Previous studies [6-9] have shown that many patients who receive treatment for Lyme disease lack firm clinical or serologic evidence of disease. Persons at increased risk for inappropriate treatment include 1) patients who develop Lyme disease and receive appropriate antibiotic therapy but have persisting fatigue, arthralgias, and myalgias that are misinterpreted by patients or physicians as evidence of partially treated or recurrent Lyme disease and 2) patients who do not have Lyme disease but seek treatment for fatigue, arthralgias, and myalgias only or other non-specific symptoms (such as insomnia, chronic headache, or low energy). Reasons for inappropriate treatment in the latter group of patients include physician reliance on serologic rather than clinical evidence of disease (false-positive results on serologic tests for Lyme disease are common [10, 11]); unfamiliarity on the part of the physician with the signs and symptoms of Lyme disease; and patient demand for treatment because of anxiety about Lyme disease [12, 13].

Although the overdiagnosis and overtreatment of Lyme disease have been recognized [6-9], the adverse consequences associated with these practices have received inadequate attention [13]. We report on the clinical presentations, laboratory findings, and specific diagnoses of 209 patients evaluated for presumptive Lyme disease at a university-based Lyme disease clinic. We provide data on the use of health services, treatment-related illness, and levels of disability and psychological distress among patients who were inappropriately given a diagnosis of or treated for Lyme disease.

Methods

Top Methods Results Discussion Author & Article Info References

Study Patients

All patients were evaluated by faculty physicians at the Yale University Lyme Disease Clinic (YLDC) between April 1994 and May 1995. Patients presented with a presumptive diagnosis of Lyme disease previously made by referring physicians, the patients themselves, or both. Data were obtained on all patients who spoke English, provided informed consent, completed a 50-item questionnaire, underwent a comprehensive physical examination and serologic testing for Lyme disease, and agreed to receive a follow-up telephone call. The human investigations committee at Yale University School of Medicine approved the study.

Data Collection

A 50-item questionnaire was mailed to all patients before their clinic appointments, along with a letter describing the purpose of the study. To ascertain the quantity of health services used, patients were asked to record the number of office visits and physicians seen for the evaluation and treatment of Lyme disease, the number of previously performed serologic tests for Lyme disease, and information about each previous treatment. To determine whether a patient had previously received a diagnosis of Lyme disease from a physician, we asked, "Has a physician ever told you that you had Lyme disease?" We also asked patients about the relative effectiveness of each antibiotic treatment previously received for Lyme disease.

The number of work days lost and days during which normal activity was restricted were determined by using questions from the National Center for Health Statistics Limitations of Activities survey [14]. A score of 16 or more (range, 0 to 28) on the previously validated 7-item Center for Epidemiologic Studies of Depression scale [15] indicated the presence of depressive symptoms. Perceived stress was measured with the 10-item Perceived Stress Scale [16]. This index reliably measures the extent to which patients perceive situations in life as stressful [16-18]. A high level of perceived stress as measured by this index has been associated with clinically relevant outcomes, such as acute upper respiratory tract infections [18]. A score of 20 or greater (range, 0 to 40) was considered indicative of a high level of perceived stress.

To identify adverse drug events, we asked patients to report the name of the drug used, the duration of therapy, the mode of administration, and whether any unanticipated reactions occurred with each course of antibiotic treatment received for Lyme disease. The last item was ascertained by asking, "During or after treatment with this specific antibiotic, did you experience any of the following reactions?" Response categories included common drug reactions caused by antibiotic medications currently used to treat Lyme disease. Patients were instructed to check off any reaction they had during or after each course of treatment. If patients thought that an antibiotic treatment had caused a reaction that was not listed, they were instructed to record those reactions as well.

Diagnostic Criteria

A diagnosis of active Lyme disease required the presence of erythema migrans or characteristic articular, neurologic, or cardiac manifestations [1, 2] with confirmation by serologic testing. Patients with previous Lyme disease fulfilled criteria for active Lyme disease in the past, had completed recommended antibiotic therapy appropriate for their stage of disease [19], and did not meet the criteria for active Lyme disease at the time of evaluation. Patients were considered to have no evidence of Lyme disease if no current or previous physician had documented erythema migrans, the medical history was inconsistent with previous infection, and the criteria for active Lyme disease were not met.

