- Email: [email protected]
Are minor echocardiographic changes associated with an increased risk of acute rheumatic fever or progression to rheumatic heart disease?
International Journal of Cardiology 198 (2015) 117–122
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Are minor echocardiographic changes associated with an increased risk of acute rheumatic fever or progression to rheumatic heart disease? Marc Rémond a,⁎,1, David Atkinson b,1, Andrew White c,1, Alex Brown d,1, Jonathan Carapetis e,1, Bo Remenyi f,1, Kathryn Roberts f,1, Graeme Maguire g,1 a
James Cook University, Cairns, QLD, Australia Rural Clinical School of Western Australia, The University of Western Australia, Broome, WA, Australia James Cook University, Townsville, QLD, Australia d South Australian Health and Medical Research Institute, Adelaide, SA, Australia e Telethon Kids Institute, University of Western Australia, Perth, WA, Australia f Menzies School of Health Research, Darwin, NT, Australia g Baker IDI Heart and Diabetes Institute, Alice Springs, NT, Australia b c
a r t i c l e
i n f o
Article history: Received 1 May 2015 Received in revised form 16 June 2015 Accepted 2 July 2015 Available online 6 July 2015 Keywords: Rheumatic heart disease Echocardiography Screening Diagnosis Acute rheumatic fever
a b s t r a c t Background: The World Heart Federation criteria for the echocardiographic diagnosis of rheumatic heart disease (RHD) include a category “Borderline” RHD which may represent the earliest evidence of RHD. We aimed to determine the significance of minor heart valve abnormalities, including Borderline RHD, in predicting the future risk of acute rheumatic fever (ARF) or RHD. Methods: A prospective cohort study of Aboriginal and Torres Strait Islander children aged 8 to 18 years was conducted. Cases comprised children with Borderline RHD or other minor non-specific valvular abnormalities (NSVAs) detected on prior echocardiography. Controls were children with a prior normal echocardiogram. Participants underwent a follow-up echocardiogram 2.5 to 5 years later to assess for progression of valvular changes and development of Definite RHD. Interval diagnoses of ARF were ascertained. Results: There were 442 participants. Cases with Borderline RHD were at significantly greater risk of ARF (incidence rate ratio 8.8, 95% CI 1.4–53.8) and any echocardiographic progression of valve lesions (relative risk 8.19, 95% CI 2.43–27.53) than their Matched Controls. Cases with Borderline RHD were at increased risk of progression to Definite RHD (1 in 6 progressed) as were Cases with NSVAs (1 in 10 progressed). Conclusions: Children with Borderline RHD had an increased risk of ARF, progression of valvular lesions, and development of Definite RHD. These findings provide support for considering secondary antibiotic prophylaxis or ongoing surveillance echocardiography in high-risk children with Borderline RHD. © 2015 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Prior to the introduction of echocardiography, diagnosis of rheumatic heart disease (RHD) was primarily based on auscultation. However, it has been shown that auscultation alone is neither sensitive [1] nor specific [2,3]. Increased availability of portable high-quality echocardiography to assess heart valve morphology and function has resulted in significant debate regarding the echocardiographic diagnosis of RHD. This debate has intensified owing to the publication of a number of RHD echocardiographic screening studies that have utilized differing diagnostic criteria [1,2,4,5]. To address this issue, in 2012 the World Heart Federation (WHF) published echocardiographic criteria for the ⁎ Corresponding author. E-mail address: [email protected] (M. Rémond). 1 This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.
http://dx.doi.org/10.1016/j.ijcard.2015.07.005 0167-5273/© 2015 Elsevier Ireland Ltd. All rights reserved.
diagnosis of RHD in the absence of a history of acute rheumatic fever (ARF) [6]. These WHF criteria include a category of “Borderline” RHD that encompasses minor heart valve abnormalities of uncertain clinical significance. The importance of such minor abnormalities in a setting of high RHD risk was highlighted by a recent Australian RHD echocardiographic screening study (gECHO (getting Every Child's Heart Okay)) [7]. Of 3946 high-risk Aboriginal and/or Torres Strait Islander children, 0.9% met the WHF criteria for Definite RHD while 1.7% met criteria for Borderline RHD. Furthermore, mitral regurgitation (MR) was detected in 22.1%, aortic regurgitation (AR) in 4.4%, morphological abnormalities of the mitral valve (MV) in 2.9%, and abnormalities of the aortic valve (AV) in 0.9%. The clinical significance of a diagnosis of Borderline RHD or other non-diagnostic valvular abnormalities in individuals without a history of ARF remains unclear and has been identified as a priority for investigation [6,8]. If these abnormalities represent the earliest changes of RHD then offering such individuals regular secondary prophylaxis may
118
M. Rémond et al. / International Journal of Cardiology 198 (2015) 117–122
prevent disease progression. In contrast, if they are simply a variant of normal echocardiographic findings then unwarranted treatment should be avoided. The Rheumatic Fever Follow-Up Study (RhFFUS) aimed to clarify the significance of minor echocardiographic changes by determining if they were associated with an increased risk of ARF or progressive heart damage consistent with the development of Definite RHD. 2. Methods The methodology of the RhFFUS study has been described previously [9]. Briefly, RhFFUS was a prospective cohort study of Aboriginal and/or Torres Strait Islander children aged 8 to 18 years residing in 32 remote Australian communities. Participants comprised a subset of children who had received an echocardiogram during the earlier gECHO study between September 2008 and November 2011 [7]. They were enrolled in RhFFUS between 2.5 and 5 years after their baseline gECHO echocardiogram. Cases were children with non-specific changes of the MV or AV detected during gECHO. Cases were subdivided into two categories: those with Borderline RHD on the WHF criteria and those with minor non-specific valvular abnormalities (NSVAs) not meeting the Borderline RHD definition (see Box 1). Criteria for Cases were more sensitive than the WHF criteria for Borderline RHD as there remains uncertainty regarding the interpretation and significance of minor echocardiographic abnormalities. Controls (two per Case) were selected who were age, gender, community and ethnicity-matched to Cases and had a prior normal gECHO echocardiogram.
participants prescribed prophylaxis were obtained from jurisdictional ARF/RHD databases. Based on timing of these injections, and on the assumption that a single dose provides 28 days of protection against Group A streptococcal (GAS) infection and possible ARF, we calculated the “days-at-risk” of ARF for each participant during the period between their baseline gECHO and subsequent RhFFUS echocardiograms. 2.2. Progression of valve lesions outcome Progression of valvular abnormalities was assessed prospectively using transthoracic echocardiography and standardized operating and reporting procedures [9]. A Vivid i/e portable cardiac ultrasound machine (GE Healthcare, Freiburg, Germany) was used with a standardized machine setup [7,9]. All echocardiograms were performed by accredited echocardiographers who were blinded to participant status as Case or Control. Progression of valve lesions was determined by a specialist pediatric cardiologist (BR). For each participant, baseline gECHO and subsequent RhFFUS echocardiograms were read individually (and categorized as Normal, NSVA, Borderline RHD [6] or Definite RHD [6]) and then compared to assess for valvular changes. The reader was blinded to the initial gECHO report for the baseline echocardiogram and to the status of participants as Cases or Controls. Reporting was based on a standardized reporting template [9]. “Progression of any valve lesion” was defined as the echocardiographically significant: • development of any significant morphological or functional abnormality [6] in a Control; or • development of a new morphological or functional abnormality [6] in a Case; or • progression of severity of a functional valve lesion (regurgitation/stenosis) based on standard severity criteria [10–12] (i.e. mild to moderate, or moderate to severe).
2.1. ARF outcome For each participant, interval diagnoses of ARF between their baseline gECHO and subsequent RhFFUS echocardiograms were assessed from jurisdictional ARF/RHD notification databases. Participants with an episode of ARF before gECHO were excluded. The use of secondary antibiotic prophylaxis was a potential confounder of ARF risk. Data relating to dates of injections of long-acting benzathine penicillin (LAB) for
Box 1 Criteria used to classify participants recruited to RhFFUS. *Criteria for morphological features of RHD of MV and AV, and pathological AR and MR as described by WHF criteria [6]. CASES Echocardiogram from gECHO that did not fulfill WHF criteria for Definite RHD [6] but which met one of the following: 1. Borderline RHD under WHF criteria [6] A. At least two morphological features of RHD of the MV* without pathological mitral regurgitation* (MR) or mitral stenosis; or B. Pathological MR with no or one morphological feature of RHD of the MV; or C. Pathological aortic regurgitation* (AR) with no or one morphological feature of RHD of the AV*. 2. Non-specific valvular abnormalities (NSVAs) • One morphological feature of RHD of the MV and/or one or more morphological feature of RHD of the AV without pathological MR or AR; or • Multiple MR jets and/or multiple AR jets (in at least two views) that do not fulfill WHF criteria for pathological MR or AR; or • MR ≥ 2 cm or AR ≥ 1 cm (that does not fulfill WHF criteria for pathological regurgitation) with no morphological features of RHD of the MV or AV. CONTROLS Normal screening echocardiogram from gECHO defined as: • • • •
No morphological features of RHD of the AV or MV; and No MR ≥ 1 cm; and No AR; and No other acquired or congenital valvular heart disease.
