• 1.

    Gyawali N, Bradbury RS, Taylor-Robinson AW, 2016. The epidemiology of dengue infection: harnessing past experience and current knowledge to support implementation of future control strategies. J Vector Borne Dis 53: 293304.

    • Search Google Scholar
    • Export Citation
  • 2.

    Instituto Nacional de Salud, 2018. Boletín Epidemiológico Semanal (BES)–Semana Epidemiológica 52 (23 al 29 de diciembre de 2018) [Internet]. Available at: https://www.ins.gov.co/buscador-eventos/BoletinEpidemiologico/2018 Boletín epidemiológico semana 52.pdf. Accessed March 8, 2019.

    • Search Google Scholar
    • Export Citation
  • 3.

    Carabali M, Lim JK, Velez DC, Trujillo A, Egurrola J, Lee KS, Kaufman JS, DaSilva LJ, Velez ID, Osorio JE, 2017. Dengue virus serologic prevalence and seroconversion rates in children and adults in Medellin, Colombia: implications for vaccine introduction. Int J Infect Dis 58: 2736.

    • Search Google Scholar
    • Export Citation
  • 4.

    Villar LA, Rojas DP, Besada-Lombana S, Sarti E, 2015. Epidemiological trends of dengue disease in Colombia (2000–2011): a systematic review. PLoS Negl Trop Dis 9: 116.

    • Search Google Scholar
    • Export Citation
  • 5.

    Martínez-Vega R, Díaz-Quijano FA, Villar-Centeno LA, 2013. Dificultad para el diagnóstico clínico temprano del dengue en un área endémica y su impacto sobre el manejo médico inicial. Rev Med Chil 134: 11531160.

    • Search Google Scholar
    • Export Citation
  • 6.

    Organización Panamericana de la Salud, 2010. Guia para la Atención Clínica Integral del Paciente con Dengue. Bogotá, Colombia: Ministerio de la Protección Social, Dirección General de Salud Pública, 45.

    • Search Google Scholar
    • Export Citation
  • 7.

    World Health Organization (WHO) and the Special Programme for Research and Tropical Diseases (TDR), 2009. Dengue Guidelines for diagnosis, tretment, prevention and control, Geneva, Switzerland: WHO, 147.

    • Search Google Scholar
    • Export Citation
  • 8.

    Goncalves A 2018. Innovative and new approaches to laboratory diagnosis of Zika and dengue: a meeting report. J Infect Dis 217: 10601068.

    • Search Google Scholar
    • Export Citation
  • 9.

    Zhang B, Salieb-Beugelaar GB, Nigo MM, Weidmann M, Hunziker P, 2015. Diagnosing dengue virus infection: rapid tests and the role of micro/nanotechnologies. Nanomedicine 11: 17451761.

    • Search Google Scholar
    • Export Citation
  • 10.

    Osorio L, Uribe M, Ardila GI, Orejuela Y, Velasco M, Bonelo A, Parra B, 2015. The use of rapid dengue diagnostic tests in a routine clinical setting in a dengue-endemic area of Colombia. Mem Inst Oswaldo Cruz 110: 510516.

    • Search Google Scholar
    • Export Citation
  • 11.

    Ahmed NH, Broor S, 2014. Comparison of NS1 antigen detection ELISA, real time RT-PCR and virus isolation for rapid diagnosis of dengue infection in acute phase. J Vector Borne Dis 51:194199.

    • Search Google Scholar
    • Export Citation
  • 12.

    Ayukekbong JA, Oyero OG, Nnukwu SE, Mesumbe HN, Fobisong CN, 2017. Value of routine dengue diagnosis in endemic countries. World J Virol 6: 9.

  • 13.

    Pal S, Dauner AL, Mitra I, Forshey BM, Garcia P, Morrison AC, Halsey ES, Kochel TJ, Wu SJ, 2014. Evaluation of dengue NS1 antigen rapid tests and ELISA kits using clinical samples. PLoS One 9: e113411.

    • Search Google Scholar
    • Export Citation
  • 14.

    Osorio L, Ramirez M, Bonelo A, Villar LA, Parra B, 2010. Comparison of the diagnostic accuracy of commercial NS1-based diagnostic tests for early dengue infection. Virol J 7: 361.

    • Search Google Scholar
    • Export Citation
  • 15.

    Bosch I 2017. Rapid antigen tests for dengue virus serotypes and Zika virus in patient serum. Sci Transl Med 9: eaan1589.

  • 16.

    Baud D, Gubler DJ, Schaub B, Lanteri MC, Musso D, 2017. Review an update on Zika virus infection. Lancet 390: 20992109.

  • 17.

    Van den Bruel A, Cleemput I, Aertgeerts B, Ramaekers D, Buntinx F, 2007. The evaluation of diagnostic tests: evidence on technical and diagnostic accuracy, impact on patient outcome and cost-effectiveness is needed. J Clin Epidemiol 60: 11161122.

    • Search Google Scholar
    • Export Citation
  • 18.

    Guyatt GH, Tugwell PX, Feeny DH, Haynes RB, Drummond M, 1986. A framework for clinical evaluation of diagnostic technologies. CMAJ 134: 587594.

  • 19.

    Sackett DL, Haynes RB, 2002. The architecture of diagnostic research. BMJ 324: 539541.

  • 20.

    Ferrante di Ruffano L, Hyde CJ, McCaffery KJ, Bossuyt PMM, Deeks JJ, 2012. Assessing the value of diagnostic tests: a framework for designing and evaluating trials. BMJ 344: e686

    • Search Google Scholar
    • Export Citation
  • 21.

    Andries AC, Duong V, Ngan C, Ong S, Huy R, Sroin KK, Te V, Y B, Try PL, Buchy P, 2012. Field evaluation and impact on clinical management of a rapid diagnostic kit that detects dengue NS1, IgM and IgG. PLoS Negl Trop Dis 6: e1993.

    • Search Google Scholar
    • Export Citation
  • 22.

    Huits R 2017. Clinical utility of the nonstructural 1 antigen rapid diagnostic test in the management of dengue in returning travelers with fever. Open Forum Infect Dis 4: 16.

    • Search Google Scholar
    • Export Citation
  • 23.

    Clemen G, Angel J, Montes C, Tovar J, Osorio L, 2019. Contribución de la prueba rapida NS1 e IgM al diagnostico de dengue en Colombia en el periodo pre-Zika. Infectio 23: 259265.

    • Search Google Scholar
    • Export Citation
  • 24.