All patients who met the criteria for active Lyme disease were informed that they had Lyme disease, whereas all patients with previous Lyme disease or no evidence of Lyme disease were told that they did not currently have Lyme disease. Alternative diagnoses for patients in the latter two groups were made on the basis of the clinical judgment of YLDC physicians.

Classification of Adverse Drug Events

A reported drug reaction was counted as a probable adverse drug event if the antibiotic in question had been previously cited [20, 21] as causing the putative reaction, the patient was not taking other medications that may have caused the adverse event, and a temporal association could be seen between the antibiotic treatment and the adverse event (for example, the reaction occurred while the patient was taking the antibiotic or within 2 weeks of discontinuing therapy with the drug). We also required that major adverse events be documented in the patient medical record.

We classified adverse drug events as minor or major. A minor event was defined as a reaction that was self-limited and did not require discontinuation of antibiotic therapy (such as epigastric distress) or that required physician intervention (such as treatment for vaginal yeast infection) but was not serious or life-threatening. A major event was defined as any drug reaction that was potentially serious or life-threatening (such as neutropenia or septic thrombophlebitis).

Reliability of Determinations of Adverse Drug Events and Patient Survey Responses

To ascertain the reliability of determinations of adverse drug events, surveys from 23 patients (5 with active Lyme disease, 6 with previous Lyme disease, and 12 with no evidence of Lyme disease) were randomly selected and independently rated by a second investigator. The proportion of observed agreement between raters was 89% ( = 0.73), indicating excellent overall agreement in the classification of adverse drug events.

To assess the reliability of patient survey responses, a second questionnaire consisting of 12 items randomly selected from the original 50-item survey was administered to 18 consecutive patients at the time of their YLDC visit. The abbreviated questionnaire contained 2 items pertinent to the results in this report (number of previous physician visits and serologic tests performed for Lyme disease). The proportions for observed agreement for these items were 72% and 83%, respectively, with corresponding intraclass correlation coefficients of 0.80 and 0.81. These results indicate substantial intrapatient reliability.

Clinical and Serologic Evaluation

For all patients, a comprehensive history was taken, medical records were reviewed, and a complete physical examination was performed. Serologic testing for Borrelia burgdorferi antibodies was done in all patients by enzyme-linked immunosorbent assay (ELISA) at the YLDC laboratory as described elsewhere [22]. A titer of 1:200 or greater for IgM or IgG was considered a positive result. All specimens with positive ELISA results were subsequently tested for IgM and IgG antibodies in our laboratory by doing Western blot analysis using a two-step procedure [23] and were evaluated according to Centers for Disease Control and Prevention criteria [24]. The most common reasons for conducting Western blot tests were suspicion of a false-positive result on ELISA (n = 88) or Western blot done at outside laboratories (n = 21).

Symptom Duration

We defined symptom duration as the time from the initial onset of symptoms to the patients' initial YLDC visit. We ascertained the time of initial onset of symptoms by asking patients, "When did you first develop signs or symptoms that suggested a diagnosis of Lyme disease?" Because some patients reported multiple episodes of Lyme disease interspersed with symptom-free periods, all patients were asked to estimate the number of symptom-free days from the time of their initial physician evaluation. We subtracted the total number of symptom-free days before calculating the symptom duration for all patients.

Patient Follow-up

All study patients received a telephone call 3 to 7 months after their YLDC visit and were asked whether they still had the symptoms that prompted the visit. Asymptomatic patients were asked to provide the approximate date on which their symptoms resolved. We also asked patients whether they had obtained additional medical evaluation or antibiotic treatment for Lyme disease and whether they agreed with the diagnosis assigned by their YLDC physician.

Results

Top Methods Results Discussion Author & Article Info References

Study Patients

Of 269 consecutive patients evaluated during the 13-month study period, 232 (86%) agreed to participate. Eighteen patients were excluded because serologic testing for Lyme disease had not been performed, and follow-up was not done in 5 patients; therefore, 209 patients (78%) were available for analysis. Nonparticipating (n = 37) and participating patients did not differ in sociodemographic or clinical attributes.

