Treatment-Related Complications in Children Hospitalized With Disseminated Lyme Disease

Abstract

We describe here treatment approaches and treatment-related complications in 138 hospitalized children with disseminated Lyme disease. The patients who received parenteral antibiotics had a higher rate of complications than those who received oral therapy (15.4 vs 4.2 per 1000 days of therapy, respectively; P < .05). Oral therapy should be used preferentially if either route is supported by current guidelines.

Lyme disease is the most common vector-borne disease in the United States. Disseminated Lyme disease has numerous manifestations, including arthritis, carditis, aseptic meningitis, and multiple erythema migrans. Both parenteral and oral regimens are recommended as treatment options for early and late disseminated disease, depending on clinical presentation [1–3]. Select oral antibiotics typically used to treat disseminated Lyme disease (eg, doxycycline) can penetrate the involved tissues to eliminate disseminated Lyme infection [4]. Expert guidelines for the management of Lyme disease recommend oral therapy for Lyme arthritis and consideration of oral doxycycline for treating Lyme carditis and Lyme meningitis [1–3]. Because oral therapy confers several theoretical advantages over parenteral therapy, such as ease of use, fewer resulting complications, and decreased cost [5, 6], it is important to determine rates of complications and clinical outcomes associated with parenteral and oral therapy for pediatric patients with disseminated Lyme disease.

We designed this study to describe the different treatment approaches used in a cohort of hospitalized pediatric patients with disseminated Lyme disease; treatments included all-parenteral (intravenous [IV]), IV-to-oral (per os [PO]) or PO-to-IV, and all-PO antibiotic therapy. We determined the frequency and types of treatment-related complications associated with the various treatment approaches.

METHODS

We performed a retrospective cohort study of children hospitalized with disseminated Lyme disease between January 2007 and December 2014 at Children’s Hospital of Philadelphia (CHOP), an urban tertiary care hospital in a region in which Lyme disease is endemic. We used data abstracted from the electronic health record (EpicCare [Epic Systems, Verona, Wisconsin]) and captured from chart review using standardized electronic data-collection forms from Research Electronic Data Capture (REDCap). For every patient, we reviewed inpatient, emergency department, and outpatient medical records within the CHOP network for the 6-week period after hospital discharge. A second investigator reabstracted data from 10% of the electronic health records to verify concordance of data accuracy of the initial abstraction. The CHOP Institutional Review Board approved this study.

Study Population

We identified patients ≤18 years of age who were admitted to the CHOP for treatment of disseminated Lyme disease with a length of stay greater than 1 day. Patients were included if (1) the index hospitalization was assigned a discharge International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis code for Lyme disease (088.81) in any of the 21 diagnosis fields, (2) the index hospitalization was assigned a discharge ICD-9-CM diagnosis code for meningitis (320.7, 320.89, 320.9, or 322.9) in any of the 21 diagnosis fields and they had positive Lyme serology results (defined as positive serum enzyme-linked immunosorbent assay results and positive confirmatory Western blot results) or positive cerebrospinal fluid (CSF) Lyme test results; or (3) they had a CSF pleocytosis (defined as a CSF white blood cell count of >10 cells/mL corrected for the presence of CSF red blood cells using a 500-to-1 red blood cell/white blood cell ratio) [7] and they had positive Lyme serology or positive CSF Lyme test results. Patients were excluded if their primary reason for admission was explained better by an alternative diagnosis as documented by the treatment team, if they received more than 48 hours of treatment for Lyme disease before admission, or if they were diagnosed as having a concurrent condition.

Study Definitions

Lyme arthritis

Positive Lyme serology results according to a Western blot immunoglobulin G (IgG) test and a physician’s clinical diagnosis of Lyme arthritis; no other alternative diagnosis.

Lyme carditis

Positive Lyme serology results according to a Western blot IgM or IgG test or both, abnormal electrocardiogram findings, and a physician’s clinical diagnosis of Lyme carditis; no other alternative diagnosis.

Lyme meningitis

Positive Lyme serology or CSF Lyme test results, a CSF pleocytosis, and a physician’s clinical diagnosis of Lyme meningitis; no other alternative diagnosis.

Multiple erythema migrans

Physician-documented multiple erythema migrans lesions in the absence of an alternative diagnosis.

Study Outcomes and Exposures

The primary outcome was treatment-related complications, defined as adverse drug reactions and central venous catheter (CVC) complications that led to harm to the patient or need for medical attention [8]. Adverse drug reactions included abnormal laboratory values, dermatitis or photosensitivity, immediate hypersensitivity reactions, gastrointestinal adverse effects (eg, antibiotic-associated diarrhea or Clostridium difficile infection), infusion-related symptoms, Jarisch-Herxheimer–like reactions, and pseudotumor cerebri (onset after initiation of antibiotic treatment for disseminated Lyme disease). CVC complications included CVC displacement, CVC malfunction, central-line–associated bloodstream infection, thrombosis, and local skin infection [9].

