a b s t r a c t

Background: Cardiomegaly on chest radiographs (CXR) in pediatric patients leads to multiple tests. We aimed to determine the positive predictive value (PPV) of cardiomegaly on CXR in predicting subsequent heart disease and to assess the utility of obtaining a B-type Natriuretic Peptide level (BNP) and/or electrocardiogram (EKG) in such patients. We hypothesized that an echocardiogram may not be appropriate in all cases of cardiomegaly on CXR, particularly in a patient with a normal EKG and BNP level.

Methods: We performed a retrospective cohort study of pediatric patients with cardiomegaly on their initial CXR between January 2015-December 2017. Patients without a subsequent echocardiogram or known congenital heart disease were excluded. A patient was deemed to have heart disease if they had structural abnormalities, functional abnormalities or a pericardial effusion on echocardiogram. The PPV of CXR and the PPV/NPV of the other tests (EKG, BNP) were calculated using contingency tables.

Results: Four hundred and eighty nine patients met inclusion criteria. The PPV of cardiomegaly on CXR alone without any other diagnostic testing in predicting subsequent heart disease was 15%. The PPV increased if there was either an abnormal EKG or a BNP N100 pg/ml and further increased if both of these were present. The PPV values were higher in patients b1 year of age.

Conclusions: Cardiomegaly on CXR can often predict the presence of heart disease, particularly in infants. Further testing with EKG and BNP can better predict who may have heart disease, but it may not eliminate the need for echocardiography.

(C) 2019

1. Introduction

The Appropriate Use Criteria (AUC) for Initial Transthoracic Echocar- diography in Outpatient Pediatric Cardiology has rated an abnormal chest radiograph suggestive of cardiovascular disease as an “Ap- propriate” indication for obtaining an echocardiogram (AUC indication # 51) [1]. However, the AUC guidelines are mostly based on expert con- sensus. The rates of CXR performance vary by pediatric emergency de- partments (ED) and their location. A study performed in a pediatric ED showed that a CXR was ordered on ~ 6% of patients [2]. Cardiomegaly on CXR in pediatric patients often leads to multiple tests including blood work, electrocardiograms (EKG) and echocardiograms. Previous studies have investigated the utility of CXR when evaluating children with po- tential heart disease [3-5], but very few studies have focused on whether current criteria for cardiomegaly by CXR can accurately predict disease [6]. A false positive diagnosis of cardiomegaly on CXR is quite frequent in children and an enlargement of the cardiac silhouette may

* Corresponding author at: 1405 Clifton Road, Atlanta, GA 30322, United States of America.

E-mail address: [email protected] (S. Dasgupta).

not always indicate heart disease. “Pseudo-cardiomegaly” may be re- lated to thoracic wall abnormalities, technical factors, mediastinal ab- normalities or pulmonary conditions [7]. A prior study suggested that a thorough physical examination and an EKG evaluation by a pediatric emergency care physician has an excellent rate of detection of cardiac- related causes, however, this was limited to patients with chest pain [8]. In addition, while an abnormal EKG may indicate heart disease, es- pecially in adults, it may not be specific for cardiac disease in the pediat- ric population. One study demonstrated that only 2% of EKG abnormalities were related to the underlying cardiac abnormality [9]. On the other hand, elevation of B-type natriuretic peptide may be seen with diseases that cause left ventricular volume and right ven- tricular volume or pressure overload, with complex congenital heart diseases having greater elevations of BNP as compared to simple cardiac defects [10]. An echocardiogram can provide qualitative, and both accu- rate and reproducible quantitative, measurements of cardiac chamber size and function [11]. However, an echocardiogram is expensive, more time consuming, and has limited availability compared to a CXR. Our goal was to determine the positive predictive value (PPV) of cardiomegaly on CXR in predicting subsequent heart disease and to as- sess the utility of obtaining a BNP and/or EKG in such patients. We

https://doi.org/10.1016/j.ajem.2019.06.045

0735-6757/(C) 2019

hypothesized that an echocardiogram may not be appropriate in all cases of cardiomegaly on CXR, particularly when a patient also had a normal EKG and a normal BNP level.

