Clinical evaluation of guidelines and two-test approach for Lyme disease

Transcription

1 Rheumatology 1999;38: Clinical evaluation of guidelines and two-test approach for Lyme disease A. A. M. Blaauw, A. M. van Loon1, J. F. P. Schellekens2 and J. W. J. Bijlsma Department of Rheumatology and Clinical Immunology and 1Department of Virology, University Medical Centre, PO Box 85500, 3508 GA Utrecht and 2National Institute of Public Health and the Environment, Diagnostic Laboratory for Infectious Diseases and Perinatal Screening, 3720 BA Bilthoven, The Netherlands Abstract Objective. The diagnosis of Lyme disease should be based on objective clinical signs and symptoms. In a clinical study, we have evaluated whether the recommended two-step approach for serodiagnosis of Lyme disease is useful in daily clinical practice and can influence clinical decision making. Methods. The signs and symptoms of patients with ongoing musculoskeletal complaints, assumed by their referring physician or themselves to be attributable to active or chronic Lyme disease, and of patients diagnosed as having Lyme disease, were evaluated. On the basis of clinical evaluation only, patients were classified into three groups: previous Lyme disease, active Lyme disease and no Lyme disease. Antibodies to Borrelia burgdorferi were determined by means of an enzyme-linked immunosorbent assay ( ELISA), followed, when positive, by immunoblotting. Results. One hundred and three patients (41 males and 62 females, mean age 48.7 yr) participated in the study. Of the 49 patients classified as previous Lyme disease, 25 (51%) had antibodies to B. burgdorferi. All 10 patients with active Lyme disease had positive antibodies and 12 of the 44 patients (27%) classified as no Lyme disease had positive antibodies. No statistically significant differences were found between the percentage of positive immunoblots from patients with previous Lyme disease (72%) and patients with active Lyme disease (100%). In the group of no Lyme disease, five out of 12 patients had a negative immunoblot. Concerning serological testing, immunoblotting could have added additional information. However, immunoblotting did not influence clinical decision making in this group of patients. Conclusion. Immunoblotting did not influence clinical decision making for the 47 patients with antibodies to B. burgdorferi in this study. KEY WORDS: Lyme disease, Lyme serology, Immunoblot. and myalgia. This has led to overdiagnosis of Lyme disease [3 5]. In Europe, no clinical diagnostic criteria for Lyme disease have been developed so far. Criteria developed in the USA by the Centers for Disease Control and Prevention (CDC) for surveillance purposes are often used as clinical criteria in the USA and in Europe. Since it is rather difficult to culture B. burgdorferi from speci- mens other than erythema migrans lesions, especially in daily clinical practice, the presence of antibodies to B. burgdorferi may confirm the clinical diagnosis. Although serological testing for Lyme disease can be performed with a high degree of sensitivity, falsenegative and false-positive results continue to be an important problem [3]. Patients with Lyme disease usually remain seropositive for IgG antibodies to B. burgdorferi for years, in some cases even permanently. The clinical manifestations of Lyme disease have been well documented since its first description as a distinct entity in 1977 [1, 2]. In Europe, at least three distinct species of Borrelia burgdorferi sensu lato can cause the clinical syndrome of Lyme disease: B. burgdorferi sensu stricto, B. garinii and B. afzelii. Clinical manifestations correlate with the genospecies causing the infection. The clinical diagnosis of Lyme disease should be based on objective clinical signs and symptoms. Too often, the diagnosis is made in patients with non-specific and atypical features, such as headache, fatigue, arthralgia Submitted 8 August 1998; revised version accepted 25 May Correspondence to: A. A. M. Blaauw, Department of Rheumatology and Clinical Immunology, F02.127, University Medical Centre, PO Box 85500, 3508 GA Utrecht, The Netherlands British Society for Rheumatology