The primary exposures were route(s) of antibiotic therapy, classified as all-PO, IV and PO, or all-IV therapy. We defined disease relapse or recurrence as unscheduled return to care for Lyme-related symptoms within 6 weeks after hospital discharge.

Statistical Analysis

Data were exported from REDCap into Excel (Microsoft, Redmond, Washington) and Stata 12 statistical software (Stata, College Station, Texas) for analysis. We summarized demographic and clinical characteristics of the patients with categorical data displayed as frequencies and percentages.

RESULTS

Study Population

The final study cohort included 138 patients; 60 had Lyme arthritis, 53 had Lyme meningitis, 14 had multiple erythema migrans without arthritis, carditis, or meningitis, and 11 had Lyme carditis. Most of the children were white (79%), were admitted to the general pediatrics inpatient service (57%), and had an infectious diseases consult (65%).

All patients were treated with 1 or more of the following antibiotics: amoxicillin, ampicillin, cefotaxime, ceftriaxone, cefuroxime, and doxycycline. Within the final cohort, the total exposures to IV and PO therapy were 1040 and 2139 days, respectively. Similar proportions of patients were treated with the 3 antibiotic treatment categories: all IV (n = 48 [35%]), all PO (n = 48 [35%]), or IV and PO (n = 42 [30%]).

Treatment-Related Complications

Twenty-five treatment-related complications occurred in 20 (14.5%) patients; 9 received all-IV therapy, 7 received IV and PO therapy, and 4 received all-PO therapy (Table 1). Of the 25 treatment-related complications, IV therapy was implicated in 16 (64%) cases and PO therapy in 9 (36%) cases. Of the 11 treatment-related complications that occurred in the IV-and-PO therapy group, IV therapy was implicated in 6 cases, and PO therapy was implicated in 5 cases. The antibiotic implicated was assigned as the medication the child was receiving on presentation for medical evaluation. The rates of treatment complications attributed to IV antibiotic administration and PO antibiotic administration were 15.4 and 4.2 per 1000 patient-days, respectively (P < .05). During the follow-up period, we identified no cases of treatment failure, although 1 patient had ongoing arthritis after completion of 1 month of oral doxycycline and required further evaluation. For 54 patients, no follow-up data were available.

DISCUSSION

A significant proportion of hospitalized children treated for disseminated Lyme disease experienced 1 or more treatment-related complications. Children treated with IV therapy experienced treatment-related complications at a higher rate than those who received PO therapy. In children with the same presentation of disseminated Lyme disease, those on IV therapy were more likely to have more complications than those on PO therapy.

This is the first descriptive study to characterize treatment approaches and associated complications in children with all presentations of disseminated Lyme disease. A previous study of treatment complications in patients with Lyme meningitis reported that 26% of the patients experienced 1 or more complications; most patients were treated with IV therapy [10]. Our overall rate of treatment complications was lower and likely reflects the lower complication rate seen with oral regimens. Our findings are similar to those of recent studies in children that reported treatment-related complications of PO and IV therapy for other infectious diseases; increasing numbers of studies are concluding that when oral options exist for the completion of antibiotic therapy, they should be used [9, 11, 12].

Our results also suggest that PO therapy for the treatment of disseminated Lyme disease in children does not have a higher rate of treatment failure than IV therapy, although our sample size was insufficient to draw appropriately powered conclusions. Study of comparative effectiveness of IV and PO therapy for disseminated Lyme disease in a prospective randomized clinical trial would be ideal, because differences in treatment regimens might be a result of confounding by indication.

Possible limitations of our study include (1) ascertainment bias (related to the use of ICD-9-CM diagnosis codes), (2) incomplete follow-up (we were unable to determine whether any patients sought follow-up care at an institution other than CHOP), and (3) limited generalizability (using a hospitalized cohort might have limited extension of these finding to patients with less severe clinical presentations who are treated in an outpatient setting). Future research is needed to fully characterize pediatric outcomes for patients with disseminated Lyme disease treated with PO antibiotics, because our follow-up data were incomplete.

CONCLUSION

In this retrospective descriptive study of the outcomes of children treated for disseminated Lyme disease, we found a higher rate of treatment-related complications in patients who received IV therapy. Given the significant morbidity associated with these complications, PO therapy should be used when appropriate to reduce treatment-related complications. These data support new recommendations for the expanded use of PO therapy in the treatment of disseminated Lyme disease.

Notes

Financial support. This work was supported by the National Institutes of Health and the Agency of Healthcare Research and Quality (grants T32-AI096345, L40-AI-107932, and U54-GM104941 to L. K. H. and K08-HS020939 to K. A. F.].

Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References