National dengue surveillance, Cambodia 2002–2020

Abstract Global dengue incidence has increased dramatically over the past few decades from approximately 500 000 reported cases in 2000 to over 5 million in 2019. This trend has been attributed to population growth in endemic areas, rapid unplanned urbanization, increasing global connectivity, and climate change expanding the geographic range of the Aedes spp. mosquito, among other factors. Reporting dengue surveillance data is key to understanding the scale of the problem, identifying important changes in the landscape of disease, and developing policies for clinical management, vector control and vaccine rollout. However, surveillance practices are not standardized, and data may be difficult to interpret particularly in low- and middle-income countries with fragmented health-care systems. The latest national dengue surveillance data for Cambodia was published in 2010. Since its publication, the country experienced marked changes in health policies, population demographics, climate and urbanization. How these changes affected dengue control remains unknown. In this article, we summarize two decades of policy changes, published literature, country statistics, and dengue case data collected by the Cambodia National Dengue Control Programme to: (i) identify important changes in the disease landscape; and (ii) derive lessons to inform future surveillance and disease control strategies. We report that while dengue case morbidity and mortality rates in Cambodia fell between 2002 and 2020, dengue incidence doubled and age at infection increased. Future national surveillance, disease prevention and treatment, and vector control policies will have to account for these changes to optimize disease control.


Introduction
The global incidence of reported dengue has increased by tenfold over the last two decades, from 505 430 cases in 2000 to 5.2 million in 2019. 1 To contain infection and manage clinical disease, accurate estimates of dengue burden are needed to ensure appropriate allocation of resources, especially during rollouts of the novel tetravalent vaccines in dengue-endemic areas. 2,3Nevertheless, countries with the highest dengue prevalence predominantly rely on resource-scarce national surveillance systems that are often based on case identification by clinical presentation of symptoms and passive reporting, leading to large underestimation of disease burden. 4ambodia is a lower-middle income country located in a belt of dengue-endemic countries in Asia that together contribute to an estimated 70% of global dengue cases. 1 National dengue surveillance data for Cambodia was last published in 2010 5 and covered a period of dynamic change from 1980 to 2008, during which post-civil war improvements in public health infrastructure led to improved systems for surveillance and vector control.Enhanced surveillance was introduced in 2001; initial analysis of complete case data from 2002-2008 found no clear trends in dengue incidence, and uncertain impact of larvicide distribution.Since 2008, changes in health policies, population demographics, and land use have occurred alongside rapid industrialization in the country.Furthermore, rising global temperatures attributed to climate change have enhanced growth conditions for the vector for dengue, the Aedes spp.mosquito.6 How these changes affect dengue control in Cambodia remains unknown.
Here we summarize two decades of policy changes, published literature, country statistics, and data from the Cambodian National Dengue Control Programme to identify important changes in the disease landscape, and derive lessons to inform future surveillance and disease control strategies.Bull World Health Organ 2023;101:605-616| doi: http://dx.doi.org/10.2471/BLT.23

Dengue and climate models
To examine whether dengue case incidence, outcome, phenotype, and age of infection changed between 2002 and 2020,we fitted generalized linear models to time series of the response variables of (i) monthly dengue case numbers; (ii) case fatality rates; (iii) case proportions of dengue haemorrhagic fever; (iv) case proportions of dengue shock syndrome: and (v) mean age of infected individuals derived from national surveillance data, using year, monthly average temperature, and monthly total precipitation at the national level as fixed predictor effects.We fitted models in the Poisson family to crude case numbers and the Gaussian family to all other response variables.Fixed predictor climate variables (temperature and precipitation) were lagged to precede case response data.To identify the most appropriate period for the climate lags, we first optimized a cross-correlation function, testing associations between lag periods spanning up to one year for mean monthly temperature and total precipitation as compared with the response variable of monthly case counts.We calculated optimal lags of −3 and −11 months between dengue cases, and temperature and precipitation, respectively; climate variables were subsequently lagged by these durations before inclusion in generalized linear models.To examine whether climate variables could be driving inter-annual changes as well, we next explored whether mean monthly temperature and total precipitation (as significant predictors of dengue case variables) changed significantly over the examined period.
To do this, we fitted generalized additive models to time series of monthly climate data, including a fixed predictor of year, and controlling for intra-annual variability via incorporation of a monthly smoothing term, with the number of smoothing knots fixed at 7, and incorporating a cyclic cubic smoothing spline.[13]