2 1122 A. A. M. Blaauw et al. In these patients, a positive test can lead to overdiagno- with musculoskeletal complaints diagnosed as Lyme sis and probably overtreatment. False-positive results disease by our staff members. have been reported for patients with rheumatoid arth- All patients were seen by one of us (AAMB) who ritis, systemic lupus erythematosus, infectious mononucleosis, recorded a medical history and performed the physical syphilis and other spirochaetal diseases [ 7]. In examination. Questions were asked about past or recent Europe, asymptomatic seropositivity determined by tick exposure and the characteristic manifestations of enzyme-linked immunosorbent assay ( ELISA) and Lyme disease. Clinical records and former medical correspondence immunofluorescence assays ( IFA) has also been demonstrated were studied. in up to 20% of the normal population as well In the daily routine of our department, it is not as people at risk, such as orienteers, hunters and foresters possible to culture B. burgdorferi from clinical speci- [8 11]. As a consequence, in the absence of typical mens. Genome amplification methods and urine antigen clinical signs and symptoms, diagnostic testing for antibodies analysis are only used in experimental studies. It was to B. burgdorferi is of limited value [10 14]. therefore decided that clinical criteria were to be used Although no consistent evidence suggests that Western as the gold standard. Patients were classified according blotting adds additional information to an accurate to clinical criteria independent of the serological test clinical history or to the results of a positive or indeterminate results of ( ELISA) antibodies to B. burgdorferi both at ELISA or indirect IFA if used in all patients, the time of (presumed) diagnosis and during the study. Dressler et al. and the CDC suggest that a two-step Because of the special referral pattern of patients seen testing strategy should be used [13, 15, 16]. This two- in our department, we expected that if these patients step strategy is also adopted in Europe. Western blotting had antibodies to B. burgdorferi, these would be IgG can detect and discriminate among antibodies to mul- antibodies. IgM antibodies were not determined. If tiple antigens of B. burgdorferi. However, it remains patients had ( ELISA) IgG antibodies to B. burgdorferi, unclear whether Western blotting can discriminate an IgG immunoblot was performed. between true-positive test results and false-positive test No exact figures about the incidence and prevalence results, and between active or previous infection with of Lyme disease in The Netherlands exist. The incidence B. burgdorferi [17, 18]. Therefore, the clinical implications is assumed to be low and is estimated to vary from of this recommended two-step approach for the 0.01% in the general population to 2 9% in populations serodiagnosis of Lyme disease remain unclear and even at risk and patients with monoarticular arthritis questionable. [8, 10, 19]. In this clinical study, we studied a patient population in which questions such as a true- or false-positive test Clinical criteria result and active or previous infection are important. On the basis of objective clinical signs and symptoms Clinical criteria were used as a gold standard, and only, the patients were divided into three groups: previ- ELISA and immunoblot test results were used and ous Lyme disease, active Lyme disease and no Lyme interpreted as in daily clinical practice. We have tried to disease. assess whether immunoblotting should be performed in The clinical criteria for Lyme disease according to all patients with a positive ELISA and whether the literature references were defined as follows [6]: eryth- suggested two-step approach can influence clinical ema migrans is defined as a red macula or papule that decision making. expanded over a period of days to weeks to form a large annular lesion, at least 5 cm in size, often with partial central clearing [1, 20, 21]. The skin lesion should have Patients and methods been confirmed by a physician or recognized by the Patients patient from pictures of erythema migrans [10]. Early neurological involvement includes lymphocytic meningitis, Consecutive patients of the out-patient clinic of the cranial neuritis and radiculopathy, accompanied Department of Rheumatology and Clinical Immunology by pleocytosis of the cerebrospinal fluid [22 24]. of the University Medical Centre Utrecht, The Chronic neuroborreliosis includes encephalopathy with Netherlands, were evaluated. Patients with a classic memory impairment, polyneuropathy with radicular presentation of Lyme disease such as erythema migrans pain or distal paraesthesiae and leucoencephalitis with are, in general, diagnosed and treated by their general spastic paraparesis [25]. Lyme arthritis is defined as practitioners. Because of our special interest in clinical recurrent, brief attacks of objective joint swelling in one Lyme disease, the department serves as a secondary or or a few joints, especially the knees, possibly followed tertiary referral centre for patients with presumable by chronic synovitis [26]. Lyme