1. Patients and methods

We performed a retrospective chart review of all pediatric patients (0-21 years) with cardiomegaly on an initial CXR between January 2015-December 2017 in a tertiary care pediatric hospital, including the ED, the outpatient clinics and the inpatient setting. Patients without a subsequent echocardiogram either at the same encounter as the CXR or within 30 days of that encounter were excluded. Additionally, those with known congenital heart disease were excluded. The institutional review board at Children’s Healthcare of Atlanta approved this study.

Chest radiography

CXR was performed by the radiology department. Patients with a finding of cardiomegaly in the impression of the Radiology report were included. Furthermore, to confirm the finding of cardiomegaly, a study investigator recorded the cardio-thoracic index (CTI) of each film in a blinded fashion (i.e., without knowledge of the echocardiogram re- sults). The CTI was calculated by measuring the widest transverse diam- eter of the heart in the postero-anterior view and dividing by the widest transverse diameter of the thoracic cavity above the diaphragm [12]. A CTI N 0.55 was used as the definition for cardiomegaly [13]. Subjects without cardiomegaly via this confirmatory method were also excluded.

Echocardiography

Echocardiography included 2-D, Doppler, and color-Doppler studies performed using either a Philips IE 33 machine or a Philips EpiQ 7 ma- chine. The echocardiogram findings were classified as normal or abnor- mal based on review of reports of the interpreting cardiologist. The abnormal findings were further divided into structural anomalies (structural defects leading to a left to right shunt, left/right sided ob- struction or significant valvular disease), functional anomalies

(depressed ventricular function including myocarditis/cardiomyopathy, significant left ventricular dilation with preserved function, hypertro- phic cardiomyopathy or restrictive cardiomyopathy) or pericardial effusion.

Additional data

In addition to CXR and echocardiogram data, BNP levels and EKG findings, if obtained, were recorded if they were obtained b30 days prior to the echocardiogram. EKG findings were recorded as normal or abnormal and the abnormal findings were further classified as abnor- mal axes, ventricular enlargement, atrial enlargement, ST segment changes or others based on review of the report of the interpreting car- diologist. A serum BNP level N100 pg/ml was considered abnormal [14]. Other information recorded included gender, weight, race/ethnicity, presenting complaint, setting of the encounter (ED, inpatient or outpa- tient), reason for obtaining CXR, current medications and the cardiovas-

cular physical examination.

Statistical analysis

Normality of continuous variables was assessed using histograms, normal probability plots and the Anderson-Darling test for normality. Descriptive statistics are presented as counts and percentages for cate- gorical variables and median with interquartile range (IQR) for continu- ous data with skewed distribution. Comparisons between heart disease groups for categorical variables were made using Chi-square tests or Fisher’s exact test when expected cell counts were b5 and for continu- ous variables using Wilcoxon rank-sum tests. The negative predictive value (NPV) and positive predictive value (PPV) were calculated for all categories of patients using contingency tables: 1) all patients with cardiomegaly on CXR, 2) cardiomegaly on CXR plus abnormal EKG prior to echocardiogram, 3) cardiomegaly on CXR plus BNP N100 pg/ml prior to echocardiogram, 4) cardiomegaly on CXR plus both abnormal EKG and BNP N100 pg/ml prior to echocardiogram, and

5) cardiomegaly on CXR plus both normal EKG and BNP b 100 pg/ml prior to echocardiogram. A subgroup analysis was replicated excluding patients with an abnormal cardiovascular physical examination.

Table 1

Patient demographics.