An additional outcome measure was a diagnosis of Definite RHD [6]. Inter-observer reliability could be assessed as each participant's baseline echocardiogram was read twice (initially during gECHO and subsequently during RhFFUS). 2.3. Sample size calculations Sample size estimations were based on projected rates of ARF based on the annual incidence of ARF in people with known RHD (2.5%) and the background annual incidence of ARF in 5 to 14 year old Aboriginal and Torres Strait Islander children in the NT (0.27%) (personal communication Northern Territory ARF/RHD register) over a five year period. Based on an assumed alpha of 0.05 and beta of 0.1 (power of 90%) detecting a difference of one or more episodes of ARF in 12.5% of Cases and 1.35% of Controls over five years of follow-up, and using a ratio of Cases to Controls of 1:2, would require a sample size of 83 Cases and 165 Controls or a total number of 248 reviews. There were no clear data to inform the powering of progression of non-specific echocardiographic changes. Nonetheless, if it is assumed that at most 5% of Controls will develop morphological and functional valvular changes on echocardiography compared with 20% of those with preexisting non-specific changes then it would require the follow-up of 77 Cases and 154 Controls to detect such a difference with the tolerances above. 2.4. Statistical analysis Statistical analysis was performed using Stata™ version 12.1 (StataCorp, Texas, USA) and SPSS version 20 (IBM Corp., Armonk, NY). Efficacy of matching of Cases and Controls was determined using χ2 analysis for categorical measures and the Mann–Whitney U tests for non-parametric numerical measures. Bivariate analysis of outcome measures was undertaken comparing all Cases to all Controls, Borderline RHD Cases to their Matched Controls, and NSVA Cases to their Matched Controls. A p-value less than 0.05 was taken to indicate statistical significance. All tests were two-sided. ARF incidence rates (IRs) were calculated both for “total time” between baseline gECHO and follow-up RhFFUS echocardiograms and for the “days-at-risk” between these dates. Incidence rate ratios (IRRs) were calculated for matched groups and an IRR with a 95% confidence interval (CI) not including one was taken to be statistically significant. Risk of progression of valve lesions, and progression to Definite RHD, was calculated for matched groups. Absolute risk difference and relative risk (RR) between groups were determined with 95% CIs. Where the 95% CI of a RR did not include one, statistical significance was inferred. Logistic regression models were developed to identify independent factors associated with progression of valve lesions and progression to Definite RHD. These models incorporated all factors associated with each outcome on bivariate analysis with a p-value b0.1. These factors comprised: age, gender, days between echocardiograms, receiving secondary antibiotic prophylaxis, and status as Borderline RHD of the MV (Borderline RHD category A or B) [6] or Borderline RHD of the AV (Borderline RHD category C) [6] or NSVA. Inter-rater reliability was assessed using the linearly weighted Kappa statistic. 2.5. Ethics The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by Human Research Ethics Committees in all jurisdictions where recruitment was undertaken [9]. Written informed consent was obtained from all participants or their guardians.
M. Rémond et al. / International Journal of Cardiology 198 (2015) 117–122
119
3. Results
3.2. Progression of valvular lesions outcome
447 individuals were enrolled. Five participants (2 Cases and 3 Controls) who had a notified episode of ARF prior to gECHO were excluded. Of the 171 potential Cases identified from gECHO, 119 (70%) were successfully enrolled. There were no significant differences in the age, gender or ethnicity of enrolled and non-enrolled Cases (data not shown). While the RhFFUS methodology prescribed two Matched Controls per Case, there was an excess of Controls enrolled as the study team was unable to locate and enroll some Cases after their Matched Controls had already been enrolled. These “unmatched” Controls were included to increase the power of the study and because their inclusion did not lead to any significant differences in the demographics of the Case and Control groups (see Table 1). Based on the original reporting of baseline gECHO echocardiograms, 55 (47%) Cases met the WHF criteria for Borderline RHD [6] while 62 (53%) had NSVAs. Of the 55 Borderline RHD Cases, 13 (24%) were subcategory A, 21 (38%) were subcategory B, and 21 (38%) were subcategory C [6]. The most common NSVA was isolated MR or AR (30/62, 46%). The remaining NSVAs (32, 54%) were morphological features of RHD alone or multiple, non-significant regurgitant jets. A comparison of Borderline and NSVA Cases revealed no significant differences in the factors listed in Table 1 apart from prescription of LAB. Borderline RHD Cases were more likely to be prescribed LAB compared with NSVA Cases (34.6% (95% CI 23.4–47.8) vs. 4.8% (95% CI 1.7– 13.3), p b 0.001).
Forty-two (9.5%, 95% CI 7.1–12.6) participants exhibited echocardiographically significant progression of valvular lesions. The majority of these (25, 60%) exhibited isolated increases in severity of valvular regurgitation, while 13 (31%) exhibited concomitant worsening of valvular regurgitation and morphology, and 4 (9%) exhibited isolated changes in valve morphology. Cases were at significantly greater risk of echocardiographic progression than Controls predominantly due to an increased risk in the Borderline RHD subgroup (see Table 2). Of the Borderline RHD subgroup, 4/13 (31%, 95% CI 9–61) of Subcategory A, 6/21 (29%, 95% CI 11–52) of Subcategory B, and 3/21 (14%, 95% CI 3–36) of Subcategory C progressed. The NSVA subgroup was not at increased risk of progression (see Table 2). Seventeen (40%) of the 42 participants who had progression of valvular lesions were diagnosed with Definite RHD [6]. The majority of these (13/17, 76%) had isolated MV disease (Definite RHD A). Two (12%) had MV and AV disease (Definite RHD D) and two (12%) had isolated AV disease (Definite RHD C). Of the 17 individuals who progressed to Definite RHD, 9 (53%) had been initially classified as Borderline RHD Cases, 6 (35%) as NSVA Cases, and 2 (12%) as Controls. Of the Borderline RHD subgroup, 2/13 (15%, 95% CI 4–42) of Subcategory A, 5/21 (24%, 95% CI 11–45) of Subcategory B, and 2/21 (10%, 95% CI 3–29) of Subcategory C progressed to Definite RHD. Cases were at significantly greater risk of progression to Definite RHD than Controls as was the NSVA subgroup (see Table 3). The Borderline RHD subgroup had a significantly increased absolute risk of progression to Definite RHD compared to their Matched Controls (p b 0.001, Fisher's exact test) but a relative risk could not be determined as none of their Matched Controls progressed to Definite RHD. Logistic regression modelling revealed three common factors that were independently associated with progression of valve lesions and development of Definite RHD (see Table 4). Both functional and morphological changes of the mitral valve were independently associated with progression and Definite RHD (data not shown) and were therefore combined. In addition, Borderline RHD of the aortic valve was found to confer a significant increase in the chance of development of Definite RHD but this was less than that seen with Borderline RHD of the mitral valve. The factors that were independently associated with progression of valve lesions and development of Definite RHD in logistic regression modelling are presented in Table 4. We stratified Cases into two subgroups based on the presence or absence of MR and compared these two groups in relation to progression outcome data. Cases with MR were not at a significantly increased risk of any progression of valvular lesions (RR 1.2, 95% CI, 0.6–2.5) or progression to definite RHD (RR 1.5, 95% CI, 0.5–4.1) compared to those Cases without MR.
3.1. ARF outcome There were 9 reported episodes of ARF. All met criteria for Definite ARF [13]. The IR of ARF was higher in Cases (4/117, 9.2 episodes/1000 person-years, 95% CI 2.5–23.5) than Controls (5/325, 4.2/1000 personyears, 95% CI 1.4–9.7), although this difference was not statistically significant (IRR 2.2, 95% CI 0.6–7.9). After adjusting for LAB use (“days-at-risk”), the IR of ARF was unchanged in Controls but increased in Cases to 9.9 episodes/1000 person-years-at-risk (95% CI 2.7–25.4). This corresponded to an IRR of 2.3 (95% CI 0.7–8.4) and again was not statistically significant. All four Cases with an episode of ARF were in the Borderline RHD subgroup. The IR of ARF in the Borderline RHD subgroup was 19.6/ 1000 person-years (95% CI 5.3–50.3) and increased to 22.9/1000 person-years-at-risk (95% CI 6.3–58.7) when controlling for LAB use. In comparison with their Matched Controls, the participants in the Borderline RHD subgroup were at a statistically significant increased risk of ARF (unadjusted IRR = 7.5, 95% CI 1.2–48.3; IRR adjusted for use of LAB = 8.8, 95% CI 1.4–53.8). Table 1 Demographic data, use of secondary prophylaxis, and time to follow-up for participants.