carditis is characterized Lyme disease or for patients with persisting complaints by fluctuating degrees of atrioventricular nodal block thought to be chronic Lyme disease. that resolved in days to weeks, possibly associated with Patients eligible for this study were (i) patients with myocarditis [27]. Uveitis is defined as intermediate persisting musculoskeletal complaints assumed by their uveitis with characteristic spider web vitritis confirmed referring physicians to be attributable to active or by an ophthalmologist [28]. Acrodermatitis chronica chronic Lyme disease, (ii) patients who believed that atrophicans ( ACA) is the characteristic bluish-red they had active or chronic Lyme disease and (iii) patients discolouration, often with a doughy infiltration and

3 Two-test approach for Lyme disease 1123 progressing to atrophy or sclerodermic changes [ 21]. late Lyme disease; the specificity of IgG immunoblot is Borrelial lymphocytoma is the typical bluish-red 98% (unpublished data, personal communication). The tumour-like infiltrate [21]. For all patients with neuro- control population in which IgG immunoblot was tested logical, cardiac, eye or joint abnormalities, all other consisted of healthy donors, pregnant women and poten- possible causes of the complaints had to be excluded. tially cross-reactive sera of patients with syphilis, leptospirosis Skin lesions that developed immediately after a tick bite or autoimmune diseases. The clinical diagnosis were not considered to be erythema migrans. Aspecific defined by a clinical expert was considered as the gold symptoms such as fatigue and arthralgia, fever, headache standard. and paraesthesiae as a single symptom, palpitations, In our hospital, the clinician receives the results of bundle branch block on electrocardiography, myocarditis immunoblotting interpreted by the laboratory as either and symmetrical polyarthritis were not accepted as positive, negative or equivocal. The clinician is unaware clinical criteria for Lyme disease unless accompanied by of which bands of the immunoblot are positive or objective characteristic Lyme manifestations mentioned negative. No follow-up samples were determined. above. A tick bite alone was also not accepted as a clinical criterion. Statistics A clinical diagnosis of previous Lyme disease was For categorical data, Fisher s exact test was used to test made if at least one of the clinical criteria was present for differences between groups. A P value of <0.05 was in the past and no objective manifestations of Lyme considered statistically significant. disease were present at the time of the study. Active Lyme disease was diagnosed when at least one of the clinical criteria was present at the time of the study. Results When no clinical criteria were present or had been In a 4 yr period (April 1994 April 1998), 105 patients present in the past, patients were assumed to have no who met either one of the three eligibility criteria were Lyme disease. seen in our department. In two patients, the ELISA for Enzyme-linked immunosorbent assay antibodies to B. burgdorferi was not performed at study entry. Of the remaining 103 patients, 41 (40%) were In daily clinical practice, our hospital uses a commercial male and 62 (60%) were female. The mean age of the ELISA for the detection of antibodies to B. burgdorferi patients was 48.7 yr (range 6 82 yr). (Dako A/S, Glostrup, Denmark) [29, 30]. In all cases, The 103 patients were grouped according to the the IgG antibody response to B. burgdorferi was deter- clinical criteria: 49 patients were classified as previous mined by this commercial ELISA. If a positive test Lyme disease (48%), 10 patients as active Lyme disease result was obtained, the Treponema pallidum haemagglu- (10%) and 44 as no Lyme disease (42%) (Table 1). tination assay was performed to exclude false positivity Forty-seven of the 103 patients ( 46%) had IgG antibod- due to antibodies to T. pallidum. ies to B. burgdorferi according to the ELISA at study Patients were tested at the time they were seen in our entry: 25 patients with previous Lyme disease (51%), all out-patient clinic. Because of our special patient popula- 10 patients with active Lyme disease (100%) and 12 tion, it is possible that patients are seen not only if they patients (27%) with no Lyme disease ( Table 1). Results have active symptoms of Lyme disease but also years are summarized in Table 2. after a (presumed) diagnosis of Lyme disease. No follow-up samples were determined. Patients classified as previous Lyme disease Western blot analysis No differences for gender or mean age were found for the 49 patients classified as previous Lyme disease who For all patients with IgG antibodies to B. burgdorferi did or did not have antibodies to B. burgdorferi. according to the ELISA assay, IgG immunoblot analysis Of the 25 patients with antibodies to B. burgdorferi, was performed as described in the literature [16, 30]. 