Patient characteristic	n	Overall (N = 489)	Heart disease, Yes (N = 153, 31.3%)	Heart disease, No (N = 336, 68.7%)	p-Value
Age, continuous	489	2.1y (54d - 11.2y)	2.0 m (19d - 1.5y)	4.5y (4.5 m - 12.8y)	b0.001
Age, categorical 0d-1 m	489	93 (19.0%)	54 (35.3%)	39 (11.6%)	b0.001
1 m-1y		131 (26.8%)	57 (37.3%)	74 (22.0%)
N 1y Weight, kg	489	265 (54.2%) 11.7 (3.5-32.0)	42 (27.5%) 3.8 (2.4-11.7)	223 (66.4%) 17.3 (5.3-41.2)	b0.001
Sex Male	489	264 (54.0%)	83 (54.2%)	181 (53.9%)	0.938
Female Encounter Setting Emergency Room	489	225 (46.0%) 252 (51.5%)	70 (45.8%) 59 (38.6%)	155 (46.1%) 193 (57.4%)	b0.001
Inpatient		223 (45.6%)	91 (59.5%)	132 (39.3%)
Outpatient Ethnicity Hispanic	489	14 (2.9%) 39 (8.0%)	3 (2.0%) 17 (11.1%)	11 (3.3%) 22 (6.5%)	0.084
Non-Hispanic		450 (92.0%)	136 (88.9%)	314 (93.5%)
Race White	489	343 (70.1%)	111 (72.5%)	232 (69.0%)	0.288
African-American		119 (24.3%)	31 (20.3%)	88 (26.2%)
Asian Cardiovascular Physical Exam Normal	488	26 (5.3%) 317 (65.0%)	11 (7.2%) 63 (41.2%)	16 (4.8%) 254 (75.8%)	b0.001
Abnormal Current medications	489	171 (35.0%)	90 (58.9%)	81 (24.2%)	0.001
None		135 (27.6%)	57 (37.3%)	78 (23.2%)
Non-cardiac		354 (72.4%)	96 (62.7%)	258 (76.8%)

Statistical significance (pb0.05) is indicated by bold value.

Table 2

Classification and sub-classification of heart disease detected by echocardiogram

Type of heart disease	Structural (106)					Functional	(32)		Pericardial effusion (15)
	L-R shunts	Coarctation	Complex congenital	Others		DCM	LV dilation	HCM RCM
	83	9	7	7		20	9	2 1

L-R: Left to right; DCM: dilated cardiomyopathy; LV: Left ventricular; HCM: Hypertrophic cardiomyopathy; RCM: Restrictive cardiomyopathy.

Statistical analysis was performed using SAS version 9.4 (Cary, NC) and statistical significance was assessed at 0.05 level.

1. Results

There were 282,618 patients with a CXR during the study timeframe, with 1544 having cardiomegaly on their initial CXR. Of these, 1055 pa- tients were excluded (909 patients did not get a subsequent echocar- diogram, 109 patients had known congenital heart disease and 37 patients did not meet criteria for cardiomegaly when measured by a blinded observer). Ultimately, 489 patients met inclusion criteria (Table 1). Additional testing included an EKG in 275 patients, a BNP level in 176, and both tests in 122 patients. The median age (25th- 75th percentile) of study patients was 2.1 years (54 days-11.2 years).

Structural heart disease, functional heart disease or a pericardial ef- fusion was detected by echocardiogram in 153 patients (31.3%). There were 106 patients with structural heart disease, 32 with functional heart disease and 15 with a pericardial effusion (Table 2). The median age (25th-75th percentile) of patients with heart disease was signifi- cantly lower as compared to those without heart disease by echocardio- gram: 2.0 months (19 days-1.5 years) vs 4.5 years (4.5 months- 12.8 years), p b 0.001. Patients b1 year of age were more likely to have heart disease detected by echocardiogram (72.5% vs 33.6%; p b 0.001) (Fig. 1). Fifty nine percent (90/153) of patients with a cardiac abnormal- ity detected by echocardiogram had an abnormal cardiovascular examination.