Total number Age (median years, IQR) Gender (% female, 95% CI) Ethnicity (%, 95% CI) Aboriginal Torres Strait Islander Aboriginal and Torres Strait Islander Prescribed secondary prophylaxis (%, 95% CI) Time between gECHO echocardiogram and RhFFUS enrolment (median days, IQR)
Cases
Controls
p value
117 13.7 (11.9–15.7)
325 13.7 (11.8–15.5)
n/a 0.733
59.0 (49.5–67.0)
57.2 (51.6–62.7)
0.787 0.965
83.8 (75.8–90.0) 6.8 (3.0–13.0) 9.4 (4.8–16.2) 18.8 (12.2–27.1)
84.3 (79.9–88.1) 7.1 (4.5–10.4) 8.6 (5.8–12.2) 2.5 (1.1–4.8)
b0.001
1346 (1171–1478)
1310 (1170–1440)
0.455
120
M. Rémond et al. / International Journal of Cardiology 198 (2015) 117–122
Table 2 Progression of valvular lesions on follow-up echocardiography. All
All Cases
All Controls
Borderline RHD Cases
Matched Controls
NSVA Cases
Matched Controls
Number
442
117
325
55
104
62
122
Progression of valve abnormalities (n, %, 95% CI)
42 9.5% (7.1–12.6%) n/a
23 19.7% (13.5–27.8%)
19 5.9% (3.8–9.0%)
13 23.6% (14.4–36.4%)
3 2.9% (1.0–8.1%)
10 16.1% (9.0–27.2%)
13 10.7% (6.3–17.4%)
Absolute risk difference (%, 95% CI) Relative risk (RR, 95% CI)
n/a
13.8% (6.2–21.5) 3.36 (1.90–5.94)
20.8% (9.1–32.4) 8.19 (2.43–27.53)
5.5% (–5.2–16.1) 1.51 (0.70–3.25)
from Uganda reported that after 2 years, 10% of children with an initial diagnosis of Borderline RHD by the WHF criteria had progressed to Definite RHD [18]. Given the shorter period of follow-up in that study this is again similar to the findings in the current study. This study addresses the issue of the utility of existing WHF criteria [6] for the echocardiographic diagnosis of RHD. Our results support the criteria used by the WHF for Borderline RHD by demonstrating that individuals who satisfy these criteria are at increased risk of ARF, progression of valvular damage, and progression to Definite RHD. However, we also demonstrated that children with less severe NSVAs may also be at a lesser but still increased risk of progression to RHD. These findings highlight that both functional and morphological changes on echocardiography form a continuum. While the current Borderline RHD category reflects an increased risk of ARF and echocardiographic progression, NSVA should also be viewed as part of this continuum. A larger cohort with longer-term follow-up will be necessary to determine exactly where, on this continuum, the cutoff between ‘normal’ and ‘abnormal’ may lie. The increased risk of ARF and progression to Definite RHD in children with Borderline RHD suggests that there may be a role for secondary antibiotic prophylaxis in such children. However, caution should be exercised. First, it is important to more accurately identify the subset of individuals at risk of ARF and progression. There were too few episodes of ARF recorded in this study to provide sufficient power to undertake a subgroup analysis of the risk of ARF in those with Borderline RHD subcategory A, B or C. However, the logistic regression models developed did reveal that children with Borderline disease of the MV (subcategories A and B) were at greatest risk of deterioration of valve lesions and progression to Definite RHD. Nonetheless, children with Borderline disease of the AV (subcategory C) still had an increased chance of development of Definite RHD. Furthermore, utilizing MR alone as a predictor of progression or development of Definite RHD was not useful even though previously it has been suggested that MR alone may have potential utility as a screening tool for RHD in high-risk populations [19]. These results suggest that functional and/or morphological features of the MV are important in detecting those at greatest chance of progression but that AV changes cannot be ignored. The second issue regarding ongoing management of Borderline RHD and NSVAs relates to the efficacy of secondary preventive initiatives in the context of this study. It is still not known whether individuals with Borderline RHD or NSVA, or indeed Definite RHD demonstrated on echocardiography without preceding ARF [6], will be equivalently
3.3. Inter-observer reliability Data outlining the different classifications assigned to each participant based on the original gECHO report of participants' baseline echocardiogram and subsequent re-report of that same echocardiogram as part of the RhFFUS study are presented in Table 5. The inter-rater reliability for reading of the baseline echocardiogram using linear weighted Kappa was 0.656 (p b 0.001, 95% CI 0.592–0.721) and consistent with substantial agreement [14]. To control for potential bias due to discordant reporting of baseline echocardiograms, Cases or Controls were reclassified based on the rereading of baseline echocardiograms during RhFFUS. Following this reclassification, Cases remained at a higher, but non-significant, risk of ARF compared with Controls (IRR 3.9, 95% CI 0.8–25.3). Furthermore, Cases remained at significantly greater risk of any progression of valvular lesions (RR 3.17, 95% CI 1.69–5.94) and progression to Definite RHD (RR 24.9, 95% CI 3.2 to 190.7).
4. Discussion We have shown, for the first time, that children diagnosed with Borderline RHD by the WHF criteria [6] are at increased risk of ARF, progression of cardiac valvular lesions, and development of Definite RHD. It is therefore clear that at least some children with Borderline RHD have true RHD. We have also shown that children with Borderline RHD of the MV, rather than the AV, are at particularly high risk of progression of valve disease. While ARF is rarely diagnosed in Australia, some disadvantaged groups, specifically Aboriginal and Torres Strait Islander populations, continue to experience high incidence rates [15]. Our reported incidence of ARF in Controls (4.2/1000 person-years) was comparable to existing data in Aboriginal Australians in this setting (2.5–3.5/1000 person-years) [15,16]. Against this already high background risk of ARF, children with Borderline RHD had a significantly increased risk of ARF (19.6–22.9/1000 person-years). While the rates of progression of valvular damage seen in this study were greater than those reported in some previous studies, this is likely to be in part explained by our longer period of follow-up. An Indian study that followed children with subclinical RHD for a mean period of 1.3 years reported that 4% progressed [17]. This is similar to, but lower than, the current study (16% over 3.6 years). Another study Table 3 Progression to Definite RHD [6]. All
All Cases
All Controls
Borderline RHD Cases
Matched Controls
NSVA Cases
Matched Controls
N
442
117
325
55
104
62
122
Progression to Definite RHD (n, %, 95% CI)
17 3.9% (2.4–6.1) n/a
15 12.8% (7.9–20.1) 12.2% (6.1–18.3) 20.8 (4.8–89.7)
2 0.6% (0.2–2.2)
9 16.4% (8.9–28.3) 16.4% (6.6–26.1) Could not be determined
0 0%
6 9.7% (4.5–19.6) 8.1% (0.3–15.7) 5.9 (1.2–28.4)
2 1.6% (0.5–5.8)
Absolute risk difference (%, 99% CI) Relative risk (RR, 95% CI)
n/a
M. Rémond et al. / International Journal of Cardiology 198 (2015) 117–122 Table 4 Factors that were independently associated with progression of valvular lesions and progression to Definite RHD in logistic regression models.
Borderline RHD of the MV — categories A and B [6] (OR, 95% CI) Borderline RHD of the AV — category C [6] (OR, 95% CI) NSVA (OR, 95% CI) In receipt of secondary antibiotic prophylaxis (OR, 95% CI) Nagelkerke R [2]
Progression of valvular lesions
Progression to Definite RHD
4.6 (1.8–12.1)
30.0 (5.4–167.3)
–
9.2 (1.0–82.6) 16.0 (3.1–81.8) 4.0 (1.1–15.3) 29.8%
3.0 (1.3–6.8) 4.2 (1.5–11.7) 14.7%
responsive to LAB as those with RHD detected in association with ARF or an incidental murmur. Our findings contribute to the ongoing debate regarding the relevance and feasibility of echocardiographic screening for RHD in highrisk populations [20]. Based on results from gECHO [7], the follow-up of Borderline RHD would triple the number of additional patients who may require management. Previously it has been shown that RHD screening activities can have an adverse impact on local health services and patient quality of life [21]. Furthermore, it has been shown that even in the relatively well resourced health care environment of Australia the uptake of LAB is often poor [22]. Despite these issues, if managing Borderline RHD Cases can be supported by local health services then the results of this study may provide an additional rationale for supporting on-going echocardiographic screening in this setting. 4.1. Study limitations One in three Borderline RHD Cases in this study was receiving LAB. Such prescription was at the discretion of clinical staff rather than dictated, or even recommended, by the gECHO study. Presumably clinicians had interpreted the relatively non-specific echocardiographic findings as being commensurate with Definite RHD. It was likely that the use of LAB had a confounding influence on the subsequent risk of ARF. Unsurprisingly, when we controlled for this potential confounder, the risk of ARF in those with Borderline RHD was increased. The influence of LAB use on echocardiographic progression was less clear. Logistic regression modeling suggested that the use of prophylaxis was associated with a greater risk of progression. It seems unlikely that LAB was conferring such an increased risk. Rather, it was more likely that this was an incidental finding resulting from the fact that, based on gECHO screening results, clinicians were identifying individuals at greatest risk of progression and prescribing LAB to them. It should also be noted that none of the participants receiving LAB in this study were covered for the entire period of time between gECHO and RhFFUS. All of these individuals commenced LAB well after their gECHO echocardiogram and, once treatment had started, none received 100% of their
121
scheduled doses in the time leading up to their RhFFUS echocardiogram. Hence all these participants were still at risk of GAS infection, recurrent episodes of ARF, and progression of valvular lesions for some part of the study period. Furthermore, there is debate over whether the recommended 4-weekly doses of LAB under Australian guidelines [13] provide protection against GAS infection for 28 days as we assumed in our analysis [23]. Hence, no conclusions can be drawn from our data in regard to the effectiveness of LAB in preventing progression of valvular lesions or development of Definite RHD in individuals with echocardiographically-diagnosed NSVAs or Borderline RHD. This is an important question that warrants further research. A further potential source of bias was inter-observer variability. We demonstrated substantial agreement between the original reading of baseline echocardiograms and subsequent re-reading of these echocardiograms during RhFFUS. Further, when we reclassified the status of participants as Cases or Controls based on the RhFFUS reading of baseline echocardiograms, this did not alter findings regarding the risks of ARF, progression of valvular lesions, or progression to Definite RHD. While results revealed that Borderline RHD Cases were at greater risk of ARF than their Matched Controls (IRR = 8.8, 95% CI 1.4–53.8), it should be noted that the number of events (episodes of ARF) in this context was small – 9 events in total, 4 in the Borderline RHD Case group – and that the confidence interval for the incidence rate ratio was broad. Hence caution needs to be exercised in interpreting this result and any attempts to generalize these findings to other contexts. Nonetheless, it should also be noted that this analysis was undertaken to test an a priori hypothesis regarding the risk of ARF in individuals with non-specific heart valve abnormalities detected on RHD screening echocardiogram and that, despite the small number of events, this finding was statistically significant. A further limitation of this study relates to potential selection bias. Participants were selected from a group of children who had been enrolled in a previous school-based RHD screening study. Hence, children who did not usually attend school at the time of this previous study were excluded. There is no evidence to suggest that in remote Indigenous Australian settings there is a difference in risk of having RHD between those that attend school and those that do not. However, this limitation needs to be considered whenever interpreting these results in the context of the natural history of Borderline RHD.