Vector control
In Cambodia, biannual applications of larvicide (temephos) from April to July and August to October occur in tandem with public education campaigns reinforcing the importance of environmental and mechanical control, such as clearing stagnant water and using jar covers on open containers. 5Both vector control and public and clinician education have been centrally coordinated by the National Dengue Control Programme since 2001, although dedicated vector control units were only created in 2016 and linked to a new system for epidemic forecasting. 14Larvicide application remains highly variable with fluctuation based on resource availability and impact remains uncertain largely due to the presence of informal breeding sites. 5More recently, increasing Aedes resistance to temephos 15 prompted evaluation of other insecticides, such as Bacillus thuringiensis israelensis and pyriproxyfen, biological control with larvivorous guppy fish, and mechanical control with mosquito traps or covers for open water containers.Multisectoral involvement has supported communityspecific mobilization. 16,17[20]

Dengue treatment guidelines
While curative therapies for dengue remain elusive, major advances in the understanding of disease pathophysiology 21 have contributed to the development of enhanced supportive strategies and improved outcomes over the last two decades.Specifically, studies on fluid resuscitation in critically ill paediatric populations with dengue [22][23][24] and other shock conditions 25 brought attention to the importance of thoughtful fluid selection and infusion rates to minimize iatrogenic harm.Although data were available from the late 1990s, updates to WHO guidelines were only made in 2009, 26  The 2018 guidelines updates 29 incorporated additional caps on fluid resuscitation, including emphasis on early fluid discontinuation, a suggested 24-hour cap on colloid volume, and adult-specific infusion rates.The update recommended identification of risk profiles specific to adult populations, including pregnancy and diabetes, requiring heightened concern for clinical deterioration.The 2018 update also recommended empiric treatment of disease complications to avoid delays in care, such as administration of oxygen, calcium gluconate and vitamin K as part of upfront management of dengue haemorrhagic fever or dengue shock syndrome or unresponsive to initial fluid resuscitation.The use of advanced organ support therapies, such as renal replacement and mechanical ventilation was added, although in reality these interventions are limited to major hospitals in Phnom Penh.Finally, the guidelines now include provisions for outbreak preparedness, such as appropriate hospital response plans, and guidelines for inter-hospital acuity-related and/or load-balancing transfers.While laudable, these provisions remain limited by lack of available and properly fitted ambulances which, when combined with exorbitant costs of medical transportation, create challenging logistics for patient access to tertiary care. 30

-Enhanced national surveillance
Standardization of dengue surveillance using WHO clinical case definitions, standardized case report forms and data entry into an electronic database across 25 provinces.Introduction of virologic surveillance in four provinces.

-National dengue guidelines published
Guidelines describe diagnosis of dengue fever, haemorrhagic fever, and shock syndrome, along with monitoring and supportive care strategies.

-Dengue reporting to WHO
Routine submission of dengue surveillance data for collation as part of WHO's monthly dengue situation updates.

-National dengue guidelines updated
Updated guidelines recommended centralized care at referral centres, and judicious fluid resuscitation of patients; and provided separate guidance for patients with complications of disease and/or higher baseline risk.

-Epidemic forecasting introduced
The National Dengue Control Programme introduced an epidemic prediction algorithm to identify early rises in case numbers beyond historic baselines that could signal an impending epidemic with 2-3 months' lead time.This algorithm was linked to a response system that deployed enhanced vector control and targeted education.

-Vector control and dengue education units established
Establishment of dedicated response teams in the National Dengue Control Programme to coordinate vector control efforts and disseminate information on dengue recognition and management to public and clinical sectors.

-National dengue guidelines updated
Updated guidelines included additional caps on fluid resuscitation, guidance for older populations, empiric and/or advanced treatment options for complicated disease, and hospital outbreak preparedness planning and response strategies.

-COVID-19 pandemic
COVID-19 led to changes in dengue transmission and detection related to societal mobility restrictions, reduced care-seeking behaviour, and disruption in local and national surveillance and vector control activities.

-Arboviral differentiation testing
The worldwide Zika virus epidemic in 2016 and a large chikungunya virus outbreak in Cambodia in 2020 led to routine testing of virologic surveillance samples for chikungunya, dengue, and Zika viruses using PCR.

