Simplified Rheumatic Heart Disease Screening Criteria for Handheld Echocardiography

Simplified Rheumatic Heart Disease Screening Criteria for Handheld Echocardiography Jimmy C. Lu, MD, Craig Sable, MD, Gregory J. Ensing, MD, Catherine Webb, MD, Janet Scheel, MD, Twalib Aliku, MD, Peter Lwabi, MD, Justin Godown, MD, and Andrea Beaton, MD, Ann Arbor, Michigan; Washington, District of Columbia; Kampala, Uganda; and Nashville, Tennessee

Background: Using 2012 World Heart Federation criteria, standard portable echocardiography (STAND) reveals a high burden of rheumatic heart disease (RHD) in resource-poor settings, but widespread screening is limited by cost and physician availability. Handheld echocardiography (HAND) may decrease costs, but World Heart Federation criteria are complicated for rapid field screening, particularly for nonphysician screeners. The aim of this study was to determine the best simplified screening strategy for RHD detection using HAND. Methods: In this prospective study, STAND (GE Vivid q or i or Philips CX-50) was performed in five schools in Gulu, Uganda; a random subset plus all children with detectable mitral regurgitation or aortic insufficiency also underwent HAND (GE Vscan). Borderline or definite RHD cases were defined by 2012 World Heart Federation criteria on STAND images, by two experienced readers. HAND studies were reviewed by cardiologists blinded to STAND results. Single and combined HAND parameters were evaluated to determine the simplified screening strategy that maximized sensitivity and specificity for case detection. Results: In 1,439 children (mean age, 10.8 6 2.6 years; 47% male) with HAND and STAND studies, morphologic criteria and the presence of any mitral regurgitation by HAND had poor specificity. The presence of aortic insufficiency was specific but not sensitive. Combined criteria of mitral regurgitation jet length $ 1.5 cm or any aortic insufficiency best balanced sensitivity (73.3%) and specificity (82.4%), with excellent sensitivity for definite RHD (97.9%). With a prevalence of 4% and subsequent STAND screening of positive HAND studies, this would reduce STAND studies by 80% from a STAND-based screening strategy. Conclusions: In resource-limited settings, HAND with simplified criteria can detect RHD with good sensitivity and specificity and decrease the need for standard echocardiography. Further study is needed to validate screening by local practitioners and long-term outcomes. (J Am Soc Echocardiogr 2015;28:463-9.) Keywords: Rheumatic heart disease, Handheld echocardiography, Mitral regurgitation, Aortic insufficiency

Rheumatic heart disease (RHD) is the most common cause of acquired heart disease in children worldwide, with the highest burden in developing countries.1 Outcomes in patients with From the University of Michigan Congenital Heart Center, Ann Arbor, Michigan (J.C.L., G.J.E., C.W.); the Children’s National Medical Center, Washington, District of Columbia (C.S., J.S., A.B.); Uganda Heart Institute, Kampala, Uganda (T.A., P.L.); and Monroe Carell Jr. Children’s Hospital at Vanderbilt, Nashville, Tennessee (J.G.). This project was supported by Award Nos. UL1TR000075 and KL2TR000076 from the NIH National Center for Advancing Translational Sciences. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Center for Advancing Translational Sciences or the National Institutes of Health. This study was also funded in part by grants from General Electric, the World Heart Federation, and the CHAMPS for Mott fund. Reprint requests: Jimmy C. Lu, MD, University of Michigan Congenital Heart Center, C.S. Mott Children’s Hospital, 1540 E Hospital Drive, Ann Arbor, MI 48109 (E-mail: [email protected]). 0894-7317/$36.00 Copyright 2015 by the American Society of Echocardiography. http://dx.doi.org/10.1016/j.echo.2015.01.001

symptomatic RHD are quite poor,2,3 although patients with disease detected at an earlier stage may have better outcomes with antibiotic prophylaxis and prevention of recurrence. Although RHD is caused by acute rheumatic fever, an autoimmune reaction that follows group A streptococcal infection, patients frequently do not recall histories of infection or symptoms of acute rheumatic fever,4 making targeted screening difficult. Echocardiography-based studies have demonstrated higher prevalence than for RHD detected by clinical evaluation.5-8 The World Heart Federation (WHF) published evidence-based guidelines for echocardiographic diagnosis of RHD in endemic areas,9 and the World Health Organization supports screening in high-prevalence areas.10 Although standard portable echocardiography (STAND) is effective, widespread screening is often not feasible because of cost and physician availability. Handheld echocardiography (HAND), a promising alternative with lower cost machines with a simple interface, has been shown to be effective in the detection of RHD, both in a smaller cohort11 and in wide-scale screening using WHF criteria.12 However, because of limited functionality of HAND devices (such as a 463

