Study Site
All data used for this comparative analysis were collected from Machala, Ecuador, a coastal port city in southern Ecuador and the capital of El Oro province; there were approximately 270,000 residents at the time of this study (2014-2015). The incidence of dengue and the density of Ae. aegypti mosquitoes in Machala is amongst the highest in Ecuador, as well as other Latin American countries and Asia (18–21). Dengue is transmitted seasonally, with more cases reported during the hot, rainy season from February to May. Dengue outbreaks have been observed to correlate with extreme climate events, such as El Niño that strongly impact local rainfall and temperatures in southern coastal Ecuador (5,18,22).
We selected four (of 23) sentinel outpatient clinics located around Machala and operated by the MoH; sites were chosen based on a high burden of dengue in the community catchment areas and their interest and ability to participate in the study (19). The clinics included Brisas del Mar, Rayito de Luz, Mabel Estupiñan, and El Paraiso. In addition, the Teófilo Dávila Hospital, the primary public hospital run by the MoH, was included as it is the province-level reference hospital where the outpatient clinics refer patients with severe dengue illness (19). Public clinics and hospitals are required to report cases of dengue-like illness (with and without warning signs) to the MoH for patients seeking care.
Active Surveillance
Figure 1 provides a flowchart of AS and PS recruitment methods for the study. The AS study design and diagnostic procedure have been described previously (19). Briefly, individuals (index subjects) were recruited into the AS research study after visiting one of the four MoH clinics or the Teofilo Davila Hospital with clinical signs and/or symptoms of dengue (see Figure 1). Index subjects were referred to our study technician or nurse; informed consent obtained, and demographic and clinical information recorded. At the time of clinical evaluation, a 20 ml blood specimen (adjusted for age and weight by United States National Institutes of Health criteria) was obtained by venipuncture from each participant. Samples were processed at the diagnostic laboratory within the Teofilo Davila Hospital. Acute dengue infections were confirmed via blood serum with NS1 rapid strip tests. A maximum of four index subjects that tested positive for dengue were randomly selected each week to participate in the community surveillance component of the study. Members of the index subject’s household and members of four neighboring households within a 200-meter radius of the index household, the typical flight range of the Ae. aegypti mosquito, were invited to participate in the study. The same demographic and clinical information were gathered from these individuals, as well as a blood sample.
Blood specimens were separated via centrifuge into serum, cells and plasma and stored at -80°C. Samples were tested for dengue using NS1 and IgM enzyme linked immunosorbent assay (ELISA) at the laboratory in Machala. Samples were then shipped to SUNY Upstate Medical University where reverse transcriptase polymerase chain reaction (RT-PCR) was used to confirm dengue infections and virus serotypes (19). Positive cases of dengue in the AS system (herein called dengue illness) were defined as individuals with the presence of one or more of the following symptoms: fever, nausea/vomiting, rash, muscle/joint pain, abdominal tenderness, bleeding, diarrhea, headache, retro-orbital pain, drowsiness/lethargy, who tested positive for dengue virus by RT-PCR, NS1 rapid test, NS1 ELISA or IgM ELISA. There were 33 individuals in 2014 and six in 2015 in the AS system who had positive laboratory tests but no symptoms. These cases were excluded from the analyses reported in this manuscript since they did not fit the definition of dengue illness (symptoms and positive lab confirmation).
Passive Surveillance
Ecuador has a mandatory PS reporting strategy for dengue with and without warning signs as well as for other mandatory reportable health conditions. According to the World Health Organization (WHO) dengue guidelines (1), dengue without warning signs is defined as fever plus any two of following symptoms: nausea/vomiting, rash, aches/pains, positive tourniquet test, and leukopenia. Dengue with warning signs includes the definition above for dengue without warning signs and at least one of following warning signs: abdominal pain/tenderness, persistent vomiting, clinical fluid accumulation, mucosal bleed, lethargy, restlessness, liver enlargement > 2 cm, and/or increased hematocrit concurrent with rapid decrease in platelets (1). If the patient is classified as having dengue with warning signs, they are admitted to the local hospital, a physical exam is administered, and a blood serum sample collected. The sample is used to confirm dengue infection via RT-PCR at the national reference laboratory of the MoH in the neighboring city of Guayaquil. Patient demographics (age, sex, pregnancy status if applicable), clinical characteristics (start date of symptoms, final clinical diagnosis), and diagnostic laboratory results are recorded. If dengue diagnosis is ruled out, an ‘other’ diagnosis is recorded. If the patient seeking care is classified as dengue without warning signs, their information is entered into a separate MoH PS dataset based on clinical symptoms and not via RT-PCR confirmation. Patients classified as dengue without warning signs are entered into the database in ‘group form’ by reporting institution or clinic, but without names or specific ages.
For this analysis, the MoH provided de-identified data on reported dengue cases from the entire city of Machala to the study team. We created two PS datasets by extracting dengue cases from these MoH reports for the same sentinel clinics/hospital utilized in the AS study from January 1, 2014 through December 31, 2015. In the primary dataset, we identified patients diagnosed with dengue with warning signs and with laboratory confirmation (referred to as dengue illness). In the second data set, we included cases with dengue-like symptoms but without warning signs and without laboratory confirmation (referred to as possible dengue).
Dengue cases in the two PS datasets were compared to dengue illness cases that entered the AS study during the same period of time. Duplicate patients were identified by matching the date, sex, and age. Variables of interest included age class (<5, 5-19, 20-64, 65+), and sex (male, female), and pregnancy status (pregnant, not pregnant). The pregnancy status variable was not available in the PS data set without warning signs. Due to the small sample size, we were not able to compare pregnancy status across surveillance systems.
Statistical Methods
Microsoft Excel (Version 16, Microsoft Corporation, Redmond, Wash, USA) was used for data quality assessment. The Statistical Package for the Social Sciences (SPSS - IBM Corp. Released 2013, IBM SPSS Statistics for Windows, Versions 25 and 26, Armonk, NY: IBM Corp.) was used to calculate overall cumulative incidence and age specific incidence proportions, to construct outbreak detection curves by epidemiologic week, and to run all analyses. Descriptive statistics were tabulated for all variables of interest. As the demographic variables were categorical, chi-square tests were performed to compare cohort characteristics between the AS and two PS data sets in 2014 and 2015. The initial comparisons were between AS and PS dengue illness, while a second set of analyses evaluated AS dengue illness, PS dengue illness and PS possible dengue across demographics. Duplicate study subjects (n = 9) were omitted from both the PS and AS systems in these chi-square analyses. Results with a p value of < 0.05 were considered statistically significant. Results were not corrected for multiple comparisons.
Cumulative incidence proportions (overall and age-specific) were calculated for cases identified by the MoH PS system (dengue illness and possible dengue) using the city of Machala population estimate for 2016 (n=276, 691) and the age-specific population figures for Machala, also for that year (23). Note that the hospital serves the entire population of Machala and the four sentinel clinics serve a subset of the city’s population. However, incidence calculations are based on the entire population of the city of Machala, to provide a common denominator. Cumulative incidence proportions displayed are per 10,000 persons per year.