4.2. Conclusions Results from this study support the contention that in some highrisk children Borderline RHD and other NSVAs represent the earliest changes of RHD or, at least, indicate a group at increased risk of ARF, progression of valvular lesions and development of Definite RHD. These findings provide a rationale for considering enhanced clinical followup of high-risk children with Borderline RHD and, to a lesser extent, other NSVAs. Whether this entails ARF education, ongoing regular echocardiographic surveillance, or prescription of LAB remains a clinical judgment.
Table 5 Inter-observer variability in reading of gECHO baseline echocardiograms. Classification based on original reading of baseline echocardiogram
Classification based upon subsequent reread of baseline echocardiogram
a
Normal
NSVA
Borderline RHD
Total
18
2
299
NSVA
279 (85.8%)a 41
7
81
Borderline RHD
4
33 (53.2%) 8
51
Definite RHD Total
1 325
3 62
39 (70.9%) 7 55
Normal
Percentages in brackets show the concordance between readers within each subcategory of participant based on original reading of baseline echocardiograms.
11 442
122
M. Rémond et al. / International Journal of Cardiology 198 (2015) 117–122
Conflicts of interest None.
[10] [11]
Acknowledgment [12]
This study was supported by an Australian National Health and Medical Research Council project grant (Grant Application 1005951) and the NHMRC Centre for Research Excellence to Reduce Inequality in Heart Disease (Grant Application 1044897).
[13]
[14]
References [15] [1] E. Marijon, P. Ou, D.S. Celermajer, et al., Prevalence of rheumatic heart disease detected by echocardiographic screening, N. Engl. J. Med. 357 (2007) 470–476. [2] J.R. Carapetis, M. Hardy, T. Fakakovikaetau, et al., Evaluation of a screening protocol using auscultation and portable echocardiography to detect asymptomatic rheumatic heart disease in Tongan schoolchildren, Nat. Clin. Pract. Cardiovasc. Med. 5 (2008) 411–417. [3] K.V. Roberts, A.D. Brown, G.P. Maguire, D.N. Atkinson, J.R. Carapetis, Utility of auscultatory screening for detecting rheumatic heart disease in high-risk children in Australia's Northern Territory, Med. J. Aust. 199 (2013) 196–199. [4] B.M. Reeves, J. Kado, M. Brook, High prevalence of rheumatic heart disease in Fiji detected by echocardiography screening, J. Paediatr. Child Health 47 (2011) 473–478. [5] R.H. Webb, N.J. Wilson, D.R. Lennon, et al., Optimising echocardiographic screening for rheumatic heart disease in New Zealand: not all valve disease is rheumatic, Cardiol. Young 21 (2011) 436–443. [6] B. Remenyi, N. Wilson, A. Steer, et al., World Heart Federation criteria for echocardiographic diagnosis of rheumatic heart disease — an evidence-based guideline, Nat. Rev. Cardiol. 9 (2012) 297–309. [7] K. Roberts, G. Maguire, A. Brown, et al., Echocardiographic screening for rheumatic heart disease in high and low risk Australian children, Circulation 129 (2014) 1953–1961. [8] M.G. Remond, G.R. Wheaton, W.F. Walsh, D.L. Prior, G.P. Maguire, Acute rheumatic fever and rheumatic heart disease — priorities in prevention, diagnosis and management. A report of the CSANZ Indigenous Cardiovascular Health Conference, Alice Springs 2011, Heart Lung Circ. 21 (2012) 632–638. [9] M.G. Remond, D. Atkinson, A. White, et al., Rheumatic fever follow-up study (RhFFUS) protocol: a cohort study investigating the significance of minor
[16]
[17]
[18] [19]
[20]
[21]
[22]
[23]
echocardiographic abnormalities in Aboriginal Australian and Torres Strait Islander children, BMC Cardiovasc. Disord. 12 (2012) 111. W.F. Armstrong, T. Ryan, Feigenbaum's Echocardiography, Lippincott Williams & Wilkins, Philadelphia, PA, 2009. W.A. Zoghbi, M. Enriquez-Sarano, E. Foster, et al., Recommendations for evaluation of the severity of native valvular regurgitation with two-dimensional and Doppler echocardiography, J. Am. Soc. Echocardiogr. 16 (2003) 777–802. H. Baumgartner, J. Hung, J. Bermejo, et al., Echocardiographic assessment of valve stenosis: EAE/ASE recommendations for clinical practice, J. Am. Soc. Echocardiogr. 22 (2009) 1–23. RHDAustralia (ARF/RHD writing group), National Heart Foundation of Australia and the Cardiac Society of Australia and New Zealand, Australian Guideline for Prevention, Diagnosis and Management of Acute Rheumatic Fever and Rheumatic Heart Disease, 2nd edition, 2012. J.R. Landis, G.G. Koch, The measurement of observer agreement for categorical data, Biometrics 33 (1977) 159–174. B. Field, Rheumatic heart disease: all but forgotten except among Aboriginal and Torres Strait Islander peoples, Bulletin No. 16. , AIHW, Canberra, 2004. J.R. Carapetis, D.R. Wolff, B.J. Currie, Acute rheumatic fever and rheumatic heart disease in the top end of Australia's Northern Territory, Med. J. Aust. 164 (1996) 146–149. A. Saxena, S. Ramakrishnan, A. Roy, et al., Prevalence and outcome of subclinical rheumatic heart disease in India: the RHEUMATIC (Rheumatic Heart Echo Utilisation and Monitoring Actuarial Trends in Indian Children) study, Heart 97 (2011) 2018–2022. A. Beaton, E. Okello, T. Aliku, et al., Latent rheumatic heart disease: outcomes 2 years after echocardiographic detection, Pediatr. Cardiol. 35 (2014) 1259–1267. M. Mirabel, R. Bacquelin, M. Tafflet, et al., Screening for rheumatic heart disease: evaluation of a focused cardiac ultrasound approach, Circ. Cardiovasc. Imaging 8 (1) (2015), http://dx.doi.org/10.1161/CIRCIMAGING.114.002324 (pii: e002324). M.G. Remond, E.K. Wark, G.P. Maguire, Screening for rheumatic heart disease in Aboriginal and Torres Strait Islander children, J. Paediatr. Child Health 49 (2013) 526–531. E.K. Wark, Y.C. Hodder, C.E. Woods, G.P. Maguire, Patient and health-care impact of a pilot rheumatic heart disease screening program, J. Paediatr. Child Health 49 (2013) 297–302. M.G. Remond, K.L. Severin, Y. Hodder, et al., Variability in disease burden and management of rheumatic fever and rheumatic heart disease in two regions of tropical Australia, Intern. Med. J. 43 (2013) 386–393. M.P. Broderick, C.J. Hansen, D.J. Faix, Factors associated with loss of penicillin G concentrations in serum after intramuscular benzathine penicillin G injection: a metaanalysis, Pediatr. Infect. Dis. J. 31 (2012) 722–725.