    Kellstein D, Fernandes L, 2019. Symptomatic treatment of dengue: should the NSAID contraindication be reconsidered? Postgrad Med 131: 109116.

    • Search Google Scholar
    • Export Citation
  • 25.

    Bird C, Winzor G, Lemon K, Moffat A, Newton T, Gray J, 2018. A pragmatic study to evaluate the use of a rapid diagnostic test to detect group A Streptococcal pharyngitis in children with the aim of reducing antibiotic use in a UK emergency department. Pediatr Emerg Care Jul 24 [Epub ahead of print]. Available at: https://www.ncbi.nlm.nih.gov/pubmed/30045356.

    • Search Google Scholar
    • Export Citation
  • 26.

    Dodd M, Adolphe A, Parada A, Brett M, Culbreath K, Mercier RC, 2018. Clinical impact of a rapid Streptococcal Antigen Test on antibiotic use in adult patients. Diagn Microbiol Infect Dis 91: 339344.

    • Search Google Scholar
    • Export Citation
  • 27.

    Boyce RM, Muiru A, Reyes R, Ntaro M, Mulogo E, Matte M, Siedner MJ, 2015. Impact of rapid diagnostic tests for the diagnosis and treatment of malaria at a peripheral health facility in western Uganda: an interrupted time series analysis. Malar J 14: 203.

    • Search Google Scholar
    • Export Citation
  • 28.

    Tun ZM, Moorthy M, Linster M, Su YC, Coker RJ, Ooi EE, Low JG, Smith GJ, Tam CC, 2016. Patterns of medication use and factors associated with antibiotic use among adult fever patients at Singapore primary care clinics. Antimicrob Resist Infect Control 5: 47.

    • Search Google Scholar
    • Export Citation
  • 29.

    Lim JK, Alexander N, Di Tanna GL, 2017. A systematic review of the economic impact of rapid diagnostic tests for dengue. BMC Health Serv Res 17: 850.

    • Search Google Scholar
    • Export Citation
  • 30.

    Tai AYC 2017. Management of dengue in Australian travellers: a retrospective multicentre analysis. Med J Aust 206: 295300.

  • 31.

    Torres-Roman JS, Díaz-Vélez C, Bazalar-Palacios J, Helguero-Santin LM, 2016. Hospital management in patients with Dengue: what challenges do we face in Latin America? Infez Med 24: 359360.

    • Search Google Scholar
    • Export Citation
  • 32.

    Robinson ML 2018. Antibiotic utilization and the role of suspected and diagnosed mosquito-borne illness among adults and children with acute febrile illness in Pune, India. Clin Infect Dis 66: 16021609.

    • Search Google Scholar
    • Export Citation
  • 33.

    Bandara SMR, Herath HMMTB, 2018. Effectiveness of corticosteroid in the treatment of dengue–a systemic review. Heliyon 4: e00816.

 

 

 

 

Impact of Dengue Rapid Diagnostic Tests on the Prescription of Antibiotics and Anti-Inflammatory Drugs by Physicians in an Endemic Area in Colombia

View More View Less
  • 1 Grupo de Epidemiología y Salud Poblacional (GESP), School of Public Health, Universidad del Valle, Cali, Colombia

There is insufficient evidence on whether dengue rapid diagnostic tests (dRDTs) influence clinical decisions in endemic areas. Therefore, our objective was to evaluate the impact of dRDTs on the prescription of antibiotics and anti-inflammatory drugs by physicians in a dengue-endemic area in Colombia. A retrospective cohort study was conducted with 330 patients in Cali, Colombia, between January 2012 and December 2017. The exposure was defined by the result of the dRDT. The outcomes were prescription of antibiotics and anti-inflammatory drugs after results of dRDT. Incidence and RR with 95% CIs were estimated. Multivariate logistic regression models were fitted separately for each outcome. Antibiotics were prescribed in 3% exposed and 14% unexposed. Anti-inflammatory drugs were prescribed in 1.2% exposed and 7.9% unexposed. A positive dRDT reduced the prescription of anti-inflammatories (AdjOR: 0.06, 95% CI: 0.008–0.5) but, by itself, had no effect on antibiotics (AdjOR: 1.1, 95% CI: 0.2–6); however, in hospitalized patients, a positive result reduced the probability of antibiotic prescription (AdjOR: 0.02, 95% CI: 0.00–0.8). Despite limitations of current dRDTs, they influence treatment decisions. Further studies are needed to assess the effect of dRDTs in patient outcomes and health-care costs.

INTRODUCTION

Dengue is considered the most important arbovirus infection in the world with around 400 million cases per year.1 The infection is endemic in Colombia with an incidence of 172.9 cases per 100,000 inhabitants in 2018 (non-epidemic year)2 and up to 220 cases per 100,000 inhabitants and 217 deaths in 2010 (epidemic year).3,4 The clinical spectrum of the infection is wide (from self-limiting fever to more severe forms with hemorrhage, shock, and organ involvement) and overlaps with other febrile diseases prevalent in the same endemic areas, which makes their diagnosis more complex.5,6 Diagnosis is relevant for the treatment actions because the physicians must decide if the patient requires antibiotics or no antibiotics (no if it is dengue because dengue viruses do not respond to them) or anti-inflammatory drugs to treat symptoms (as they are not recommended by the WHO in dengue cases because of the potential risk of bleeding).7 However, in other differential diagnoses, such as leptospirosis, malaria, or typhoid fever, the use of antimicrobial agents to treat the infection could be indicated, as well as the use of anti-inflammatory drugs for comfort management in some cases, so this decision may affect the outcome in the patient. This difficulty in clinical diagnosis explains why part of the research in dengue has focused on the development of laboratory tests, particularly rapid diagnostics.8

Currently, dengue rapid diagnostic tests (dRDTs) are based on immunochromatographic techniques that measure the immunoglobulin response and the presence of nonstructural antigens of the virus such as non-structural protein 1 (NS1). This technique yields results in 15 to 20 minutes, is easy to perform, is relatively inexpensive, and is more available in clinical laboratories than reference tests based on reverse-transcription polymerase chain reaction or ELISA.9,10 The rapid tests that detect dengue-specific IgM and IgG have disadvantages that false positives occur because of cross-reactions with antibodies to other flaviviruses (including recent yellow fever vaccination), they have lower sensitivity than ELISA-based tests, and their performance varies between manufacturers.9,11 The sensitivity of the rapid tests that detect NS1 antigen ranges from 40% to 100%, with specificity from 76% to 100%.1214 There is evidence that sensitivity decreases after 3 days of fever onset, in secondary infections and dengue virus 2 and 4. In these situations, the simultaneous detection of NS1 and IgM improves the sensitivity (78.4% 95% CI: 72.2–83.7) without compromising its specificity (91.3% 95% CI: 83.6–96.2); however, a negative result does not rule out dengue.14 In regard to positive results, these may not confirm dengue in the context of co-circulation of other flaviviruses, such as the recent introduction of Zika virus in the American continent, because of cross-reaction.15,16 Therefore, clinicians in endemic areas who use currently available dRDTs should consider that a positive result may not confirm and a negative result does not exclude the diagnosis of dengue.9 In Colombia, dRDTs are not recommended by national guidelines,6 but they have become a tool of frequent use in health services.10 Moreover, whether these dRDTs are helping the clinicians in their treatment decisions is unknown.

