Acute Kidney Injury Among Children Admitted With Viral Rhabdomyolysis

We conducted a single-center retrospective cohort study via chart review of medical records filed between May 1, 2010, and December 31, 2018, of children with rhabdomyolysis who were admitted to a tertiary-care urban pediatric hospital. Patients included in the study were <21 years of age; had a primary or secondary discharge diagnosis of rhabdomyolysis (International Classification of Diseases, 10th Revision [code M62.82] or International Classification of Diseases, Ninth Revision [code 728.88]) or infective myositis (International Classification of Diseases, 10th Revision [code M60.009] or International Classification of Diseases, Ninth Revision [code 728.0]); or had a CK level of 5 times the upper limit of normal (>1000 IU/L). In concordance with previous studies, patients were excluded if they had (1) a documented history of muscular dystrophy or other metabolic muscle disorders, (2) a history of chronic kidney disease as documented in the admission history and physical or progress notes, or (3) cardiac arrest or myocardial damage with a documented CK-MB fraction of >3%.²⁵  We also excluded patients without adequate data including documented height and serum creatinine. In subjects with multiple hospital admissions that fulfilled inclusion criteria, the first admission to the study hospital for rhabdomyolysis was included with subsequent admissions excluded. All charts underwent duplicate review by 2 (H.M.G. and A.H.) of the 5 authors. Discrepancies were resolved through team consensus before data analysis. The institutional review board of the medical center determined this study to be exempt (approval IRB-300003410).

We defined rhabdomyolysis as a serum CK level of >1000 IU/L in the absence of elevated CK-MB >3%. The primary outcome was the development of AKI. AKI was defined as serum creatinine >1.5 times the baseline creatinine in line with KDIGO guidelines.²⁹  Because the majority of patients did not have a serum creatinine within 6 months of hospitalization, the baseline creatinine was back-calculated by using the admission height and bedside Schwartz formula (120 = 0.413 × height in centimeters/serum creatinine) for an estimated glomerular filtration rate of 120 mL/minute per 1.73 m² as previously described.³⁰  Viral etiology was determined via chart review in 2 ways: (1) whether the provider’s history and physical assessment named viral rhabdomyolysis as the cause of the symptoms and/or (2) if there were documented symptoms currently or within 2 weeks before admission consistent with a viral infection such as fever, vomiting, diarrhea, upper respiratory symptoms, or malaise. In addition, if a patient had a positive reverse transcription polymerase chain reaction viral respiratory panel testing result during admission, they were determined to have a viral etiology.

We abstracted historical and laboratory data from the inpatient records. Historical data included age, sex, race, insurance (defined as government, private, or other), weight, height, viral symptoms (presence of fever, cough, sore throat, rhinitis, diarrhea, or vomiting), history of recent virus testing outside of hospitalization, and documented etiology of rhabdomyolysis (viral, seizures, trauma, exertional, oncologic, sepsis, medication or drug related, muscular infections, malnutrition, rheumatologic, or idiopathic). Laboratory data included CK (initial, peak, and discharge); serum urea nitrogen (SUN) (initial and peak); creatinine (admission, peak, and discharge); initial urinalysis results including gross blood, microscopic red blood cells, urine pH, protein, and specific gravity; urine toxicology screens (amphetamines, methamphetamines, opiates, barbiturates, phencyclidine, cocaine, methadone, cannabis, benzodiazepines, and tricyclic antidepressants); serum or urine myoglobin; and viral respiratory panel results.

Bivariate analyses were performed with the Kruskal-Wallis test for continuous variables and χ² test for categorical data. Logistic regression modeling with fixed effects was used to predict the primary outcome of AKI. Model covariates included all demographic variables and variables that satisfied an α < .15. For analysis, an age cutoff of 13 years was used to differentiate between school age and teenage years because infectious causes of rhabdomyolysis are more common in younger children compared with trauma- and exercise-induced rhabdomyolysis being more common in teenagers.²⁵  A CK cutoff of 15 000 IU/L was used because CKs less than this have been cited to have a low association with AKI.⁴  Lastly, an SUN cutoff of 15 mg/dL was chosen on the basis of clinical suspicion. All data were analyzed by using SAS 9.4 (SAS Institute, Inc, Cary, NC). A P value <.05 was considered statistically significant.

