Study Design
This pragmatic prospective cohort study was conducted from April 2019 to March 2020 at a primary care public hospital in Cali, Colombia, to assess the role of point-of-care ultrasound in dengue diagnosis. Clinical staff were instructed to alert the research team (comprising a study physician and a field assistant) upon identifying patients suspected of having dengue. Clinicians used the 2009 World Health Organization (WHO) criteria [8] for probable dengue (fever plus two of: nausea/vomiting, rash, aches/pains, positive tourniquet test, leukopenia, any warning sign), dengue with or without warning signs (abdominal pain/tenderness, persistent vomiting, clinical fluid accumulation, mucosal bleeding, lethargy/restlessness, liver enlargement > 2 cm, increased hematocrit with decreased platelet count), or severe dengue (severe plasma leakage leading to shock or respiratory distress, severe bleeding, and severe organ involvement). Importantly, no dengue confirmatory tests were performed for study purposes, mirroring the constraints of real-world primary care settings.
Men and women older than 2 years with a reported fever lasting less than 10 days and a physician’s diagnosis of dengue were consecutively enrolled. Exclusion criteria encompassed current pregnancy, contraindications to ultrasound (e.g., skin injury at the scan site), fever attributed to other causes, comorbidities predisposing to third-spacing (e.g., liver failure, heart failure, cancer), and conditions requiring immediate intervention that ultrasound examination might delay. Eligible participants underwent point of care ultrasound (POCUS) examination and were followed for two weeks post-enrollment to determine clinical outcomes. The target sample size was 369 individuals, calculated based on an anticipated 40% prevalence of plasma leakage at enrollment—lower than the 60% reported in tertiary care settings [9], with a 5% significance level and 5% precision. Recruitment concluded in March 2020 due to national lockdown measures enacted in response to the COVID-19 pandemic. To maintain clinical equipoise, ultrasound findings were not routinely disclosed to treating physicians unless suggestive of the need for further imaging. Ethical approvals were secured from the Ethics Committee of Universidad del Valle and the Hospital Joaquín Paz Borrero (HJPB), as well as the University of Minnesota (STUDY00004437). Written informed consent was obtained from all participants, and from parents/guardians for minors.
Data Collection
Data collection was facilitated using REDCap electronic data capture tools, hosted at the University of Minnesota [10], employing a pre-designed, encrypted case record form. Initially, a study physician gathered demographic details, conducted clinical interviews, and performed physical examinations to classify participants’ dengue status according to WHO guidelines [8]. Subsequently, point of care ultrasound (POCUS) was performed using a standardized protocol derived from the Focused Assessment with Sonography for Trauma (FAST) exam [11]. Laboratory results obtained during any hospital admission were recorded in REDCap. These included physician-ordered tests like rapid dengue IgM and IgG tests (SD BIOLINE Dengue IgG/IgM, Standard Diagnostics, Republic of Korea), along with maximum hematocrit and monocyte values, and minimum leukocyte, lymphocyte, and platelet counts during the current illness episode. NS1-based tests were not available at this facility. Participant clinical progression, need for hospitalization or referral, and final diagnoses were ascertained through telephone follow-up roughly 14 days post-enrollment. For participants unreachable by phone, medical records were reviewed to determine outcomes.
Ultrasound Examination
Training
The study’s general physician, without prior ultrasound experience, underwent focused training in point of care ultrasound (POCUS) performance and interpretation. This training adhered to guidelines from the Society of Point of Care Ultrasound (SPOCUS) and the American College of Emergency Physicians (ACEP) [12, 13]. The training consisted of a 3-hour online module covering basic ultrasound principles, proper technique, and image interpretation, followed by 30 hours (over 5 days) of hands-on practice. Practice sessions were conducted on 4 healthy volunteers and 53 hospitalized patients, primarily with cardiovascular conditions including cardiac failure but excluding dengue. The trainee acquired a total of 627 POCUS images during this period. Training was provided in Cali, Colombia, by a single instructor: a U.S.-based emergency medicine physician and ultrasound trainer from Hennepin County Medical Center and the University of Minnesota. An additional 6-hour practice session, led by the study’s expert radiologist, was conducted 10 months into the study to refine technique and image interpretation.