Health technology assessment models demand that diagnostic tests not only be studied for their ability to accurately classify the patient’s disease state but also for their ability to reduce the uncertainty of diagnosis, with a consistent and rational treatment decision.17 Several authors have proposed frameworks for the evaluation of technologies in health and the validation of diagnostic tests conceiving the process by hierarchical steps.1719 Likewise, Ferrante di Ruffano et al.20 propose to evaluate the impact of diagnostic tests focused on treatment decision and established management. Considering the aforementioned, we sought to evaluate the impact of dRDTs on the prescription of antibiotics and anti-inflammatory drugs by physicians in an endemic area in Colombia.

MATERIALS AND METHODS

Population and study design.

A retrospective cohort study was conducted with patients who attended nine health-care institutions in the city of Cali (southwest Colombia) between January 2012 and December 2017 and who underwent a dRDT. The eligible participants were men or women of any age, in whose clinical record the therapeutic behavior taken immediately after the result of the dRDT was known. Pregnant patients were not excluded. The main exposure was defined by the result of the rapid test. The patients who had a positive result in either NS1 or IgM or both were the exposed and those who had a negative result in both markers were the unexposed. The tests used by the institution’s clinical laboratory were SD Bioline Dengue Duo (Abbott, Santa Clara, CA, formerly Alere Inc., Waltham, MA) that measures the NS1 antigen plus the immunoglobulins IgM and IgG, and SD Bioline Dengue that measures the immunoglobulins IgM and IgG; both tests yield qualitative results. These rapid tests are routinely used by the institution since 2012. The events under the study were the prescription of two groups of drugs, antibiotics, and anti-inflammatories; the latter included nonsteroidal anti-inflammatory drugs (NSAIDs) and steroids. This study was approved by the Research Ethics Committees of Comfandi and Universidad del Valle.

Sample size, sampling, and data collection.

The sample size was estimated in 330 individuals, 165 in each exposure group, based on 17% antibiotics prescription in exposed, 33% in unexposed,21 90% statistical power, and 95% confidence level. A sampling frame was constructed with the information of the results of all the rapid tests carried out between January 2012 and December 2017 provided by the clinical laboratory of the institution. A list of 4,404 patients with a positive result of the rapid tests and another 16,383 patients with a negative result were obtained. Participants were selected by simple random sampling without replacement in each exposure group.

Data were collected by a single researcher who reviewed clinical records and registered data in Epi Info 7.2.2.2 (US, CDC). The same researcher double-entered 10% of the records each time for quality control. In addition to exposure and outcomes, age, gender, care service, date of onset of symptoms, date of consultation, date of completion of the rapid test, admission diagnoses (as per the International Classification of Diseases, 10th Revision [ICD-10]), differential diagnoses, comorbidities, clinical classification of dengue, leukocyte and platelet counts before or simultaneously with the rapid test, and the presence of respiratory, gastroenteric, or urinary symptoms were recorded.

Statistical analysis.

Absolute and relative frequencies were estimated for categorical variables and means with the corresponding SD or median and range for quantitative variables according to their distribution. Incidences and RR were calculated with their corresponding 95% CI and compared with the chi-squared or Fisher’s exact test when appropriate. Quantitative variables were compared by using the Student’s t-test if the distribution was normal or Mann–Whitney U-test if the distribution was skewed. Stratified analyses were performed to determine the presence of potential confounders and effect modifiers. Multivariate logistic regression models were fitted for each event separately (one for antibiotics and one for anti-inflammatory prescription) using the backward approach and likelihood ratio test. Adjusted OR with 95% CIs and Wald tests were estimated. A P-value < 0.05 was considered statistically significant. The goodness of fit of the regression models was assessed with the Hosmer and Lemeshow test, statistical discrimination tables, and the receiver operating characteristic (ROC) curve; in addition, the regression diagnosis was made for each model through residual analysis. All analyzes were performed in STATA 14 (Stata-Corp., College Station, TX).

RESULTS

A total of 488 clinical records (241 exposed and 247 unexposed) were reviewed, from which 352 (183 exposed and 169 unexposed) were eligible. Of those eligible, 22 were excluded because of errors in the classification of the exposure in the sampling frames (n = 20) and incorrect interpretation of test results by the treating physician (n = 2). The estimated sample size of 165 participants for each exposure group was included in the analysis (Figure 1).

Figure 1.
Figure 1.

Recruitment diagram of the cohort participants.

Citation: The American Journal of Tropical Medicine and Hygiene 101, 3; 10.4269/ajtmh.19-0222

Description of the cohort.

Most participants were young adults, were without comorbidities, were with ≤ 5 days of fever, had non-severe dengue, and were attending ambulatory services. However, those exposed were younger (median = 23 years), more frequently classified as having dengue (n = 133, 80.6%), had alarm signs (n = 75, 45.4%), and had lower platelet (median = 120,000/μL) and leukocyte (median = 4,730/μL) counts than those unexposed. Most participants, both exposed (n = 75, 45.4%) and unexposed (n = 49, 29.7%) were treated at the “H” institution, which is the one with the highest level of care. There were more exposed in the year 2013 (n = 74, 44.8%) and more unexposed in 2015 (n = 65, 39.4%). Despite the sampling technique, no unexposed patients were included in the year 2017 (Table 1). The frequency of prescription of antibiotics in the entire cohort was 8.5% (95% CI: 6–12) and of anti-inflammatory drugs was 4.5% (95% CI: 2.7–7.4). The antibiotics were penicillin (n = 8), cephalosporins (n = 7), fluoroquinolones (n = 2), macrolides (n = 2), and others (n = 8). Whereas, NSAIDs were prescribed only in the unexposed group (dipyrone in five, diclofenac in two, and ibuprofen, naproxen, or ketorolac in four patients), steroids, such as dexamethasone, hydrocortisone, betamethasone, or methylprednisolone, were prescribed in both exposed (n = 2) and unexposed (n = 2).