464 Lu et al

lack of spectral Doppler), some WHF criteria designed for AI = Aortic insufficiency STAND may be less useful with HAND, and some criteria CI = Confidence interval may require different cutoff HAND = Handheld points. Furthermore, WHF echocardiography criteria may be difficult to apply, particularly for MR = Mitral regurgitation screening by local practitioners RHD = Rheumatic heart or nonphysicians. Screening disease guidelines have not yet been STAND = Standard portable validated specifically for echocardiography HAND. The aim of this study was to determine the best simplified screening criteria for RHD detection using HAND.

Journal of the American Society of Echocardiography April 2015

Abbreviations

METHODS Study Population This prospective study included five primary schools in Gulu, Uganda. All students 5 to 17 years of age were eligible for inclusion. Parents of minors provided written informed consent; adolescents >15 years of age provided informed consent, as is customary in Uganda. This study was approved by the institutional review boards at the University of Michigan, Children’s National Medical Center, and Makerere University. All enrolled children underwent STAND. A random subset of 10% of all subjects undergoing STAND were designated to undergo HAND. To evaluate the ability of HAND to differentiate pathologic from physiologic regurgitation, any child with visible mitral regurgitation (MR) or aortic insufficiency (AI) was also referred for HAND. The cohort of the present study includes all patients who underwent both STAND and HAND (Figure 1). STAND STAND (Figures 2–4) was performed by a team of experienced imagers (pediatric cardiologists, fellows, and sonographers) with Vivid q or i (GE Medical Systems, Milwaukee, WI) or CX-50 (Philips Medical Systems, Best, The Netherlands) machines. A 3.6MHz transducer was used without harmonics, because harmonic imaging can increase apparent tissue thickness.9 Frame rates ranged from 50 to 80 Hz for two-dimensional imaging and from 17 to 30 Hz for color Doppler. Depth, gain, and compression were optimized by the imager. The imaging protocol consisted of 13 clips 2 sec in length: parasternal long-axis view through the mitral and aortic valves, color Doppler over the mitral valve, color Doppler over the aortic valve, apical four-chamber view, apical four-chamber view with color Doppler over the mitral valve, apical five-chamber or three-chamber view, color Doppler over the aortic valve, continuous-wave Doppler of any MR or AI, parasternal short-axis view at the levels of the mitral and aortic valves, and color Doppler across the mitral and aortic valves. All images were read by six experienced pediatric cardiologists using 2012 WHF criteria (Table 1).9 Measurements of anterior mitral leaflet thickness were made to the nearest 10th of a millimeter; measurements of MR jet length were made to the nearest millimeter. Because STAND interpretation was used as the gold standard, studies read as depicting borderline or definite RHD were confirmed by a second reader, with disagreements adjudicated by a third reader blinded to results of both prior readers.

