Reassessment of analytical sensitivity and specificity demonstrated that the two PCR assays performed at least as well in our laboratory as in the original publications 
. The LOD was comparable between the two assays in our hands, although the initial publications reported that the LOD was lower for rrs
than for lipL32
(5 vs. 20 GE/reaction, respectively). L. interrogans
has two copies of rrs
but a single copy of lipL32
, and this may have been responsible for the very subtle difference in the LOD of the two assays (for the rrs
assay, 5/13 and 7/13 runs were positive for 1 GE/reaction in single and duplicate samples, respectively, compared with 3/13 and 9/13 runs for the lipL
32 assay). An obvious difference between the two assays is that the target for the rrs
assay is ubiquitous among Leptospira
spp. while the lipL32
assay would only be predicted to be positive for the pathogenic group and not the intermediate and non-pathogenic groups. We are currently defining the species of Leptospira
causing disease in Thailand and cannot yet exclude the possibility that culture-negative patients who were positive by rrs
but negative by lipL32
were infected with a species belonging to the intermediate group.
Our finding that PCR had a lower diagnostic sensitivity than MAT is consistent with previous reports 
. Plausible explanations include late presentation associated with absence of Leptospira
in blood, and pre-treatment with antimicrobial drugs prior to admission. A wide range of oral antibiotics is available over the counter in Thailand, and self-medication prior to hospital presentation is common.
The quantification of Leptospira in blood during this study was a useful exercise, since this can provide critical baseline information during the development of point of care antigen detection tests. The finding that the bacterial count was higher in patients who were culture positive compared with those who were culture negative was intuitive.
Our data on the window of PCR or culture positivity after the onset of symptoms suggest that these tests only have clinical utility within the first week of clinical manifestations, as reported previously 
. We observed that the period over which PCR was positive after the start of symptoms was longer than that for culture. A small number of patients were positive by culture but negative by PCR. However, the difficulty and expense of culture combined with the prolonged delay before culture becomes positive means that culture results will not influence individual patient care.
The basis for a negative PCR result but positive culture remains unexplained, but possible explanations include a very low count in the initial sample associated with a stochastic effect in which the organism was present in the aliquot taken for culture but not for PCR. It is also possible that PCR inhibitors were present that interfered with the detection of a very low copy number but did not affect the detection of the positive control DNA (rnaseP
), which would be present in abundance. The basis for positive PCR results in patients who were negative by culture and MAT and who had another diagnosis or unknown diagnosis is also uncertain. One possibility is that some patients had more than one infection and false negative diagnostic tests for leptospirosis. It is quite possible that patients could develop both leptospirosis and scrub typhus in the same timeframe since agricultural workers are often exposed to the pathogens causing both infections. Previous studies have documented patients with serological evidence for concurrent leptospirosis and scrub typhus 
, but the putative situation in which patients have more than one infection but negative diagnostic tests for leptospirosis is speculative and extremely difficult to prove. An alternative explanation is laboratory contamination, although the negative controls remained negative throughout the study and the stage of any contamination event would have to have been at an earlier part of the study pathway (for example, during DNA extraction).
Making an accurate diagnosis of leptospirosis contributes to both the characterization of disease epidemiology and to individual patient care. The diagnosis of leptospirosis across much of Thailand continues to be made on the basis of clinical features because of a lack of inexpensive and easy to use diagnostics tests. MAT is performed by the National Institute for Health, Thailand and is available as a reference test, but is used for a minority of suspected cases overall to underpin epidemiological data and provides a retrospective diagnosis. Leptospira culture is largely a research activity, and has no clinical utility in relation to immediate patient care. The PCR assays evaluated in this study confirmed the diagnosis of leptospirosis in half of definite cases, and further studies are now required to determine whether such information would have altered patient morbidity and mortality, together with the effect of false positive test results. The feasibility of introducing PCR tests, however, rests on affordability; the cost of introducing a test into laboratories that do not currently perform PCR would be high both in terms of equipment and training.
In conclusion, Leptospira detection using PCR could improve the management of patients presenting to hospital within the first few days of the onset of symptoms of leptospirosis, although cost represents a barrier to its implementation in resource-restricted countries. An on-going study is currently evaluating the diagnostic sensitivity and specificity of LAMP (loop-mediated isothermal amplification), a technique that requires minimal equipment and modest training.