Original Article

The Yield of Neuroimaging in Children Presenting to the Emergency Department With Acute Ataxia in the Post–Varicella Vaccine Era

Journal of Child Neurology 1-7 ª The Author(s) 2014 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/0883073814561300 jcn.sagepub.com

Tiffany Rudloe, MD1, Sanjay P. Prabhu, MBBS, FRCR2, Mark P. Gorman, MD3, Lise E. Nigrovic, MD, MPH1, Marvin B. Harper, MD1,4, Assaf Landschaft, BSc1, and Amir A. Kimia, MD1

Abstract To determine the yield of neuroimaging in children presenting to the emergency department with acute ataxia in the post– varicella vaccine era, we conducted a cross-sectional study between 1995 and 2013 at a single pediatric tertiary care center. We included children aged 1-18 years evaluated for acute ataxia of 0.4)] in the multivariable analysis.18 For these analyses, we used the Statistical Program for the Social Sciences (IBM SPSS Statistic Version 21, IMB Inc, Chicago, IL).

Recursive Partitioning We used binary recursive partitioning with Gini splitting rules to identify children at low risk of clinically urgent intracranial pathology and could therefore have neuroimaging safely deferred at the initial evaluation. For this analysis, we used Salford Predictive Modeler v7.0 (Salford Systems, San Diego, CA). To maximize the sensitivity of the resulting model, we assigned a misclassification cost of 100:1 for the classification of a child with clinically significant intracranial pathology as low risk. A 10-fold internal crossvalidation was used to avoid model overfitting. We report the test performance (sensitivity, specificity, and predictive values) for resulting models’ clinical prediction rules, with 95% confidence intervals using exact methods.

Classification of Neuroimaging Studies The study neuroradiologist (SP), blinded to clinical outcome, reviewed every neuroimaging study (cranial computed tomography [CT] or magnetic resonance imaging [MRI]) performed within 72 hours of initial presentation to the study site emergency department. When study radiologist’s interpretations differed from the clinical interpretation, the study radiologist’s interpretation was used.

Outcome Measure Our primary outcome is to determine the yield of neuroimaging among these patients. For the purposes of this study, clinically urgent intracranial pathology on neuroimaging is defined as any finding that would routinely require medical or surgical intervention. Examples of findings considered to represent clinically significant intracranial pathology include any mass lesion causing mass effect, infarction, hemorrhage, arteriovenous malformation, hydrocephalus, abscess, or acute disseminated encephalomyelitis. For those children who did not have neuroimaging performed, children who had a normal follow-up

Results Demographics During the 18-year study period, 961 882 patients visited the study emergency department. We identified 364 children with acute ataxia who met study criteria (0.04% of emergency department visits) (Figure 1). The median patient age was 2.8 years (interquartile range [IQR] 1.7-5.5) and 46% were female. The median duration of symptoms prior to emergency department evaluation was 2 days (IQR 1-2 days). One hundred ninety-one children (53% of study patients) were admitted to the hospital. Use and results of neuroimaging. Neuroimaging was obtained in 284 children (78%). Of those with neuroimaging, 261/284 (92%) were performed within 24 hours of presentation. One hundred thirty-three had a CT scan (47%), 53 had an MRI

Downloaded from jcn.sagepub.com at FRESNO PACIFIC UNIV on December 27, 2014

Rudloe et al

3 Comparison of children with and without clinically urgent neuroimaging findings. Univariate comparison of clinical history, physical exam findings, and laboratory results of patients with and without clinically urgent intracranial pathology are shown in Table 2. This yielded age, headache, reported weakness, ataxia duration, and additional neurologic signs as candidate predictors for evaluation in multivariate analysis.

Figure 1. Case identification.

