Correspondence
Antimicrobial Prophylaxis after Tick Bites
N Engl J Med 1993; 328:1418-1420May 13, 1993
- Article
To the Editor:
We believe Shapiro et al. (Dec. 17 issue)1 had inadequate evidence to support their recommendation for nontreatment of deer-tick bites. The population samples were much too small for a meaningful comparison of the risks of serious late sequelae with the risks and benefits of prophylactic antibiotic therapy.
Consider the hypothetical situation of 10,000 deer-tick bites in Connecticut, in which 120 people would acquire the disease and 15 to 36 would not manifest erythema migrans. Magid et al.2 estimate that 60 percent of these people would acquire chronic arthritis, 17 percent meningoencephalitis or neuropathy, and 6 percent cardiac sequelae. Assuming half this estimate, between 4 and 14 people would face serious, debilitating diseases, which, if recognized as related to Borrelia burgdorferi, might or might not respond to high-dose parenteral antibiotic treatment. These people should not be dismissed as a “tiny” fraction.
An adequate trial to draw conclusions about antibiotic prophylaxis for deer-tick bites might require 8000 to 10,000 subjects. The trend in the data so far suggests that treatment may be indicated. We would personally insist on prophylactic treatment were we to receive deer-tick bites.
2 ReferencesJohn Genter, M.D.
Nancy G. Berman, Ph.D.
Harbor-UCLA Medical Center, Torrance, CA 90501Roberta E. Madison, Dr.P.H.
California State University, Northridge, CA 913301
Shapiro ED ,Gerber MA ,Holabird NB , et al. A controlled trial of antimicrobial prophylaxis for Lyme disease after deer-tick bites. N Engl J Med 1992;327:1769-1773
Full Text | Web of Science | Medline2
Magid D ,Schwartz B ,Craft J ,Schwartz JS . Prevention of Lyme disease after tick bites -- a cost-effectiveness analysis. N Engl J Med 1992;327:534-541
Full Text | Web of Science | Medline
To the Editor:
The report by Shapiro et al. raises two critical questions: first, is amoxicillin prophylaxis ineffective in preventing Lyme disease, and second, on what grounds do the authors conclude that antibiotic treatment is not indicated?
The answer to the first question is easy. The sample is too small and the power of the study is inadequate to demonstrate statistically that amoxicillin either does or does not prevent Lyme disease. With a similar sample size, many valuable measures for preventing disease would not be proved effective. The failure to disprove the null hypothesis does not show that antibiotics are ineffective in preventing Lyme disease.
The answer to the second question is more complex. Without having determined by how much antibiotics reduce the risk of Lyme disease, and assuming that most patients who contract Lyme disease after tick bites will have and recognize erythema migrans, the authors conclude that preventive antibiotics are not indicated and that treatment at the time of the rash will suffice. Based on the authors' own two cases and their reinterpretation of two published prospective studies, this conclusion ignores the previous findings of a rate of subclinical seroconversion of about 50 percent1,2. Also, it hinges on a set of unstated presumptions: erythema migrans is not serious; a small risk of serious complications is tolerable; and the prevention of Lyme disease is not “cost effective.” Cost effectiveness was determined by modifying the conservative assumptions of a previous analysis by Magid et al.3 that reached an essentially opposite conclusion on the basis of medical expenses alone -- not including the costs of disability, work lost, and other factors.
What is the downside of prophylactic treatment of tick bites? Of course, if antibiotics did not work, there would be no point in giving them. The cost of 10 days of amoxicillin treatment is trivial., and there is a very small incidence of side effects (zero in the present study).
What will happen when the first patient with an intentionally untreated tick bite is found to have the Guillain-Barre syndrome, chronic progressive encephalomyelitis, myocarditis, or chronic arthritis? Will this Journal article be cited as a moral or legal defense? In a medical environment in which the catch phrases “outcomes research” and “cost effectiveness” can set the standards of future practice, there is a serious risk in drawing negative conclusions from inadequate data, when an effective means of preventing a potentially serious disease may exist.
3 ReferencesDavid A. Drachman, M.D.
University of Massachusetts Medical School, Worcester, MA 016551
Steere AC ,Taylor E ,Wilson ML ,Levine JF ,Spielman A . Longitudinal assessment of the clinical and epidemiological features of Lyme disease in a defined population. J Infect Dis 1986;154:295-300
CrossRef | Web of Science | Medline2
Hanrahan JP ,Benach JL ,Coleman JL , et al. Incidence and cumulative frequency of endemic Lyme disease in a community. J Infect Dis 1984;150:489-496
CrossRef | Web of Science | Medline3
Magid D ,Schwartz B ,Craft J ,Schwartz JS . Prevention of Lyme disease after tick bites -- a cost-effectiveness analysis. N Engl J Med 1992;327:534-541
Full Text | Web of Science | Medline
To the Editor:
Shapiro et al.1 incorrectly take issue with several assumptions and conclusions of our recently published model for the care of patients bitten by ixodes ticks2. Here we respond to some of the most salient issues raised.