Table 1

Baseline characteristics of participants in the exposed and unexposed groups

CharacteristicIgM/NS1 positiveIgM/NS1 negative
(Exposed), n = 165(Unexposed), n = 165
Gender (%)
 Female80 (48.5)87 (52.7)
 Male85 (51.5)78 (47.3)
Age in years (Me, IQR)23 (14–39)33 (19–51)
Age groups (%)
 ≤ 15 years53 (32.2)35 (21.2)
 16–44 years75 (45.4)75 (45.4)
 45–64 years29 (17.6)43 (26.1)
 ≥ 65 years8 (4.8)12 (7.3)
Comorbidities (%)
 No148 (89.7)153 (92.7)
 Yes17 (10.3)12 (7.3)
Care service (%)
 Ambulatory113 (68.5)142 (86.1)
 Hospitalization49 (29.7)22 (13.3)
 Intensive care unit3 (1.8)1 (0.6)
Classification of dengue (%)
 Without alarm signs50 (30.3)66 (40)
 With alarm signs75 (45.5)37 (22.4)
 Severe1 (0.6)
 No case39 (23.6)62 (37.6)
Days of fever (X¯, SD)5.13 (±2.3)5.69 (±4.3)
 ≤ 5 (%)101 (61.2)94 (57)
 ≥ 6 (%)64 (38.8)71 (43)
Platelets —103/μL (Me, IQR)120 (56–169)180 (130–253.5)
Categories of platelets (%)n = 164
 150 to ≥ 45056 (34)106 (64.6)
 100–14940 (24.2)31 (18.9)
 ≤ 99.969 (41.8)27 (16.5)
Leukocytes—103/μL (Me, IQR)4.73 (3.45–6.39)5.4 (4.08–7.72)
Categories of leukocytes (%)n = 163n = 161
 4–1098 (60.1)108 (67)
 ≤ 3.9959 (36.2)36 (22.4)
 ≥ 10.16 (3.7)17 (10.6)
Respiratory symptoms (%)
 No150 (90.9)127 (77)
 Yes15 (9.1)38 (23)
Gastrointestinal symptoms (%)
 No113 (68.5)122 (74)
 Yes52 (31.5)43 (26)
Urinary symptoms (%)
 No158 (95.8)161 (97.6)
 Yes7 (4.2)4 (2.4)
ICD-10 on admission (%)
 Dengue133 (80.6)59 (35.8)
 Other32 (19.4)106 (64.2)
Prescription of antibiotics (%)
 No160 (97)142 (86)
 Yes5 (3)23 (14)
Prescription of anti-inflammatories (%)
 No163 (98.8)152 (92.1)
 Yes2 (1.2)13 (7.9)
Institution (%)
 A18 (11)19 (11.5)
 B19 (11.5)30 (18.2)
 C8 (5)6 (3.6)
 D2 (1.2)1 (0.6)
 E8 (4.8)7 (4.2)
 F20 (12.1)23 (13.9)
 G5 (3)5 (3)
 H75 (45.4)49 (29.7)
 I10 (6)25 (15.3)
Year of the test (%)
 20124 (2.4)8 (4.8)
 201374 (44.8)23 (14)
 201418 (11)27 (16.4)
 201532 (19.4)65 (39.4)
 201636 (21.8)42 (25.4)
 20171 (0.6)

IQR = interquartile range; Me = median; X¯ = mean.

Prescription of antibiotics.

Antibiotics were prescribed in 5 (3%) exposed and 23 (14%) unexposed, resulting in a decreased risk of being prescribed antibiotics (RR: 0.2; 95% CI: 0.08–0.5) with a positive test. By contrast, having urinary symptoms (RR: 6.3; 95% CI: 3–13.4), having leukocyte count ≥ 10,000/μL (RR: 5.7; 95% CI: 2.8–11.8), having an admission diagnosis different from dengue (RR: 4.2; 95% CI: 1.8–9.5), having respiratory symptoms (RR: 3.4; 95% CI: 1.7–6.8), not meeting the case definition for dengue (RR: 3.1; 95% CI: 1.2–7.5), and being hospitalized or admitted to the intensive care unit (ICU) (RR: 2.2; 95% CI: 1.1–4.5) had an increased risk for the prescription of antibiotics (Table 2).

Table 2

Bivariate analysis of factors associated with the prescription of antibiotics

CharacteristicAntibioticsAntibioticsRRP-value
NoYes
n (%)n (%)(95% CI)
Rapid test result
 Negative142 (86)23 (14)1< 0.0001
 Positive160 (97)5 (3)0.2 (0.08–0.5)
Gender
 Female156 (93.4)11 (6.6)10.2
 Male146 (89.6)17 (10.4)1.58 (0.8–3.3)
Age in years X¯ (SD)30.9 (±19.7)37.1 (±24.1)0.1
Age groups
 ≤ 15 years83 (94.3)5 (5.7)10.2
 16–44 years136 (90.7)14 (9.3)1.6 (0.6–4.4)
 45–64 years67 (93)5 (7)1.2 (0.4–4)
 ≥ 65 years16 (80)4 (20)3.5 (1.03–12)
Comorbidities
 No277 (92)24 (8)10.3
 Yes25 (86.2)4 (13.8)1.7 (0.6–4.6)
Care service
 Ambulatory238 (93.3)17 (6.7)10.03
 Hospitalization/ICU64 (85.3)11 (14.7)2.2 (1.07–4.5)
Classification of dengue
 Without alarm110 (94.8)6 (5.2)10.01
 With alarm106 (94.6)6 (5.4)1 (0.3–3.1)
 Severe1 (100)0 (0)0
 No case85 (84.2)16 (15.8)3 (1.2–7.5)
Days of fever X¯ (SD)5.4 (±3.4)5.2 (±3)0.7
 ≤ 5177 (90.8)18 (9.2)10.5
 ≥ 6125 (92.6)10 (7.4)0.8 (0.4–1.7)
Platelets—103/μL Me (IQR)147 (79–206)203 (101–289.5)0.01
Categories of platelets (103/μL) 
 150–≥ 450143 (88.3)19 (11.7)10.1
 100–14968 (95.8)3 (4.2)0.4 (0.1–1.2)
 ≤ 99.990 (93.7)6 (6.3)0.5 (0.2–1.3)
Leukocytes—103/μL Me (IQR)5 (4–6.7)7.1 (5–12.1)0.0002
Categories of leukocytes (103/μL)
 4–10192 (93.2)14 (6.8)1< 0.0001
 ≤ 3.9992 (96.8)3 (3.2)0.5 (0.1–1.6)
 ≥ 10.114 (60.9)9 (39.1)5.7 (2.8–11.8)
Respiratory symptoms
 No260 (93.9)17 (6.1)1< 0.0001
 Yes42 (79.2)11 (20.8)3.4 (1.7–6.8)
Gastrointestinal symptoms
 No219 (93.2)16 (6.8)10.08
 Yes83 (87.4)12 (12.6)1.8 (0.9–3.8)
Urinary symptoms
 No296 (92.8)23 (7.2)1< 0.0001
 Yes6 (54.5)5 (45.5)6.3 (3–13.4)
ICD-10 before the dengue test
 Dengue185 (96.3)7 (3.7)1< 0.0001
 Other117 (84.8)21 (15.2)4.2 (1.8–9.5)
Institution’s level of care
 Outpatient109 (94)7 (6)10.07
 Priority/emergency76 (95)4 (5)0.8 (0.2–2.7)
 Hospitalization117 (87.3)17 (12.7)2.1 (0.9–4.9)
Year of the test
 201211 (91.7)1 (8.3)10.08
 201389 (91.7)8 (8.3)1 (0.1–7.2)
 201440 (88.9)5 (11.1)1.3 (0.2–10.4)
 201594 (96.9)3 (3.1)0.4 (0.04–3.3)
 2016–201768 (86.1)11 (13.9)1.7 (0.2–11.8)