Assigned Diagnoses

Active Lyme Disease

Forty-four patients (21%) met the criteria for active Lyme disease. The median age of these patients was 40 years, slightly more than half were women, and most had high levels of education (Table 1). Most patients with active Lyme disease were referred by a physician (63%), 28% were self-referred, and 9% cited other sources of referral. The median duration of symptoms for patients with active Lyme disease was 1 month. On presentation, 31% had a positive test result for Lyme disease from an outside laboratory, 68% had previously received a diagnosis of Lyme disease, and 77% had recently completed or were currently receiving antibiotic therapy.

Of the 44 patients with active Lyme disease, 33 (75%) had erythema migrans on presentation and a recent history of flu-like symptoms or tick bite; 5 (11%) presented with distinctive neurologic or articular symptoms, had documentation of erythema migrans in their medical records, and were seropositive for Lyme disease as tested by our laboratory; and 6 (14%) had active Lyme arthritis and positive serologic test results at our evaluation.

Previous Lyme Disease

Forty patients (19%) fulfilled the criteria for previously active Lyme disease. Patients had a median age of 41 years, were predominantly women (67%), and had high levels of education (Table 1). Fifty-six percent were referred by a physician, 30% were self-referred, and 14% reported other sources of referral. The median duration of symptoms was 19 months. Eighty percent of patients had previous serologic evidence of infection, and all had received a diagnosis of Lyme disease from a physician.

At the time of the YLDC evaluation, all patients with previous Lyme disease had completed at least one full course of antibiotic therapy (Table 1). Patients received a total of 130 courses of antibiotic therapy (113 oral courses and 17 parenteral courses) and reported partial resolution of symptoms after 42% of these treatments, complete resolution followed by the emergence of new or similar symptoms after 36% of the treatments, or no change in symptoms after 22% of the treatments (data not shown).

Twenty-four patients with previously active Lyme disease (60%) sought evaluation because of nonspecific fatigue with arthralgias or myalgias, 4 (10%) reported focal joint symptoms, 6 (15%) had single or multiple somatic symptoms, and 6 (15%) cited other reasons for evaluation. Fifteen patients (38%) had positive results for Lyme disease on ELISA in our laboratory, but only 8 had positive results on Western blot. The fatigue-arthralgia-myalgia syndrome was the final diagnosis for 21 (53%) patients (Table 2). This symptom complex was present in 14 patients shortly after completion of antibiotic therapy and may have been a post-Lyme disease sequela. In the remaining 7 patients, the symptoms were more chronic, developed months to years after completion of antibiotic treatment, and were consistent with a fibromyalgia syndrome. Nine patients (23%) were thought to be depressed, 4 (10%) had osteoarthritis or rheumatoid arthritis, 2 (5%) were thought to have a viral syndrome, 3 (8%) received a diagnosis of peripheral neuropathy, and 1 (2%) had Graves disease (Table 2).

No Evidence of Lyme Disease

One hundred twenty-five patients (60%) lacked evidence of previous or current infection. Patients were young and predominately female, had high levels of education, and had similar patterns of referral (Table 1). The median duration of symptoms was 13 months. Sixty-one percent had previously had a positive test result for Lyme disease, and 70% had been previously told by a physician that they had Lyme disease.

Ninety-four patients (75%) reported previous use of antibiotics for Lyme disease (Table 1) and received a total of 232 courses of antibiotic treatment (171 oral courses and 61 parenteral courses [data not shown]). Patients with no evidence of Lyme disease reported partial resolution of symptoms after 57% of treatments, no change in symptoms after 34% of treatments, and worsening of symptoms after 9% of treatments.

Of the 125 patients with no evidence of Lyme disease, 43 (34%) sought evaluation because of fatigue in combination with arthralgias or myalgias, 31 (25%) had focal joint problems, 27 (21%) had one or more somatic symptoms, 7 (6%) had deer tick bites, and 17 (14%) cited other reasons for evaluation (Table 3). Only 2 patients (1.6%) had positive results on ELISA in our laboratory; both of these patients had negative results on Western blot. Thirty-nine patients (31%) received a diagnosis of the fatigue-arthralgia-myalgia syndrome, 20 (16%) were thought to have primary depression, 7 (6%) had asymptomatic deer tick bites, 29 (23%) had osteoarthritis or rheumatoid arthritis, 3 (2%) had evidence of bursitis, and 27 (22%) were thought to have other disorders.