Contents lists available at ScienceDirect
International Journal of Cardiology journal homepage: www.elsevier.com/locate/ijcard
Are minor echocardiographic changes associated with an increased risk of acute rheumatic fever or progression to rheumatic heart disease? Marc Rémond a,⁎,1, David Atkinson b,1, Andrew White c,1, Alex Brown d,1, Jonathan Carapetis e,1, Bo Remenyi f,1, Kathryn Roberts f,1, Graeme Maguire g,1 a
James Cook University, Cairns, QLD, Australia Rural Clinical School of Western Australia, The University of Western Australia, Broome, WA, Australia James Cook University, Townsville, QLD, Australia d South Australian Health and Medical Research Institute, Adelaide, SA, Australia e Telethon Kids Institute, University of Western Australia, Perth, WA, Australia f Menzies School of Health Research, Darwin, NT, Australia g Baker IDI Heart and Diabetes Institute, Alice Springs, NT, Australia b c
a r t i c l e
i n f o
Article history: Received 1 May 2015 Received in revised form 16 June 2015 Accepted 2 July 2015 Available online 6 July 2015 Keywords: Rheumatic heart disease Echocardiography Screening Diagnosis Acute rheumatic fever
a b s t r a c t Background: The World Heart Federation criteria for the echocardiographic diagnosis of rheumatic heart disease (RHD) include a category “Borderline” RHD which may represent the earliest evidence of RHD. We aimed to determine the significance of minor heart valve abnormalities, including Borderline RHD, in predicting the future risk of acute rheumatic fever (ARF) or RHD. Methods: A prospective cohort study of Aboriginal and Torres Strait Islander children aged 8 to 18 years was conducted. Cases comprised children with Borderline RHD or other minor non-specific valvular abnormalities (NSVAs) detected on prior echocardiography. Controls were children with a prior normal echocardiogram. Participants underwent a follow-up echocardiogram 2.5 to 5 years later to assess for progression of valvular changes and development of Definite RHD. Interval diagnoses of ARF were ascertained. Results: There were 442 participants. Cases with Borderline RHD were at significantly greater risk of ARF (incidence rate ratio 8.8, 95% CI 1.4–53.8) and any echocardiographic progression of valve lesions (relative risk 8.19, 95% CI 2.43–27.53) than their Matched Controls. Cases with Borderline RHD were at increased risk of progression to Definite RHD (1 in 6 progressed) as were Cases with NSVAs (1 in 10 progressed). Conclusions: Children with Borderline RHD had an increased risk of ARF, progression of valvular lesions, and development of Definite RHD. These findings provide support for considering secondary antibiotic prophylaxis or ongoing surveillance echocardiography in high-risk children with Borderline RHD. © 2015 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Prior to the introduction of echocardiography, diagnosis of rheumatic heart disease (RHD) was primarily based on auscultation. However, it has been shown that auscultation alone is neither sensitive [1] nor specific [2,3]. Increased availability of portable high-quality echocardiography to assess heart valve morphology and function has resulted in significant debate regarding the echocardiographic diagnosis of RHD. This debate has intensified owing to the publication of a number of RHD echocardiographic screening studies that have utilized differing diagnostic criteria [1,2,4,5]. To address this issue, in 2012 the World Heart Federation (WHF) published echocardiographic criteria for the ⁎ Corresponding author. E-mail address: [email protected] (M. Rémond). 1 This author takes responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.
http://dx.doi.org/10.1016/j.ijcard.2015.07.005 0167-5273/© 2015 Elsevier Ireland Ltd. All rights reserved.
diagnosis of RHD in the absence of a history of acute rheumatic fever (ARF) [6]. These WHF criteria include a category of “Borderline” RHD that encompasses minor heart valve abnormalities of uncertain clinical significance. The importance of such minor abnormalities in a setting of high RHD risk was highlighted by a recent Australian RHD echocardiographic screening study (gECHO (getting Every Child's Heart Okay)) [7]. Of 3946 high-risk Aboriginal and/or Torres Strait Islander children, 0.9% met the WHF criteria for Definite RHD while 1.7% met criteria for Borderline RHD. Furthermore, mitral regurgitation (MR) was detected in 22.1%, aortic regurgitation (AR) in 4.4%, morphological abnormalities of the mitral valve (MV) in 2.9%, and abnormalities of the aortic valve (AV) in 0.9%. The clinical significance of a diagnosis of Borderline RHD or other non-diagnostic valvular abnormalities in individuals without a history of ARF remains unclear and has been identified as a priority for investigation [6,8]. If these abnormalities represent the earliest changes of RHD then offering such individuals regular secondary prophylaxis may
118
M. Rémond et al. / International Journal of Cardiology 198 (2015) 117–122
prevent disease progression. In contrast, if they are simply a variant of normal echocardiographic findings then unwarranted treatment should be avoided. The Rheumatic Fever Follow-Up Study (RhFFUS) aimed to clarify the significance of minor echocardiographic changes by determining if they were associated with an increased risk of ARF or progressive heart damage consistent with the development of Definite RHD. 2. Methods The methodology of the RhFFUS study has been described previously [9]. Briefly, RhFFUS was a prospective cohort study of Aboriginal and/or Torres Strait Islander children aged 8 to 18 years residing in 32 remote Australian communities. Participants comprised a subset of children who had received an echocardiogram during the earlier gECHO study between September 2008 and November 2011 [7]. They were enrolled in RhFFUS between 2.5 and 5 years after their baseline gECHO echocardiogram. Cases were children with non-specific changes of the MV or AV detected during gECHO. Cases were subdivided into two categories: those with Borderline RHD on the WHF criteria and those with minor non-specific valvular abnormalities (NSVAs) not meeting the Borderline RHD definition (see Box 1). Criteria for Cases were more sensitive than the WHF criteria for Borderline RHD as there remains uncertainty regarding the interpretation and significance of minor echocardiographic abnormalities. Controls (two per Case) were selected who were age, gender, community and ethnicity-matched to Cases and had a prior normal gECHO echocardiogram.
participants prescribed prophylaxis were obtained from jurisdictional ARF/RHD databases. Based on timing of these injections, and on the assumption that a single dose provides 28 days of protection against Group A streptococcal (GAS) infection and possible ARF, we calculated the “days-at-risk” of ARF for each participant during the period between their baseline gECHO and subsequent RhFFUS echocardiograms. 2.2. Progression of valve lesions outcome Progression of valvular abnormalities was assessed prospectively using transthoracic echocardiography and standardized operating and reporting procedures [9]. A Vivid i/e portable cardiac ultrasound machine (GE Healthcare, Freiburg, Germany) was used with a standardized machine setup [7,9]. All echocardiograms were performed by accredited echocardiographers who were blinded to participant status as Case or Control. Progression of valve lesions was determined by a specialist pediatric cardiologist (BR). For each participant, baseline gECHO and subsequent RhFFUS echocardiograms were read individually (and categorized as Normal, NSVA, Borderline RHD [6] or Definite RHD [6]) and then compared to assess for valvular changes. The reader was blinded to the initial gECHO report for the baseline echocardiogram and to the status of participants as Cases or Controls. Reporting was based on a standardized reporting template [9]. “Progression of any valve lesion” was defined as the echocardiographically significant: • development of any significant morphological or functional abnormality [6] in a Control; or • development of a new morphological or functional abnormality [6] in a Case; or • progression of severity of a functional valve lesion (regurgitation/stenosis) based on standard severity criteria [10–12] (i.e. mild to moderate, or moderate to severe).
2.1. ARF outcome For each participant, interval diagnoses of ARF between their baseline gECHO and subsequent RhFFUS echocardiograms were assessed from jurisdictional ARF/RHD notification databases. Participants with an episode of ARF before gECHO were excluded. The use of secondary antibiotic prophylaxis was a potential confounder of ARF risk. Data relating to dates of injections of long-acting benzathine penicillin (LAB) for
Box 1 Criteria used to classify participants recruited to RhFFUS. *Criteria for morphological features of RHD of MV and AV, and pathological AR and MR as described by WHF criteria [6]. CASES Echocardiogram from gECHO that did not fulfill WHF criteria for Definite RHD [6] but which met one of the following: 1. Borderline RHD under WHF criteria [6] A. At least two morphological features of RHD of the MV* without pathological mitral regurgitation* (MR) or mitral stenosis; or B. Pathological MR with no or one morphological feature of RHD of the MV; or C. Pathological aortic regurgitation* (AR) with no or one morphological feature of RHD of the AV*. 2. Non-specific valvular abnormalities (NSVAs) • One morphological feature of RHD of the MV and/or one or more morphological feature of RHD of the AV without pathological MR or AR; or • Multiple MR jets and/or multiple AR jets (in at least two views) that do not fulfill WHF criteria for pathological MR or AR; or • MR ≥ 2 cm or AR ≥ 1 cm (that does not fulfill WHF criteria for pathological regurgitation) with no morphological features of RHD of the MV or AV. CONTROLS Normal screening echocardiogram from gECHO defined as: • • • •
No morphological features of RHD of the AV or MV; and No MR ≥ 1 cm; and No AR; and No other acquired or congenital valvular heart disease.
