Cardiac Manifestation Among Children with Hemolytic Uremic Syndrome

Abstract

Objective:

The primary objectives of the study were to describe the association between cardiac manifestations and in-hospital mortality among children with Hemolytic Uremic Syndrome (HUS).

Study design:

Using the Pediatric Health Information System database, this retrospective, multi-center, cohort study identified the first HUS-related inpatient visit among children <=18 years (years 2004 to 2018). Frequency of selected cardiac manifestations and mortality rates were calculated. Multivariate analysis identified the association of specific cardiac manifestations and the risk of in-hospital mortality.

Results:

Among 3,915 patients in the analysis, 238 (6.1%) had cardiac manifestations. A majority of patients (82.8%; n=197) had 1 cardiac condition, 17.2% (n=41) had 2, or or more cardiac conditions. The most common cardiac conditions was pericardial disease (n=102), followed by congestive heart failure (n=46), and cardiomyopathy/myocarditis (n=34). The percent mortality for patients with 0, 1, 2 or more cardiac conditions was 2.1%, 17.3%, and 19.5%, respectively. Patients with any cardiac condition had an increased odds of mortality (OR 9.74; P = .0001). In additional models, the presence of 2 or more cardiac conditions (OR-9.90; p<0.001), cardiac arrest (OR- 38.25; p<0.001), or ECMO deployment (OR- 11.61; p<0.001) were associated with increased risk of in-hospital mortality.

Conclusion(s):

This study identified differences in in-hospital mortality based on the type of cardiac manifestations, with increased risk observed for patients with multiple cardiac involvement, cardiac arrest, and ECMO deployment.

Hemolytic Uremic Syndrome (HUS) is associated with a constellation of features, including microangiopathic hemolytic anemia, thrombocytopenia, and renal impairment [1]. Besides a reported mortality rate of approximately 3% during the acute phase of the disease [2,3], HUS is associated with significant morbidities related to ongoing renal impairment, cardiovascular and neurological dysfunction, and less commonly recurrent disease [4].

Cardiac involvement reported in children with HUS includes electrophysiological disturbances, pericardial effusions, cardiomyopathy, and heart failure [5–13]. Cardiac involvement has also been mentioned as a rare contributor to mortality [5,6]; however, these reports on cardiac involvement in HUS are primarily limited to isolated case reports and single center experiences with small sample sizes [5–10]. We analyzed the Pediatric Health Information System (PHIS) database, a large multi-center validated pediatric database to investigate cardiac involvement in children with HUS. The large size of the dataset allowed for comparisons between outcomes of various cardiac manifestations. As such, the primary objective of the study was to describe cardiac involvement in HUS and identify independent cardiac risk factors associated with in-hospital mortality among children with HUS.

METHODS

This study was determined and approved by our by the first author’s Institutional Review Board. The data for this study comes from PHIS, an administrative database that contains inpatient data from >45 U.S. not-for-profit, tertiary care children’s hospitals in North America [14]. Institutions are affiliated with the Child Health Corporation (Shawnee Mission, KS) and account for 20% of all tertiary care children’s hospitals in the United States. Participating hospitals provide hospital discharge data, which includes diagnoses and procedural information using the International Classification of Diseases, Ninth Revision (ICD-9) or Tenth Revision (ICD-10), as well as procedural information using Correct Procedures Terminology (CPT) codes. Data are de-identified at the time of submission and are subjected to a number of reliability and validity checks before being added into the database.

The study sample included pediatric patients who were discharged from a PHIS participating hospital between January 1, 2004 and December 31, 2018 with a diagnosis code of HUS. Our final study population included the first hospitalization among 3,915 children with HUS. All exclusions are indicated in Table I (available at www.jpeds.com).

Table 1.

Inclusion and exclusion criteria for the final study sample

Sample	Exclusion reason	Number excluded
10875
	Hospitalizations with non-unique MRN values	160
10715
	Hospitalizations that were not the first visit with HUS	6075
4640
	Patients > age 18	56
4584
	Renal disease, including the hypertensive CKD codes	495
4089
	Organ transplant	86
4003
	Selected congenital cardiac disease	88
3915

Data for demographics, clinical characteristics, and various diagnoses and procedures were abstracted. Cardiac condition categories investigated included 1) pericardial disease, 2) endocarditis, 3) cardiomyopathy/myocarditis, 4) congestive heart failure, 5) cardiac arrest, 6) valvar disease, 7) rhythm disorders, and 8) heart block. Cardiac covariates investigated were hypertension, extracorporeal membrane oxygenator (ECMO) deployment and pericardiocentesis. The primary outcome measure was in-hospital mortality, and the secondary outcome measure was overall hospital length of stay (LOS). Corresponding codes for all diagnoses and procedures included in this study are provided in Table 2 (available at www.jpeds.com).