Equipment, Image Recording and Interpretation
A Philips Lumify C5-2 broadband curved array transducer, connected to a Samsung Galaxy Tab S5 running the Philips Lumify Ultrasound app, was used for both training and study data collection. Three-second video clips and still images from ultrasound scans were stored locally on the tablet’s password-protected hard drive and subsequently uploaded to Box for Healthcare, a HIPAA-compliant cloud storage platform. Ultrasound video recordings and still images were independently reviewed by both the trained general physician and the study’s expert radiologist. The radiologist assessed image quality using the ACEP grading system [14] (detailed ACEP grading system available in Additional file 1).
Standardized Protocol
The standardized POCUS protocol employed in this study was developed based on preliminary findings regarding ultrasound manifestations in dengue patients [5]. Patients were positioned semi-recumbent at 45° or supine (if 45° was not tolerated), irrespective of fasting status. The protocol began with assessing for pulmonary B-lines at the lung apices (intersection of midclavicular line and second intercostal space) and bases (intersection of midclavicular line and fourth to fifth intercostal space). Pleural effusions were evaluated at the intersection of the posterior axillary line and sixth to ninth intercostal space, optimizing visualization of the costophrenic angle. Pericardial effusion was assessed via a subxiphoid approach with a transverse probe orientation. Free fluid in the hepatorenal space was examined at the mid- to posterior axillary line, typically at the eighth to eleventh rib spaces; in the splenorenal space at the mid- to posterior axillary line, typically at the sixth to ninth rib spaces; and in the pelvic space at the midline superior to the pubic symphysis with a sagittal probe orientation. The gallbladder’s anterior wall appearance and thickness were assessed in the right upper quadrant, using in-app calipers for measurement. The protocol sequence proceeded from right to left and then to the body’s center to enhance sonographer and patient comfort and minimize examination duration (Fig. 1).
Sites examined by point-of-care ultrasound for evidence of plasma leakage in patients with suspected dengue. This image illustrates the key anatomical locations assessed using POCUS to detect fluid accumulation indicative of plasma leakage in dengue fever cases.
Statistical Analysis
Absolute and relative frequencies were calculated for categorical variables. Means with standard deviations or medians and ranges were used for quantitative variables, depending on their distribution. Thrombocytopenia was defined as platelet counts < 100,000/µl, and leukopenia as leukocyte counts < 4000 cells/µl, and monocytopenia as monocyte counts < 200 monocytes/µl and monocytosis as > 1200 monocytes/µl. Hemoconcentration degree was calculated as the percentage increase from minimum to peak hematocrit. The dengue classification assigned by the study physician was used for analysis. Only images with a quality score of 3 (“Minimal criteria met for diagnosis, recognizable structures but with some technical or other flaws”) or higher were included in the analysis. The Kappa coefficient was employed to evaluate interobserver agreement between the study physician and the expert radiologist regarding plasma leakage overall and at each anatomical site. Kappa coefficient interpretation was as follows: 0.01–0.20 = “slight”, 0.21–0.40 = “fair”, 0.41–0.60 = “moderate”, 0.61–0.80 = “substantial”, and 0.81–0.99 = “almost perfect” [15].
Plasma leakage frequency was determined as the proportion of participants exhibiting ultrasonographic evidence of pleural effusion (any volume), B-lines (> 3 in either apex or base of both lungs), pericardial effusion, ascites, pericholecystic fluid, and thickened gallbladder wall (> 3 mm) as reported by the expert radiologist, among all participants with at least one image of suitable quality (score ≥ 3). Associations between plasma leakage evidence (as interpreted by the expert radiologist) and sociodemographic, clinical, and laboratory characteristics were assessed by calculating crude and adjusted odds ratios (cORs and aORs) with 95% confidence intervals (CIs). The same methodology was used to identify factors predicting hospital admission or referral versus outpatient care. Categorical variables were compared using chi-squared or Fisher’s exact tests, as appropriate, and quantitative variables were compared using Student’s t-test or non-parametric tests. Multiple logistic regression models were developed using backward selection, considering variable statistical and clinical relevance, and likelihood-ratio tests. Regression model goodness-of-fit was evaluated using the Hosmer and Lemeshow test, and predicted probabilities were estimated using average marginal effects. A P value