ICU = intensive care unit; X¯ = mean; IQR = interquartile range; Me = median; SD = standard deviation.

During the stratified analysis, an interaction was found between hospital admission and the dRDT result. This interaction was statistically significant in the multivariate model, which showed that the rapid test result by itself was not associated with the prescription of antibiotics; however, a positive test was an independent protective factor in hospitalized subjects (OR: 0.02; 95% CI: 0.0–0.8). By contrast, being hospitalized independently increased the risk of antibiotic prescription (OR: 9.8; 95% CI: 2.3–42.1). Leukocyte count ≥ 10.000/μL, urinary symptoms, and an ICD-10 diagnosis other than dengue were also risk factors for antibiotic prescription (Table 3). This model fitted relatively well (ROC: 0.93, Hosmer and Lemeshow test P = 0.68, pseudo R2: 0.47).

Table 3

Multivariate analysis of factors associated with the prescription of antibiotics

CharacteristicCrude ORP-valueAdjusted ORP-value
(95% CI)(95% CI)
Rapid test result
 Negative10.00110.9
 Positive0.2 (0.07–0.5)1.1 (0.2–6)
Service
 Ambulatory10.0310.002
 Hospitalization/ICU2.4 (1.1–5.4)9.8 (2.3–42.1)
Interaction term
 Exposed and service0.02 (0–0.8)0.04
Classification of dengue
 Without alarm signs11
 With alarm/severe1 (0.3–3.3)0.91.5 (0.3–9.2)0.6
 No case3.4 (1.3–9.2)0.014 (0.8–21)0.1
Categories of leukocytes (103/μL)
 4–1011
 ≤ 3.990.4 (0.1–1.6)0.20.2 (0.04–1.2)0.09
 ≥ 10.18.8 (3.2–24)< 0.00017.3 (1.9–28.4)0.004
Urinary symptoms
 No11
 Yes10.7 (3–37.8)< 0.0001108.8 (15.3–771)< 0.0001
ICD-10 before the dengue test
 Dengue11
 Other4.7 (2–11.5)0.00114.3 (2.7–75.7)0.002

Prescription of anti-inflammatory drugs.

Regarding anti-inflammatory drugs, they were prescribed in 2 (1.2%) exposed and 13 (7.9%) unexposed, resulting in a decreased risk of prescription with a positive test result (RR: 0.15; 95% CI: 0.03–0.7). By contrast, the presence of comorbidities such as respiratory and rheumatological diseases (RR: 3.8; 95% CI: 1.3–11.1) and having respiratory symptoms (RR: 3.5; 95% CI: 1.3–9.4) were associated with an increased risk of being prescribed anti-inflammatory drugs. Gastrointestinal symptoms were also associated with the prescription (P = 0.007), but because no subject with gastrointestinal symptoms was prescribed anti-inflammatory drugs, the relative risk was zero (Table 4).

Table 4

Bivariate analysis of factors associated with the prescription of anti-inflammatory drugs

CharacteristicAnti-inflammatoriesAnti-inflammatoriesRRP-value
NoYes
n (%)n (%)(95% CI)
Rapid test result
 Negative152 (92.1)13 (7.9)10.006
 Positive163 (98.8)2 (1.2)0.15 (0.03–0.7)
Gender
 Female157 (94)10 (6)10.3
 Male158 (97)5 (3)0.5 (0.2–1.5)
Age in years X¯ (SD)31.2 (±20.4)36.4 (±14.7)0.3
Age groups
 ≤ 15 years88 (100)0 (0)00.03
 16–44 years140 (93.3)10 (6.7)1
 45–64 years67 (93)5 (7)1 (0.5–2.2)
 ≥ 65 years20 (100)0 (0)0
Comorbidities
 No290 (96.3)11 (3.7)10.03
 Yes25 (86.2)4 (13.8)3.8 (1.3–11.1)
Service
 Ambulatory245 (96)10 (4)10.3
 Hospitalization/ICU70 (93.3)5 (6.7)1.7 (0.6–4.8)
Classification of dengue
 Without alarm signs109 (94)7 (6)10.5
 With alarm signs109 (97.3)3 (2.7)0.4 (0.1–1.7)
 Severe1 (100)0 (0)0
 No case96 (95)5 (5)0.8 (0.3–2.5)
Days of fever X¯ (SD)5.3 (±3.4)7.3 (±3.7)0.02
 ≤ 5188 (96.4)7 (3.6)10.3
 ≥ 6127 (94)8 (6)1.6 (0.6–4.4)
Platelets—103/μL Me (IQR)158 (±96.7)189.5 (±99.2)0.6
Categories of platelets (103/μL)
 150 to ≥ 450152 (93.8)10 (6.2)10.3
 100–14970 (98.6)1 (1.4)0.2 (0.02–1.7)
 ≤ 99.992 (95.8)4 (4.2)0.7 (0.2–2.1)
Leukocytes—103/μL Me (IQR)5 (3.6–6.7)8 (3.4-9)0.05
Categories of leukocytes (103/μL)
 4–10198 (96.1)8 (3.9)10.2
 ≤ 3.9991 (95.8)4 (4.2)1.1 (0.3-3.5)
 ≥ 1020 (87)3 (13)3.3 (0.9-11.8)
Respiratory symptoms
 No268 (96.7)9 (3.3)10.02
 Yes47 (88.7)6 (11.3)3.5 (1.3–9.4)
Gastrointestinal symptoms
 No220 (93.6)15 (6.4)10.007
 Yes95 (100)0 (0)0
Urinary symptoms
 No304 (95.3)15 (4.7)11
 Yes11 (100)0 (0)0
ICD-10 before dengue test
 Dengue186 (98.9)6 (3.1)10.1
 Other129 (93.5)9 (6.5)2.1 (0.8–5.7)
Institution’s level of care
 Outpatient112 (96.5)4 (3.5)10.3
 Priority/emergency78 (97.5)2 (2.5)0.72 (0.1–3.9)
 Hospitalization125 (93.4)9 (6.6)2 (0.6–6.1)
Year of the test
 201211 (91.7)1 (8.3)10.2
 201394 (96.9)3 (3.1)0.4 (0.04–3.3)
 201445 (100)0 (0)0
 201592 (94.8)5 (5.2)0.6 (0.07–4.9)
 2016–201773 (92.4)6 (7.6)0.9 (0.1–7)

