comes from underreporting of symptomatic DENV infections, followed by the type of treatment of episodes. Probabilistic sensitivity analyses and tornado diagrams are helpful to understand the proportion of a confidence interval that arises from various sources of uncertainty , . The biggest payoff for burden of dengue estimates would come from studies that can link and analyze existing data. For example, data from cohort studies and clinical trials could be re-analyzed and compared with officially reported dengue episodes to estimate EFs  and population-based economic burden. Understanding the health-seeking behavior of people with symptomatic DENV infections would, for example, allow researchers to estimate the probability that a dengue episode is reported as a function of setting (inpatient or outpatient), sector (public or private), case severity, age, type of facility, access to healthcare, and other variables in the surveillance system. We also expect that neglected impacts of dengue, such as decreases in tourism or health system congestion, would represent substantial costs during outbreaks.