Over the past decade, echocardiographic screening for rheumatic heart disease (RHD) has emerged as a potentially transformative strategy for global RHD control. School-aged children from nearly every continent have been screened under the premise of epidemiology or research. New Caledonia has established a national echocardiographic screening program and several other countries (including Tonga and Samoa) have embraced population-based screening. Echocardiographic screening has revealed a large burden of undiagnosed RHD, broadened our understanding of the spectrum of RHD within endemic communities, and inspired increased research and advocacy that has refocused global attention on RHD. However, many questions remain before echocardiographic screening can be endorsed as an effective strategy in a global RHD control program. In this chapter, we review the rationale behind echocardiographic screening for RHD, what we have learned from screening, provide a review of research in this area, and outline future directions and important outstanding questions.
The Rationale for Rheumatic Heart Disease Screening
Despite the importance of the recognition and treatment of streptococcal infection and the recognition of acute rheumatic fever (ARF) and initiation of secondary prophylaxis, RHD cannot be entirely prevented through these strategies. Up to one-third of children with ARF cannot recall a preceding sore throat and up to three quarters of children with RHD cannot recall an episode consistent with ARF. In many resource-limited settings, RHD is frequently diagnosed at a late stage, for example, in Uganda, 85% of newly diagnosed RHD patients presented with severe valvular involvement and cardiovascular complications.
This is particularly unfortunate, as severe RHD carries a high risk of morbidity and mortality and can only be treated through cardiac surgical or catheterization procedures, which are often cost prohibitive and unavailable in the vast majority of RHD endemic areas. In contrast, when mild RHD is identified following ARF, initiation of secondary prophylaxis can prevent disease progression and, in some cases, allow for resolution of valvular damage. RHD is most typically the manifestation of a cumulative disease process. It stands to reason then, that strategies that improve early detection of RHD may improve outcomes.
Screening aims to identify RHD before an individual develops clinical disease, to intervene to improve health outcomes. The pathogenesis of RHD means that there is often time after an initial ARF episode and the development of advanced cardiac disease, which can occur as late as decades later. Screening for RHD by auscultation has been undertaken in various parts of the world from the 1960s to the 1980s. Examples are reports from Pakistan, South Africa, Tonga, India, Samoa, and New Zealand. Active case finding was championed in the 1980s by the World Health Organization as part of the Global Programme for the Prevention of Rheumatic Fever and Rheumatic Heart Disease in 16 countries, utilizing cardiac auscultation to identify pathological heart murmurs; however, case detection was low and scale-up of screening outside of pilot sites was not achieved before the program being discontinued in 2001.
The Evolution of Rheumatic Heart Disease Screening
The first study on echocardiographic screening for RHD was published in 1996 by Anabwani and Bonhoeffer. They demonstrated that the use of echocardiography detected more children with RHD than auscultation. At that time, the technological capacities of portable echocardiograms were limited. The team needed to carry a generator to provide power in remote villages in Kenya. Nearly a decade later, a collaborative study led in Mozambique and Cambodia demonstrated that echocardiography detected 10–20 times more valvular lesions in schoolchildren in these two endemic countries. This landmark study by Marijon and colleagues, reported in 2007, heralded the beginning of widespread interest in echocardiographic screening of RHD. Within 2 years of publication, several teams, using various diagnostic criteria, showed similar results across different epidemiological settings.
Cases of RHD detected through echocardiographic screening have been termed latent RHD. There has been inconsistent use of vocabulary in the literature ( Table 13.1 ). Latent RHD includes both previously undiagnosed clinical RHD and subclinical RHD, that is, where there is no audible murmur on auscultation. However, latent RHD is most often mild. Although most children with latent RHD on echocardiographic screening have subclinical disease, between 1/3 (Uganda ) and 2/3 (Fiji ) of children with latent, definite RHD (using 2012 World Heart Federation (WHF) criteria, see later) already have moderate or severe mitral or aortic regurgitation or mitral stenosis. Conversely, the vast majority of children with latent RHD had subclinical disease in higher income countries such as New Caledonia and New Zealand (both before WHF criteria). This heterogenous mix of early and advanced disease detected through echocardiographic screening is important when assessing the value of an echocardiography-based screening program.
|Latent RHD||All cases of RHD diagnosed through echocardiographic screening, to include previously unrecognized clinical RHD and subclinical RHD.|
|Clinical RHD||All cases of RHD that have clinical signs or symptoms including heart murmur a diagnosed either through echocardiographic screening or clinical evaluation. Clinical RHD is typically more advanced than subclinical RHD.|
|Subclinical RHD||All cases of RHD that do not have clinical signs or symptoms including heart murmur a . Subclinical RHD is typically less advanced than clinical RHD.|
a It is important to again point out that detection of a pathological heart murmur without echocardiography has been shown to be poorly sensitive and specific in echocardiographic screening studies for RHD.