X¯ = mean; IQR = interquartile range; Me = median; SD = standard deviation.

In the multivariate model, a positive test result was a protective factor for the prescription of anti-inflammatory drugs (OR: 0.06; 95% CI: 0.008–0.5). Conversely, comorbidities behaved as a risk factor (OR: 6.2; 95% CI: 1.5–24.5). The leukocyte counts also entered the model, but showed no association with the event (Table 5). The goodness of fit of this model was not as good as the model for antibiotics (ROC: 0.78, Hosmer and Lemeshow test P = 0.46, pseudo R2: 0.17) even after an influential observation identified during model post estimation of residuals was withdrawn.

Table 5

Multivariate analysis of factors associated with the prescription of anti-inflammatory drugs

CharacteristicCrude ORP-valueAdjusted ORP-value
(95% CI)(95% CI)
Result of the test
 Negative10.0110.01
 Positive0.14 (0.03–0.6)0.06 (0.008–0.5)
Comorbidities
 No10.0210.01
 Yes4.2 (1.25–14.21)6.2 (1.5–24.5)
Leukocytes (103/μL)1 (1–1)0.031 (1–1)0.3

DISCUSSION

The present study evaluated the effect of dRDT on the decision to use antibiotics and anti-inflammatory drugs by physicians in a dengue-endemic area in Colombia. The results showed that a positive result on a dRDT reduced the risk of prescribing anti-inflammatory drugs and antibiotics, the latter only in the context of hospitalized patients. The converse interpretation is also valid: that a negative result in the dRDT increased the risk of prescribing anti-inflammatory drugs and increased the risk of prescribing antibiotics, the latter only in hospitalized patients. These findings suggest that in our setting, dRDTs are being used to make significant clinical decisions (i.e., use of antibiotics and anti-inflammatory drugs). This study joins others that have assessed the impact of dRDTs in clinical management with contrasting results. For instance, in hospitalized children in Cambodia, non-statistically significant differences were found in the frequency of antibiotics prescription among dRDT positives and negatives; however, the sample size was relative small to detect differences (66 positives and 39 negatives) and potential confounders were not taken into account.21 In returning travelers to Belgium, empirical treatment with antibiotics and hospital admission were less frequent when NS1-based tests (2/43 and 3/43, respectively) were introduced than in historical controls using tests that detect only dengue-specific antibodies (11/43 and 14/43, respectively).22 This effect was attributed to the capacity of NS1-based dRDTs to confirm dengue in non-endemic areas where the subjects are more likely to harbor primary infection and seek care in the first days of fever.22 Hence, in that setting, physicians would trust that a positive result confirms dengue and would decrease the use of antibiotics. There were few NS1 tests in our study (n = 19) to assess the differences between tests that detect NS1 or antibodies. In Colombia, we have also shown that a positive result in a dRDT in patients with a clinical diagnosis of dengue confirms it, but this was before Zika introduction.23 Therefore, performance of dRDTs in the post-Zika era needs to be assessed to inform physicians on the interpretation of these tests. In any case, in both endemic and non-endemic areas, a negative dRDT (NS1 and IgM single or combined) result does not rule out dengue and should not be used by itself to inform clinical decisions.

To ascertain whether prescriptions of antibiotics and anti-inflammatory drugs were appropriate or not was beyond the scope of the present study. However, the fact that dRDTs are not useful to rule out dengue means that a proportion of negative results are actually false negatives. Hence, some patients with false negative dRDTs would be put at risk of bleeding with the use of NSAIDs such as dipyrone.24 The use of dRDTs in the decision to prescribe antibiotics in the hospital context is particularly relevant because of the potential impact on increased antibiotic drug consumption and resistance with false-negative results in dRDTs. Conversely, false-positive results in the dRDT could prevent patients from receiving targeted management for their disease. The impact of rapid diagnostic tests on the use of antibiotics has been observed in other infectious diseases such as group A streptococcal pharyngitis, with reductions from 79% to 27% in children with pain of the throat in emergency services,25 or from 22% to 8.3% in adults.26 In febrile syndromes similar to dengue such as malaria, a reduction of antibiotic use from 75% to 53.6% was found after the introduction of a malaria rapid diagnostic test in rural hospitals in Uganda.27 Hence, the various contexts in which rapid diagnostic tests contribute to the rational use of antibiotics could be further explored.

