Methods

A retrospective cohort study was performed, using the Pediatric Health Information System (PHIS) database, to determine if the presence of arrhythmias was associated with in-hospital mortality, length of stay, and costs per day for children admitted with acute myocarditis. This study was classified by the Columbia University Medical Center Institutional Review Board as nonhuman subjects research and was exempted from further review.

Data for this study were obtained from PHIS, an administrative database that contains in-patient, emergency department, ambulatory surgery, and observation data from 44 not-for-profit, tertiary care pediatric hospitals in the United States. These hospitals are affiliated with the Children’s Hospital Association (Overland Park, Kansas). Data quality and reliability are ensured through a joint effort between the Children’s Hospital Association and participating hospitals. The data warehouse for the PHIS database is managed by Truven Health Analytics (Ann Arbor, Michigan). For the purposes of external benchmarking, participating hospitals provide discharge and encounter data, including demographics, diagnoses, and procedures. All data are deidentified at the time of data submission, and data are subject to a number of reliability and validity checks before inclusion in the database.

The database was queried for all children ≤18 years of age, discharged from January 1, 2004 to March 31, 2013, with primary or secondary diagnoses of myocarditis. Subjects were considered to have had myocarditis on the basis of the International Classification of Disease, Ninth Revision, Clinical Modification codes for Coxsackie myocarditis (74.23), acute myocarditis in disease classified elsewhere (422.0), acute myocarditis unspecified (422.90), idiopathic myocarditis (422.91), other acute myocarditis (422.99), or myocarditis not elsewhere specified (429.0). Subjects with myocarditis secondary to sepsis, toxins, and acute rheumatic fever were not included, as treatment and disease progression differ for these groups. Subjects with rheumatologic disorders (codes 710 to 710.9, 695.4, 373.34, 390 to 393, 394.1, 395.0 to 395.2, 395.9, 397.9, 398.0, 398.90, 398.91, and 398.99) and those who had undergone heart transplantation (codes 375.1 and V42.1), were excluded, as one is not able to determine the temporal relation between transplantation and the onset of myocarditis or arrhythmia in this database. Subjects for whom medication billing was not reported were also excluded.

The primary outcomes were (1) mortality, defined as in-hospital death; (2) total length of hospital stay; and (3) costs per day. Costs are estimated in the PHIS database using billing records and hospital- and department-specific cost-to-charge ratios, as submitted to the Centers for Medicare and Medicaid Services. Costs are adjusted for costs of living by hospital location, using the Centers for Medicare and Medicaid Services Wage Index ( http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/Wage-Index-Files.html ). All costs were further adjusted for inflation (to 2013 dollars), using the Medical Consumer Price Index ( http://www.bls.gov/cpi/#tables ).

The primary predictors of interest were the presence of clinically significant tachyarrhythmias and bradyarrhythmias. Subjects were assumed to have clinically significant tachyarrhythmias if they had codes for supraventricular tachycardia (427.0), atrial fibrillation (427.31), atrial flutter (427.32), or ventricular tachycardia or fibrillation (427.1 and 427.41 to 427.42) and received antiarrhythmic medications or underwent electrical cardioversion. Antiarrhythmic medications considered included adenosine and drugs based on the Vaughan-Williams classification: class I (proprafenone, procainamide, and mexiletine), class II (nadolol, propranolol, atenolol, esmolol, and metoprolol), class III (amiodarone and sotalol), and class IV (diltiazem and verapamil). Digoxin and carvedilol were not included because their use could have been associated with heart failure management instead of arrhythmia treatment. Lidocaine was not included because the route of administration is not reliably coded in the PHIS database, and it could have been used as a local anesthetic medication. The use of digoxin, carvedilol, and lidocaine was quantified for descriptive purposes only. Subjects were considered to have clinically significant bradyarrhythmias if they had codes for second-degree (426.12 to 426.13), complete (426.0 or 746.86), or other (426.6) heart block and received pacemakers (378.0 to 378.3).

Other variables considered included gender, age at admission, geographic region (Midwest, Northeast, South, and West), year and month of admission, presence of congenital heart disease, coding for an identified virus, and the use of intravenous immunoglobulins (IVIGs), steroids, inotropes, and/or nonsteroidal anti-inflammatory drugs (NSAIDS; ibuprofen was excluded from analyses so as not to include medications potentially used as antipyretics). Congenital heart disease was defined as coding for any congenital cardiac defect, excluding patent foramen ovale, atrial septal defect, patent ductus arteriosus, and mitral or tricuspid regurgitation (codes 416.0, 416.8 to 416.9, 417 to 417.9, 424 to 424.3, 745.0 to 745.4, 745.6, and 746.0 to 746.9). Month of admission was included as a series of 12 categorical variables. Locally weighted scatterplot smoothing (LOWESS) was used to determine the best models for the relations between age at admission and year of admission and the outcomes of interest. It was determined these relations were best modeled as linear. There were no missing data. The need for mechanical ventilation, extracorporeal membrane oxygenation, or ventricular assist devices was quantified for descriptive purposes only; they were not included either as outcomes or as predictor variables, as the temporal relations between these variables and arrhythmia onset could not be determined.

