Differentiating Bell’s Palsy From Lyme-Related Facial Palsy

We performed a retrospective chart review of children with acute-onset unilateral peripheral FP who received care at a Children’s Hospital of Philadelphia (CHOP) care network facility between July 1, 2013 and June 30, 2018. The CHOP care network is located throughout the Lyme endemic region of Southeastern Pennsylvania and New Jersey and had an estimated 750 000 primary care visits, 88 000 emergency department visits, and 31 000 urgent care visits in 2018. The CHOP Institutional Review Board approved this study with a waiver of informed consent and Health Insurance Portability and Accountability Act authorization. This study followed guidance for high-quality chart review,²³  and the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines for cross-sectional studies.²⁴ 

We queried the CHOP electronic medical record (EMR) for children between 1 and 18 years old with an International Classification of Diseases, 10th Revision diagnostic code reflecting facial nerve dysfunction (G51.0, G51.8, G51.9, G52.7, G52.8, G52.9, G53, R29.810, M79.2, A69.21, A69.22, S04.5). Study data were collected from the EMR and managed by using Research Electronic Data Capture tools hosted at CHOP. We excluded children with (1) no documented FP, (2) facial asymmetry present from birth, (3) recurrent FP, (4) bilateral FP, (5) multiple cranial neuropathies or other unrelated acute neurologic examination abnormalities, or (6) history of other neurologic disorders that may affect facial nerve function such as multiple sclerosis or intracranial neoplasms.

We defined Bell’s palsy as acute-onset unilateral peripheral FP of unknown etiology. We attributed FP to otitis media when FP onset occurred within 14 days of acute otitis media without alternative explanation. We defined LRFP as acute-onset unilateral peripheral FP in a child who also met the Centers for Disease Control and Prevention criteria for diagnosis of Lyme disease within 2 months of the onset of FP (Table 1).

We defined a systemic prodrome in the 6 weeks preceding the onset of the FP as including at least 3 of the following: fever, headache, malaise, arthralgias, or myalgias.

We defined neuroimaging at initial presentation as those studies that were ordered on FP presentation and completed within 2 weeks of symptom onset.

We considered “standard” Lyme antibiotic therapy to include doxycycline (4.4 mg/kg per day divided twice daily up to 100 mg per dose for 10–21 days), amoxicillin (50 mg/kg per day divided 3 times daily up to 500 mg per dose for 14–21 days), or cefuroxime (30 mg/kg per day divided twice daily up to 500 mg per dose for 14–21 days). We considered a standard corticosteroid course to be prednisone ≥2 mg/kg per day (up to 60 mg daily; or alternative glucocorticoid equivalent) for at least 5 days.⁷  Patients who were prescribed lower doses were recorded as receiving corticosteroids, but not at standard dose. Antivirals considered included acyclovir or valacyclovir.

We defined recovery date on the basis of family report if documented in the EMR or as the first date in the EMR after the encounter for FP on which a physician documented a normal exam.

We conducted analyses using Stata 14.2 with 2-sided tests and a P value <.05 as the criteria for statistical significance, using nonparametric analyses given the small sample size. We described continuous variables (such as age) using medians and interquartile range or range values and we compared them using the Wilcoxon rank test or the Kruskal-Wallis equality-of-populations rank test. We described categorical variables (such as sex) using counts, frequencies, and proportions and we compared them using the χ² or Fisher exact test.

Demographics of the overall cohort and subgroups by FP etiology (Bell’s palsy, Lyme-related, and Other) are summarized in Table 2. Overall, our EMR query yielded 498 records. After applying exclusion criteria, 306 children remained eligible for inclusion in our cohort, with a median age of 10.3 years (range 1.1–17.9). Most children were diagnosed with either Bell’s palsy (209 of 306, 68%) or LRFP (82 of 306, 27%). Only 15 children had other etiologies of their FP, including infectious (otitis media [n = 8], parotitis [n = 2], mastoiditis [n = 1], lymphadenitis [n = 1]), oncologic (leukemia [n = 1], medulloblastoma [n = 1]), and trauma (n = 1).