Accuracy of Auscultation for Rheumatic Heart Disease
The advent of echocardiography has shown auscultation to be unreliable for RHD screening. Specificity is consistently low, as first reported by the sentinel study by Marijon et al., who confirmed rheumatic valvular lesions in only 1 out of 9 children with a murmur. In context, low specificity in screening would result in many false positive RHD diagnoses, which could overwhelm resource-limited health systems and lead to inappropriate use of secondary prophylaxis. Sensitivity is also low, as consistently supported in studies comparing echocardiography and auscultation from several countries. A meta-analysis of screening in endemic regions showed a pooled prevalence of RHD by auscultation of 2.9 per 1000 compared to 12.9 per 1000 by echocardiography. Low sensitivity means that auscultatory screening cannot be used even in a two-step design with echocardiographic confirmation, as many cases would be missed in the first screening round. As noted below, these powerful data on the poor performance of auscultation led to modifications in the criteria to removal of the presence or absence of “heart murmur” from the decision around RHD diagnostic certainty.
Standardization of Echocardiographic Criteria
At the time the first studies were designed and published, there were no published guidelines or recommendations on echocardiographic criteria for the diagnosis of subclinical RHD and thus different criteria were used. Standardization of criteria for RHD became a key issue. A research group was formed and agreed on the diagnostic criteria based on the best level of evidence based on available echocardiographic, surgical, and pathologic descriptions of RHD. These WHF endorsed guidelines allow classification of an echocardiogram in a patient without a history of ARF into three groups: (1) normal; (2) borderline RHD; (3) definite RHD. The WHF criteria include Doppler criteria for the diagnosis of pathological mitral stenosis, mitral regurgitation, and aortic regurgitation. Additional morphological changes of the mitral and aortic valves were introduced. The combination of pathological left-sided regurgitation and morphological changes allows the diagnosis of definite RHD; whereas, if found in isolation, borderline RHD may be diagnosed (see Table 5.1 , Chapter 5 ).
Technical Considerations and Interpretation of the Guidelines
Although the 2012 WHF criteria represent a significant advancement from previous guidelines, there remain important limitations and practical issues to consider for those contemplating or participating in echocardiographic screening. First, the WHF criteria do not include a borderline category in adults. Although there is a solid rationale for this within the details of the guidelines, an unintended consequence has been that, in research studies assessing change in echocardiography over time, transition to adulthood has been recorded as an “improvement” from borderline RHD to normal, when the functional mitral or aortic findings are actually unchanged, with only age causing the change in categorization. This issue should be carefully considered with expert consensus to standardize reporting.
Second, the distinction between physiological and pathological regurgitation is more nuanced than the WHF criteria are able to describe. For example, the length of a mitral regurgitation jet, a critical component of the WHF definition of pathological regurgitation, is fraught with complications. Length can vary substantially based on cardiac loading conditions and blood pressure, resulting in rapid differences within the same individual over short time periods, most important for those “on the edge” of the 2 cm cut-off point. These individuals then waver between a normal and borderline RHD diagnosis despite no clinically significant cardiac change. There is also no provision within the WHF criteria for age- or height-standardized measurement (z-scores) of regurgitation, which are used as standard practice for other pediatric measurements. A jet length of 2 cm in a small child represents a considerably higher mitral regurgitation to left atrial ratio than in an adult, and shorter jets, that may be pathological, are excluded from classification as RHD. Further, the proportion of mitral regurgitation that is physiological is age dependent. Of those aged 10-12 years, 15% have physiological mitral regurgitation, with central, thin jets of mitral regurgitation, irrespective of length, more likely to be physiologic. Conversely, eccentric, posterior jets of mitral regurgitation, a classical finding of RHD, may not measure 2 cm as the jets hit the back of the left atrium. Table 13.2 summaries the differential diagnosis of pathological regurgitation of the mitral and aortic valves.