Table 2:

List of Cardiac ICD-9 and ICD-10 Diagnosis and Procedure

Pericardial Disease:	4200, 42090, 42091, 42099, 4230, 4232, 4233, 4238, 4239, I300, I301, I309, I310, I311, I314, I313, I32
Hypertension:	4019, 40291, 40501, 40509, 40519, 40591, 40599, I10, I110, I119, I150, I151, I152, I158, I159, I160, I161, I169
Congestive heart failure:	4280, 4281, 42820, 42821, 42823, 42830, 42840, 42841, 42843, 4289, I501, I5020, I5021, I5031, I5033, I5040, I5041, I5042, I50810, I5082, I509
Endocarditis:	4210, 4219, 42490, I330
Cardiomyopathy/Myocarditis:	42511, 42518, 4254, 4258, 4259, 4253, I421, I424, I425, I427, I428, I429, I43, I420, 4291, 42290, 42291, 42292, 4290, I400, I514
Heart block:	4260, 42610, 42613, 42652, I440, I441, I442, I4510, I4519
Rhythm disorders:	42731, 42732, 42741, 4270, 4271, I472, I4901, I4902, I4892, I471
Valve Disease:	I430, 4240, I350, I351, I352, 4241, I361, 4242, 4243
ECMO	3961, 3965, 5A15223; CPT- 33946, 33947, 33948, 33955, 33956, 33960, 33961
Pericardial drainage:	70, 3712, 3731, CPT 33010, 33011

Descriptive analyses to test for differences between patients with and without a cardiac condition included t-tests for continuous variables and chi-squared tests for categorical variables. To adjust for outliers and to improve data interpretation, hospital LOS (ie, 88 days) was truncated at the 99^th percentile such that patients with values greater than the 99^th percentile (n=38) were given a LOS equal to the 99^th percentile. Multivariate logistic regression analyses were conducted to assess the association of cardiac conditions with in-hospital mortality. Specifically, three sets of logistic regressions were conducted to assess 1) a binary variable indicating any cardiac condition, 2) all cardiac conditions as independent predictors, and 3) a categorical variable indicating the number of cardiac conditions (0, 1, >=2). Sensitivity analyses were conducted to assess the robustness of the models. Covariates selected for inclusion in the multivariable models included demographic characteristics and procedures associated with increased risk of in-hospital mortality. Two sets of survival curves were constructed. These curves included a graph indicating survival among patients categorized by the number of cardiac conditions (0, 1, >=2) and by individual cardiac conditions. Regressions and demographic tests were analyzed using Stata, version 12. Visualizations were created using Tableau Software, version 2019.2. All tests were 2-sided assuming a p value of 0.05.

RESULTS

Among 3,915 analyzed patients, cardiac involvement was observed in 238 (6.1%) patients (Table 3). The median age among patients with cardiac involvement was 5.3 years, with 49.6% of patients being between 1 and 5 years of age.

Table 3:

Demographic characteristics and unadjusted outcomes, by cardiac status

			Overall (n=3915)	Non-cardiac (n=3677)	Cardiac (n=238)	p-valuea
Covariate Variables
Age in yearsb		Mean Median (IQR)	5.3 3.8 (2.0, 7.5)	5.3 3.9 (2.0, 7.5)	5.3 3.7 (1.8, 8.1)	0.927
Age		%				0.247
	<30 days		2.9	2.8	3.4
	30 days to <1 year		5.3	5.1	8.0
	1 to 5 years		52.2	52.4	49.6
	>5 yearsc		39.7	39.7	39.1
Gender		%				0.43
	Male		47.5	47.4	50.0
	Femalec		52.5	52.7	50.0
Race		%
	Whitec		75.5	76.2	65.1	<0.001
	Black		6.2	5.7	13.5
	Asian		2.5	2.3	4.2
	Other/Missing		15.8	15.8	17.2
Year of Discharge						0.10
	2004–2008	%	27.0	27.3	22.3
	2009–2013	%	36.5	36.6	34.8
	2014–2018	%	36.6	36.2	42.9
ECMO		%	0.8	0.3	8.8	<0.001
Pericardiocentesis		%	0.6	0.0	8.8	<0.001
Hypertension		%	36.3	34.9	57.6	<0.001
Shigatoxin		%	18.9	21.9	16.7	0.23
Enteritis		%	44.0	43.9	45.4	0.66
Outcomes
Hospital LOS		Mean(SD)	15.3(0.2)	14.3	30.7	<0.001
In-hospital mortality		%	3.0	2.1	17.7	>0.001

IQR =Interquartile range; SD=Standard deviation; ECMO= Extracorporeal membrane oxygenation; LOS= length of stay; ICU- Intensive care unit

^aP-values from t-tests for continuous variables and chi-square tests for categorical variables. Test for continuous variables evaluated differences in mean values.