Figure 1 Flowchart of patient assignment. All children with concern for borderline or definite RHD by STAND were referred for complete echocardiography and evaluation by a pediatric cardiologist based in Gulu. Children with confirmed definite RHD were started on penicillin prophylaxis; children with confirmed borderline RHD were enrolled in biannual follow-up. HAND HAND (Figures 2–4) was performed by experienced imagers using Vscan (GE Medical Systems) at a station isolated from STAND stations. The imager using HAND was unaware of the reason for referral for HAND. Imaging was performed with the single, attached 1.7- to 3.4-MHz transducer, with frame rates ranging from 25 to 30 Hz for two-dimensional imaging and from 12 to 16 Hz for color Doppler. Depth and gain were optimized by the imager; other imaging settings are not adjustable. The location but not the size of the color box is adjustable. This device offers an image-based ‘‘auto-cycle’’ function for the automatic detection of a full heart cycle beginning with end-diastole. The same imaging protocol was used for HAND, with the exception of spectral Doppler, which is not available on the Vscan. Images were interpreted offline using Gateway software (GE Medical Systems). All measurements were made to the nearest millimeter, because the software cannot measure tenths of a millimeter. Because of the lack of spectral Doppler or data evaluating discrimination of physiologic and pathologic MR on HAND, pansystolic MR was defined as MR present in two consecutive frames to avoid measurement of closing volumes or physiologic regurgitation. Studies were read by the same six experienced readers, blinded to STAND results. A random subset of 10% was reevaluated for interobserver agreement. Statistical Analysis Data are presented as mean 6 SD or as median (interquartile range) as appropriate. Sensitivity and specificity of individual HAND parameters were calculated for any RHD (borderline or definite), as well as sensitivity for definite RHD, as defined by the gold standard of STAND using the 2012 WHF criteria. Receiver operating characteristic curve analysis was performed to determine the optimal cutoff for MR jet length for differentiation of patients with borderline or definite RHD from normal patients. MR jet lengths by STAND and HAND were compared by Wilcoxon matched-pairs signed rank tests. Interobserver agreement for individual HAND parameters was assessed with k coefficients, with 95% confidence intervals (CIs). P values <.05 were considered to indicate statistical significance. HAND parameters were combined into potential screening strategies, to optimize sensitivity for RHD, while maintaining

Journal of the American Society of Echocardiography Volume 28 Number 4

Lu et al 465

Figure 2 Representative STAND and HAND images from the same patient with mitral stenosis. specificity $80%. The percentage of patients still requiring STAND in a HAND-based screening strategy, assuming confirmation of positive HAND results with STAND, was calculated using the prevalence of the entire cohort screened with STAND. The truepositive fraction was calculated as (sensitivity in the HAND cohort)  (prevalence of borderline or definite RHD in the entire screened cohort). The false-positive fraction was calculated as (1 specificity)  (1 prevalence). The percentage requiring STAND was then calculated as (true-positive fraction) + (false-positive fraction). For confirma-

Figure 3 Representative STAND and HAND images from the same patient with MR, with jet length measurement. Note the underestimation of the HAND measurement. tion of these estimates, sensitivity and specificity were also calculated only in the cohort of patients randomly assigned to HAND. The percentage of patients requiring STAND was calculated as the number of true positives and false positives divided by the number of this cohort.

466 Lu et al

Journal of the American Society of Echocardiography April 2015

RHD, and six (0.4%) had other diagnoses. The subset of children randomly assigned to HAND had a similar prevalence of disease to the overall cohort, with borderline RHD in 11 of 447 (2.5%) and definite RHD in seven of 447 (1.6%), with a total prevalence of 4.0%. Mitral Valve Criteria With the exception of anterior mitral leaflet thickening, mitral valve morphologic criteria had poor sensitivity but good specificity for RHD (Table 2). Sensitivity for definite RHD was particularly poor for morphologic criteria. Thickened anterior mitral leaflet ($3 mm) and thickened mitral valve chordae were common findings using HAND in this cohort. The presence of any one positive morphologic criterion had sensitivity of 71.7% and specificity of 63.1% for RHD. Requiring two positive morphologic criteria, similar to WHF criteria, improved specificity to 88.3% but decreased sensitivity to 46.4%. Mitral stenosis was rare in this cohort, with high specificity but poor sensitivity for RHD. Detection of any MR by HAND was common, with good sensitivity but poor specificity for RHD. MR seen in two views or pansystolic MR had higher sensitivity. MR jet length by HAND was correlated with length by STAND (r = 0.54, P < .0001). However, in patients with measurable MR, jet length by HAND (median, 1.3 cm; interquartile range, 1.0– 1.8 cm) was mildly underestimated relative to STAND (median, 1.4 cm; interquartile range, 1.1–1.8 cm) (P = .002). MR jet length by HAND differentiated patients with borderline or definite RHD from normal subjects (Figure 5). The standard WHF cutoff of $2.0 cm for pathologic MR had poor sensitivity (45.2%) but excellent specificity (95.9%). Receiver operating characteristic curve analysis identified an optimal MR jet length cutoff of $1.5 cm by HAND, with sensitivity of 64.4% and specificity of 85.6% for any RHD. Aortic Valve Criteria Aortic morphologic criteria were very rare in this population, with poor sensitivity for RHD (Table 3). Using any one positive morphologic criterion increased sensitivity to 10%. Aortic stenosis was not detected by either STAND or HAND in any patients. AI was very specific but had low sensitivity when used alone to detect RHD. Screening Criteria