An additional outcome measure was a diagnosis of Definite RHD [6]. Inter-observer reliability could be assessed as each participant's baseline echocardiogram was read twice (initially during gECHO and subsequently during RhFFUS). 2.3. Sample size calculations Sample size estimations were based on projected rates of ARF based on the annual incidence of ARF in people with known RHD (2.5%) and the background annual incidence of ARF in 5 to 14 year old Aboriginal and Torres Strait Islander children in the NT (0.27%) (personal communication Northern Territory ARF/RHD register) over a five year period. Based on an assumed alpha of 0.05 and beta of 0.1 (power of 90%) detecting a difference of one or more episodes of ARF in 12.5% of Cases and 1.35% of Controls over five years of follow-up, and using a ratio of Cases to Controls of 1:2, would require a sample size of 83 Cases and 165 Controls or a total number of 248 reviews. There were no clear data to inform the powering of progression of non-specific echocardiographic changes. Nonetheless, if it is assumed that at most 5% of Controls will develop morphological and functional valvular changes on echocardiography compared with 20% of those with preexisting non-specific changes then it would require the follow-up of 77 Cases and 154 Controls to detect such a difference with the tolerances above. 2.4. Statistical analysis Statistical analysis was performed using Stata™ version 12.1 (StataCorp, Texas, USA) and SPSS version 20 (IBM Corp., Armonk, NY). Efficacy of matching of Cases and Controls was determined using χ2 analysis for categorical measures and the Mann–Whitney U tests for non-parametric numerical measures. Bivariate analysis of outcome measures was undertaken comparing all Cases to all Controls, Borderline RHD Cases to their Matched Controls, and NSVA Cases to their Matched Controls. A p-value less than 0.05 was taken to indicate statistical significance. All tests were two-sided. ARF incidence rates (IRs) were calculated both for “total time” between baseline gECHO and follow-up RhFFUS echocardiograms and for the “days-at-risk” between these dates. Incidence rate ratios (IRRs) were calculated for matched groups and an IRR with a 95% confidence interval (CI) not including one was taken to be statistically significant. Risk of progression of valve lesions, and progression to Definite RHD, was calculated for matched groups. Absolute risk difference and relative risk (RR) between groups were determined with 95% CIs. Where the 95% CI of a RR did not include one, statistical significance was inferred. Logistic regression models were developed to identify independent factors associated with progression of valve lesions and progression to Definite RHD. These models incorporated all factors associated with each outcome on bivariate analysis with a p-value b0.1. These factors comprised: age, gender, days between echocardiograms, receiving secondary antibiotic prophylaxis, and status as Borderline RHD of the MV (Borderline RHD category A or B) [6] or Borderline RHD of the AV (Borderline RHD category C) [6] or NSVA. Inter-rater reliability was assessed using the linearly weighted Kappa statistic. 2.5. Ethics The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by Human Research Ethics Committees in all jurisdictions where recruitment was undertaken [9]. Written informed consent was obtained from all participants or their guardians.
M. Rémond et al. / International Journal of Cardiology 198 (2015) 117–122
119
3. Results
3.2. Progression of valvular lesions outcome
447 individuals were enrolled. Five participants (2 Cases and 3 Controls) who had a notified episode of ARF prior to gECHO were excluded. Of the 171 potential Cases identified from gECHO, 119 (70%) were successfully enrolled. There were no significant differences in the age, gender or ethnicity of enrolled and non-enrolled Cases (data not shown). While the RhFFUS methodology prescribed two Matched Controls per Case, there was an excess of Controls enrolled as the study team was unable to locate and enroll some Cases after their Matched Controls had already been enrolled. These “unmatched” Controls were included to increase the power of the study and because their inclusion did not lead to any significant differences in the demographics of the Case and Control groups (see Table 1). Based on the original reporting of baseline gECHO echocardiograms, 55 (47%) Cases met the WHF criteria for Borderline RHD [6] while 62 (53%) had NSVAs. Of the 55 Borderline RHD Cases, 13 (24%) were subcategory A, 21 (38%) were subcategory B, and 21 (38%) were subcategory C [6]. The most common NSVA was isolated MR or AR (30/62, 46%). The remaining NSVAs (32, 54%) were morphological features of RHD alone or multiple, non-significant regurgitant jets. A comparison of Borderline and NSVA Cases revealed no significant differences in the factors listed in Table 1 apart from prescription of LAB. Borderline RHD Cases were more likely to be prescribed LAB compared with NSVA Cases (34.6% (95% CI 23.4–47.8) vs. 4.8% (95% CI 1.7– 13.3), p b 0.001).
Forty-two (9.5%, 95% CI 7.1–12.6) participants exhibited echocardiographically significant progression of valvular lesions. The majority of these (25, 60%) exhibited isolated increases in severity of valvular regurgitation, while 13 (31%) exhibited concomitant worsening of valvular regurgitation and morphology, and 4 (9%) exhibited isolated changes in valve morphology. Cases were at significantly greater risk of echocardiographic progression than Controls predominantly due to an increased risk in the Borderline RHD subgroup (see Table 2). Of the Borderline RHD subgroup, 4/13 (31%, 95% CI 9–61) of Subcategory A, 6/21 (29%, 95% CI 11–52) of Subcategory B, and 3/21 (14%, 95% CI 3–36) of Subcategory C progressed. The NSVA subgroup was not at increased risk of progression (see Table 2). Seventeen (40%) of the 42 participants who had progression of valvular lesions were diagnosed with Definite RHD [6]. The majority of these (13/17, 76%) had isolated MV disease (Definite RHD A). Two (12%) had MV and AV disease (Definite RHD D) and two (12%) had isolated AV disease (Definite RHD C). Of the 17 individuals who progressed to Definite RHD, 9 (53%) had been initially classified as Borderline RHD Cases, 6 (35%) as NSVA Cases, and 2 (12%) as Controls. Of the Borderline RHD subgroup, 2/13 (15%, 95% CI 4–42) of Subcategory A, 5/21 (24%, 95% CI 11–45) of Subcategory B, and 2/21 (10%, 95% CI 3–29) of Subcategory C progressed to Definite RHD. Cases were at significantly greater risk of progression to Definite RHD than Controls as was the NSVA subgroup (see Table 3). The Borderline RHD subgroup had a significantly increased absolute risk of progression to Definite RHD compared to their Matched Controls (p b 0.001, Fisher's exact test) but a relative risk could not be determined as none of their Matched Controls progressed to Definite RHD. Logistic regression modelling revealed three common factors that were independently associated with progression of valve lesions and development of Definite RHD (see Table 4). Both functional and morphological changes of the mitral valve were independently associated with progression and Definite RHD (data not shown) and were therefore combined. In addition, Borderline RHD of the aortic valve was found to confer a significant increase in the chance of development of Definite RHD but this was less than that seen with Borderline RHD of the mitral valve. The factors that were independently associated with progression of valve lesions and development of Definite RHD in logistic regression modelling are presented in Table 4. We stratified Cases into two subgroups based on the presence or absence of MR and compared these two groups in relation to progression outcome data. Cases with MR were not at a significantly increased risk of any progression of valvular lesions (RR 1.2, 95% CI, 0.6–2.5) or progression to definite RHD (RR 1.5, 95% CI, 0.5–4.1) compared to those Cases without MR.
3.1. ARF outcome There were 9 reported episodes of ARF. All met criteria for Definite ARF [13]. The IR of ARF was higher in Cases (4/117, 9.2 episodes/1000 person-years, 95% CI 2.5–23.5) than Controls (5/325, 4.2/1000 personyears, 95% CI 1.4–9.7), although this difference was not statistically significant (IRR 2.2, 95% CI 0.6–7.9). After adjusting for LAB use (“days-at-risk”), the IR of ARF was unchanged in Controls but increased in Cases to 9.9 episodes/1000 person-years-at-risk (95% CI 2.7–25.4). This corresponded to an IRR of 2.3 (95% CI 0.7–8.4) and again was not statistically significant. All four Cases with an episode of ARF were in the Borderline RHD subgroup. The IR of ARF in the Borderline RHD subgroup was 19.6/ 1000 person-years (95% CI 5.3–50.3) and increased to 22.9/1000 person-years-at-risk (95% CI 6.3–58.7) when controlling for LAB use. In comparison with their Matched Controls, the participants in the Borderline RHD subgroup were at a statistically significant increased risk of ARF (unadjusted IRR = 7.5, 95% CI 1.2–48.3; IRR adjusted for use of LAB = 8.8, 95% CI 1.4–53.8). Table 1 Demographic data, use of secondary prophylaxis, and time to follow-up for participants.