The PPV of cardiomegaly on CXR in predicting subsequent heart dis- ease overall was approximately 31%. The PPV increased if there was ei- ther an abnormal EKG (67.1%) or a BNP N100 pg/ml (59.0%) and further increased if both of these were abnormal (76.7%) (Fig. 2). In this cohort of patients with cardiomegaly on CXR, the NPV of having heart disease was 14.5% for those with a normal EKG and 37.5% for those with a BNP b100 pg/ml.

When stratified by age, the PPV of a patient with cardiomegaly on CXR with either an abnormal EKG or a BNP N 100 pg/ml or both were higher in patients b1 year of age as compared to those N1 year of age. A similar trend in data was seen when patients with an abnormal car- diovascular physical examination were excluded (Fig. 3). Further strat- ified results are illustrated in supplementary figures (Figs. S1 and S2).

1. Discussion

We hypothesized that an echocardiogram may not be appropriate in patients with cardiomegaly on CXR and a normal EKG and a normal BNP level. Our study demonstrates that cardiomegaly on CXR in pediatric pa- tients is an overall rare finding. However, when present it may predict the presence of heart disease, particularly in infants. Our findings sug- gest that it may be appropriate to consider an echocardiogram in this patient population with cardiomegaly on CXR and particularly in those with an abnormal EKG and BNP level in addition to cardiomegaly. However, our data suggest that a normal EKG and a normal BNP level may not completely eliminate the need for an echocardiogram, since 19% of these children still had heart disease.

We were unable to demonstrate any significant variation in the rela- tionship of cardiomegaly and heart disease based on gender, race or eth- nicity. While approximately 59% of patients with heart disease detected by echocardiogram had an abnormal cardiovascular examination, 41% of patients of with heart disease had a normal physical exam. This was

indeed surprising, suggesting that a normal physical examination may not rule out heart disease in the pediatric population in the presence of cardiomegaly.

Satou et al. prospectively evaluated 95 patients in order to deter- mine the usefulness of cardiomegaly on CXR in being able to predict ac- tual cardiac size by echocardiogram. Their results suggested that the assessment of cardiomegaly on CXR had a relatively high specificity and negative predictive value, but a low sensitivity and positive predic- tive value in predicting cardiac enlargement by echocardiogram [6]. Their study also demonstrated that all patients with structural/func- tional heart disease by echocardiogram had either a concerning history, an abnormal physical examination, or an abnormal EKG. Compared to the findings of Satou et al., our study demonstrated that cardiomegaly on CXR had a lower PPV overall in being able to predict subsequent heart disease and that a significant number of patients with a normal physical examination had heart disease. However, Satou et al. did not stratify their findings by age and they did not evaluate the PPV or the NPV of an abnormal EKG or a biomarker such as BNP in being able to predict heart disease when coupled with cardiomegaly.

BNP is a biomarker which is synthesized and secreted mainly by the ventricular myocardium [15]. Under conditions of sustained ventricular expansion and pressure overload, proBNP is released into the blood, where it is cleaved into BNP, the physiologically active hormone [16]. It is often used as a marker of functional heart disease in the pediatric population. Maher et al. demonstrated that an elevated BNP level (N100 pg/ml) was able to identify heart disease with 100% sensitivity and 98% specificity with no overlap with patients without heart disease [17]. Our study demonstrated that infants with an abnormal BNP level were more likely to have heart disease but the ability of an abnormal BNP level to predict heart disease was lower in children above 1 year of age. Importantly, heart disease was found by echocardiography in ap- proximately 37% of patients with a normal BNP level.

Even though our results suggested that a normal EKG was able to ac- curately predict absence of heart disease in approximately 85% patients overall (90% of patients N1 year age), EKG findings are highly variable, dependent on the technique used to obtain it and, on the physician interpreting it. While certain EKG findings are more concerning for heart disease, other non-specific findings may not necessarily indicate heart disease. Gorla et al. demonstrated that approximately 36% of neo- nates without heart disease have non-specific ST-T wave changes on an

Fig. 1. Presence of heart disease stratified by age.