(18%), and 98 had both CT and MRI (35%). Two hundred twenty-four children had their primary imaging performed at our institution (79%). There were 78 patients who were transferred from an outside facility, of whom 60 (77%) were imaged prior to transfer. One patient had an MRI at the outside facility (which was not repeated) and 59 had CT scans, of which 6 (10%) were repeated in our facility and 35 (59%) proceeded to having an MRI. The prevalence of significant intracranial pathology identified on neuroimaging was similar when limited to those with primary imaging performed at our facility (26/259 [10%], 95% confidence interval 7%-15%). Children who did not have neuroimaging. Of the 80 children who did not have initial neuroimaging, 51 (64%) had long-term clinical follow-up at the study institution. None were found to have clinically significant intracranial pathology. The clinical diagnoses for the 80 children without neuroimaging were 49 (63%) acute cerebellar ataxia, 11 (12%) postviral ataxia (2 with positive Epstein Barr virus serology), 8 (10%) labyrinthitis, 3 seizures, 2 Guillain-Barre´ syndrome or other demyelinating disorders, 2 myositis, 2 complicated migraine, 2 ingestions, and 1 case of fecal impaction. Therefore, 335 children were assessed for clinically urgent intracranial injury: those with emergent neuroimaging or long-term follow-up. Please see Figure 2 for patient assessment and inclusion is model. Forty-two children (13%, 95% confidence interval 9%-17%) presenting with acute ataxia had clinically urgent intracranial pathology. The clinical characteristics and final diagnoses for the 42 children with clinically significant intracranial pathology identified on neuroimaging are presented in Table 1.

Developing a prediction model for acute ataxia patients that are most likely or least likely to benefit from acute neuroimaging. Fifty-nine patients (59/355 [17%]) with additional neurologic signs were assumed to require acute neuroimaging, so these patients were removed from model development (Figure 2). Therefore, the model included 276 children with no additional neurologic signs other than ataxia, of whom 20 had clinically urgent intracranial pathology. The following candidate predictors were entered into a recursive partitioning model: age at presentation (less than or equal to 3 years of age versus greater than 3 years), duration of ataxia (less than or equal to 3 days versus greater than 3 days of ataxia and localization on neurologic exam (ie, falling to one side after attempting to stand). The model is presented in Figure 3: Figure 3A includes all children, while Figure 3B is a subanalysis of patients who had imaging obtained. Children who were less than or equal to 3 years of age and had a duration of ataxia for 3 days or less were at a much lower risk of clinically urgent intracranial pathology (1/142 [0.7%], 95% confidence interval 0%-4.4%). This model had a sensitivity of 95% (95% confidence interval 75%-100%), specificity of 56% (95% confidence interval 49%-62%), and a negative predictive value of 99% (95% confidence interval 96%-100%).

Discussion We present a large retrospective cohort of children with acute ataxia in the post varicella vaccine era. In our cohort, the proportion of children with clinically urgent intracranial pathology is higher than previously reported. Nonetheless, children less than 3 years of age and with duration of ataxia of less than 3 days may represent a low-risk group where neuroimaging can be deferred if contingent upon close clinical follow-up and reassessment. Most children with acute ataxia will not have clinically urgent findings on neuroimaging. Prior studies have reported that0 to 5% will have a significant diagnosis based on the clinical setting.1-3,14,20,21 Our finding that 13% will have identified significant intracranial pathology is likely due to the substantial decline of varicella-associated ataxia, no mumps-associated cases, and an a priori exclusion of patients with a known ingestion. The decline in varicella-associated cases is consistent with the reports of others that varicella vaccination has reduced the number of cases of acute ataxia from 1 to 5 cases per 20 00015 with wild-type varicella infections14,19,20 to an estimated 1 to 2 cases per million doses21 post–varicella vaccine.

Downloaded from jcn.sagepub.com at FRESNO PACIFIC UNIV on December 27, 2014

4

Journal of Child Neurology

Figure 2. Cases included in analysis.

