Study design and oversight
This was an observational cross-sectional study designed to measure the prevalence of RHD in school children in Lusaka, the capital of Zambia. The study was registered on clinicaltrials.gov (#NCT02661763) and approved by the Biomedical Research Ethics Committee at the University of Zambia, the Ministry of Health, the Ministry of Education, and the Ministry of Community Development, Mother and Child Health. Data collection took place between September and December in 2015.
Study population
The study population was drawn from a random sample of children in grades seven through twelve attending basic, secondary, and high schools in Lusaka. Screening was focused on older school children (rather than younger school children) based on evidence that they are at higher risk of RHD [9, 29, 30].
Written informed consent for participation in the study was required from parents or legal guardians. Information about the study and consent forms were sent home with students several weeks before each screening session. Consent forms were written in English, Nyanja and Bemba, which are the most common languages in Lusaka. Children were also verbally informed about the study at the time of screening and were required to provide written assent with their name or thumbprint in order to participate.
Inclusion criteria were all children whose parents consented, who provided assent, and who were present at the time the screening was conducted. Exclusion criteria included conditions that precluded the performance of transthoracic echocardiography (e.g., skin infection of the chest).
A member of the study team approached the administrative head of each selected school to invite their school to participate. Awareness messages (flyers and posters) that described the study were distributed, and study staff briefed teachers whose classrooms were to be involved. The classrooms selected to participate in each school were determined at random and approximately 60–80 students were enrolled from each school.
Study procedures
Prior to screening, each participant’s parent or guardian completed a questionnaire that collected information on demographics and the participant’s general health history. On the day of screening, the study team (consisting of a field coordinator, pediatric nurses, radiographers, data entry staff, and sometimes a physician) organized a private screening area and ensured that a member of the school staff would be present during interactions between study staff and participants.
Study staff conducted a focused history and physical examination with each participant, specifically to assess for pharyingitis, RF, or RHD. Study staff then performed a screening echocardiogram using the abbreviated screening protocol and a handheld ultrasound device (Fig. 1). The overall process generally lasted 15–20 min for each participant; the echocardiographic screening component lasted about 5 min per participant. Refreshments were provided after the screening. The study staff prioritized flexibility in scheduling so that interruption of regular school activities was minimized.
“eRegister” database
Data obtained at the time of screening was entered through mobile electronic devices (Samsung Galaxy 4 Tab computer tablets) into a digital system (“eRegister”) that was developed specifically for the study [31]. The system used a well-established e-health platform (“CommCare” by Dimagi) [32, 33], which includes a customizable mobile health data collection application and an online portal for secure data viewing and reporting. All access to the digital platform including mobile submissions was done over HTTPS (hypertext transfer protocol secure) and was cryptographically secure. Mobile devices were password-locked. Echocardiograms were also saved in the system; images and videos were temporarily stored on the hard drive of the ultrasound device and then transferred to the eRegister.
Echocardiographer training
Six general radiographers from UTH were recruited to participate in the school screening study, and three ultimately completed training and remained part of the study team throughout the study period. Training comprised classroom-style and hands-on workshops involving actual RHD patients at UTH and supportive supervision during school screening [28]. Training activities took place periodically over 12 months beginning May 2014. Training was structured to accommodate the practical considerations of trainees, who continued to work as general radiographers at UTH.
Radiographers were trained to use a handheld machine (General Electric V-Scan, General Electric Company, Fairfield, CT, USA) and to screen participants for evidence of RHD using a recently validated, simplified protocol called FOCUS (an acronym for FOCused method Utilizing hand-held echocardiography in Screening for RHD) [24]. This protocol incorporates key aspects of the much more complex (and definitive) WHF protocol for echocardiography-based RHD diagnosis.
The abbreviated protocol utilizes a single view (the parasternal long-axis view) and a single measurement (regurgitation jet in the mitral valve, which is implicated in most RHD cases). According to the protocol, the mitral valve is interrogated, first without and then with color Doppler. A regurgitation jet-length greater or equal to 2 cm (measured from the vena contracta to the last pixel of the regurgitant jet using the V-Scan radius measuring tool) constitutes a positive result.
Following the first intensive training of radiographers that started in May 2014, school screening initially began in October 2014. However, quality assurance checks by study investigators (JM, LZ) revealed suboptimal performance by the newly trained radiographers. The study was put on hold pending additional training and practice. The study re-started in September 2015 after the radiographers’ proficiency was confirmed through quality assurance checks by a blinded external expert (described below). Improvement in radiographers’ performance was attributed to repeated practice, real-time guidance during periodic visits by volunteer echocardiography technicians from abroad, and training with the assistance of RHD patients at UTH.
When the study re-started, two mechanisms were implemented to help ensure successful screening: (1) four international echocardiography technicians spent a total of 8 weeks accompanying study staff during school screening to support the work of local radiographers; and (2) all screening echocardiograms were reviewed remotely by a professional third-party echocardiographer (VirtualScopics, University of Rochester, USA) who accessed de-identified scans through the eRegister and conducted independent review of the echocardiograms recorded by the field team.
Screening and confirmatory echocardiography
Echocardiographic screening was performed using V-Scan handheld machines. If a participant screened positive at the school or through the VirtualScopics quality assurance (VSQA) review, study staff informed the participant’s parents or guardians and arranged follow-up at UTH where a member of the study team (JM) performed comprehensive echocardiography using a standard portable ultrasound machine (Philips CX 50 or Sonosite M-Turbo).