Children with Bell’s palsy presented consistently throughout the year (Fig 1). Approximately one-fifth had preceding upper respiratory symptoms (rhinorrhea, congestion, or “a cold”) before the onset of FP (Table 2). However, a systemic prodrome in the 6 weeks before presentation was rare in children with Bell’s palsy (6%).

In contrast, children with LRFP presented predominantly between June and November, consistent with previous literature on seasonal trends in Lyme disease and risk of tick exposure.^4,25,26  None of the children with LRFP had preceding upper respiratory symptoms, but more than half had a systemic prodrome.

Among the 306 total children, 244 (80%) were tested for Lyme disease (Table 2). Among the 82 children with LRFP (Table 1), 25 (30%) had a rash consistent with erythema migrans (EM), noted a median of 2 days before FP onset (range: 35 days before 3 days after FP onset). Six children had an onset of EM the same day as FP, 2 had an onset of EM after FP. Only 3 children in our cohort had a documented history of a tick bite.

A total of 39 children (13%) underwent at least 1 neuroimaging study for the evaluation of their FP (Table 2). At initial presentation, 14 children underwent head computed tomography (HCT) (median of 0.5 days after FP onset, interquartile range [IQR] 0–1). One HCT scan revealed otomastoiditis in a child with eye pain and recent sinus infection, and 1 revealed lymphadenitis in a child with pain and swelling ipsilateral to FP. The remaining HCTs did not reveal pertinent acute abnormalities. Twelve children underwent brain MRIs ordered at initial presentation (median of 6.5 days after FP onset, IQR 1–8), 7 of whom had isolated inflammation of the affected facial nerve consistent with LRFP or Bell’s palsy; the remainder of initial MRIs did not reveal pertinent acute abnormalities.

After the initial FP diagnostic evaluation, 15 children had 16 additional neuroimaging studies ordered later because of a lack of improvement or the development of new clinical signs or symptoms. This included 4 HCTs performed a median of 8 days after FP onset (IQR 6.75–9.25) and 12 MRIs performed a median of 22 days after FP onset (IQR 18.75–30). An intracranial neoplasm was found on 1 HCT scan of a child with new coordination and gait disturbances that began 9 days after FP presentation. Stigmata of pseudotumor cerebri were found on an MRI scan of a child who developed new headaches and pulsatile tinnitus after initial presentation; these findings were attributed to doxycycline therapy for Lyme disease. Eight other MRIs revealed isolated inflammation of the affected facial nerve consistent with Bell’s palsy or LRFP. The remaining studies were normal. No child with isolated peripheral FP had neuroimaging findings that impacted clinical management.

Only 3 of 306 children in our cohort had a lumbar puncture (LP) performed for FP evaluation. One LP was performed at initial presentation to evaluate for meningitis given concurrent fever, headache, and neck pain; cerebrospinal fluid (CSF) cell counts, protein, and glucose were normal, and CSF Lyme total antibody and polymerase chain reaction test results were negative. A second LP was performed later to confirm an MRI diagnosis of pseudotumor cerebri (see above); CSF Lyme polymerase chain reaction test results were negative. A third LP was performed after 3 weeks without FP improvement; CSF cell counts were normal. CSF Lyme testing was not sent.

Seventy-eight children (95%) with LRFP completed standard Lyme-appropriate antibiotics. The remaining 4 children with LRFP also received antibiotics, but the dose was not documented (n = 2), the dose was lower than recommended (n = 1), or the course was shorter than recommended (n = 1). Thirty-six children (17%) with Bell’s palsy also completed a full course of standard Lyme-appropriate antibiotics, despite negative serum Lyme test results in 25 (Table 3). Twenty-six children (32%) with LRFP and 130 children (62%) with Bell’s palsy completed a standard course of corticosteroids.

