One reason for such a poor performance is the sheer number of factors that can interfere with the accuracy of serological tests (see box). Also, a patient can be infected and have a variety of symptoms, yet show no antibodies to the bacteria in the blood, as antibody response can wax and wane over time. The spirochaetes can also be released into the blood and, for reasons as yet not understood, retreat into host body tissues until conditions are optimal for their spread and reproduction. These phenomena are known as seronegative disease.

In such patients, it is often possible to culture live spirochaetes from the blood (Infection, 1989; 17: 355-9). In one study of transplacental transmission, over half the mothers with adverse pregnancy outcomes and fetuses or infants with B. burgdorferi were sero negative (Rheum Dis Clin N Am, 1989; 15: 657-77).

Furthermore, while other tests such as urinalysis are not given routinely, examination of a patient's urine may reveal B. burgdorferi specific antigen in symptomatic patients given the all clear on the basis of seronegative test results (J Clin Microbiol, 1993; 31: 1961-3).

What all of this means, say Lyme disease specialists, is that a diagnosis of the infection simply cannot be made on the basis of blood tests alone. In fact, the US National Institutes of Health (NIH) guidelines are clear that Lyme disease is a clinical diagnosis one made on the basis of patient history and symptoms (Clin Courier, 1991; 9 [Aug]: 5-8).

However, there may be problems even with a clinical diagnosis unless the process is thorough. For instance, it has been widely assumed that all patients bitten by the tick will get a rash resembling a bull's eye, or erythema migrans. In fact, only 40-60 per cent of light skinned patients develop such a rash. On dark skinned subjects, it may look more like a bruise. Because the tick itself is so small and the rash is not a given, many people never know they have been bitten.

Even more frustrating is that an individual may be infected and yet remain asymptomatic, often for a long time. When scientists analysed blood samples obtained through the Blood Transfusion Service in Ireland, they found that the overall prevalence of positive samples in donors with no symptoms was 9.75 per cent (Zentralbl Bakteriol, 1991; 275: 382-9). In blood samples from those living in high risk areas, the prevalence of positive samples was 15 per cent.

In a survey of Swiss orienteers (people who pursue a sport involving map reading and running through the countryside), 26.1 per cent were positive for antibodies, but only 2-3 per cent had a history of probable or definite Lyme borreliosis (J Infect Dis, 1991; 163: 305-10). On a second blood sample taken six months later, 8 per cent of those who originally tested negative had become positive, though only 2.2 per cent developed clinical signs of the disease.

The refusal to acknowledge the possibility of seronegative disease has other implications. When Lyme is studied, generally only those with a positive blood test diagnosis of Lyme are included. But this denies Lyme patients who lack confirmatory blood tests treatment and encourages the underreporting of the true incidence of the disease. Only when a study population is defined by more than just blood test measures can the true incidence of Lyme disease become clear.

These diagnostic difficulties have also meant that many patients are misdiagnosed with other diseases. The literature is wellsupplied with case studies of the similarities and misdiagnoses of Lyme as Parkinson's disease, trigeminal neuralgia, Guillain-Barr syndrome, demy elination and multiple sclerosis (Rev Neurol, 1997; 25: 1919-21; J Orthop Sports Phys Ther, 1996; 24: 268-78; Eur J Pediatr, 1993; 152: 810-2; Neurology, 1982; 32: 1302-5; Wien Med Wochenschr, 1995; 145: 188-90; Acta Neurol [Napoli], 1993; 15: 253-7) while lupus, juvenile chronic arthritis, ADHD and Alzheimer's disease have been mistaken for Lyme disease (Lupus, 1995; 4: 131-7; J Pediatr, 1986; 109: 753-8; Psychiatr Clin North Am, 1998; 21: 693-703; Neuro report, 1993; 4: 841-8).