15 had had early symptoms of Lyme disease defined as The IgG immunoblot was considered positive if at least signs or symptoms within <1 yr after possible exposure. five of the following eight bands were present: 17 kda, Fourteen of these patients had erythema migrans and 22 kda, 31 kda, 34 kda, 39 kda, 41 kda, kda one patient had neuroborreliosis. Ten patients had late and 92 kda, or if four bands were positive including symptoms defined as signs and symptoms >1 yr after 17 kda or 22 kda or 39 kda or 92 kda. One or more possible exposure. Two of these 10 patients had ACA, bands in the kda region were considered as one one had uveitis and seven had mono- or oligoarthritis band [31, 32]. IgG immunoblot was considered negative of the knees (six patients) and elbow (one patient). if three bands were positive excluding 17 kda, 22 kda, Of the 24 patients without antibodies to B. burgdorferi, 39 kda or 92 kda, and if less than three bands were 19 had erythema migrans, one had erythema migrans found to be positive. IgG immunoblot was considered and neuroborreliosis, and three had neuroborreliosis. equivocal if four positive bands (31 kda, 34 kda, One patient had late symptoms: arthritis of the knee. 41 kda and kda) were found, or only three bands Patients with antibodies to B. burgdorferi were seen including 17 kda or 22 kda or 39 kda or 92 kda. Using on average 3.9 yr after the diagnosis of Lyme disease the above criteria, the sensitivity of IgG immunoblot is was made elsewhere, patients without antibodies on 45% in early Lyme disease and 95% in disseminated or average after 4.6 yr. At study entry, none of these 49

4 1124 A. A. M. Blaauw et al. TABLE 1. Patient characteristics Number of IgG antibodies to Number of Gender Classificationa patients B. burgdorferi (ELISA) patients (%) male/female Mean age (S.D.) Previous Lyme disease 49 Positive 25 (51%) 9/ (15.45) Negative 24 (49%) 9/15 49 (13.82) Active Lyme disease 10 Positive 10 (100%) 3/7 53 (21.7) No Lyme disease 44 Positive 12 (27%) 7/5 40 (18.61) Negative 32 (73%) 13/ (15.8) Total 103 Positive 47 (46%) 19/28 48 (17.8) Negative 56 (54%) 22/34 49 (14.9) aclassification based on clinical criteria only (independent of serological results); see the text for a definition of criteria. TABLE 2. Characteristics of patients with IgG antibodies to B. burgdorferi (ELISA) and results of IgG immunoblot Number of patients Positive Total number with IgG IgG Classification of patients antibodies Early symptomsa Late symptomsb immunoblot Previous Lyme disease (72%) Active Lyme disease (100%) No Lyme disease (58%) Total (74%) aearly symptoms: like erythema migrans and neuroborreliosis within <1 yr of possible exposure. blate symptoms: like arthritis and ACA >1 yr after early symptoms of possible exposure. patients had objective signs of Lyme disease. In 51% of these patients, antibodies to B. burgdorferi could be detected after all these years. Patients with active Lyme disease Ten patients were classified as having active Lyme disease based on clinical symptoms: one patient had neuroborreliosis with radiculopathy and lymphocytic meningitis, three patients had ACA, six patients had episodes of recurrent arthritis of the knee, one of them after erythema migrans. All patients received appropriate antibiotic treatment. All patients with active Lyme disease had IgG antibodies to B. burgdorferi. All patients contracted Lyme disease in Europe, except for the patient with neuroborreliosis who developed signs and symptoms after a trip to the east coast of the USA. Patients with no Lyme disease According to our clinical criteria, 44 patients did not have objective signs of Lyme disease at study entry. Based on clinical evaluation and chart and correspondence review, they did not have objective signs or symptoms either, at the time a presumable diagnosis of Lyme disease was made elsewhere. Thirty-two patients did not have IgG antibodies to B. burgdorferi at the time of study entry. However, 12 of these 44 patients (27%) did have IgG antibodies to B. burgdorferi. Immunoblot An IgG immunoblot was performed in all 47 patients with IgG antibodies to B. burgdorferi. Of the 25 patients classified as previous Lyme disease, 18 had a positive immunoblot ( 72%) and seven had a negative immunoblot. The seven patients with a negative immunoblot did have objective signs and symptoms of Lyme disease in the past. ELISA