Figure 4 Representative STAND and HAND images from the same patient with AI. RESULTS Of 4,773 subjects screened with STAND, 140 (2.9%) had borderline RHD by 2012 WHF criteria, and 52 (1.1%) had definite RHD, for a total prevalence of 4.0%. HAND was performed in 1,439 children (mean age, 10.8 6 2.6 years; 47% male). Within this cohort, 133 (9.2%) had borderline RHD, 47 (3.3%) had definite

Using HAND criteria of either MR jet length $1.5 cm or any AI had the best combination of sensitivity and specificity for borderline or definite RHD, with excellent sensitivity for definite RHD (Table 4). Pansystolic MR could be substituted for MR jet length, with similar sensitivity and specificity but slightly lower sensitivity for definite RHD. Using MR in two views or AI significantly decreased specificity, although sensitivity for borderline or definite RHD and for definite RHD were maintained. When limited to children randomly assigned to HAND, sensitivity and specificity were similar (Table 5). On the basis of the 4.0% prevalence of any RHD in the overall, nonselected cohort of 4,773 children, the screening strategy of MR jet length $ 1.5 cm or any AI on HAND would be expected to be positive in 19.8% of the nonselected population. Pansystolic MR or any AI would be positive in 22.2%, while MR in two views or any AI would positive in 36.4%. When limiting analysis of these screening strategies to the subset of children randomly assigned to HAND, the prevalence of positive HAND screening (true positive + false positive) was even lower (Table 5).

Lu et al 467

Journal of the American Society of Echocardiography Volume 28 Number 4

Table 1 WHF criteria for diagnosis of RHD in patients # 20 years of age Definite: A, B, C, or D A. Pathologic MR and at least two morphologic features of RHD of the mitral valve B. Mitral stenosis with mean gradient $ 4 mm Hg C. Pathologic AI and at least two morphologic features of RHD of the aortic valve D. Borderline disease of both the aortic and mitral valves Borderline: A, B, or C A. At least two morphologic features of RHD of the mitral valve B. Pathologic MR C. Pathologic AI Pathologic MR (all criteria must be met)

Pathologic AR (all criteria must be met)

Seen in two views

Seen in two views

Jet length $ 2 cm (in at least one view)

Jet length $ 1 cm (in at least one view)

Velocity $ 3 m/sec for one complete envelope

Velocity $ 3 m/sec for one complete envelope

Pansystolic jet in at least one envelope

Pandiastolic jet in at least one envelope

Morphologic features of the mitral valve

Morphologic features of the aortic valve

Anterior leaflet thickening $ 3 mm

Irregular or focal thickening

Chordal thickening

Coaptation defect

Restricted leaflet motion

Restricted leaflet motion

Excessive leaflet tip motion during systole

Prolapse

Table 2 Mitral valve criteria using HAND for detection of any (borderline or definite) RHD and sensitivity for definite RHD Criterion

Prevalence (%)

Sensitivity for any disease (%)

Specificity for any disease (%)

Sensitivity for definite RHD (%)