Fig. 2. Positive predictive value (PPV) of cardiomegaly and positive/negative predictive value of cardiomegaly on chest radiographs when combined with EKG and BNP results in predicting subsequent heart disease stratified by age.

EKG [18]. In another study, we demonstrated that certain EKG findings such as isolated left atrial enlargement, premature atrial complexes and non-specific ST-T wave changes are not associated with subsequent heart disease on echocardiogram [19]. The lack of guidelines and litera- ture on the likelihood of an abnormal echocardiogram which is strati- fied by the type of EKG abnormality, makes it difficult to predict the possibility of heart disease purely based on an abnormal EKG. As one would expect, our study demonstrated that an abnormal EKG coupled with an abnormal BNP level was highly predictive of heart disease spe- cially in infants. The higher predictive value in infants may be explained by the higher likelihood of undiagnosed congenital heart disease in this population as compared to those above 1 year of age. However, surpris- ingly, a normal EKG and a normal BNP level would miss heart disease in 19% of patients with cardiomegaly on CXR if an echocardiogram was not

performed. This suggests that while an echocardiogram should be strongly considered in patient with abnormalities in both, normal find- ings may not effectively rule out the need for an echocardiogram.

We found similar results in our sub analysis of patients with a nor- mal physical examination. This sub analysis was performed because we believed that an abnormal cardiovascular physical examination coupled with cardiomegaly on CXR should be an “automatic” indication for an echocardiogram and would not lead to the same diagnostic di- lemma as compared to a patient with cardiomegaly and a normal exam- ination. Our data demonstrated that an abnormal physical examination did little to add to the predictive value of cardiomegaly on CXR.

Our study had some inherent limitations. It was a retrospective study, so we were unable to evaluate the factors which led to a CXR and an echocardiogram being ordered. Another limitation of this study

Fig. 3. Positive predictive value (PPV) of cardiomegaly and positive/negative predictive value of cardiomegaly on chest radiographs when combined with EKG and BNP results in predicting subsequent heart disease stratified by age after excluding patients with an abnormal cardiovascular examination.

was that we included only those patients who had both a CXR and echo- cardiogram performed as a part of their clinical care. We were thus un- able to evaluate the performance of the CXR to detect cardiomegaly in patients in whom clinicians did not feel an echocardiogram was war- ranted. Our study does have an inherent bias when ordering an EKG or BNP on patients since they were not ordered in all. We would expect this to falsely raise the PPV of cardiomegaly in predicting heart disease if these tests were more likely to be ordered when the ordering physician suspected heart disease. A final limitation was that we were unable to assess the likelihood of a normal CXR in being able to rule out heart dis- ease, based on our inclusion criteria.

1. Conclusions

Cardiomegaly on CXR is an overall rare finding. However, when present, it often predicts the presence of structural or functional heart disease, particularly in infants. Further testing with EKG and BNP can better predict who may have heart disease. However, a normal EKG and a normal BNP level may not eliminate the need for echocardiogra- phy. This remains true even in patients with a normal cardiovascular physical examination.

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2019.06.045.

Funding source

This research did not receive any specific grant from funding agen- cies in the public, commercial, or not-for-profit sectors.

Declaration of Competing Interest

None.

References

1. Campbell RM, Douglas PS, Eidem BW, Lai WW, Lopez L, Sachdeva R. ACC/AAP/AHA/ ASE/HRS/SCAI/SCCT/SCMR/SOPE 2014 appropriate use criteria for initial transtho- racic echocardiography in outpatient pediatric cardiology: a report of the American College of Cardiology Appropriate Use Criteria Task Force, American Acad- emy of Pediatrics, American Heart Association, American Society of

Echocardiography, Heart Rhythm Society, Society for Cardiovascular Angiography and Interventions, Society of Cardiovascular Computed Tomography, Society for Car- diovascular Magnetic Resonance, and Society of Pediatric Echocardiography. J Am Soc Echoardiogr 2014;27(12):1247-66 Dec.