test results of these patients cannot be regarded as false positive in these seven patients. Because of the time interval between the diagnosis of Lyme disease and the time of the study, a waning humoral immunity with a rest-response to the 41 kda protein is possible. All 10 patients with active Lyme disease had a positive immunoblot (100%). No statistically significant difference was found between the percentage of positive immunoblots from patients with previous Lyme disease and those with active Lyme disease. Based on clinical criteria, however, it was possible to differentiate between previous and active Lyme disease. Therefore, the use of the immunoblot assay in these two patient groups did not add any additional information to the results of the ELISA, and did not influence clinical decision making. In the group of no Lyme disease, seven of 12 patients (58%) had a positive immunoblot and five patients (42%) had a negative immunoblot. Concerning serological testing, immunoblotting could have added additional information for five patients in this group of no Lyme disease: ELISA results were probably false positive. However, based on the clinical criteria (independent of any serological results), these patients did not have signs of Lyme disease. Therefore, immunoblotting did not influence clinical decision making for this group of patients. Seven patients classified as no Lyme disease had positive ELISA IgG antibodies to B. burgdorferi and a positive IgG immunoblot. Clinical characteristics of these patients are discussed briefly. A 37-yr-old male had systemic sclerosis and arthralgia. Because of anti- bodies to B. burgdorferi, he was treated with several courses of ceftriaxone i.v. without any clinical effect. A 40-yr-old male had symmetrical, rheumatoid factor-

5 Two-test approach for Lyme disease 1125 negative polyarthritis. Treatment with doxycycline and of Lyme disease. No generally accepted criteria have ceftriaxone had no effect. He was started on sulphasalazine been developed, although practice parameters and as a second-line anti-rheumatic drug with excellent guidelines for the diagnosis of patients with Lyme response. A 72-yr-old male had symmetrical, rheumatoid borreliosis of the nervous system and Lyme arthritis factor-positive, erosive polyarthritis classified as typical have been proposed and developed [22, 35]. rheumatoid arthritis. A 56-yr-old male had eczema and Overdiagnosis as well as underdiagnosis of erythema arthralgia. He never had objective signs of arthritis. migrans, considered as the hallmark of the disease, have A 17-yr-old male was referred because of a tick bite in been described [36, 37]. For this particular study, we the past and antibodies to B. burgdorferi. He did not used diagnostic criteria based on the literature as well have any complaints or objective signs or symptoms. as our own experience with Lyme disease patients [2, 3, A 52-yr-old female had arthralgia without objective 5, 12, 13, 21, 22]. We grouped our patients on the basis arthritis. A 23-yr-old female had arthralgia and painful of these clinical criteria independently of the serological knees without signs of arthritis, classified as chondro- results, in an attempt to identify the possible surplus malacia patellae. value of serology and immunoblotting. Patients could The positive results of ELISA and immunoblot in easily be assigned to one of the three groups previous these seven patients have to be considered as true, but Lyme disease, active Lyme disease and no Lyme disease. asymptomatic, positive. On clinical criteria, we do not Second, our study was performed in a referral hospital feel that these patients have or had Lyme disease. with a population of patients which is not representative of the total population of patients with disease due to infection with B. burgdorferi. However, these patients Discussion are a reflection of the problems which are encountered In this study, we evaluated recommendations for in the daily management of patients with (presumed) the use of a two-test approach for Lyme disease in a Lyme disease: active infection, past infection with clinical situation. We especially assessed whether clinical persistent antibodies to B. burgdorferi, false-positive or decision making was influenced by the suggested asymptomatic IgG antibodies to B. burgdorferi. We have two-test approach: a positive or undetermined ELISA not provided any data on sensitivity, specificity, and the should be followed by Western blot. The results indi- positive or negative predictive values of the ELISA and cate that immunoblotting does not reliably discriminate immunoblotting, because these data