Substantial discrepancies in dengue case estimates between the Global Burden of Disease Study and Taiwan Centers for Disease Control

Taiwan’s dengue cases vary annually, peaking in infrequent epidemics, which differ substantially from the Global Burden of Disease Study’s (GBD’s) projections. Although the GBD study provides invaluable insights into global health trends, its modelling approach fails to capture the dynamic change of dengue transmission.

The Global Burden of Disease Study (GBD), launched in 1991, is an ongoing effort to quantify mortality, morbidity and disability arising from major diseases, injuries and risk factors. 1 ,2 GBD estimates, based on modelling because of scarce reliable data, have become crucial for research and policy-making.However, despite advancements in modelling techniques, their credibility is limited by inconsistent availability and quality of underlying data. 3engue poses a significant public health challenge.Yang et al. 4 , using GBD 2019 data, reported that Taiwan experienced the second-fastest rise in dengue-related age-standardized death rate from 1990 to 2019, with an estimated annual percentage change of 24.47.This report warrants closer examination, as the Taiwan Centers for Disease Control (Taiwan CDC) recorded only 228 deaths during the 2015 outbreak, the most severe dengue epidemic in recent decades. 5Therefore, this study aimed to compare the number of laboratory-confirmed dengue cases recorded by Taiwan CDC with the GBD estimates for Taiwan.
In Taiwan, physicians are mandated to report patients with symptoms suspected to be dengue infection to the CDC and take blood samples for laboratory confirmation.The laboratory confirmation criteria include viral isolation, positive real-time reverse transcription-polymerase chain reaction, 4fold increase in immunoglobulin G (IgG) titre, detection of the non-structural protein 1 (NS1) and detection of dengue-specific immunoglobulin M (IgM) and IgG in a single serum sample (applicable before 2009).The annual numbers of suspected and laboratory-confirmed dengue cases were obtained from Taiwan CDC websites (available since 1998). 6The GBD estimates of annual dengue case count were obtained from the Global Health Data Exchange website.
Figure 1 shows that the number of laboratory-confirmed cases recorded by Taiwan CDC fluctuates greatly each year, with notable spikes of 15 732 and 43 784 in 2014 and 2015.In contrast, GBD estimates depict a trend of increasing cases, escalating from 341 187 in 1998 to a peak of 658 864 in 2017, and then slightly receding to 615 567 in 2019.These estimates are in marked disparity with official Taiwanese data, conspicuously missing the intensity of the 2014 and 2015 outbreaks.The discrepancy between GBD estimates and Taiwan CDC data ranges from 12-to 5240-fold.
Taiwan, located on the Tropic of Cancer, hosts Aedes aegypti, an efficient transmission vector in the south, and Aedes albopictus island-wide.Despite these vectors, dengue has not become endemic in Taiwan, with outbreaks totally driven by imported cases, particularly where A. aegypti prevails. 7Fever screening at airports, initiated in 2003, aims to reduce case importation. 7nlike Southeast Asia's hyperendemic countries with all dengue serotypes and children mainly affected, Taiwan usually experiences outbreaks of a single serotype affecting adults.Between 2016 and 2022, most confirmed cases were imported, but a decline was noted post-2020 because of stringent coronavirus disease 2019 (COVID-19) quarantine policies and reduced international travel.Therefore, the annual variability of dengue cases in Taiwan, often peaking during infrequent severe epidemics, is not accurately captured by GBD estimates.We were unable to compare dengue death data between the Taiwan CDC and the GBD, as Taiwan CDC only provides death counts for selected years. 5engue cases are often underreported because of symptom similarity with other febrile illnesses and mild cases not seeking medical care.In Taiwan, before 2015, mandatory reporting of suspected cases by physicians, along with laboratory confirmation, was potentially hampered by delays in obtaining test results due limited availability of certified laboratories. 8This issue was exacerbated by the practice of pre-emptive insecticide spraying at patients' homes, workplaces and schools, leading to complaints if subsequent tests proved negative.The widespread adoption of NS1 antigen rapid diagnostic tests in and post-2015 has markedly reduced underreporting, prompting physicians to quickly report and test suspected dengue cases to contain outbreaks swiftly.Every suspected case reported to Taiwan CDC undergoes testing, allowing the calculation of a positivity rate (Table 1)-an indicator broadly employed during the COVID-19 pandemic to assess transmission intensity. 9High positivity rates in 2014 and 2015 highlighted severe dengue epidemics in those years.In response, intensified vector control and surveillance by the local governments in southern Taiwan since 2016 have led to a notable decline in confirmed cases, evidenced by low positivity rates despite extensive testing.Therefore, it is recommended that governments should regularly disclose both confirmed case numbers and total tests administered.This practice facilitates a thorough analysis of surveillance efficacy and enables the determination of positivity rates.Consequently, this aids in assessing local transmission activity and distinguishing whether a rise in case numbers is because of increased transmission or intensified testing efforts.
The GBD 2019 utilizes specific covariates, including the Health Care Access and Quality Index, the population-weighted probability of dengue infection, the cause-specific mortality rate and population density, for modelling dengue incidence. 2owever, the complex epidemiology of dengue, characterized by intricate interactions amongst humans, vectors and the environment, renders this limited set of covariates insufficient for accurately capturing the disease's variability globally.In dengue non-endemic regions like Taiwan, the occurrence of outbreaks is more stochastic, making it even more challenging for the GBD's model that employs a smoothing approach to evaluate the yearly variation. 2Another notable limitation in the GBD's model is their reliance on 34 data points derived from 17 studies identified in their literature review to model the under-reporting adjustment factor for each country. 2 However, this estimation may not fully account for the effects of the widespread adoption of rapid diagnostic tests in recent years, which could have notably reduced under-reporting. 2In fact, a serosurvey conducted at the end of the intense 2015 dengue epidemic in Tainan City, Taiwan, revealed minimal under-reporting, likely attributed to the widespread use of rapid NS1 tests and increased awareness from comprehensive public and medical campaigns in this severe epidemic. 10n summary, whilst the GBD study provides invaluable insights into global health trends, its estimates for dengue case numbers in Taiwan appear completely implausible.The GBD study's modelling approaches have limitations, particularly for diseases like dengue, where transmission dynamics vary within and between countries and where a mix of seasonal epidemics, periodic or multi-year interval epidemics can occur.

Figure 1
Figure 1 Annual counts of laboratory-confirmed dengue cases in Taiwan as reported by Taiwan CDC, compared with annual dengue case numbers in Taiwan estimated by the GBD, accompanied by 95% uncertainty intervals, Taiwan, 1998-2019

Table 1
Annual numbers of reported, laboratory-confirmed dengue cases and test positivity rates for dengue virus infection amongst all cases, indigenous cases and imported cases inTaiwan, 1998-2022