^bVariable not included in regressions

^cReference variable in regressions

A majority of patients (82.8%; n=197) with cardiac involvement had 1 cardiac condition, and 17.2% (n=41) had 2 or more cardiac conditions (Table 4 and Table 5). The most common cardiac conditions were pericardial disease in 102 patients, congestive heart failure in 46 patients, and cardiomyopathy/myocarditis in 34 patients. Pericardial disease was present for 102 patients, representing 2.6% of the overall cohort and 42.9% among those in the cardiac group. Isolated pericardial disease was noted in only 34.9% (n=83) among those with pericardial disease. Among investigated cardiac procedures, pericardiocentesis was noted in 20.6% (n=21) of those patients with pericardial disease and ECMO was deployed in 33 patients (results not shown). Cross-tabulation of subjects with cardiac and cerebral manifestations among children with HUS is provided in Table 4. The mean hospital LOS was 14.3 days (median: 11d), 29.9 days (median: 25d), and 34.7 days (median: 27d) for patients with 0, 1, 2 or more cardiac conditions and the percent mortality was 2.6% (n=77/3677), 17.3% (n=34/197), and 19.5% (n=8/41), respectively (Figure 1; available at www.jpeds.com).

Table IV.

Cross-tabulation of patients with cardiac and cerebral manifestations among children with HUS

	Arrhythmia (n = 27)		Pericardial disease (n = 102)		Cardiomyopathy (n = 34)		CHF (n = 46)
Variables	Survivors	Nonsurvivors	Survivors	Nonsurvivors	Survivors	Nonsurvivors	Survivors	Nonsurvivors
Seizure disorder
No	19	6	80	11	28	2	42	3
Yes	2	0	9	2	3	1	0	1
Encephalopathy
No	19	4	64	9	23	3	35	4
Cerebral edema/brain compression
No	21	5	84	12	31	2	41	3
Yes	0	1	5	1	0	1	1	1

Table 5:

Combinations of cardiac manifestations and associated mortality rates^a

Pericardi al disease (n=102)	Endocarditis (n=5)	Cardiomyopathy (n=34)	Congestiv e heart failure (n=46)	Cardiac Arrest (n=53)	Valve disease (n=17)	Rhyth m disorder (n=27)	Heart block (n=7)	# of conditions	Total Patients	# Died	% Died
.	.	.	.	.	.	.	.	0	3677	77	2.1
X	.	.	.	.	.	.	.	1	83	7	8.4
.	.	.	.	X	.	.	.	1	35	21	60.0
.	.	.	X	.	.	.	.	1	29	0	0.0
.	.	.	.	.	.	X	.	1	20	4	20.0
.	.	X	.	.	.	.	.	1	14	0	0.0
.	.	.	.	.	X	.	.	1	7	1	14.3
.	.	X	X	.	.	.	.	2	7	0	0.0
X	.	.	.	X	.	.	.	2	6	2	33.3
.	.	.	.	X	.	X	.	2	5	0	0.0

^aTop 10 columns shown here. Rest of the data shown in (Table 2)

Figure 1.

Mortality and hospital length of stay by cardiac condition count.

In-hospital mortality differed widely between various cardiac conditions (Table 5 and Table 6, right most column [available at www.jpeds.com]). Among cardiac condition combinations with at least 10 patients, the categories with the highest mortality included patients with cardiac arrest (60.0%) followed by patients with only pericardial effusion and cardiac arrest (33.3%). Figure 2 (available at www.jpeds.com) shows the Kaplan –Meier survival estimates for various cardiac conditions, which further highlight the increased risk of mortality among patients with multiple cardiac conditions. Among the 238 patients with cardiac involvement, 35 received eculizumab therapy on the index visit. The mortality for those who received eculizumab therapy was 12.9% (4/35) versus 22.2% (30/165) in those who did not receive eculizumab therapy (p=0.46).