would only be between previous infection and active infection with B. applicable to a similar population and cannot be generalized burgdorferi. Clinical decision making was not influenced for other populations. We used a standard commerburgdorferi. by the results. cial ELISA for this study as will be used in daily clinical Twelve out of 44 patients classified as no Lyme disease practice by most physicians. We are aware that more at study entry had IgG antibodies to B. burgdorferi. sensitive assays will be available in the future. However, This percentage is in contrast with the percentage of the possible surplus value of all these assays should be antibodies to B. burgdorferi found in the Dutch popula- evaluated against clinical criteria. tion and patients at risk [8, 10, 11]. This high percentage Third, based on clinical criteria, we classified 44 is probably due to referral bias. It is likely that patients patients as no Lyme disease. We used only objective with antibodies are referred, as patients without antibodies signs and symptoms in our criteria. None of these 44 are not. Seven of these 12 patients did have a positive patients ever had any of these symptoms. This would immunoblot. Based on their medical history and the have made their pre-test likelihood of Lyme disease clinical picture, these patients had not had Lyme disease. <20%. These patients should not have been tested for In this group of patients classified as no Lyme disease, the presence of antibodies to B. burgdorferi in the first the use of immunoblotting could have added additional place because a positive test is more likely to be clinically information for the five patients whose immunoblot was false positive than true positive [10, 13]. We assumed negative. However, based on clinical criteria, these that it is likely that referring physicians use a negative patients did not have Lyme disease at study entry or in result of ELISA for antibodies to B. burgdorferi to rule the past. Thus, in this group of patients with no Lyme out the presence of Lyme disease and refer patients with disease, the immunoblot did not influence clinical a positive test result to our out-patient clinic. decision making either. Our study supports and Fourth, we studied a relatively small patient population strengthens the suggestion that the diagnosis of Lyme with either positive or negative ELISA results. disease should be made primarily on clinical signs and Larger study populations are needed for definite conclusions symptoms [33]. Neither the results of serological testing and for patients with possible indeterminate nor the results of immunoblotting can be interpreted ELISA results. However, it can be expected that if adequately without knowledge of clinical manifestations. clinicians determine the pre-test probability of Lyme Our results confirm the findings of Gern et al. [34] who disease in the diagnostic evaluation of a patient for stated that immunoblotting is of little help in diagnosing Lyme disease, based on findings of a thorough clinical Lyme disease in populations at risk as well as in endemic examination and knowledge of the incidence of Lyme areas where seropositivity for B. burgdorferi is common. disease in the population represented by the patient, A few points should be stressed. First, there is no testing for antibodies to B. burgdorferi can be avoided gold standard for the clinical and laboratory diagnosis in patients with non-specific symptoms [13].

6 1126 A. A. M. Blaauw et al. In view of the results of this study, we cannot recomburgdorferi 18. Cooke WD, Bartenhagen NH. Seroreactivity to Borrelia mend that, in a clinical situation, a positive ELISA for antigens in the absence of Lyme disease. antibodies to B. burgdorferi should always be followed J Rheumatol 1994;21: by immunoblotting. Immunoblotting did not influence 19. Blaauw AAM, Dijkmans BAC, Bouma PAD, van der Linden S. Rational diagnosis and treatment in unclassified clinical decision making for 47 patients with positive arthritis: how clinical data may guide requests for Lyme ELISA antibodies to B. burgdorferi. Results of seroserology and antibiotic treatment. Ann Rheum Dis logical testing and immunoblot analysis should not be 1993;52: interpreted without sufficient knowledge of the clinical 20. Kuiper H. Erythema migrans in Nederland. [Erythema picture of the patients tested. migrans in the Netherlands: a clinical and epidemiological References study of 77 patients] Ned Tijdschr Geneeskd 1995;139: Berglund J, Eitrem R, Ornstein K et al. An epidemiologic 1. Steere AC, Malawista SE, Snydham DR et al. Lyme study of Lyme disease in Southern Sweden. N Engl J Med arthritis. An epidemic of oligoarticular arthritis in children 1995;333: and adults in three Connecticut communities. Arthritis 22. Halperin JJ, Logigian EL, Finkel MF, Pearl RA. Practice Rheum 1977;20:7 17. parameters for the diagnosis of patients with nervous 2. Steere AC. Lyme disease. N Engl J Med 1989;321: system Lyme borreliosis (Lyme disease). Neurology 3. Steere AC, Taylor E, McHugh GL, Logigian EL. The 1996;46: overdiagnosis of Lyme disease. J Am Med Assoc 23. Halperin JJ, Little BW, Coyle PK, Dattwyler RJ. Lyme 1993;269: disease: cause of a treatable peripheral neuropathy. 