FP was resolved by parent/clinician assessment in all but 2 of the 240 children with documented follow-up in our EMR (Table 3). One child had only partial improvement of Bell’s palsy in the context of an MRI significant for isolated inflammation of the facial nerve, and 1 child developed additional neurologic symptoms before being ultimately diagnosed with medulloblastoma. Among those with a resolution, this was documented a median of 76 days (IQR 33–176) after FP onset; this was not significantly different between those with Bell’s palsy or LRFP (84 days vs 58 days, P = .08). The proportion of children who received any corticosteroid was similar among those who recovered and those who were lost to follow-up in both Bell’s palsy (76% vs 76%, P = .85) and LRFP (38% vs 56%, P = .32).

In this large single-center cohort of children with acute-onset unilateral peripheral FP, Bell’s palsy and Lyme disease were by far the most common causes of FP, similar to previous reports in Lyme-endemic regions.^4,25,27  Distinguishing clinical factors between Bell’s palsy and LRFP included an expected seasonal distribution of LRFP, and a systemic prodrome preceding the onset of LRFP more often than Bell’s palsy (55% vs 6%). These factors confirm and expand on the clinical predictors described by Nigrovic et al,⁴  which include season of onset, presence of fever, and history of headache. We also found that neuroimaging and lumbar puncture did not add diagnostic value in isolated FP. Overall, almost all children with LRFP or Bell’s palsy who had documented follow-up recovered by parent/clinician assessment regardless of corticosteroid treatment. However, subtle residual deficits may not have been appreciated; we were unable to determine if there was any impact of therapy on the time course to recovery given our retrospective study design and we cannot comment on the recovery of children who did not receive follow-up care at our institution.

We were interested to note that only 30% of children with LRFP had a recent history of EM rash (16 of 25 of which appeared within 1 week of FP presentation), and only 3 children in our cohort had a documented tick bite. Clinical documentation often expressed a lower suspicion of LRFP when there was no history of EM rash or tick exposure. However, although EM is traditionally considered a sign of early localized disease, multiple EM lesions may also be present in early disseminated disease, when Lyme-related cranial neuropathies typically occur,²⁸  and a short interval between LRFP and EM rash has been reported.¹⁸  In addition, only a small fraction of patients with LRFP historically report a preceding tick bite.^1,26,29  Therefore, LRFP should still be considered in children presenting with FP in Lyme-endemic regions between June and November,⁴  particularly in the context of a systemic prodrome, even in the absence of an EM rash or a known tick bite.

More than 80% of children with FP in this cohort were diagnosed and managed without neuroimaging or CSF studies, which differs from several previous reports in which LP was performed in up to 90% of children presenting with peripheral FP.^1,27  Among the 39 children in our cohort who underwent neuroimaging and/or LP at any point in their FP course, only 4 children had results that impacted their clinical management, all of whom had additional signs or symptoms beyond isolated FP. Neuroimaging findings did not impact clinical management in any child with isolated peripheral FP. These data suggest that neuroimaging and/or CSF testing is not necessary for the evaluation of all isolated acute-onset unilateral peripheral FP,⁷  but can be considered on a case-by-case basis to rule out other non-Lyme causes of symptoms.

Treatment practices for FP varied broadly in this cohort with respect to corticosteroids and antibiotic therapy. A sizeable portion of our cohort completed a full course of standard Lyme-directed antibiotics despite a diagnosis of Bell’s palsy, even with negative Lyme serology results. The risks of unnecessary antibiotic use are well-documented,³⁰  and 1 patient in our cohort developed pseudotumor cerebri secondary to doxycycline therapy. Moreover, it is important to recognize that antibiotics are not thought to hasten the resolution of the LRFP,¹⁸  so waiting for the results of testing before starting antibiotics should not impact the speed of FP recovery. Thus, we advocate waiting for results of Lyme testing before starting antibiotics, unless there is high clinical suspicion for Lyme disease, such as when the patient has EM rash, systemic prodrome, or in the high season in a Lyme-endemic area.