|A. Mitral Regurgitation and Structural Nonrheumatic Mitral Valve Abnormalities|
|1. Upper limit physiological MR|
|2. Congenital heart disease with MR, for example, secundum ASD or primum ASD|
|3. Isolated congenital abnormalities of the mitral valve: Double orifice MV, parachute MV, hammock MV, cleft MV|
|4. Congenital mitral valve prolapse associated with Marfan’s syndrome; other connective tissue disorders|
|5. Interscallop separations of the posterior mitral valve leaflet b|
|B. The Differential Diagnosis of Pathological Aortic Regurgitation|
|1. Bicuspid (also termed bicommissural) aortic valves c|
|2. Dilated aortic root|
|3. Subaortic membrane (which can be very subtle and challenging to detect in field-based screening)|
a The WHF criteria are designed for diagnosis and interpretation by cardiologists or echocardiographers with specialized training and expertise in RHD evaluation. It is important to avoid over and underdiagnosis of RHD.
Despite these issues, jet direction is not incorporated into the guidelines. Caution must also be exercised when interpreting freeze-frame color images of mitral or aortic regurgitation, as these often appear more substantial than those in moving loops. High quality reporting and interreviewer reliability necessitates provision of cine loops and avoidance of bias through still-frame images. Finally, it can be subjective whether a jet of regurgitation is pan-systolic or pan-diastolic. Image gain, equipment, and both regurgitation and Doppler angle can result in suboptimal Doppler signals, even in the presence of otherwise pathological regurgitation. Provision for these considerations should be discussed in research and guideline revisions.
Those interpreting echocardiograms must also consider age-related normal for abnormal thickening of the anterior mitral valve leaflet defined as follows: ≥3 mm for individuals aged ≤20 years ; ≥4 mm for individuals aged 21–40 years; ≥5 mm for individuals aged >40 years. Valve thickness measurements obtained using harmonic imaging should be cautiously interpreted and a thickness up to 4 mm should be considered normal in those aged >20 years. Normal mitral and aortic valve thickness data in health children have been published. In practice, reproducibly measuring anterior mitral valve leaflet (AMVL) thickness, in particular in the hands of nonexperts, has been challenging. AMVL thickness should be measured during diastole at full excursion. Measurement should be taken at the thickest portion of the leaflet, including focal thickening, beading, and nodularity. Measurement should be performed on a frame with maximal separation of chordae from the leaflet tissue. Valve thickness should only be assessed if the images were acquired at optimal gain settings without harmonics. Further work is needed to improve reliability of this measurement across populations and practitioners.
Contribution of Echocardiographic Screening for Rheumatic Heart Disease Epidemiology
Echocardiographic screening has dramatically improved our understanding of the global epidemiology of RHD. The most recent estimates from The Global Burden of Disease 2015 study included a systematic review of all RHD estimates, including echocardiography-based screening studies. Borderline RHD cases were not included in the model, thereby restricting estimates to patients with clinical RHD and definite echocardiographic RHD. Importantly, in the lowest resource settings, where medical records are often sparse or nonexistent, echocardiographic screening studies filled in a major gap of disease estimates. In Africa, only 15 of 54 countries had any fatal or nonfatal RHD estimates. Of the 15 countries with at least some data, eight were based on echocardiographic screening studies. These new estimates placed the prevalence of RHD at 33.4 million people. RHD-related death of 319,400 deaths per annum remained unchanged compared to previous estimates. The novelty in the latest estimates is the computation of the number of years lived with disability. Watkins and coworkers demonstrated the burden that RHD puts on developing countries, including the potentially negative economic impact of RHD (see Chapter 1 ).
The advent of echocardiography-led active surveillance has also changed the epidemiological concept of RHD. As mentioned earlier, the use of echocardiography unveiled a large number of asymptomatic individuals who are potentially at risk of developing severe RHD. Currently the epidemiology of RHD may be considered as a pyramid of increasing severity, with a smaller number of individuals with severe, symptomatic RHD at the top and a larger base of individuals with borderline, asymptomatic RHD ( Fig. 13.1 ). Contemporary data originating both from population- and hospital-based studies in a captive island-bound population reinforced this model, showing a 2- to 3-fold incidence/prevalence ratio in active screening compared to symptomatic cases admitted to the hospital. The authors appropriately concluded that significant numbers of patients with subclinical disease never become symptomatic. Important questions remain regarding the chance of progression up the RHD pyramid, at what rate advancement occurs, risk factors, and finally whether early diagnosis can prevent such progression (discussed further later, Future Directions ).