Other Outcomes in Patients with Previous Lyme Disease and Patients with No Evidence of Lyme Disease

The following sections provide data only for patients with previous Lyme disease and patients with no evidence of Lyme disease. Primary outcomes are not reported for patients with active Lyme disease because of clinically important differences in the duration of illness; these data are available upon request.

Use of Health Services

Table 4 shows the median number and range of health services for the evaluation or treatment of Lyme disease used by patients with previous Lyme disease and patients with no evidence of Lyme disease. Patients in the two groups saw equal numbers of physicians (4 and 4). Compared with patients who had no evidence of Lyme disease, patients with previous Lyme disease reported more office visits (7 and 11), the same number of serologic tests (4 and 4), and more days of antibiotic use (42 and 75). The distribution of antibiotic use shows that 28% of patients with previous Lyme disease and 31% of patients with no evidence of Lyme disease received antibiotic therapy for more than 100 days (Table 4). Eleven patients with previous Lyme disease (28%) and 28 patients with no evidence of Lyme disease (22%) received at least one course of parenteral therapy; for both subgroups combined, the median duration of parenteral treatment was 28 days (range, 14 to 196 days).

The use of health services did not vary by pattern of referral among the two groups of patients: That is, self-referred patients did not report greater use of health services than physician-referred patients and patients referred by other sources. However, patients with no evidence of Lyme disease who previously received a physician diagnosis of Lyme disease had a substantially greater median number of office visits (8 and 4) and days of antibiotic use (63 and 0) than patients in this diagnostic category who had not been previously told by a physician that they had Lyme disease.

Treatment-Related Illness

Patients in both groups reported a total of 199 minor adverse drug reactions; of these, 162 (81%) were classified as probable adverse drug events (Table 5). Most of the 37 disqualified reports were rejected because they were not recognized as adverse drug reactions in standard references [17, 18]. Twenty-one patients with previous Lyme disease (53%) and 69 patients with no evidence of Lyme disease (55%) had at least one minor adverse drug event, including diarrhea, yeast infections, and gastrointestinal irritation (Table 5).

Of the 13 major adverse drug reactions reported, 10 (77%) were classified as major adverse drug events and 3 were disqualified because of a lack of documentation. The prevalence of major adverse drug events varied from 8% among patients with previous Lyme disease to 6% among patients with no evidence of Lyme disease. Major events occurred only in patients who received parenteral therapy and included antibiotic-associated (Clostridium difficile) colitis (4 cases), septic thrombophlebitis (2 cases), neutropenia (2 cases), serum sickness (1 case), and jaundice (1 case).

Disability and Psychological Distress

The proportion of days on which patients were unable to perform normal activities, adjusted for the total number of symptomatic days, ranged from 3% to 18% among patients with previous Lyme disease and 2% to 16% among patients with no evidence of Lyme disease. The prevalence of symptoms of depression and high levels of perceived stress were substantial in both groups: The proportion of patients with positive scores on the Center for Epidemiologic Studies of Depression scale ranged from 38% among patients with previous Lyme disease to 42% among patients with no evidence of Lyme disease, and 52% of patients in the former group and 45% of patients in the latter group were found to have high levels of perceived stress. The level of disability and distress did not vary by referral pattern. Among patients with no evidence of Lyme disease, however, those who had previously received a physician diagnosis of Lyme disease were far more likely than those who had not received a physician diagnosis to have symptoms of depression (51% and 18%, respectively) and high levels of perceived stress (52% and 32%).