Thickened chordae

19.6

42.8

83.7

53.2

Restrictive motion

4.9

16.7

96.8

36.2

31.7

62.2

72.7

78.7

6.3

11.7

94.4

17.0

Any regurgitation

55.5

87.8

49.2

91.5

Seen in 2 views

37.3

76.0

68.3

85.1

Pansystolic

22.8

67.2

83.6

85.1

1.1

5.6

99.6

19.1

Morphologic

Thickened anterior leaflet Excessive motion MR

Mitral stenosis

Reproducibility In a random subset of 302 children reevaluated for interrater reliability of criteria on HAND, detection of any AI had the highest agreement (k = 0.65; 95% CI, 0.46–0.83). MR jet length $ 1.5 cm (k = 0.59; 95% CI, 0.48–0.70) had better agreement than pansystolic MR (k = 0.49; 95% CI, 0.44–0.63) or MR in two views (k = 0.53; 95% CI, 0.44–0.63). In a subset of 313 children with second readings on STAND, agreement was higher compared with HAND for AI (k = 0.81; 95% CI, 0.70–0.92), MR jet length $ 1.5 cm (k = 0.81; 95% CI, 0.75–0.88), and pansystolic MR (k = 0.55; 95% CI, 0.46–0.65), although not for MR seen in two views (k = 0.33; 95% CI, 0.20–0.46).

DISCUSSION We have shown that HAND, using screening criteria of MR jet length $ 1.5 cm or any AI, can detect any (borderline or definite) RHD with

good sensitivity and specificity, with excellent sensitivity for definite RHD. To our knowledge, this is the first study to evaluate screening criteria for RHD using HAND in a large endemic population. HAND is an attractive approach to screening for RHD because of the highest prevalence of RHD in resource-limited populations. HAND machines are less expensive, smaller, and simpler to use than STAND machines. HAND is battery powered, although electricity would likely still be necessary for prolonged screening; in this field study, multiple battery packs were charging while each device was in use. However, because of technical differences in image quality and limited options for image optimization, screening and diagnostic criteria validated for standard echocardiography may not be directly translatable to this platform. In a smaller Ugandan cohort, HAND with modified WHF criteria had higher sensitivity and specificity.11 However, application to widespread screening, such as the present cohort, requires adaptation to multiple issues in suboptimal settings. For example, in the prior study, all imaging was performed by a single experienced imager; variability in image acquisition likely contributes to the lower sensitivity and specificity compared with more ideal settings. We previously

468 Lu et al

Journal of the American Society of Echocardiography April 2015

Table 4 Performance of screening criteria for detection of any (borderline or definite) RHD and sensitivity for definite RHD

Criteria

MR jet length $ 1.5 cm or any AI

Figure 5 Receiver operating characteristic curve of MR jet length by HAND, for detection of any (borderline or definite) RHD. AUC, Area under the curve.

Table 3 Aortic valve criteria using HAND for detection of any (borderline or definite) RHD and sensitivity for definite RHD

Criterion

Sensitivity Specificity Sensitivity for definite for any for any Prevalence disease (%) disease (%) RHD (%) (%)

Morphologic Thickened

1.2

3.9

99.2

1.8

Restrictive motion

0.1

0.6

99.9

1.4

Prolapse

0.4

1.1

99.8

1.5

Noncoaptation

1.1

5.6

99.6

1.8

AI Any insufficiency

5.6

15.6

95.8

25.5

Seen in two views

3.0

12.2

98.4

25.5

Pandiastolic

3.0

13.9

98.6

25.5

demonstrated that modified WHF criteria in this larger cohort had sensitivity of 78.9% and specificity of 87.2% for borderline or definite RHD.12 The similar performance of these simplified criteria to the more extensive WHF criteria suggests that tailoring criteria to HAND can make widespread screening simpler and more practical, with little sacrifice in sensitivity or specificity. In a cohort in Mozambique, Mirabel et al.13 validated a single criterion of MR jet length $ 2 cm for STAND, with sensitivity of 73%. The sensitivity of simplified HAND criteria compares favorably, with sensitivity of 73.3% for borderline or definite RHD and 97.9% for definite RHD. The lower threshold for MR jet length by HAND identified by receiver operating characteristic curve analysis likely reflects differences in modality. The lower frame rate and shorter (one heartbeat) clip length could lead to missing the frame of maximal color jet length, which may explain the mild systematic underestimation of MR jet length by HAND in the present cohort. However, the consistent factor appears to be the utility of MR jet length measurement for discriminating physiologic from pathologic regurgitation. Although isolated aortic involvement is less common than MR, the inclusion

Sensitivity Specificity Sensitivity for any for any for definite disease (%) disease (%) RHD (%)