Table 6:

Combinations of cardiac manifestations and associated mortality rates^a (continuation of Table 2)

Pericardial disease (n=102)	Endocarditis (n=5)	Myopathy (n=34)	CHF (n=46)	Cardiac Arrest (n=53)	Valve disease (n=17)	Rhythm disorder (n=27)	Heart block (n=7)	# of conditions	Total Patients	# Died	% Died
.	X	.	.	.	.	.	.	1	5	1	20.0
.	.	.	.	.	.	.	X	1	4	0	0.0
X	.	.	.	.	X	.	.	2	4	0	0.0
X	.	X	X	.	.	.	.	3	3	1	33.3
.	.	.	X	.	.	.	X	2	2	0	0.0
.	.	X	.	.	X	.	.	2	2	0	0.0
.	.	X	.	X	.	.	.	2	2	0	0.0
X	.	.	.	X	.	X	.	3	2	2	100.0
.	.	.	X	X	.	.	.	2	1	1	100.0
.	.	.	X	X	X	.	.	3	1	0	0.0
.	.	X	.	.	.	.	X	2	1	0	0.0
.	.	X	X	.	X	.	.	3	1	1	100.0
X	.	X	.	.	.	.	.	2	1	0	0.0
X	.	X	.	.	X	.	.	3	1	0	0.0
X	.	X	X	.	X	.	.	4	1	0	0.0
X	.	X	X	X	.	.	.	4	1	1	100.0

^aThe table included in the main text is limited to only those rows with >=10 patients. Rows with fewer than 10 patients are available in the online appendix.

Figure 2.

Kaplan-Meier survival estimates depicting survival by, A, number of cardiac conditions and, B, survival by specific cardiac condition. For B, the cardiac condition categories are not mutually exclusive, such that a patient may be in more than 1 category.

Table 7 provides the results of the multivariate logistic regressions that assessed the association of cardiac conditions with in-hospital mortality. Model 1 indicates that the presence of any cardiac condition [odds ratio (OR) 9.74 (95% confidence interval (CI): 5.95, 15.96; p<0.001)] and ECMO deployment (OR-11.46; 95% CI-5.11, 25.68; p<0.001) were significantly associated with in-hospital mortality. Model 2, in which cardiac patients are categorized as having either 1, 2 or more cardiac conditions, indicates that presence of 2 cardiac conditions (OR-9.70; 95% CI: 5.74, 16.41; p<0.001), 3 or more cardiac conditions (OR-4.72; 95% CI 1.36, 16.37; p=0.014), and ECMO deployment (OR-11.43; 95% CI-5.08, 25.76; p<0.001) were significantly associated with in-hospital mortality. In Model 3, which analyzes individual cardiac conditions are separately, indicated that cardiac arrest (OR- 38.25; 95% CI: 19.33, 75.71; p<0.001) and ECMO deployment (OR- 11.61; 95% CI- 4.58, 29.42; p<0.001) were associated with increased risk of in-hospital mortality. In all three models, younger age (age less than 30 days and between 30 days and 1 year) and black race, when compared with their respective reference groups, were also independently associated with increased risk of in-hospital mortality. We conducted a sensitivity analysis which indicated that the findings were robust to excluding patients who transferred or who had missing racial information (Table 8; available at www.jpeds.com).

Table 7:

Multivariate logistic regression to assess the association of cardiac manifestation with in-hospital mortality