Patient Follow-up

The median interval between patients' YLDC evaluations and telephone follow-up was 4 months (Table 6). Selected follow-up data on patients with active Lyme disease are provided for comparative purposes. About one quarter of patients with active Lyme disease reported persistent symptoms at follow-up; the most notable symptoms were fatigue, arthralgias, and myalgias. Among the 34 patients with active Lyme disease who were asymptomatic at follow-up, 82% reported complete resolution of symptoms by 12 weeks. In contrast, 71% of patients with previous Lyme disease and 82% of patients with no evidence of Lyme disease reported persistent symptoms at follow-up. Moreover, only 47% of patients with previous Lyme disease and 44% of patients with no evidence of Lyme disease agreed with the diagnosis assigned by their YLDC physicians (Table 6). Among patients in the latter two groups who disagreed with their diagnosis, all reported persistent symptoms and most expressed concern that their symptoms could be attributable to ongoing Lyme disease. Almost all patients with active Lyme disease (98%) agreed with their assigned diagnoses.

Table 6 also shows that 31% of patients with previous Lyme disease and 20% of patients with no evidence of Lyme disease obtained additional medical evaluations for Lyme disease during the follow-up period, and 21% and 11%, respectively, received additional antibiotic therapy.

Discussion

Top Methods Results Discussion Author & Article Info References

In our study, only 21% of patients evaluated at the YLDC had active Lyme disease. Most study patients had previous but not currently active Lyme disease (19%) or lacked any evidence of previous or current infection (60%). Our findings are consistent with those of other studies [6-9] and suggest that in endemic areas, both early and late Lyme disease may be frequently overdiagnosed and overtreated.

Among patients with previously active Lyme disease, most had clinical histories and physical examination findings consistent with the post-Lyme disease syndrome or fibromyalgia. These patients reported substantial use of health services for the evaluation or treatment of Lyme disease: The median number of days spent receiving antibiotic treatment and total number of office visits were 75 and 11, respectively. Some of this use was appropriate because these patients had had active Lyme disease in the past. We contend, however, that health services utilization by patients with previous Lyme disease was excessive and contributed to unnecessary treatment-related illness. Patients with previous Lyme disease received a total of 130 courses of antibiotic therapy (an average of 3.25 courses per patient), more than half (53%) had at least one minor adverse drug event, and 8% had a major adverse drug event. We did not survey treating physicians to determine their reasons for administering repeated courses of antibiotic therapy to these patients. Patient survey responses, however, indicate that most recurrent therapy was probably administered after nonspecific symptoms improved but did not resolve completely. Because no evidence shows that protracted antibiotic therapy hastens the resolution of these symptoms, the risks and costs associated with this practice far outweigh any theoretical benefits that may accrue.

In our study, patients with previous Lyme disease reported substantial disability and distress: After adjustment for the total number of symptomatic days, the proportion of days during which normal activities could not be done was 18%, 38% of patients had symptoms of depression, and 52% of patients had high levels of perceived stress. Lack of complete resolution of symptoms despite protracted courses of antibiotic therapy may have contributed to the high prevalence of depression and stress among this group of patients.

Most study patients (60%) lacked any evidence of previous or active Lyme disease. Most of these patients had previously been told that they had Lyme disease, and 75% had received at least one course of antibiotic therapy. Approximately one third of these patients were thought to have a non-specific fatigue-arthralgia-myalgia syndrome. However, more than 50% had treatable disorders, such as depression, rheumatoid arthritis, bursitis, and myasthenia gravis (Table 3). We think that failure to diagnose and treat these conditions represents an additional type of treatment-related illness in this group of patients.

Inappropriate use of health services was substantial among patients with no evidence of Lyme disease. These patients had a median of four serologic tests, seven office visits, and 42 days of antibiotic treatment for Lyme disease. Antibiotic treatment resulted in avoidable treatment-related illness: More than half (55%) of these patients had at least one minor adverse drug event, and 6% had a major adverse drug reaction.

We did not ascertain physicians' reasons for administering antibiotic treatment to patients with no evidence of Lyme disease. Because more than 60% of these patients had at least one positive serologic test result and most came from areas in which Lyme disease is endemic, we suspect that physician reliance on laboratory test results and patient demand for treatment were responsible for a substantial portion of the inappropriate diagnosis and treatment.