In spite of dRDTs influencing patient management, the overall frequency of prescription of antibiotics (8.5%) in the present study was relative low compared with 17% in dengue positives and 33% in dengue negatives in hospitalized children aged less than 12 years in Cambodia21 and 16% acute febrile adults in Singapore.28 Other measuring patient outcomes, such as mortality, hospital length of stay, and severity of disease, are required to continue the evaluation of dRDTs in Colombia and other endemic and non-endemic areas. Further health-economic studies would shed light on the contribution of dRDTs to patient outcomes, rational use of antibiotics, and other health resources.29

The prescription of anti-inflammatory drugs was even lower (4.5%) than 22% in Australia30 and 83% in Peru.31 Patient-, physician-, and institutional-related factors would further influence drug prescription patterns. The urinary symptoms were the strongest risk factor for the use of antibiotics in this study, as was found with the white blood cell count ≥ 10,000 cells/μL; the presence of these factors could suggest bacterial infection. Likewise, a clinical diagnosis different to dengue could point toward a bacterial etiology as found by Robinson et al. in India where patients with diagnosis of dengue or malaria on admission were less likely to be prescribed antibiotics (P < 0.01), whereas those with pneumonia (P = 0.01) were more likely.32 Others have reported that besides leukocytosis (> 11.0 × 109 cells/L) and patient symptoms, year of enrolment, health-care institution, body mass index, housing type, and headache were independently associated with antibiotic prescription in acute febrile adults.28 We did not find associations between the prescription of antibiotics or anti-inflammatories and the year when tests were performed or health-care institution, probably because our study was conducted in institutions belonging to a single health insurer with some degree of care standardization. Relative homogeneity would be expected among our study participants in socioeconomical conditions because of insurer population coverage, but we did not explore this. All these characteristics plus emerging symptoms during follow-up and other laboratory results (including other diagnostic tests) could be assessed in future studies. This is particularly relevant to improve the knowledge of factors associated with the prescription of anti-inflammatory drugs as our model had low explanatory power. The use of steroids is still controversial and not widely recommended, as several studies have shown that they do not affect clinical outcomes such as mortality.33 The observed increased prescription of anti-inflammatories in patients with comorbidities (adjOR: 6.2, 95% CI: 1.5–24.5) may be due to the fact that most comorbidities were respiratory (3.3%) and rheumatological (1.5%), in which the use of anti-inflammatory drugs could be indicated, as is the case of exacerbated asthma or arthritis.

One limitation in the present study was that the two events were less frequent than what was previously reported and used to estimate sample size; therefore, the sample was insufficient to achieve greater accuracy of the estimates and affect statistical power to detect differences in variables, such as age groups, the clinical classification of dengue, and the presence of gastrointestinal symptoms. Children (aged < 15 years) were not adequately represented, so the results of this study cannot necessarily be applied to pediatric patients. The anti-inflammatory model has limitations in its ability to explain and predict the behavior of the prescription of NSAIDs and steroids; therefore, more research with additional variables is needed.

In conclusion, in spite of the limitations of currently available dRDTs to rule in and rule out dengue, they are being used and their results are influencing medical decisions such as the prescription of antibiotics and anti-inflammatory drugs. Further studies are needed to assess the effect of dengue rapid tests in patient outcomes and health-care costs.

Acknowledgments:

We would like to thank Comfandi for facilitating the research scenario and their collaboration to obtain the data.

REFERENCES

  • 1.

    Gyawali N, Bradbury RS, Taylor-Robinson AW, 2016. The epidemiology of dengue infection: harnessing past experience and current knowledge to support implementation of future control strategies. J Vector Borne Dis 53: 293304.

    • Search Google Scholar
    • Export Citation
  • 2.

    Instituto Nacional de Salud, 2018. Boletín Epidemiológico Semanal (BES)–Semana Epidemiológica 52 (23 al 29 de diciembre de 2018) [Internet]. Available at: https://www.ins.gov.co/buscador-eventos/BoletinEpidemiologico/2018 Boletín epidemiológico semana 52.pdf. Accessed March 8, 2019.

    • Search Google Scholar
    • Export Citation
  • 3.

    Carabali M, Lim JK, Velez DC, Trujillo A, Egurrola J, Lee KS, Kaufman JS, DaSilva LJ, Velez ID, Osorio JE, 2017. Dengue virus serologic prevalence and seroconversion rates in children and adults in Medellin, Colombia: implications for vaccine introduction. Int J Infect Dis 58: 2736.

    • Search Google Scholar
    • Export Citation
  • 4.

    Villar LA, Rojas DP, Besada-Lombana S, Sarti E, 2015. Epidemiological trends of dengue disease in Colombia (2000–2011): a systematic review. PLoS Negl Trop Dis 9: 116.

    • Search Google Scholar
    • Export Citation
  • 5.

    Martínez-Vega R, Díaz-Quijano FA, Villar-Centeno LA, 2013. Dificultad para el diagnóstico clínico temprano del dengue en un área endémica y su impacto sobre el manejo médico inicial. Rev Med Chil 134: 11531160.

    • Search Google Scholar
    • Export Citation
  • 6.

    Organización Panamericana de la Salud, 2010. Guia para la Atención Clínica Integral del Paciente con Dengue. Bogotá, Colombia: Ministerio de la Protección Social, Dirección General de Salud Pública, 45.

    • Search Google Scholar
    • Export Citation
  • 7.

    World Health Organization (WHO) and the Special Programme for Research and Tropical Diseases (TDR), 2009. Dengue Guidelines for diagnosis, tretment, prevention and control, Geneva, Switzerland: WHO, 147.

    • Search Google Scholar
    • Export Citation
  • 8.

    Goncalves A 2018. Innovative and new approaches to laboratory diagnosis of Zika and dengue: a meeting report. J Infect Dis 217: 10601068.

    • Search Google Scholar
    • Export Citation
  • 9.

    Zhang B, Salieb-Beugelaar GB, Nigo MM, Weidmann M, Hunziker P, 2015. Diagnosing dengue virus infection: rapid tests and the role of micro/nanotechnologies. Nanomedicine 11: 17451761.

    • Search Google Scholar
    • Export Citation
  • 10.

    Osorio L, Uribe M, Ardila GI, Orejuela Y, Velasco M, Bonelo A, Parra B, 2015. The use of rapid dengue diagnostic tests in a routine clinical setting in a dengue-endemic area of Colombia. Mem Inst Oswaldo Cruz 110: 510516.

    • Search Google Scholar
    • Export Citation
  • 11.

    Ahmed NH, Broor S, 2014. Comparison of NS1 antigen detection ELISA, real time RT-PCR and virus isolation for rapid diagnosis of dengue infection in acute phase. J Vector Borne Dis 51:194199.

    • Search Google Scholar
    • Export Citation
  • 12.

    Ayukekbong JA, Oyero OG, Nnukwu SE, Mesumbe HN, Fobisong CN, 2017. Value of routine dengue diagnosis in endemic countries. World J Virol 6: 9.