73.3

82.4

97.9

Pansystolic MR or any AI

75.0

80.4

93.6

MR seen in two views or any AI

51.1

65.5

91.5

of AI as a second criterion improved sensitivity in our cohort and can be easily evaluated in a screening protocol. Although HAND performed fairly well in our cohort, positive results should be confirmed with STAND. Because of the high prevalence of RHD in resource-limited populations, specificity of HAND screening must remain high to limit the use of STAND. Although the positive rate will vary with populations, the estimated 20% combined true- and false-positive rate, when projecting the sensitivity and specificity to the entire nonselected cohort evaluated with STAND, suggests that initial screening with HAND could reduce the need for STAND by approximately 80%, compared with strategies requiring STAND for screening, while detecting nearly all cases of definite RHD. The subset of patients randomly assigned to HAND demonstrated even lower true- and false-positive rates, suggesting that this estimated reduction of STAND may even be conservative. The sensitivity for definite RHD remained high; the apparent decrease compared with the larger cohort likely reflects the smaller sample size, as only one in seven patients with definite RHD had negative findings. The prioritization of specificity comes at a cost of missing some cases of borderline RHD, although the potential difference in outcomes for undiagnosed borderline RHD is unclear. In a small cohort, 42% of children with borderline RHD improved after 2-year follow-up, although 10% worsened to definite RHD.14 Further study on long-term outcomes is under way, and serial evaluation may be necessary. In addition to maintaining specificity, an ideal strategy must be simple for widespread adoption, particularly if nonphysicians, such as school nurses, are to participate in a screening strategy. Although the WHF criteria are well suited for diagnosis in the hands of trained experts, the complexity limits the utility for large population-based screening. We thus focused on strategies using one or two criteria. Morphologic criteria can be difficult to judge, and discerning normal from thickened chordae or subtle differences in motion can be difficult with limited reproducibility for echocardiographers, much less nonphysician screeners. HAND is not well suited to the measurement of leaflet thickness, as measurements can be performed only to the nearest millimeter (with a WHF cutoff of 3 mm for abnormal thickening). We thus focused on MR and AI. For evaluation in the field, measurement of dimensions or jet length requires either accessing options from a menu on the handheld device or downloading images into separate software for interpretation. The current study design used the software to measure jet length after screening. If future screening efforts involve screening by nonphysician imagers to determine the need for STAND, measurement at the time of image acquisition could slow screening evaluation, and qualitative criteria, such as pansystolic MR or MR seen in two views, could be easier to use without measurement. However, the decrease in sensitivity and specificity and the lower interreader

Lu et al 469

Journal of the American Society of Echocardiography Volume 28 Number 4

Table 5 Performance of screening criteria for detection of any (borderline or definite) RHD and sensitivity for definite RHD in the cohort randomly assigned to HAND Criteria

Sensitivity for any disease (%)

Specificity for any disease (%)

Sensitivity for definite RHD (%)

Positive screens (%)

MR jet length $ 1.5 cm or any AI

72.2

89.0

85.7

13.5

Pansystolic MR or any AI

66.7

89.4

85.7

12.9

MR seen in two views or any AI

66.7

77.6

85.7

24.2

agreement of these criteria, even when interpreted by experienced echocardiographers, suggest that this may be further limited in less experienced hands. The lower agreement for criteria using HAND compared with STAND suggests that there is a component related to modality or image quality, which could potentially be exacerbated with less experienced screeners. Limitations The prospective design of this study ensured adequate image quality to evaluate these screening criteria. However, the cross-sectional nature of this study, with a single evaluation for each patient, prevents comment on serial evaluation to detect progression or new onset of disease. Given the magnitude of the study, echocardiographic diagnosis was used as the gold standard; clinical details and laboratory results were not available, although laboratory tests are not likely to be useful adjuncts for screening.5 All subjects with positive results were subsequently referred for clinical evaluation, but evaluation of all subjects, including those with normal findings, was not feasible. Because this study assessed adapted WHF criteria on HAND, results should not be interpreted as need to change the existing WHF criteria for standard echocardiography. Technical limitations of the HAND device required adaptation of WHF criteria; other definitions for findings such as pansystolic MR could be used but would further decrease sensitivity. Because specificity was still high, we believe that the current definitions are reasonable. All imaging was performed by experienced imagers; performance in the hands of nonphysicians may be different and will be evaluated in an ongoing study. Sensitivity and specificity can be slightly affected by prevalence within the cohort. We intentionally oversampled patients with MR or AI to determine the efficacy of HAND in discriminating physiologic and pathologic findings, but the size of the population, particularly the overall cohort to determine prevalence of disease, should minimize any potential bias. The similar performance of these criteria in the randomly assigned patients suggests that any differences due to altered prevalence of disease are minimal.