2020-2021 -More virologic surveillance sites
Sentinel surveillance sites in 11 provinces were added to original sites in 4 provinces between 2020 and 2021; virologic surveillance now spans 15 of 25 provinces.

-National strategic plan published
In consultation with WHO, the National Dengue Control Programme created a 10-year plan for sustainable prevention and control of dengue and other Aedes spp.-transmitted arboviral diseases.
Dengue surveillance, Cambodia Christina Yek et al.
][34] Two of these publications estimated burden of additional dengue cases 3.9to 29.0-fold higher than that captured by national surveillance (1.1-5.7 per 1000 person per season). 32,34   clinicopathologic criteria to define dengue cases.Similar limitations exist for empirical studies on underreporting published elsewhere, 4 and data should be interpreted with caution; ultimately, nationally representative longitudinal cohort studies with careful selection of case definitions are needed to better understand degree of incomplete case capture by national passive surveillance, but such studies are resource-intensive and may be impractical in the long term.

Age structure of cases
Before 2014, children aged 5 to 9 years represented the majority of dengue cases, followed by children aged 0 to 4 years.From 2014 to 2019, children aged 10 to 14 years represented a larger proportion of dengue cases, including more severe cases of dengue haemorrhagic fever and dengue shock syndrome, than children aged 0 to 4 years.By 2020, children aged 10 to 14 years were the most commonly represented age group among dengue patients (Fig. 4).Overall, the mean age of infected individuals increased significantly from 5.8 years (SE: 0.3) in 2002 to 9.1 years (SE: 0.4) in 2020 (slope: 0.20; SE: 0.0094; P -value: < 0.001; Fig. 5).This change corresponds to the increase in median age of the overall population as indicated in the 1998, 2008 and 2018 census surveys (online repository), 31 reflecting Cambodia's demographic transitions as a result of industrialization.

Trends in viral serotypes
The National Dengue Control Programme performs dengue serotypespecific RT-PCR on a monthly basis in a subset of samples from sentinel sites (Fig. 6).Before 2008, DENV-2 and DENV-3 serotypes were the predominant circulating serotypes, including an epidemic in 2007 driven by DENV-3 that preceded near-extinction of this serotype.Since 2008, DENV-1 and DENV-2 have dominated, with lower proportions of DENV-4 constituting approximately 10% of circulating variants.The 2012 epidemic was driven by DENV-1, most likely due to a genotype replacement of genotype IV to I that increased mosquito-virus transmission potential. 36During the large 2019 epidemic, DENV-1 and DENV-2 dominated; modelling studies are ongoing to investigate possible immunological, anthropological and ecological mechanisms of the outbreak.Studies in other countries variably attributed the high case numbers in 2019 to extreme weather events, 37 human migration, 38 and waning cross-protective immunity from prior arboviral outbreaks 39 or with re-emergence of extinct genotypes. 37

Climate factors
Annual peaks in dengue cases occurred in June through August, approximately 3 months after peak temperatures and preceding peak precipitation by 1-2 months (online repository). 31Both precipitation and temperature demonstrated significant intra-annual seasonality but no significant change between 2002 and 2020 (online repository). 31These climate factors strongly predicted seasonality and annual dengue cases (online repository); 31 however, increases in dengue incidence remained significant (P-value: < 0.001) even after adjustment for both factors.These results suggest that while climate changes contributed to rising dengue incidence in the last two decades, other factors may also be responsible for the increase.Additionally, the seasonal effects of temperature and precipitation reinforce the utility of including climate variables in epidemic forecasting systems.Notably, potential micro-effects at the village, district and/or province level cannot be evaluated in this analysis. 40,41

Vector control
The  where data are available, larval and adult vector density does not correlate with measures of mosquito exposure. 43

Other Aedes-borne outbreaks
While autochthonous transmission occurred in Singapore during the 2015-2016 Zika disease outbreaks, 44 only sporadic Zika virus cases were reported between 2007 and 2020 in Cambodia. 12,45,46Chikungunya virus re-emerged in Cambodia in 2011 with introduction of the east-central South African genotype, 47 but burden was not routinely assessed in national surveillance and outbreaks may have gone undetected until a large epidemic in 2020. 12,48During this outbreak, the National Dengue Control Programme introduced arboviral differentiation RT-PCR in its sentinel surveillance programme.The 2020 outbreak consisted of 7014 suspected cases across 23 provinces; in 2021, a total of 1421 cases were reported across 15 provinces. 49To date, yellow fever has not been reported in Cambodia.