Clinical and demographic variables were described with standard summary statistics. To assess the marginal associations between predictor variables and mortality, chi-square tests were used for categorical variables, and Student’s t tests or Wilcoxon’s rank-sum tests were used for continuous variables. To assess the marginal associations between the predictor variables and length of stay and adjusted inpatient costs per day, Wilcoxon’s rank-sum or Kruskal-Wallis tests were used for categorical variables, and Spearman’s correlations were used for continuous variables. Variables with p values ≤0.10 in univariable analyses were evaluated together in multivariable analyses. To assess the associations between predictor variables and mortality, we used generalized estimating equations with an exchangeable working correlation structure. To assess the associations between predictor variables and the logarithms of length of stay and costs per day, we used linear regression. Length of stay and costs per day were log transformed, as they were right skewed. Standard errors were clustered to account for possible correlation between children treated within each institution. Because some children with myocarditis die before discharge, and because hospital days and costs do not accumulate post mortem, in the assessment of length of stay and costs, we used censored regression models that censored the log of length of stay and the log of costs for patients who died. A p value ≤0.05 was considered statistically significant. Final multivariable models were determined by using a forward stepwise procedure, in which clinically significant tachyarrhythmias and bradyarrhythmia were forced into the models a priori . Other variables were included in the final models if their p values met the significance criterion or if their inclusion changed the magnitude of the coefficient for clinically significant tachyarrhythmias or bradyarrhythmia by ≥10%. All statistical analyses were conducted in Stata version 13 (StataCorp LP, College Station, Texas).

Results

A total of 2,041 children from 44 hospitals were identified with acute myocarditis during the study period. The distribution of age was bimodal, with 1/4 of patients admitted at <1 year of age (25.6% [n = 522]) and slightly more than 1/4 admitted at >15 years of age (27.8% [n = 568]) (see Figure 1 ). Congenital heart disease was reported in 132 subjects (6.4%). Viral infection was indicated in 284 subjects (13.9%). The most frequently reported viruses were influenza, rhinovirus, and parvovirus, with the frequency of influenza reporting peaking between November and March. For more detailed patient characteristics, see Table 1 .

Age at admission and associated probability of mortality. The histogram depicts the age of patients at admission and superimposes a plot of the probability of mortality as a function of age, using locally weighted scatterplot smoothing (LOWESS).

Clinically significant tachyarrhythmias were found in 234 patients (11.5%). Most tachyarrhythmias were ventricular (79.5% [n = 186]). Forty-four patients (2.2% of the total population) reported atrial fibrillation or atrial flutter; 15 (0.7%) reported another supraventricular tachycardia. Eight patients (0.4%) reported ventricular arrhythmias and atrial fibrillation or flutter; 3 patients (0.1%) reported ventricular arrhythmias and supraventricular tachycardia. Female patients were more likely than male patients to have tachyarrhythmias (13.8% vs 9.9%, p = 0.007), as were those receiving IVIGs (13.1% vs 8.7%, p = 0.002), steroids (15.8% vs 7.8%, p <0.001), or inotropes (14.6% vs 5.0%, p <0.001). NSAID use was associated with a lower incidence of tachyarrhythmia (9.1% vs 14.5%, p <0.001). Other baseline characteristics, including age, did not differ between those with and those without tachyarrhythmias. Of those with tachyarrhythmias, 86 (36.8%) underwent cardioversion, and 196 (83.8%) received ≥1 antiarrhythmic medication (47.4% [n = 111] received ≥2). Clinically significant bradyarrhythmias were reported less frequently (1.1% [n = 22]), with most of these reporting complete heart block (81.8% [n = 18]). Five patients (0.2%) reported both significant tachyarrhythmias and bradyarrhythmias. Table 2 describes the relative frequencies with which treatment modalities were used in survivors and nonsurvivors.

The overall mortality rate was 8.7% (n = 178). Tachyarrhythmias were significantly associated with mortality in univariable and multivariable analyses. In univariable analyses, the presence of a tachyarrhythmia was associated with a 3.1 times increase in the odds of mortality (95% confidence interval [CI] 2.2 to 4.5, p <0.001). In multivariable analyses, the presence of a tachyarrhythmia was associated with a 2.3 times increase in the odds of mortality (95% CI 1.6 to 3.3, p <0.001), after controlling for the use of NSAIDs, steroids, and inotropes and clustering by institution. No children with bradyarrhythmias died. See Table 3 .

Stay updated, free articles. Join our Telegram channel

Join