Total number Age (median years, IQR) Gender (% female, 95% CI) Ethnicity (%, 95% CI) Aboriginal Torres Strait Islander Aboriginal and Torres Strait Islander Prescribed secondary prophylaxis (%, 95% CI) Time between gECHO echocardiogram and RhFFUS enrolment (median days, IQR)
Cases
Controls
p value
117 13.7 (11.9–15.7)
325 13.7 (11.8–15.5)
n/a 0.733
59.0 (49.5–67.0)
57.2 (51.6–62.7)
0.787 0.965
83.8 (75.8–90.0) 6.8 (3.0–13.0) 9.4 (4.8–16.2) 18.8 (12.2–27.1)
84.3 (79.9–88.1) 7.1 (4.5–10.4) 8.6 (5.8–12.2) 2.5 (1.1–4.8)
b0.001
1346 (1171–1478)
1310 (1170–1440)
0.455
120
M. Rémond et al. / International Journal of Cardiology 198 (2015) 117–122
Table 2 Progression of valvular lesions on follow-up echocardiography. All
All Cases
All Controls
Borderline RHD Cases
Matched Controls
NSVA Cases
Matched Controls
Number
442
117
325
55
104
62
122
Progression of valve abnormalities (n, %, 95% CI)
42 9.5% (7.1–12.6%) n/a
23 19.7% (13.5–27.8%)
19 5.9% (3.8–9.0%)
13 23.6% (14.4–36.4%)
3 2.9% (1.0–8.1%)
10 16.1% (9.0–27.2%)
13 10.7% (6.3–17.4%)
Absolute risk difference (%, 95% CI) Relative risk (RR, 95% CI)
n/a
13.8% (6.2–21.5) 3.36 (1.90–5.94)
20.8% (9.1–32.4) 8.19 (2.43–27.53)
5.5% (–5.2–16.1) 1.51 (0.70–3.25)
from Uganda reported that after 2 years, 10% of children with an initial diagnosis of Borderline RHD by the WHF criteria had progressed to Definite RHD [18]. Given the shorter period of follow-up in that study this is again similar to the findings in the current study. This study addresses the issue of the utility of existing WHF criteria [6] for the echocardiographic diagnosis of RHD. Our results support the criteria used by the WHF for Borderline RHD by demonstrating that individuals who satisfy these criteria are at increased risk of ARF, progression of valvular damage, and progression to Definite RHD. However, we also demonstrated that children with less severe NSVAs may also be at a lesser but still increased risk of progression to RHD. These findings highlight that both functional and morphological changes on echocardiography form a continuum. While the current Borderline RHD category reflects an increased risk of ARF and echocardiographic progression, NSVA should also be viewed as part of this continuum. A larger cohort with longer-term follow-up will be necessary to determine exactly where, on this continuum, the cutoff between ‘normal’ and ‘abnormal’ may lie. The increased risk of ARF and progression to Definite RHD in children with Borderline RHD suggests that there may be a role for secondary antibiotic prophylaxis in such children. However, caution should be exercised. First, it is important to more accurately identify the subset of individuals at risk of ARF and progression. There were too few episodes of ARF recorded in this study to provide sufficient power to undertake a subgroup analysis of the risk of ARF in those with Borderline RHD subcategory A, B or C. However, the logistic regression models developed did reveal that children with Borderline disease of the MV (subcategories A and B) were at greatest risk of deterioration of valve lesions and progression to Definite RHD. Nonetheless, children with Borderline disease of the AV (subcategory C) still had an increased chance of development of Definite RHD. Furthermore, utilizing MR alone as a predictor of progression or development of Definite RHD was not useful even though previously it has been suggested that MR alone may have potential utility as a screening tool for RHD in high-risk populations [19]. These results suggest that functional and/or morphological features of the MV are important in detecting those at greatest chance of progression but that AV changes cannot be ignored. The second issue regarding ongoing management of Borderline RHD and NSVAs relates to the efficacy of secondary preventive initiatives in the context of this study. It is still not known whether individuals with Borderline RHD or NSVA, or indeed Definite RHD demonstrated on echocardiography without preceding ARF [6], will be equivalently
3.3. Inter-observer reliability Data outlining the different classifications assigned to each participant based on the original gECHO report of participants' baseline echocardiogram and subsequent re-report of that same echocardiogram as part of the RhFFUS study are presented in Table 5. The inter-rater reliability for reading of the baseline echocardiogram using linear weighted Kappa was 0.656 (p b 0.001, 95% CI 0.592–0.721) and consistent with substantial agreement [14]. To control for potential bias due to discordant reporting of baseline echocardiograms, Cases or Controls were reclassified based on the rereading of baseline echocardiograms during RhFFUS. Following this reclassification, Cases remained at a higher, but non-significant, risk of ARF compared with Controls (IRR 3.9, 95% CI 0.8–25.3). Furthermore, Cases remained at significantly greater risk of any progression of valvular lesions (RR 3.17, 95% CI 1.69–5.94) and progression to Definite RHD (RR 24.9, 95% CI 3.2 to 190.7).
4. Discussion We have shown, for the first time, that children diagnosed with Borderline RHD by the WHF criteria [6] are at increased risk of ARF, progression of cardiac valvular lesions, and development of Definite RHD. It is therefore clear that at least some children with Borderline RHD have true RHD. We have also shown that children with Borderline RHD of the MV, rather than the AV, are at particularly high risk of progression of valve disease. While ARF is rarely diagnosed in Australia, some disadvantaged groups, specifically Aboriginal and Torres Strait Islander populations, continue to experience high incidence rates [15]. Our reported incidence of ARF in Controls (4.2/1000 person-years) was comparable to existing data in Aboriginal Australians in this setting (2.5–3.5/1000 person-years) [15,16]. Against this already high background risk of ARF, children with Borderline RHD had a significantly increased risk of ARF (19.6–22.9/1000 person-years). While the rates of progression of valvular damage seen in this study were greater than those reported in some previous studies, this is likely to be in part explained by our longer period of follow-up. An Indian study that followed children with subclinical RHD for a mean period of 1.3 years reported that 4% progressed [17]. This is similar to, but lower than, the current study (16% over 3.6 years). Another study Table 3 Progression to Definite RHD [6]. All
All Cases
All Controls
Borderline RHD Cases
Matched Controls
NSVA Cases
Matched Controls
N
442
117
325
55
104
62
122
Progression to Definite RHD (n, %, 95% CI)
17 3.9% (2.4–6.1) n/a
15 12.8% (7.9–20.1) 12.2% (6.1–18.3) 20.8 (4.8–89.7)
2 0.6% (0.2–2.2)
9 16.4% (8.9–28.3) 16.4% (6.6–26.1) Could not be determined
0 0%
6 9.7% (4.5–19.6) 8.1% (0.3–15.7) 5.9 (1.2–28.4)
2 1.6% (0.5–5.8)
Absolute risk difference (%, 99% CI) Relative risk (RR, 95% CI)
n/a
M. Rémond et al. / International Journal of Cardiology 198 (2015) 117–122 Table 4 Factors that were independently associated with progression of valvular lesions and progression to Definite RHD in logistic regression models.
Borderline RHD of the MV — categories A and B [6] (OR, 95% CI) Borderline RHD of the AV — category C [6] (OR, 95% CI) NSVA (OR, 95% CI) In receipt of secondary antibiotic prophylaxis (OR, 95% CI) Nagelkerke R [2]
Progression of valvular lesions
Progression to Definite RHD
4.6 (1.8–12.1)
30.0 (5.4–167.3)
–
9.2 (1.0–82.6) 16.0 (3.1–81.8) 4.0 (1.1–15.3) 29.8%
3.0 (1.3–6.8) 4.2 (1.5–11.7) 14.7%
responsive to LAB as those with RHD detected in association with ARF or an incidental murmur. Our findings contribute to the ongoing debate regarding the relevance and feasibility of echocardiographic screening for RHD in highrisk populations [20]. Based on results from gECHO [7], the follow-up of Borderline RHD would triple the number of additional patients who may require management. Previously it has been shown that RHD screening activities can have an adverse impact on local health services and patient quality of life [21]. Furthermore, it has been shown that even in the relatively well resourced health care environment of Australia the uptake of LAB is often poor [22]. Despite these issues, if managing Borderline RHD Cases can be supported by local health services then the results of this study may provide an additional rationale for supporting on-going echocardiographic screening in this setting. 4.1. Study limitations One in three Borderline RHD Cases in this study was receiving LAB. Such prescription was at the discretion of clinical staff rather than dictated, or even recommended, by the gECHO study. Presumably clinicians had interpreted the relatively non-specific echocardiographic findings as being commensurate with Definite RHD. It was likely that the use of LAB had a confounding influence on the subsequent risk of ARF. Unsurprisingly, when we controlled for this potential confounder, the risk of ARF in those with Borderline RHD was increased. The influence of LAB use on echocardiographic progression was less clear. Logistic regression modeling suggested that the use of prophylaxis was associated with a greater risk of progression. It seems unlikely that LAB was conferring such an increased risk. Rather, it was more likely that this was an incidental finding resulting from the fact that, based on gECHO screening results, clinicians were identifying individuals at greatest risk of progression and prescribing LAB to them. It should also be noted that none of the participants receiving LAB in this study were covered for the entire period of time between gECHO and RhFFUS. All of these individuals commenced LAB well after their gECHO echocardiogram and, once treatment had started, none received 100% of their
121
scheduled doses in the time leading up to their RhFFUS echocardiogram. Hence all these participants were still at risk of GAS infection, recurrent episodes of ARF, and progression of valvular lesions for some part of the study period. Furthermore, there is debate over whether the recommended 4-weekly doses of LAB under Australian guidelines [13] provide protection against GAS infection for 28 days as we assumed in our analysis [23]. Hence, no conclusions can be drawn from our data in regard to the effectiveness of LAB in preventing progression of valvular lesions or development of Definite RHD in individuals with echocardiographically-diagnosed NSVAs or Borderline RHD. This is an important question that warrants further research. A further potential source of bias was inter-observer variability. We demonstrated substantial agreement between the original reading of baseline echocardiograms and subsequent re-reading of these echocardiograms during RhFFUS. Further, when we reclassified the status of participants as Cases or Controls based on the RhFFUS reading of baseline echocardiograms, this did not alter findings regarding the risks of ARF, progression of valvular lesions, or progression to Definite RHD. While results revealed that Borderline RHD Cases were at greater risk of ARF than their Matched Controls (IRR = 8.8, 95% CI 1.4–53.8), it should be noted that the number of events (episodes of ARF) in this context was small – 9 events in total, 4 in the Borderline RHD Case group – and that the confidence interval for the incidence rate ratio was broad. Hence caution needs to be exercised in interpreting this result and any attempts to generalize these findings to other contexts. Nonetheless, it should also be noted that this analysis was undertaken to test an a priori hypothesis regarding the risk of ARF in individuals with non-specific heart valve abnormalities detected on RHD screening echocardiogram and that, despite the small number of events, this finding was statistically significant. A further limitation of this study relates to potential selection bias. Participants were selected from a group of children who had been enrolled in a previous school-based RHD screening study. Hence, children who did not usually attend school at the time of this previous study were excluded. There is no evidence to suggest that in remote Indigenous Australian settings there is a difference in risk of having RHD between those that attend school and those that do not. However, this limitation needs to be considered whenever interpreting these results in the context of the natural history of Borderline RHD.