We reviewed 531 medical records of patients discharged with the diagnosis of rhabdomyolysis from May 1, 2010, through December 31, 2018. Of these, 212 patient encounters were excluded (because of exclusion criteria, noninitial visits, and missing data). The remaining 319 patients were included in the study, with 78 patients in the viral group and 241 patients in the nonviral group as seen in Fig 1. Forty patients had a virus identified on testing and 38 patients had a clinically diagnosed viral infection.

Table 1 reveals a comparison of basic demographics and laboratory results between the viral group and the nonviral group of patients. The cohort was predominantly male and non-Hispanic Black and had a government payer. Patients in the viral group were younger and had higher initial CK. Median initial CK was 2924 IU/L and median peak CK was 4549 IU/L. Median admission SUN was 12.50 mg/dL.

Non–viral-induced causes of rhabdomyolysis were more common than viral-induced causes (241 of 319 [75.5%] vs 78 of 319 [24.1%]). Of the 241 cases of nonviral rhabdomyolysis, 77 (31.9%) were due to exertion, 73 (30.2%) were due to trauma, 45 (18.6%) were medication and/or drug related, 34 (14.1%) were due to sepsis, 29 (12.0%) were idiopathic, 22 (9.1%) were due to seizures, 10 (4.1%) had a rheumatologic diagnosis, 6 (2.4%) were due to a muscular infection, and 4 (1.6%) had an oncologic diagnosis. Of those with a viral etiology, the viruses identified included influenza A and B (59%), respiratory syncytial virus (7%), parainfluenza virus (5%), adenovirus (5%), Epstein-Barr virus (2%), coronavirus (5%), enterovirus (5%), and rhinovirus (10%). In the cohort, 38 of 78 (49%) patients were identified with a virus on the basis of symptoms alone. Several patients within both the viral and nonviral groups as well as those who had AKI and no AKI had chronic underlying medical conditions, which can be viewed in Supplemental Table 4.

In the total cohort, 45% (141 patients) developed AKI as described by the KDIGO consensus definition. In bivariate analysis of outcomes, there was no significant difference in development of AKI between the exposure groups (nonviral 47% versus viral 39%; P = .19) (see Table 2, which compares the differences between patients with AKI and no AKI).

Children aged ≥13 years of age and male patients were significantly more likely to develop AKI (P < .01, respectively). There were no differences in outcomes for race, insurance, or myoglobinuria (P = .11, .19, and .48, respectively). Patients with proteinuria and uremia on admission had statistically higher rates of AKI (P < .01). The median initial CK was higher in the non-AKI group compared with the AKI group (4035 vs 2404 IU/L; P = .01).

After controlling for etiology, age, sex, race, admission proteinuria, admission SUN, and initial CK, there was no difference in the odds of developing AKI between the nonviral and viral etiologies of rhabdomyolysis (adjusted odds ratio 1.00; 95% confidence interval [CI]0.50–2.00; P = .99). Children >13 years of age, those with proteinuria on admission, and patients with admission SUN >15 mg/dL had higher odds of developing AKI (Table 3).

In this study, our hypothesis that viral causes of rhabdomyolysis had lower rates of AKI was not supported. The incidence of AKI associated with viral-induced rhabdomyolysis was 39%. After controlling for demographic and clinical factors, we observed no difference in the rate of AKI between viral and nonviral etiologies of rhabdomyolysis. Patients with viral-induced rhabdomyolysis who were ≥13 years of age and had proteinuria and elevated SUN on admission were more likely to have an AKI.

To our knowledge, this is the largest pediatric study of rhabdomyolysis using a contemporary definition of AKI. Our results were similar to a smaller study by Lim et al¹³  who evaluated 39 patients. They found that 35.9% of pediatric patients with rhabdomyolysis had AKI using the KDIGO definition. Of the 12 patients with viral-induced rhabdomyolysis, they reported an AKI rate of 42% (5 of 12). Similar to our study, they noted that high SUN, proteinuria, and comorbid conditions were common in patients with rhabdomyolysis.¹³ 