4. Sigal LH. Summary of the first 100 patients seen at a Lyme disease referral center. Am J Med 1990;88: Neurology 1987;37: Blaauw AAM, Bijlsma JWJ. Musculoskeletale klachten 24. Halperin JJ, Luft BJ, Volkman DJ, Dattwyler RJ. Lyme vaak ten onrechte aan Lyme-ziekte toegeschreven. [Lyme neuroborreliosis, peripheral nervous system manifesta- disease, overdiagnosis?] Ned Tijdschr Geneeskd tions. Brain 1990;113: ;141: Logigian EL, Kaplan RF, Steere AC. Chronic neurologic 6. Wharton M, Chorba TL, Vogt RL, Morse DL, Buehler manifestations of Lyme disease. N Engl J Med JW. Case definitions for Public Health Surveillance. 1990;323: MMWR Morbid Mortal Weekly Rep 26. Steere AC, Schoen RT, Taylor E. The clinical evolution 1990;39(RR-13): of Lyme arthritis. Ann Intern Med 1987;107: Magnarelli LA, Miller JN, Anderson JF, Rivere GR. 27. Sigal LH. Early disseminated Lyme disease: cardiac mani- Cross-reactivity of nonspecific treponemal antibody in festations. Am J Med 1995;98(suppl. 4A):25S 9S. serologic tests for Lyme disease. J Clin Microbiol 28. Rothova A, Kuiper H, Spanjaard L, Dankert J, Breebaart 1990;28: AC. Spiderweb vitritis in Lyme borreliosis. Lancet 8. Nohlmans MKE, van den Boogaard AEJM, Blaauw 1991;i: AAM, van Boven CPA. Prevalentie van Lyme borreliosis 29. Nohlmans MKE, Blaauw AAM, van den Bogaard AEJM, in Nederland. [Prevalence of Lyme borreliosis in the van Boven CPA. Evaluation of nine serological tests for Netherlands] Ned Tijdschr Geneeskd 1991;135: diagnosis of Lyme borreliosis. Eur J Clin Microbiol Infect 9. Fahrer H, van der Linden SM, Sauvain MJ, Gern L, Dis 1994; Zhiuoa A, Aeschlimann A. The prevalence and incidence 30. Rijpkema SJ, Groen J, Molkenboer M, Herbrink P, of clinical and asymptomatic Lyme borreliosis in a populaflagellum of Borrelia burgdorferi measured with an inhibi- Osterhaus A, Schellekens J. Serum antibodies to the tion at risk. J Infect Dis 1991;163: Blaauw AAM, Nohlmans MKE, van den Boogaard tion enzyme-linked immunosorbent assay are diagnostic AEJM, van der Linden S. Diagnostic tools in Lyme for Lyme borreliosis. Serodiagn Immunother Infect Dis borreliosis: clinical history compared with serology. J Clin 1994;6:61 7. Epidemiol 1992;11: Engstrom SM, Shoop E, Johnson RC. Immunoblot inter- 11. Kuiper H, de Jongh BM, Nauta AP et al. Lyme borreliosis pretation criteria for serodiagnosis of early Lyme disease. in Dutch forestry workers. J Infect 1991;23: J Clin Microbiol 1995;33: American College of Physicians. Guidelines for laboratory 32. Rijpkema SJ. Diagnosis and transmission of Lyme borreevaluation in the diagnosis of Lyme disease. Ann Intern liosis. Thesis, University of Utrecht, ISBN: Med 1997;127: Tugwell P, Dennis DT, Weinstein A et al. Laboratory 33. Sigal LH. The Lyme disease controversy. Social and evaluation in the diagnosis of Lyme disease. Ann Intern financial costs of misdiagnosis and mismanagement. Arch Med 1997;127: Intern Med 1996;156: Nichol G, Dennis D, Steere AC et al. Test-treatment 34. Gern L, Leuba-Garcia S, Frossard E. Characterization strategies for patients suspected of having Lyme disease: and follow-up of the IgG antibody response against a cost-effectiveness analysis. Ann Intern Med Borrelia burgdorferi using Western blot in a seropositive 1998;128: ( ELISA) population from an endemic area. Bull Soc 15. Anonymous. Recommendations for test performance and Neuchatoise Sci Nat 1993;116:5 14. interpretation from the Second National Conference on 35. Steere AC. Diagnosis and treatment of Lyme arthritis. serologic Diagnosis of Lyme disease. MMWR Med Clin North Am 1997;81: ;31: Feder HM, Whitaker DL. Misdiagnosis of erythema 16. Dressler F, Whelan JA, Reinhart BN, Steere AC. Western migrans. Am J Med 1995;99: blotting in the serodiagnosis of Lyme disease. J Infect Dis 37. Blaauw AAM, Schuwirth L, van der Vleuten C, Smits F, 1993;167: van der Linden S. How well do general practitioners, 17. Nohlmans MKE. Lyme borreliosis in The Netherlands. rheumatologists and dermatologists recognize Lyme borre- Thesis, University of Limburg, ISBN liosis? Arthritis Rheum 1992;35:S177 (Abstract).