  • 13.

    Pal S, Dauner AL, Mitra I, Forshey BM, Garcia P, Morrison AC, Halsey ES, Kochel TJ, Wu SJ, 2014. Evaluation of dengue NS1 antigen rapid tests and ELISA kits using clinical samples. PLoS One 9: e113411.

    • Search Google Scholar
    • Export Citation
  • 14.

    Osorio L, Ramirez M, Bonelo A, Villar LA, Parra B, 2010. Comparison of the diagnostic accuracy of commercial NS1-based diagnostic tests for early dengue infection. Virol J 7: 361.

    • Search Google Scholar
    • Export Citation
  • 15.

    Bosch I 2017. Rapid antigen tests for dengue virus serotypes and Zika virus in patient serum. Sci Transl Med 9: eaan1589.

  • 16.

    Baud D, Gubler DJ, Schaub B, Lanteri MC, Musso D, 2017. Review an update on Zika virus infection. Lancet 390: 20992109.

  • 17.

    Van den Bruel A, Cleemput I, Aertgeerts B, Ramaekers D, Buntinx F, 2007. The evaluation of diagnostic tests: evidence on technical and diagnostic accuracy, impact on patient outcome and cost-effectiveness is needed. J Clin Epidemiol 60: 11161122.

    • Search Google Scholar
    • Export Citation
  • 18.

    Guyatt GH, Tugwell PX, Feeny DH, Haynes RB, Drummond M, 1986. A framework for clinical evaluation of diagnostic technologies. CMAJ 134: 587594.

  • 19.

    Sackett DL, Haynes RB, 2002. The architecture of diagnostic research. BMJ 324: 539541.

  • 20.

    Ferrante di Ruffano L, Hyde CJ, McCaffery KJ, Bossuyt PMM, Deeks JJ, 2012. Assessing the value of diagnostic tests: a framework for designing and evaluating trials. BMJ 344: e686

    • Search Google Scholar
    • Export Citation
  • 21.

    Andries AC, Duong V, Ngan C, Ong S, Huy R, Sroin KK, Te V, Y B, Try PL, Buchy P, 2012. Field evaluation and impact on clinical management of a rapid diagnostic kit that detects dengue NS1, IgM and IgG. PLoS Negl Trop Dis 6: e1993.

    • Search Google Scholar
    • Export Citation
  • 22.

    Huits R 2017. Clinical utility of the nonstructural 1 antigen rapid diagnostic test in the management of dengue in returning travelers with fever. Open Forum Infect Dis 4: 16.

    • Search Google Scholar
    • Export Citation
  • 23.

    Clemen G, Angel J, Montes C, Tovar J, Osorio L, 2019. Contribución de la prueba rapida NS1 e IgM al diagnostico de dengue en Colombia en el periodo pre-Zika. Infectio 23: 259265.

    • Search Google Scholar
    • Export Citation
  • 24.

    Kellstein D, Fernandes L, 2019. Symptomatic treatment of dengue: should the NSAID contraindication be reconsidered? Postgrad Med 131: 109116.

    • Search Google Scholar
    • Export Citation
  • 25.

    Bird C, Winzor G, Lemon K, Moffat A, Newton T, Gray J, 2018. A pragmatic study to evaluate the use of a rapid diagnostic test to detect group A Streptococcal pharyngitis in children with the aim of reducing antibiotic use in a UK emergency department. Pediatr Emerg Care Jul 24 [Epub ahead of print]. Available at: https://www.ncbi.nlm.nih.gov/pubmed/30045356.

    • Search Google Scholar
    • Export Citation
  • 26.

    Dodd M, Adolphe A, Parada A, Brett M, Culbreath K, Mercier RC, 2018. Clinical impact of a rapid Streptococcal Antigen Test on antibiotic use in adult patients. Diagn Microbiol Infect Dis 91: 339344.

    • Search Google Scholar
    • Export Citation
  • 27.

    Boyce RM, Muiru A, Reyes R, Ntaro M, Mulogo E, Matte M, Siedner MJ, 2015. Impact of rapid diagnostic tests for the diagnosis and treatment of malaria at a peripheral health facility in western Uganda: an interrupted time series analysis. Malar J 14: 203.

    • Search Google Scholar
    • Export Citation
  • 28.

    Tun ZM, Moorthy M, Linster M, Su YC, Coker RJ, Ooi EE, Low JG, Smith GJ, Tam CC, 2016. Patterns of medication use and factors associated with antibiotic use among adult fever patients at Singapore primary care clinics. Antimicrob Resist Infect Control 5: 47.

    • Search Google Scholar
    • Export Citation
  • 29.

    Lim JK, Alexander N, Di Tanna GL, 2017. A systematic review of the economic impact of rapid diagnostic tests for dengue. BMC Health Serv Res 17: 850.

    • Search Google Scholar
    • Export Citation
  • 30.

    Tai AYC 2017. Management of dengue in Australian travellers: a retrospective multicentre analysis. Med J Aust 206: 295300.

  • 31.

    Torres-Roman JS, Díaz-Vélez C, Bazalar-Palacios J, Helguero-Santin LM, 2016. Hospital management in patients with Dengue: what challenges do we face in Latin America? Infez Med 24: 359360.

    • Search Google Scholar
    • Export Citation
  • 32.

    Robinson ML 2018. Antibiotic utilization and the role of suspected and diagnosed mosquito-borne illness among adults and children with acute febrile illness in Pune, India. Clin Infect Dis 66: 16021609.

    • Search Google Scholar
    • Export Citation
  • 33.

    Bandara SMR, Herath HMMTB, 2018. Effectiveness of corticosteroid in the treatment of dengue–a systemic review. Heliyon 4: e00816.

Author Notes

Address correspondence to María Elena Tello-Cajiao, Universidad del Valle, Bldg. 118, Calle 4B 36-00, San Fernando Campus, Cali, Colombia. E-mail: tello.maria@correounivalle.edu.co

Financial support: This study was partially funded by Fondo de Ciencia, Tecnología e Innovación-FCTeI de SGR, Colombia-BPIN 2013000100011, Red AEDES, and Universidad del Valle.

Authors’ addresses: María Elena Tello-Cajiao and Lyda Osorio, Grupo de Epidemiología y Salud Poblacional (GESP), School of Public Health, Universidad del Valle, Cali, Colombia, E-mails: tello.maria@correounivalle.edu.co and lyda.osorio@correounivalle.edu.co.

Save