Our study improves our understanding of previous pediatric studies on rhabdomyolysis and AKI. Mannix et al²⁵  is commonly quoted in literature as the largest study of pediatric rhabdomyolysis to date, with 191 subjects, of which 38% had viral rhabdomyolysis and only 1.6% had kidney injury attributable solely to rhabdomyolysis. However, Mannix et al²⁵  was published before KDIGO criteria, and they used acute renal failure with a serum creatinine of greater than the 97.5 percentile for age and sex as their definition of kidney injury. In a systematic literature review of 72 articles from 1973 to 2017, Capoferri et al³¹  previously found that viral rhabdomyolysis was never reported to be complicated by AKI. Rosenberg et al²⁶  similarly showed that of the 54 patients in their study with benign acute childhood myositis, all had normal creatinine. The KDIGO definition used in our cohort identifies more patients with AKI. The previous definitions for kidney injury used very severe persistent serum creatinine elevations, anuria, or the need for dialysis to define acute renal failure. Given the KDIGO definition identifies AKI with a serum creatinine of only 1.5 times the baseline, the significant difference in incidence of kidney injury between our study and these previous reports is possibly due to the lower serum creatinine cutoffs for AKI.

Within our cohort, CK revealed an inverse relationship with the rate of AKI, which is contrary to typical thought. This could be related to the timing of patient presentation, timing of initial CK draw, timing of treatment, and possible treatment before being admitted. However, the degree of CK elevation has been reported to have a weak correlation with risk of kidney injury.^{4,11,12,32,33}  Other literature is contradictory, and some pediatric studies have suggested that there is a direct correlation between CK levels and AKI.^3,14  However, our study is more in line with large adult studies such as the study by McMahon et al³²  of 2371 patients, in which there was no linear association of CK and mortality or need for renal replacement therapy until CK levels rose above 40 000 IU/L. Simpson et al¹⁰  compared the McMahon score to using CK alone and found that CK had a limited prognostic value. Peak CK may increase after renal injury has already occurred, limiting its usefulness in guiding care in the individual patient. Within our study context, we do not recommend use of initial CK elevation as a predictor for AKI.

Patients in this study were categorized into viral or nonviral groups on the basis of the presence of positive viral testing results or clinical diagnosis. However, patients with viral infections could also have additional reasons for rhabdomyolysis and AKI (eg, viral sepsis, capillary leak, renal hypoperfusion, dehydration, secondary bacterial infection or evolving diagnosis, and/or lowered seizure threshold due to the viral infection). To limit study bias and reduce the chances of committing a type 1 error, these patients were kept in the viral group, and the additional rhabdomyolysis etiologies were not mutually exclusive. Secondary analysis of patients with only a viral infection and no additional identifiable rhabdomyolysis etiologies reveals that 2.8% (1 of 36) had AKI.

This is the largest study in pediatric literature thus far with 319 patients with rhabdomyolysis. It is also the first study comparing viral-induced rhabdomyolysis to all other rhabdomyolysis etiologies with AKI as an outcome and using the KDIGO definition, which is the most up-to-date guideline for defining AKI.

Our study has limitations that require noting. This is a single-center study that makes generalizability difficult to the broader population. Our study design is also limited to retrospective chart review and relies on measures to be documented. Therefore, certain variables were not feasible to obtain such as the impact of hydration status, timing of symptoms, and timing of presentation. Further work should incorporate these measures. Next, because 71% of subjects did not have a previous serum creatinine measure in the medical record, their baseline serum creatinine value had to be calculated by using their height and assuming an estimated glomerular filtration rate of 120 mL/minute per 1.73 m². This could have affected our AKI rates. Next, the etiologies of rhabdomyolysis are not mutually exclusive as discussed above. Lastly, because we do not have protocolized serial measurements of CK and serum creatinine levels for multiple days before and after presentation, we are not able to determine the threshold of CK that would rule in or rule out an AKI; therefore, our inverse CK results should not change future clinical action.

Our results reveal that children with viral-induced rhabdomyolysis warrant cautious monitoring because they can have secondary AKI. Special attention should be given to older children with comorbidities presenting with viral-induced rhabdomyolysis and those who present with elevated SUN and proteinuria because these biological factors were more likely to be associated with AKI. With this study, we conclude that viral causes of rhabdomyolysis have similar rates of AKI compared with other etiologies of rhabdomyolysis. However, additional larger prospective studies are necessary to further investigate this topic.