The follow-up echocardiogram was conducted according to evidence-based consensus WHF guidelines [25] using 2D, continuous-wave, and color-Doppler to assess for morphological abnormalities or pathological mitral or aortic valve regurgitation. The echocardiograms of participants that underwent follow-up echocardiography were classified as “normal,” “definite RHD,” or “borderline RHD.”
As described by WHF, definite RHD corresponds to echocardiographic changes that are likely to be rheumatic in origin assuming the clinical context is consistent. Borderline RHD corresponds to echocardiographic changes that may represent early RHD in individuals aged 20 years or less. The borderline category was established by WHF because young people may benefit from early detection of RHD and initiation of antibiotic prophylaxis [25]; progression of heart disease in patients with borderline RHD has been documented [34, 35].
Student disposition
Participant disposition was determined by the results of school screening and confirmatory echocardiography (Fig. 2). No follow-up was required if the focused history and physical examination, the screening echocardiogram, and the VSQA review were normal.
When the physical examination revealed pharyngitis and the screening echocardiogram/VSQA review was normal, the participant was referred to UTH or a local health clinic for further evaluation and management, including primary prevention of RHD. When the screening echocardiogram revealed any abnormality, or there was a clinical suspicion of ARF, the participant was referred to UTH for a comprehensive evaluation and quantitative echocardiogram in order to assess for ARF or RHD. Participants with definite or borderline RHD were referred to the pediatric cardiology clinic at UTH or a local health clinic for long-term follow-up including antibiotic prophylaxis and repeat echocardiography on a 6- or 12-monthly basis [14, 34].
Participants with cardiac abnormalities due to congenital heart disease or other non-rheumatic problems (for example, skin or dental problems) were referred for follow-up evaluation and treatment at UTH or their local clinic in accordance with standard medical care in Zambia.
RHD control program in health clinics
Participants that were identified to have borderline or definite RHD were referred for follow-up management which included the recommendation for antibiotic prophylaxis (with oral penicillin or monthly injections of benzathine penicillin) according to national guidelines. Penicillin-allergic individuals were alternatively offered erythromycin. UTH was an inconvenient location for some participants to routinely travel for follow-up and therefore selected health clinics in Lusaka that were located near the study schools were enrolled in an RHD control program [27].
As part of this program, study nurses and physicians conducted an introductory on-site training workshop, disseminated educational materials to staff and patients, and performed monthly or bi-monthly supportive supervisory visits. The training taught knowledge and skills to support effective primary prevention (i.e., treatment of bacterial pharyngitis with a single dose of benzathine penicillin to decrease risk of RF), secondary prevention (i.e., continuous antibiotic prophylaxis in children with RHD to mitigate progression of heart disease), and pharmacovigilance. Details of the training are described previously [26]. Reliable availability of high-quality benzathine penicillin at UTH and participating clinics in Lusaka was faciliated by a product grant to the Ministry of Health by Sandoz, the generics division of Novartis.
Sample size considerations
The total school population in Lusaka within which the study took place is approximately 500,000. Since previous studies demonstrated that RHD prevalence increases in older children [9, 29, 30], this study focused on prevalence in children attending grades 7 and above. A minimum sample size of 1024 participants in 15 randomly selected schools was calculated based on precision around a point estimate of 1% RHD prevalence in Lusaka. This was felt to be a conservative estimate in light of similar studies from other countries in sub-Saharan Africa [8, 29, 30, 36]. The sample size also accounted for a presumed rejection rate of 9%. With this sample, it was determined that the prevalence of RHD could be estimated within 0.7 percentage points of precision with 95% confidence. A 1.2 design effect due to clustering in the sampling design was assumed given that some participants were enrolled from within clusters (i.e., participants enrolled from the same classrooms and schools were more likely to be similar in unintended ways relative to a completely random population).
Variation in the total number of eligible individuals in each school was expected, as was the total size of each school. We accounted for this variation in the sampling process by weighting the primary sampling unit (the school) and selecting according to size, using probability proportional to size (PPS) sampling [37]. The sampling weights were also used in the analysis to control for possible misspecifications of standard errors in each cluster.
As discussed above, the study was temporarily halted early on to allow for improved training of radiographers. A formal assessment of radiographers’ performance using the screening protocol was then conducted by VSQA review, and schools were re-randomized prior to re-starting the study. Fifteen schools included in the study were randomly selected from a Ministry of Education database of all 119 basic, secondary, and high schools in Lusaka. Within each randomly selected school, one grade was randomly selected (grades 7–12). Multiple classrooms in the selected grade were then randomly selected with the aim to enroll a minimum of 60 students per school. All students in the selected classrooms were eligible to participate if consent and assent were completed.
Analysis plan
The study measured prevalence expressed as a proportion per 1000 individuals using standard statistical analysis tools. The estimate of the standard error accounted for clustering effects and was considered in the confidence interval computation. Sub-analyses using a similar approach were conducted by age, gender, and selected additional demographic parameters, though it was not expected that these variables would have adequate power for precise estimates of prevalence for these subgroups. Categorical data were compared by chi-squared analyses, chi-squared for trend, or expressed as risk ratios with 95% confidence intervals as appropriate. To assess the reliability of the training program in skills-building for the local radiographers, a pairwise kappa test was performed.