Echocardiographic screening has also, to a lesser extent, improved our epidemiological understanding of RHD in populations outside of the school ages. A community-based study in Ethiopia examined 987 children and young adults, revealing a peak prevalence (60 per 1000 cases of definite RHD) in those between 16 and 20 years. These data confirm Nicaraguan community data, which demonstrated a higher prevalence of RHD among teens and young adults, when compared to schoolchildren, though utilizing old guidelines. A community-based study in Uganda, including a house-to-house echocardiographic survey of those between 5 and 50 years, demonstrated a lifetime prevalence of 2.45%, but showed a substantial portion of adults with mild forms of latent RHD, the implications of which will require additional follow-up. Finally, in what could be considered a pilot study, a community-based screening of women presenting for antenatal care in Uganda demonstrated a 1.5% prevalence of RHD among pregnant women, with significant adverse maternal and neonatal outcomes.
Innovations in Echocardiographic Screening
Screening by Nonexpert Operators
Screening surveys or population-based screening programs require considerable organization and human resources. Echocardiographic screening requires personnel with specialized skills. The WHF criteria are designed for diagnosis, not as a screening test, and include specialized imaging modalities, intended to be performed and interpreted by cardiologists or echocardiographers with specialized training and expertise. However, in most settings where RHD remains prevalent, there are inadequate numbers of expert operators to perform echocardiograms. Standard training in echocardiography requires considerable time, and it is unlikely to be feasible to train a sufficient workforce to screen at mass scale.
Therefore, shorter, focused training programs have been developed, to reallocate a simplified cardiac ultrasound test to nonexpert health workers, known as “task shifting”.
In parallel, focused cardiac ultrasound protocols, rather than standard comprehensive echocardiography, have been increasingly implemented. The technical requirements for RHD echocardiography screening are competence in basic two-dimensional and color-Doppler imaging in limited views. Initial pilot studies found that training nonexpert nurses or medical students to perform echocardiography for RHD screening was feasible. Four subsequent studies have examined this in more detail, with larger sample sizes ( Table 13.3 ). Heterogeneity of study design and methods precludes direct comparison of the results of these studies. However, consistent themes have emerged from these studies. There is now convincing evidence that nonexpert health workers can be trained to perform focused echocardiography and measure regurgitation as a risk marker for RHD. The accuracy of this screening strategy is in the range of 80% sensitivity and 85% specificity. Nurses have been trained most commonly, but other cadres of health workers also show proficiency after brief training.
|Setting, Year||Fiji, 2012–13||New Caledonia, 2013||Uganda, 2014||Brazil, 2015|
|Reference||Engelman et al. 2016.||Mirabel et al. 2015.||Ploutz et al. 2015.||Beaton et al. 2016.|
|Operators||7 nurses||2 nurses||2 nurses||2 nurses |
2 medical students
|Previous echocardiography experience||None||None||6 months||6 weeks (4 operators) |
12 months (2 operators)
|Sample||2004 students |
|1217 students |
|956 students |
|397 students |
|RHD prevalence||2.8%||4%||4.5%||13% (artificially high sample)|
|Duration of screening||12 months||5 months||3 months||4 days|
|Ratios||1–2 trainers to 7 students |
2 machines to 7 students
|1 trainer to 1–2 students |
|1 trainer to 1–2 students |
|Overview of training||1 week theoretical |
6 weeks supervised practice
No hurdle assessment
|3 days theoretical |
30 h practical
Further 1 to 1 training to address specific issues
|4 h theoretical |
2 days practical
|3 weeks |
Weekly hurdle assessments
|Machine||Portable (Sonosite M-Turbo)||Handheld (GE Vscan)||Handheld (GE Vscan)||Handheld (GE Vscan)|
|Screen-positive criteria||Any MR or any AR||MR ≥1.5 cm or any AR||MR ≥2 cm or any AR||MR ≥1.5 cm or any AR||MR ≥1.5 cm or any AR|
|Screen-positive proportion||16%||9%||11%–12%||24%||n/a a|
|Sensitivity (95% CI)||84% (72–92)||77% (64–87)||Operator A: 84% |
Operator B: 78%
|74% (59–86)||83% (76–89)|
|Specificity (95% CI)||85% (84–87)||93% (92–94)||Operator A: 91% |
Operator B: 92%
|79% (76–81)||85% (82–87)|
|Positive predictive value||15%||25%||29%–39%||14%||n/a a|
|Negative predictive value||99%||99%||99%||98%||n/a a|