	Any Cardiac Condition			Cardiac Condition Categories			Specific Cardiac Conditions
Variable	OR	p	95%CI	OR	p	95%CI	OR	p	95%CI
Male	0.74	0.154	(0.50,1.12)	0.75	0.156	(0.50,1.12)	0.71	0.116	(0.46,1.09)
Female	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
<30 days	13.88	<0.001	(7.56,25.48)	13.89	<0.001	(7.56,25.53)	16.72	<0.001	(8.98,31.15)
30 days to <1 year	2.54	0.013	(1.21,5.33)	2.54	0.013	(1.21,5.33)	3.35	0.002	(1.59,7.07)
1 to 5 years	1.03	0.906	(0.64,1.66)	1.03	0.904	(0.64,1.66)	1.00	0.986	(0.60,1.67)
>5 years	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
White	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Black	1.64	0.107	(0.90,3.00)	1.64	0.107	(0.90,3.00)	1.98	0.029	(1.07,3.67)
Asian	0.30	0.248	(0.04,2.30)	0.30	0.248	(0.04,2.30)	0.34	0.310	(0.04,2.71)
Other/Missing	1.44	0.157	(0.87,2.38)	1.44	0.160	(0.87,2.38)	1.52	0.119	(0.90,2.58)
ECMO	11.46	<0.001	(5.11,25.68)	11.43	<0.001	(5.08,25.76)	11.61	<0.001	(4.58,29.42)
Pericardiocentesis	0.98	0.981	(0.24,4.02)	0.99	0.985	(0.24,4.07)	2.76	0.242	(0.50,15.11)
Hypertension	0.54	0.013	(0.33,0.88)	0.54	0.013	(0.33,0.88)	0.66	0.106	(0.40,1.09)
Any Cardiac Condition	9.74	<0.001	(5.95,15.96)
0 Cardiac Conditions				Ref	Ref	Ref
1 Cardiac Condition				9.70	<0.001	(5.74,16.41)
2+ Cardiac Conditions				9.90	<0.001	(3.90,25.13)
Pericardial disease							2.30	0.098	(0.86,6.16)
Endocarditis							4.76	0.257	(0.32,70.41)
Cardiomyopathy							0.97	0.971	(0.19,4.99)
CHF							0.74	0.716	(0.14,3.77)
Cardiac arrest							38.25	<0.001	(19.33,75.71)
Valve disease							2.48	0.375	(0.33,18.60)
Rhythm disorder							2.81	0.106	(0.80,9.85)

OR=Odds Ratio; CI=Confidence interval; ECMO=Extracorporeal membrane oxygenation; CHF=Congestive heart failure

Table 8:

Sensitivity analysis: logistic regression assessing the association between cardiac manifestation and in-hospital mortality

	Removing Transfers (n=98)			Removing Patients with Missing Race (n=180)
Variable	OR	p	95%CI	OR	p	95%CI
Male	0.75	0.169	(0.50,1.13)	0.77	0.227	(0.51,1.17)
Female	x	x	x	x	x	x
<30 days	14.70	<0.001	(7.93,27.25)	15.74	<0.001	(8.39,29.52)
30 days to <1 year	2.56	0.014	(1.21,5.41)	3.14	0.003	(1.48,6.67)
1 to 5 years	1.04	0.868	(0.65,1.68)	1.17	0.543	(0.71,1.92)
>5 years	x	x	x	x	x	x
White	x	x	x	x	x	x
Black	1.65	0.104	(0.90,3.03)	1.61	0.122	(0.88,2.94)
Asian	0.30	0.243	(0.04,2.28)	0.30	0.240	(0.04,2.25)
Other/Missing	1.44	0.162	(0.86,2.39)	1.50	0.158	(0.85,2.65)
ECMO	14.62	<0.001	(6.15,34.75)	10.95	<0.001	(4.78,25.08)
Pericardiocentesis	1.37	0.658	(0.34,5.62)	1.03	0.964	(0.25,4.22)
Hypertension	0.56	0.018	(0.34,0.90)	0.52	0.011	(0.32,0.86)
Any Cardiac Condition	9.77	<0.001	(5.93,16.09)	9.61	<0.001	(5.76,16.04)

OR=Odds Ratio; CI=Confidence interval; ECMO=Extracorporeal membrane oxygenation

DISCUSSION

This study highlights the differences in in-hospital mortality based on the type of cardiac involvement and the identification of specific cardiac conditions as independent risk factors for in-hospital mortality.

Our results for overall mortality (17.7%) among those with cardiac involvement are significantly higher than the 3% overall mortality reported from the Foodborne Disease Active Surveillance Network [3,5]. However, the pediatric sample in that analysis was small (only 4 deaths reported) and mortality among those with cardiac conditions was not explored. Mortality associated with cardiac manifestations has been higher in other studies, ranging from 5.9% to 37% [5,6, 13]. Rigamoti et al, in a case report and review of the literature, identified 19 cases of childhood hemolytic-uremic syndrome complicated by a direct cardiac involvement.[13] The overall mortality was 37% with a significantly higher case fatality ratio (83%) in those with onset of cardiac compromise within 3 days of diagnosis of HUS.

By using a large dataset that spans 44 pediatric hospitals, our study provides more stable estimates with the ability to evaluate specific cardiac conditions and combinations of the various cardiac conditions. In our study, the mortality among those with cardiac involvement increased from 3.8% to 17.6% when patients with isolated hypertension were excluded as a type of cardiac manifestation suggesting that the type of cardiac involvement impacts the associated risk of mortality. Mody et al listed cardiac complications as the reason for mortality in 7.6% of patients who died from HUS [5]. Similarly, Oakes et al, in a 2006 case-control study from the Intermountain HUS Patient Registry reported 1 death from congestive heart failure among the 17 reported deaths [6]. The mechanism for cardiac involvement is unclear but may be in part related to the development of thrombotic microangiopathy within the myocardial tissue [13].