The role of corticosteroids in influencing the rate or extent of recovery in Bell’s palsy or LRFP in children remains unclear.²²  Similar to others,¹⁴  we did not find an association between corticosteroid use and parent/clinician assessment of recovery in either Bell’s palsy or LRFP, although it is possible that subtle changes or deficits may have been missed with this retrospective outcome measure. Although it is also possible that our results may have been impacted by the number of children lost to follow-up, a similar proportion of those who recovered and those lost to follow-up received corticosteroids in both Bell’s palsy and LRFP. In addition, we feel it is likely that most children with ongoing facial weakness would have returned for follow-up care. Unfortunately, the data on time to recovery was flawed in this retrospective study, so differences by FP etiology and/or the effect of corticosteroids on the speed of recovery may have been missed. Actual time to resolution may have been shorter than the medians we report; most charts did not include a specific date of recovery, so time to resolution was largely based on physician documentation of a normal examination at a subsequent encounter, which may lag from actual resolution. Ongoing prospective, randomized, double-blind, placebo-controlled trials examining corticosteroid use are underway in Bell’s palsy via the Pediatric Research in Emergency Departments International Collaborative network¹¹  and in a cohort of children with both Bell’s palsy and LRFP³¹ ; the results of these 2 studies will be critical in determining best practice regarding corticosteroid use in pediatric facial palsy.

Notably, the Pediatric Research in Emergency Departments International Collaborative network recently suggested the consideration of screening complete blood count before the initiation of corticosteroids for facial palsy treatment.²  They have identified 5 children among a total of 833 children (0.6%) with peripheral FP because of previously undiagnosed leukemia. In 2 of these 5 children, corticosteroid treatment likely delayed their leukemia diagnosis, delaying chemotherapy. Our study also included 1 child who similarly had FP because of as-yet-undiagnosed leukemia, and who received corticosteroid treatment. The leukemia was later diagnosed when she represented with fever and weight loss after her corticosteroid course was complete. We, therefore, agree that a complete blood count to screen for occult hematologic malignancy before corticosteroid initiation could reduce the potential harm from this treatment course.

Our study had several limitations. First, information regarding clinical history, treatment course, and outcomes were identified retrospectively rather than through a prospective standardized assessment. As a result, long-term outcomes may appear more favorable when assessed by retrospective chart review as compared with in-person assessment,^32–34  resulting in possible misclassification bias. Similarly, documented physical exams did not use standardized FP rating scales, may not have commented on all associated findings (eg, synkinesis, tearing, pain, and others), and clear time to recovery data were not consistently available. In addition, some historical features may have been missed or misclassified, and the dosing of corticosteroids and types of antibiotics used varied. Future prospective studies would address this limitation. Second, the study was conducted in a Lyme-endemic region. Thus, although caution should be used in generalizing data on LRFP broadly, data on Bell’s palsy and other causes of FP should still apply across a variety of settings. Finally, not all children underwent neuroimaging to rule out other alternative structural causes of FP. However, the fact that most children ultimately recovered argues against an alternative etiology that would require different therapy.

Bell’s palsy and LFRP are common causes of pediatric acute-onset peripheral FP in Lyme-endemic regions. Location, month of presentation, and systemic prodrome may help clinicians distinguish children with LRFP from those with Bell’s palsy at symptom onset, which may help guide appropriate antibiotic use. Corticosteroid use in the first week of symptoms did not impact our measures of parent/clinician assessment of recovery in Bell’s palsy or LRFP, and almost all children with Bell’s palsy of LRFP that followed up in our care network ultimately recovered. However, subtle abnormalities may not have been appreciated, and time to recovery was not assessed. Thus, the role of corticosteroids in influencing the rate or full extent of recovery in pediatric Bell’s palsy or LRFP remains uncertain after this retrospective project. Future prospective studies using standardized assessment tools at regular follow-up intervals will help further address these questions.

CHOP

Children’s Hospital of Philadelphia

CSF

cerebrospinal fluid

EM

erythema migrans

EMR

electronic medical record

FP

facial palsy

HCT

head computed tomography

IQR

interquartile range

LP

lumbar puncture

LRFP

Lyme-related facial palsy