High levels of disability and distress were found among patients with no evidence of Lyme disease: Symptoms of depression and stress were present in 42% and 45% of these patients, respectively. In this group of patients, the prevalences of depression and stress were substantially higher among those who had previously received a physician diagnosis of Lyme disease (51% and 18%) than among those who had diagnosed the condition themselves (52% and 32%). We think that many patients and physicians incorrectly view Lyme disease as a chronic and often incurable illness that requires multiple courses of antibiotic treatment to cure or control disease. A physician diagnosis as opposed to self-diagnosis of Lyme disease may reinforce this erroneous belief and contribute to increased depression and stress when the patient's symptoms do not improve despite protracted courses of antibiotic therapy.

The patient's belief in the chronicity of Lyme disease and the need for multiple courses of antibiotic treatment was evident in patients with previous Lyme disease and patients with no evidence of Lyme disease. Although these patients were informed by YLDC physicians that they did not have Lyme disease and would not benefit from additional antibiotic therapy, 20% to 31% obtained additional evaluations for Lyme disease and 11% to 21% received additional antibiotic treatments during the follow-up period. In addition, more than 50% of these patients disagreed with the diagnosis assigned by their YLDC physicians, and most expressed concern that they may have had ongoing Lyme disease infection.

Although we focused on patients who did not have active Lyme disease, it is important to note that among patients with active Lyme disease, 84% received oral antibiotic therapy for 28 days or less and 77% were free from symptoms at follow-up. This therapeutic success supports the findings of earlier studies [3, 4] and suggests that a single course of antibiotic therapy results in complete resolution of symptoms and cure for most patients with Lyme disease. The remaining patients with active Lyme disease (23%) reported persistence of nonspecific fatigue, arthralgias, and myalgias at follow-up. This symptom complex-the post-Lyme disease syndrome-is common [3, 5, 25] and may take several months to resolve; however, the complex does not respond to repeated courses of antibiotic therapy [8, 26].

We think that our findings correctly reflect a widespread phenomenon in the current management of patients suspected of having Lyme disease. However, our study has several possible limitations. First, patients were drawn from a large university-based Lyme disease clinic. The most refractory patients with previous Lyme disease or no evidence of Lyme disease may have been self-referred or sent by their physicians for evaluation. Although we cannot exclude this possibility, such referral bias cannot discount the significant degree of avoidable treatment-related illness in and excessive use of health services by these patients. Second, we used self-reported data to ascertain primary outcomes. Although patients were consistent in their responses about previous use of health services, as determined by test-retest appraisals, we did not establish the validity of self-reported information (with the exception of major adverse drug events). Finally, given our study design, it is impossible to know the extent to which the high prevalence of depression and stress seen in patients with previous Lyme disease or with no evidence of Lyme disease predated the "diagnosis" of Lyme disease or whether misdiagnosis or treatment contributed to the prevalence of these conditions. Although the proportion of these patients with recorded diagnoses of depression or stress disorders was less than 2%, it is possible that these conditions were not caused by but rather preceded the "diagnosis" of Lyme disease.

The following recommendations may help avoid inappropriate use of health services and adverse clinical outcomes related to the overdiagnosis and overtreatment of Lyme disease. First, serologic tests should be used only to confirm a well-formulated clinical suspicion of Lyme disease. Even in highly endemic regions, the incidence of this infectious disease is low [27]; therefore, most positive results for Lyme disease are falsely positive. Second, the use of protracted courses of antibiotic therapy should be avoided because active Lyme disease is effectively treated, with few exceptions, by a single course of antibiotics [3, 4]. Third, patients with active Lyme disease should be assured that complete resolution of symptoms is expected but may not occur for several months. Fourth, among patients who present with nonspecific symptoms, a diagnosis of other treatable conditions (such as depression or inflammatory arthritis) should be considered; Lyme disease rarely, if ever, presents in this manner. Finally, for many patients who present with a presumptive diagnosis of Lyme disease and lack evidence of active infection, customary reassurance may not suffice. As we show, simply removing the label of "Lyme disease" was not an effective therapeutic intervention for most patients with previous Lyme disease or no evidence of Lyme disease. Future research should be directed toward more effective therapies for these patients.

Drs. Schoen, Evans, and Horwitz: Department of Medicine, Yale University School of Medicine, 333 Cedar Street, New Haven, CT 06510.

Ms. Rosenberg: Northwestern University Medical School, 303 East Chicago Avenue, Chicago, IL 60611.