Our model assumes a 1 percent rate of serious late sequelae among untreated patients only if the base-line risk of infection after a tick bite is 3.0 percent. With adjustment for actual tick-carriage and infection rates in the four studies cited, our estimate of late sequelae falls within the 95 percent confidence interval calculated. Sensitivity analyses did not alter our results even when the rate of late sequelae was assumed to be 0.4 percent, well within the calculated confidence interval (0 to 0.7 percent).
Our model does not assume that nearly all untreated infected patients have sequelae, but rather that the incidence of sequelae is a function of infection and unrelated to the development of erythema migrans. No data exist to support the authors' suggestion of a substantially lower risk of late sequelae among untreated infected persons without erythema migrans.
Our base-line estimate of a 70 percent incidence of erythema migrans after infection is based on the best available data and most authoritative review articles. The authors' 86 percent estimate is based on two small studies with several methodologic limitations. Our model's results remain unchanged even if the rate of erythema migrans exceeds 90 percent.
The study by Shapiro et al. underscores several of our model's important points. First, even a rigorous randomized trial of several hundred subjects is too small to address adequately the issue of the efficacy of empirical therapy in preventing acute disease and late sequelae. Cost-effectiveness analysis with decision models is a valid and useful alternative method. Second, the conclusions based on a single study cannot be generalized to other communities, given the highly variable rates of ixodes tick carriage in different geographic areas. Finally, the results of Shapiro et al. are virtually identical to those predicted by our model. With 12 percent tick-carriage rates, our model predicts incidence rates of 1.2 percent for Lyme disease and 0.84 percent for erythema migrans in untreated subjects (1.8 predicted vs. 2 observed cases) and 0.46 percent for late sequelae (well within the 95 percent confidence interval of 0 to 0.7 percent).
Our recommendations (empirical treatment when the probability of infection is 3.6 percent or more; no empirical treatment when the probability is 1.0 percent or less) do not differ from the authors' conclusions for a population like the one they studied. The 1.2 percent infection rate in this study falls at the lower end of the model's discretionary range, leaning strongly against empirical treatment.
2 ReferencesDavid Magid, M.D.
University of Colorado, Denver, CO 80262Brian S. Schwartz, M.D.
Johns Hopkins University, Baltimore, MD 21205Joseph Craft, M.D.
Yale University, New Haven, CT 06510J. Sanford Schwartz, M.D.
University of Pennsylvania, Philadelphia, PA 191041
Shapiro ED ,Gerber MA ,Holabird NB , et al. A controlled trial of antimicrobial prophylaxis for Lyme disease after deer-tick bites. N Engl J Med 1992;327:1769-1773
Full Text | Web of Science | Medline2
Magid D ,Schwartz B ,Craft J ,Schwartz JS . Prevention of Lyme disease after tick bites -- a cost-effectiveness analysis. N Engl J Med 1992;327:534-541
Full Text | Web of Science | Medline
To the Editor:
. . . Shapiro et al. seem oblivious to the phenomenon of latency, well recognized to occur in spirochetal diseases. That no illness is apparent (on telephone evaluation) at one year after tick attachment does not necessarily mean that illness will not occur later. Thus, it will be interesting and important to know the outcomes of the study subjects after 2, 5, 10, 15, and 20 years of follow-up.
Avoidable infection may occur in some patients not treated for known deer-tick bites. In some bitten persons, classic disease will develop early. Ideally, such illness will be correctly diagnosed and adequately treated. Even this is not certain, however, for the mild influenza-like illness and other nonspecific symptoms that may follow deer-tick bites and may indicate true infection may not be correctly recognized by either the patient or the physician. In other cases, infection is subclinical. The outcome will depend on the interaction of spirochete and host factors, on whether or not the correct diagnosis is made at a later time, and on whether effective treatment is given.
Cases of meningomyeloencephalitis are becoming commonplace in the New York counties of Westchester, Putnam, Dutchess, Orange, and Rockland as the epidemic of Lyme disease enters its second decade. Patients with this condition may be misdiagnosed as having multiple sclerosis1,2 and treated inappropriately with corticosteroids or cytotoxic agents. The application of even intensive antibiotic therapy at this stage may not halt the neurologic injury.