1. Zitek T, Lunn J, Ma A, Dadon N, Fisher J. Nondiagnostic Pediatric Chest X Rays are Common: Correlate clinically for pneumonia. J Emerg Med Care;1:101.
2. Birkebaek NH, Hansen LK, Elle B, Andersen PE, Friis M, Egeblad M, et al. Chest roent- genogram in the evaluation of heart defects in asymptomatic infants and children with a cardiac murmur: reproducibility and accuracy. Pediatrics 1999;103(2):E15.
3. Birkebaek NH, Hansen LK, Oxhoj H. Diagnostic value of chest radiography and elec- trocardiography in the evaluation of asymptomatic children with a cardiac murmur. Acta Paediatr 1995;84:1379-81.
4. Newburger JW, Rosenthal A, Williams RG, Fellows K, Miettinen OS. Noninvasive tests in the initial evaluation of heart murmurs in children. N Engl J Med 1983; 308:61-4.
5. Satou GM, Lacro RV, Chung T, Gauvreau K, Jenkins KKJ. Heart size on chest X-ray as a predictor of cardiac enlargement by echocardiography in children. Pediatr Cardiol 2001;22:218-22.
6. Hoeffel JC, Mainard L, Galloy MA, Claudon M. radiological study of false cardiomegaly in pediatric patients. Feuillets de Radiologie;41(6):497-508.
7. Drossner DM, Hirsh DA, Sturm JJ, Mahle WT, Goo DJ, Massey R, et al. Cardiac disease in pediatric patients presenting to a pediatric ED with chest pain. Am J Emerg Med 2011;29(6):632-8 Jul.
8. Selbst SM, Ruddy RM, Clark BJ, Henretig FM, Santulli T. Pediatric chest pain: a pro- spective study. Pediatrics 1988;82(3).
9. Fernandes BA, Maher KO, Deshpande SR. Cardiac biomarkers in pediatric heart dis- ease: a state of art review. World J Cardiol 2016;8(12):719-27.
10. Silverman NH. Quantitative methods to enhance morphologic information using M- mode, Doppler, and cross-sectional ultrasound. Pediatric echocardiography. Balti- more: Williams and Wilkins; 1993. p. 35-42.
11. Ovitt TW. The chest roentgenogram. In: Moss AJ, Adams FH, editors. 1995 heart dis- ease in infants, children, and adolescents. Baltimore: Williams and Wilkins; 1995.

p. 182-3.

1. Abdulla R, Luxenberg D. Cardiac interpretation of pediatric chest X-ray. Heart dis- eases in children: a pediatrician’s guide. 17-34.
2. Doust J, Lehman R, Glasziou P. The role of BNP testing in heart failure. Am Fam Phy- sician 2006;74(11):1893-8 Dec 1.
3. Suttner SW, Boldt J. Natriuretic peptide system: physiology and clinical utility. Curr Opin Crit Care 2004;10(5):336-41.
4. Levin ER, Gardner DG, Samson WK. Natriuretic peptides. N Engl J Med 1998;339(5): 321-32.
5. Maher KO, Reed H, Cuadrado A, Simsic J, Mahle WT, DeGuzman M. B-type natriuretic peptide in the emergency diagnosis of critical heart disease in children. Pediatrics 2008;121(6):1484-8.
6. Gorla SR, Hsu DT, Kukarni A. Lack of association of ST-T wave abnormalities to con- genital heart disease in neonates. Congenit Heart Dis 2016;11(5):403-8.
7. Dasgupta S, Anderson S, Kelleman M, Sachdeva R. Application of pediatric appropri- ate use criteria for initial outPatient evaluation of asymptomatic patients with ab- normal electrocardiograms. Congenit Heart Dis 2019;14(2):230-5 [Oct 23 Epub].