CONCLUSIONS HAND offers promise in extending the reach of screening for RHD in resource-limited populations. Simplified criteria of MR jet length $ 1.5 cm or any AI can detect RHD with good sensitivity and specificity and greatly reduce the need for standard echocardiography. Further studies are needed to evaluate differences in outcomes with early detection, as well as the accuracy of handheld echocardiographic screening in the hands of nonphysicians, particularly local caregivers such as nurses.

ACKNOWLEDGMENTS The authors thank Alison Reese, Ashley Shrestha-Astudillo, Peter Dean, Lasya Gaur, and Jacqueline Weinberg for assistance in performing echocardiograms; the Rotary Club of Gulu for organizational and logistical support throughout this project; as well as the children and families who consented for participation.

REFERENCES 1. Carapetis JR, Steer AC, Mulholland EK, Weber M. The global burden of group a streptococcal diseases. Lancet Infect Dis 2005;5:685-94. 2. Gunther G, Asmera J, Parry E. Death from rheumatic heart disease in rural Ethiopia. Lancet 2006;367:391. 3. Oli K, Asmera J. Rheumatic heart disease in Ethiopia: could it be more malignant? Ethiop Med J 2004;42:1-8. 4. Bland EF, Duckett Jones T. Rheumatic fever and rheumatic heart disease; a twenty year report on 1000 patients followed since childhood. Circulation 1951;4:836-43. 5. Beaton A, Okello E, Lwabi P, Mondo C, McCarter R, Sable C. Echocardiography screening for rheumatic heart disease in Ugandan schoolchildren. Circulation 2012;125:3127-32. 6. Bhaya M, Panwar S, Beniwal R, Panwar RB. High prevalence of rheumatic heart disease detected by echocardiography in school children. Echocardiography 2010;27:448-53. 7. Colquhoun SM, Kado JH, Remenyi B, Wilson NJ, Carapetis JR, Steer AC. Echocardiographic screening in a resource poor setting: borderline rheumatic heart disease could be a normal variant. Int J Cardiol 2014;173:284-9. 8. Marijon E, Ou P, Celermajer DS, Ferreira B, Mocumbi AO, Jani D, et al. Prevalence of rheumatic heart disease detected by echocardiographic screening. N Engl J Med 2007;357:470-6. 9. Remenyi B, Wilson N, Steer A, Ferreira B, Kado J, Kumar K, et al. World Heart Federation criteria for echocardiographic diagnosis of rheumatic heart disease—an evidence-based guideline. Nat Rev Cardiol 2012;9:297-309. 10. Rheumatic fever and rheumatic heart disease. World Health Organ Tech Rep Ser 2004;923:1-122. back cover. 11. Beaton A, Aliku T, Okello E, Lubega S, McCarter R, Lwabi P, et al. The utility of handheld echocardiography for early diagnosis of rheumatic heart disease. J Am Soc Echocardiogr 2014;27:42-9. 12. Beaton A, Lu JC, Aliku T, Dean P, Gaur L, Weinberg J, et al. The utility of handheld echocardiography for early rheumatic heart disease diagnosis: a field study [published online ahead of print January 6, 2015]. Eur Heart J Cardiovasc Imaging. http://dx.doi.org/10.1093/ehjci/jeu296. 13. Mirabel M, Celermajer DS, Ferreira B, Tafflet M, Perier MC, Karam N, et al. Screening for rheumatic heart disease: evaluation of a simplified echocardiography-based approach. Eur Heart J Cardiovasc Imaging 2012; 13:1024-9. 14. Beaton A, Okello E, Aliku T, Lubega S, Lwabi P, Mondo C, et al. Latent rheumatic heart disease: outcomes 2 years after echocardiographic detection. Pediatr Cardiol 2014;35:1259-67.