2 ReferencesKenneth B. Liegner, M.D.
8 Barnard Rd., Armonk, NY 105041
Liegner KB. Evidence for borrelial etiology and pathogenesis in a series of patients carrying a diagnosis of multiple sclerosis. Presented at the 45th International Northwestern Conference on Diseases in Nature Communicable to Man, Hamilton, Montana, August 1990. abstract.
2
Liegner KB. Difficulty of distinction between borrelial (Lyme) encephalomyelitis and multiple sclerosis. Presented at the 47th International Northwestern Conference on Diseases in Nature Communicable to Man, Vancouver, British Columbia, August 1992. abstract.
Author/Editor Response
The authors reply:
To the Editor: As we stated1 the power of our study was inadequate to assess the efficacy of prophylaxis because the risk of Lyme disease in untreated patients is so low. Although we agree with Drs. Drachman, Genter et al., and Liegner that prevention of Lyme disease is desirable, it does not follow that prevention should be pursued regardless of the costs. Decisions about treatment must balance the benefits against the risks and monetary costs of specific strategies. Amoxicillin has a well-described potential for adverse effects. Contrary to Dr. Drachman's argument, given the risk we found of Lyme disease after a recognized deer-tick bite, the cost-effectiveness analysis of Magid et al.2 supports our conclusion that antimicrobial prophylaxis is not routinely indicated.
We agree with Magid et al. that cost-effectiveness analysis with decision models is a useful approach to the question of how to care for patients with recognized deer-tick bites. However, we disagree with some of the “best estimates” and with the ranges of these estimates for the sensitivity analyses used in their model2 -- specifically, the proportion of infected persons who do not have erythema migrans (30 percent) and the proportion of these (83 percent, which we think is “nearly all”) in whom late sequelae develop (most of which, contrary to the suggestions of Genter et al. and Liegner, are not chronic and do respond to antimicrobial treatment). Both these assumptions bias the model toward routine antimicrobial prophylaxis at a lower risk of Lyme disease after a deer-tick bite. The “two small studies” to which the authors refer are the only published prospective studies3,4 and, we believe, are more reliable than the retrospective reviews they cite.
Additional support for our contention comes from an unpublished prospective study of Lyme disease in an active surveillance system that includes 84 percent of the primary care physicians in the Lyme, Connecticut, area. Of the 630 cases of Lyme disease reported in 1992, 95 percent involved erythema migrans (Hamilton D, Connecticut Department of Health Services: personal communication). In their model the risk of late sequelae is clearly related to the development of erythema migrans, since those who have erythema migrans are treated (with 95 percent efficacy) before late sequelae develop. Indeed, the risk of late sequelae in their model is more than eight times greater for infected persons who do not have erythema migrans than for those who do.
As Magid et al. point out, our results are a reflection of the risk in our study area. Risk varies in different areas. Our study was conducted in an area of southeastern Connecticut in which the incidence of Lyme disease is among the highest in the country; there are very few areas in which the risk of Lyme disease after a recognized deer-tick bite is likely to be higher. We are gratified that Magid et al., despite their protests, agree with our conclusions.
Liegner ignores the fact that no one in our study had serologic evidence of asymptomatic infection, although all were tested three months or more after the bite. He also ignores the fact that persons who have latent infection with Treponema pallidum have generally had signs and symptoms of primary or secondary syphilis that was never treated. In Connecticut, meningoencephalitis due to Lyme disease is not “commonplace.” We would be surprised if it is in an adjacent county in New York.
4 ReferencesEugene D. Shapiro, M.D.
Anne T. Berg, Ph.D.
Yale University School of Medicine, New Haven, CT 06510Michael A. Gerber, M.D.
Henry M. Feder, Jr., M.D.
University of Connecticut Health Center, Farmington, CT 060321
Shapiro ED ,Gerber MA ,Holabird NB , et al. A controlled trial of antimicrobial prophylaxis for Lyme disease after deer-tick bites. N Engl J Med 1992;327:1769-1773
Full Text | Web of Science | Medline2
Magid D ,Schwartz B ,Craft J ,Schwartz JS . Prevention of Lyme disease after tick bites -- a cost-effectiveness analysis. N Engl J Med 1992;327:534-541
Full Text | Web of Science | Medline3
Hanrahan JP ,Benach JL ,Coleman JL , et al. Incidence and cumulative frequency of endemic Lyme disease in a community. J Infect Dis 1984;150:489-496
CrossRef | Web of Science | Medline4
Steere AC ,Taylor E ,Wilson ML ,Levine JF ,Spielman A . Longitudinal assessment of the clinical and epidemiological features of Lyme disease in a defined population. J Infect Dis 1986;154:295-300
CrossRef | Web of Science | Medline