The literature on pericardial disease and HUS are scant and limited to case reports [9,16], with a majority of studies being case reports describing cardiac tamponade. Our study suggests, either in isolation or concurrently with other cardiac manifestations that pericardial disease is more common among pediatric HUS patients than previously reported and may serve as a contributor to increased risk of mortality.

In our study, 34 patients were classified as having cardiomyopathy/myocarditis with a mortality rate of 8.8%. Myocarditis as a complication of HUS was first described in a case series in 1971 in two patients < 3 years of age; both patients had significant ECG changes and clinical evidence of myocarditis without histological evidence [7]; however, more recent studies have identified myocardial involvement in older pediatric patients [13,17,18]. Although cardiomyopathy, with or without congestive heart failure, is described infrequently in pediatric patients [10,11,19–21], our data would suggest that cardiomyopathy/myocarditis is a more common cardiac condition than what has been reported in previous literature. A recent study of 220 adult patients with thrombotic microangiopathies reported that acute heart failure develops in 9.5% of patients and its presence more than doubled the in-hospital mortality rates (38% vs. 17%, p=0.04) compared with those without acute heart failure [20]. Our study demonstrates a lower prevalence of acute heart failure in the pediatric population, with about 3.0% of the population with any cardiac condition having a CHF diagnosis and 1.2% of the overall study sample with a CHF diagnosis. Additionally, ECMO deployment was noted in 10.5% (N=7/67) of children with either the cardiomyopathy/myocarditis code or the CHF code. Survival among this subgroup who received ECMO was 57.1% (N=4/7) compared with 98.3% (N=59/60) of patients with cardiomyopathy/myocarditis or CHF who did not receive ECMO. These results support the reports from the few published cases that suggest the potential reversibility of cardiomyopathy/myocarditis in pediatric patients with HUS with mechanical cardiac support [10,11]. Additional analysis is needed to fully understand the associations of ECMO support with mortality among populations of pediatric HUS patients with different types of cardiac manifestations.

Cardiac arrest among patients with HUS was an independent cardiac risk factor associated with mortality. Cardiac arrest represents the final pathway following cardiovascular collapse from other cardiovascular etiologies such as myocarditis, cardiac tamponade or rhythm disturbances. Moreover, isolated cases do suggest that life threatening ventricular arrhythmias [8] may occur as a result of thrombotic microangiopathy within the conduction system [22] and thus highlights the need for close cardiac monitoring while in the inpatient setting for these patients.

Hypertension was noted in approximately one-third of pediatric patients with HUS during their acute hospitalization, which aligns with previous reports of 25% of patients developing hypertension during the acute phase of HUS [4]. Similarly, follow-up data from the 2011 E. Coli O104:H4 outbreak in Germany also reports that 19% of the patients had persistent hypertension on a 3-year follow-up after their acute HUS hospitalization [23]. Our study would suggest that hypertension is more common in the acute phase than previously reported.

There are a number of limitations to our study. The study used a multi-institutional pediatric administrative database that was not collected for research purposes. Though the use of the PHIS database offers the advantage of a large sample size, like all administrative analyses it may be limited by classification errors. PHIS subjects all the data released to a number of stringent reliability and validity quality checks to mitigate this limitation. We have attempted to mitigate this issue by using clear diagnostic and procedural codes that are important for reimbursement and likely to be appropriately documented in the data. Another limitation relates to the granularity of the dataset as specific laboratory and diagnostic test values are not available and deductions have to be made based on the available information. Thus we did not subcategorize HUS into diarrheal and no diarrheal (pneumococcal) cases for analysis.

Cardiac conditions associated with HUS are a more common and underreported occurrence than reported in previous literature. They are important contributors to HUS mortality. Given the high mortality associated with specific cardiac involvement, clinicians should evaluate for cardiac involvement in hospitalized pediatric patients with HUS. This study points to the need to further investigate these questions in a prospective study design.

Abbreviations:

HUS

Hemolytic uremic syndrome

PHIS

Pediatric Health Information System

Ninth Revision

International Classification of Diseases

ICD-9

ICD-10- Tenth Revision

CPT

Correct Procedures Terminology

LOS

Length of stay

ECMO

Extracorporeal membrane oxygenation

OR

Odds ratio

CI

Confidence interval