Current clinical references offer varying recommendations regarding neuroimaging for children with acute ataxia.22,23 The 2009 edition of Clinical Pediatric Neurology states, ‘‘Brain imaging is recommended for most children with acute cerebellar ataxia.’’24 Conversely, another published review states that ‘‘in the absence of altered consciousness, focal neurologic signs, or marked asymmetry of ataxia, the yield of scans is low.’’25 Finally, at the time of writing this manuscript, the online reference UpToDate1 suggests that no emergent neuroimaging needs to be performed in the absence of signs of elevated intracranial pressure, focal neurologic exam, or fever with meningismus. Our data supports an approach that includes neuroimaging except among children younger than 3 years of age having symptoms for less than 3 days. Our study has some important limitations. First, it is retrospective and the predictors were assessed by medical chart

review. We minimize this potential bias by selecting objective parameters likely to be recorded accurately in the medical record and we tested and found good interrater reliability. Second, the risk of significant pathology in our low-risk group will require validation in another study, as the upper bounds for our 95% confidence interval is almost 5%. Regardless of whether neuroimaging is performed initially, these children merit close clinical follow-up.

Conclusions The prevalence of clinically urgent intracranial pathology is higher than reported in the pre–varicella vaccination era and imaging may be indicated for most patients presenting with acute ataxia. Neuroimaging may be deferred for younger

Downloaded from jcn.sagepub.com at FRESNO PACIFIC UNIV on December 27, 2014

Rudloe et al

5

Table 1. Clinical Diagnoses, Neuroimaging Modality and Long-Term Follow-Up for the 42 Children With Clinically Significant Intracranial Pathology. Findings

n

Evident on CT scan a

Evident on MRI

Tumors Brain tumors Pilocytic astrocytoma Medulloblastoma Pontine glioma Ependymoma Bony tumor Temporal bone histiocytosis ADEM

22 21 6 6 6 1 1 1 12

19/20 18/19 5/6 6/6 5/5 2/2

20/20 19/19 6/6 6/6 5/5 2/2

1/1 2/10

1/1 12/12

Infarct

3

2/3

3/3

Encephalitis Transverse myelitis Intracranial bleed Syrinx with hydrocephalus Mastoiditis with an epidural abscess

1 1 1 1 1

1/1 0/1 1/1 1/1 1/1

1/1 1/1 1/1 1/1 1/1

Comments and follow-up

1 with metastatic disease, 2 with neurologic sequelae, 1 lost to follow-up 4 survivors 2 of them with sequelae, 1 deceased, 1 lost for follow-up 5 deceased with 6 mo 1 y, one with residual disease at 10 mo Deceased Survivor 10 year follow-up 8 are asymptomatic, 2 later diagnosed with multiple sclerosis, 2 lost to follow-up Three survived with mild sequelae, 2 had evidence of dissection of left vertebral artery One patient later diagnosed with Leigh disease Minimal sequelae—12-year follow-up Secondary to cavernous hemangioma Operated on emergently, mild sequelae on follow-up Survived asymptomatically

Abbreviations: ADEM, acute disseminated encephalomyelitis; CT, computed tomography; MRI, magnetic resonance imaging. a Some patients were examined with 1 imaging modality only.

Table 2. Univariate comparison of clinical history, physical exam findings, and laboratory results of patients with and without clinically urgent intracranial pathology. Characteristics Demographics Female gender Age greater than 3 y History Recent febrile illness Vomiting Headache Weakness (reported) Duration longer than 3 d Temperature >38 C Physical exam Otitis media on exam/history Additional neurologic signsb Falling onto one side Deep tendon reflex abnormality Laboratory findings Peripheral WBC count k/mm3 -median, (IQR) Positive blood culture CSF WBC count/mm3; median (IQR) Positive CSF culture (pathogen)

Clinically significant intracranial pathology (n ¼ 42)

No clinically significant intracranial pathologya (n ¼ 293)

Significance (P value)

22/42; 52% 34/42; 81%

132/293; 45% 127/293; 43%

.41

The Yield of Neuroimaging in Children Presenting to the Emergency Department With Acute Ataxia in the Post-Varicella Vaccine Era.

To determine the yield of neuroimaging in children presenting to the emergency department with acute ataxia in the post-varicella vaccine era, we cond...
412KB Sizes 1 Downloads 4 Views