Planning for the future
Despite expanded surveillance, enhanced data integration and improved disease management, national dengue surveillance in Cambodia continues to have several limitations including reliance on patient self-referral; predominant clinical syndrome-based identification of disease; exclusion of patients seeking care at private healthsector facilities; and limited integration with effective vector control efforts.In the 2021-2030 National strategic plan on sustainable prevention and control of dengue and other Aedes-transmitted arboviral disease through a comprehensive integrated approach, 49 the National Dengue Control Programme identified several specific objectives (Table 1) to achieve three main targets by 2030: (i) reduce case fatality rates to goal of 0%; (ii) provide 100% detection and response to anticipated outbreaks; and (iii) reduce disease incidence by 50% (from 25 000 to 12 500 cases per year).

Additional considerations
Dengue control relies upon close monitoring of each component in the hostvector-virus triad.Current surveillance in Cambodia centres around description of the host and virus, with limited focus on the vector.Integration of vector surveillance and control measures will provide a new, important dimension to the National Dengue Control Programme,  with benefits to both containment of dengue and other Aedes-borne diseases.However, rising temephos resistance may require adoption of alternative methods of vector control.Viral serotyping is available and performed on a subset of surveillance samples; widespread adoption of advanced serologic assays during the coronavirus disease 2019 (COVID-19) pandemic may help boost understanding and anticipation of unusual fluctuations in population susceptibility preceding large epidemics.Similarly, in-country access to pathogen-agnostic technologies such as metagenomic sequencing 50 can be harnessed for future responses to outbreaks caused by a novel viral genotype or a look-alike pathogen.
To develop effective interventions, improvements directed at better capture and characterization of at-risk populations are needed.Dengue has been classically described as a disease of the young, but recent trends in Cambodia and other countries may indicate the need for a par-adigm shift. 51This transition to older age groups could indicate waning exposures to vectors, attributed to improved living conditions and sanitation.Movement of the host-vector interface away from the home and into public spaces, such as schools and workplaces, has important implications for where vector control measures are implemented.
In the clinical area, disease recognition in non-paediatric populations remains poor 12 and may contribute to diagnostic delays.In addition, less familiarity with treating adult patients with distinct co-morbidities and risk profiles may contribute to higher morbidity and mortality in this population. 52While Cambodian treatment guidelines were updated in 2018 to inform management of adult patients with distinct co-morbidities and risk profiles, dedicated clinical studies in this age group along with educating providers and the public in symptom recognition and management will be important to ensure appropriate diagnosis and treatment.At a population level, accurate and timely capture of affected populations through continued passive surveillance combined with periodic febrile and serologic surveillance will continue to provide valuable insights to guide rollout of control measures.

Conclusion
While the incidence of reported dengue cases in Cambodia increased between 2002 and 2020, the true burden of disease remains underestimated.Despite this, the National Dengue Control Programme has had notable successes reflected by reductions in cases of severe dengue and case fatality rates.For future interventions to reach susceptible populations at the appropriate scale, they will need to consider disease underestimation, shifting demographics, fluctuating viral serotypes, changing climate and land use, and novel emerging infectious threats.■

Table 1 . Specific objectives and proposed actions of 2021-2030 national strategic plan on sustainable prevention and control of dengue and other Aedes-transmitted arboviral diseases, Cambodia
CambodiaChristinaYek et al.entourant la maladie; mais aussi (ii) tirer des leçons en vue d'élaborer, à l'avenir, des stratégies de surveillance et de lutte contre la maladie.Nous signalons qu'en dépit d'une baisse des taux de morbidité et de mortalité liés aux cas de dengue entre 2002 et 2020 au Cambodge, son incidence a doublé et l'âge des patients au moment de l'infection a augmenté.Les futures politiques nationales de surveillance, de prévention et de traitement de la dengue, mais aussi de lutte contre ses vecteurs, devront tenir compte de ces changements de façon à mieux maîtriser la maladie.