4.2. Conclusions Results from this study support the contention that in some highrisk children Borderline RHD and other NSVAs represent the earliest changes of RHD or, at least, indicate a group at increased risk of ARF, progression of valvular lesions and development of Definite RHD. These findings provide a rationale for considering enhanced clinical followup of high-risk children with Borderline RHD and, to a lesser extent, other NSVAs. Whether this entails ARF education, ongoing regular echocardiographic surveillance, or prescription of LAB remains a clinical judgment.
Table 5 Inter-observer variability in reading of gECHO baseline echocardiograms. Classification based on original reading of baseline echocardiogram
Classification based upon subsequent reread of baseline echocardiogram
a
Normal
NSVA
Borderline RHD
Total
18
2
299
NSVA
279 (85.8%)a 41
7
81
Borderline RHD
4
33 (53.2%) 8
51
Definite RHD Total
1 325
3 62
39 (70.9%) 7 55
Normal
Percentages in brackets show the concordance between readers within each subcategory of participant based on original reading of baseline echocardiograms.
11 442
122
M. Rémond et al. / International Journal of Cardiology 198 (2015) 117–122
Conflicts of interest None.
[10] [11]
Acknowledgment [12]
This study was supported by an Australian National Health and Medical Research Council project grant (Grant Application 1005951) and the NHMRC Centre for Research Excellence to Reduce Inequality in Heart Disease (Grant Application 1044897).
[13]
[14]
References [15] [1] E. Marijon, P. Ou, D.S. Celermajer, et al., Prevalence of rheumatic heart disease detected by echocardiographic screening, N. Engl. J. Med. 357 (2007) 470–476. [2] J.R. Carapetis, M. Hardy, T. Fakakovikaetau, et al., Evaluation of a screening protocol using auscultation and portable echocardiography to detect asymptomatic rheumatic heart disease in Tongan schoolchildren, Nat. Clin. Pract. Cardiovasc. Med. 5 (2008) 411–417. [3] K.V. Roberts, A.D. Brown, G.P. Maguire, D.N. Atkinson, J.R. Carapetis, Utility of auscultatory screening for detecting rheumatic heart disease in high-risk children in Australia's Northern Territory, Med. J. Aust. 199 (2013) 196–199. [4] B.M. Reeves, J. Kado, M. Brook, High prevalence of rheumatic heart disease in Fiji detected by echocardiography screening, J. Paediatr. Child Health 47 (2011) 473–478. [5] R.H. Webb, N.J. Wilson, D.R. Lennon, et al., Optimising echocardiographic screening for rheumatic heart disease in New Zealand: not all valve disease is rheumatic, Cardiol. Young 21 (2011) 436–443. [6] B. Remenyi, N. Wilson, A. Steer, et al., World Heart Federation criteria for echocardiographic diagnosis of rheumatic heart disease — an evidence-based guideline, Nat. Rev. Cardiol. 9 (2012) 297–309. [7] K. Roberts, G. Maguire, A. Brown, et al., Echocardiographic screening for rheumatic heart disease in high and low risk Australian children, Circulation 129 (2014) 1953–1961. [8] M.G. Remond, G.R. Wheaton, W.F. Walsh, D.L. Prior, G.P. Maguire, Acute rheumatic fever and rheumatic heart disease — priorities in prevention, diagnosis and management. A report of the CSANZ Indigenous Cardiovascular Health Conference, Alice Springs 2011, Heart Lung Circ. 21 (2012) 632–638. [9] M.G. Remond, D. Atkinson, A. White, et al., Rheumatic fever follow-up study (RhFFUS) protocol: a cohort study investigating the significance of minor
[16]
[17]
[18] [19]
[20]
[21]
[22]
[23]
echocardiographic abnormalities in Aboriginal Australian and Torres Strait Islander children, BMC Cardiovasc. Disord. 12 (2012) 111. W.F. Armstrong, T. Ryan, Feigenbaum's Echocardiography, Lippincott Williams & Wilkins, Philadelphia, PA, 2009. W.A. Zoghbi, M. Enriquez-Sarano, E. Foster, et al., Recommendations for evaluation of the severity of native valvular regurgitation with two-dimensional and Doppler echocardiography, J. Am. Soc. Echocardiogr. 16 (2003) 777–802. H. Baumgartner, J. Hung, J. Bermejo, et al., Echocardiographic assessment of valve stenosis: EAE/ASE recommendations for clinical practice, J. Am. Soc. Echocardiogr. 22 (2009) 1–23. RHDAustralia (ARF/RHD writing group), National Heart Foundation of Australia and the Cardiac Society of Australia and New Zealand, Australian Guideline for Prevention, Diagnosis and Management of Acute Rheumatic Fever and Rheumatic Heart Disease, 2nd edition, 2012. J.R. Landis, G.G. Koch, The measurement of observer agreement for categorical data, Biometrics 33 (1977) 159–174. B. Field, Rheumatic heart disease: all but forgotten except among Aboriginal and Torres Strait Islander peoples, Bulletin No. 16. , AIHW, Canberra, 2004. J.R. Carapetis, D.R. Wolff, B.J. Currie, Acute rheumatic fever and rheumatic heart disease in the top end of Australia's Northern Territory, Med. J. Aust. 164 (1996) 146–149. A. Saxena, S. Ramakrishnan, A. Roy, et al., Prevalence and outcome of subclinical rheumatic heart disease in India: the RHEUMATIC (Rheumatic Heart Echo Utilisation and Monitoring Actuarial Trends in Indian Children) study, Heart 97 (2011) 2018–2022. A. Beaton, E. Okello, T. Aliku, et al., Latent rheumatic heart disease: outcomes 2 years after echocardiographic detection, Pediatr. Cardiol. 35 (2014) 1259–1267. M. Mirabel, R. Bacquelin, M. Tafflet, et al., Screening for rheumatic heart disease: evaluation of a focused cardiac ultrasound approach, Circ. Cardiovasc. Imaging 8 (1) (2015), http://dx.doi.org/10.1161/CIRCIMAGING.114.002324 (pii: e002324). M.G. Remond, E.K. Wark, G.P. Maguire, Screening for rheumatic heart disease in Aboriginal and Torres Strait Islander children, J. Paediatr. Child Health 49 (2013) 526–531. E.K. Wark, Y.C. Hodder, C.E. Woods, G.P. Maguire, Patient and health-care impact of a pilot rheumatic heart disease screening program, J. Paediatr. Child Health 49 (2013) 297–302. M.G. Remond, K.L. Severin, Y. Hodder, et al., Variability in disease burden and management of rheumatic fever and rheumatic heart disease in two regions of tropical Australia, Intern. Med. J. 43 (2013) 386–393. M.P. Broderick, C.J. Hansen, D.J. Faix, Factors associated with loss of penicillin G concentrations in serum after intramuscular benzathine penicillin G injection: a metaanalysis, Pediatr. Infect. Dis. J. 31 (2012) 722–725.