To the Editor:

Robertson and colleagues [Reference Robertson, Nelson and Stephen1] analysed the spatial epidemiology of leptospirosis in Sri Lanka and showed a probable correlation between occurrence of leptospirosis and rainfall patterns in Sri Lanka. They also identified risk clusters for leptospirosis based on the spatial distribution of the reported cases. We appreciate this work which filled several knowledge gaps in the epidemiology of leptospirosis in Sri Lanka. However, we are concerned about some facts in the paper which need further explanation and corrections.

The disease rates in the paper were analysed using routinely reported data. Authors have correctly described that these were not confirmed cases. They further provide evidence to suggest that the reported cases would be valid for this analysis. The supportive evidence [Reference Dassanayake2] provided for validity of reported cases was not actually for the reported cases. We carried out the particular study to validate the WHO proposed surveillance case definition. Although this case definition is recommended for case reporting, the clinical practice and reporting is completely different from this case definition. Since the laboratory diagnosis is not routinely available, reporting is entirely based on clinical judgement. We previously analysed this error in our study conducted during the 2008 epidemic, in Kandy, Kegalle and Matale, districts which showed that only 52·6% of the patients treated for leptospirosis actually had the disease. Furthermore, 46·2% of the patients who were treated for other conditions were retrospectively confirmed as leptosprosis cases [Reference Agampodi3]. To show the diversity of the clinical judgement we further analysed these cases by reporting hospitals. It showed that only 28·3% of cases from Matale were confirmed positive, compared to 58·3% and 55·0% in Kandy and Kegalle hospitals, respectively. According to the routinely reported data, Matale had the highest incidence of leptospirosis during the 2008 epidemic, which we proved as not the correct figure. It is clear that Matale had a higher number of cases during the 2008 outbreak. However, fewer than one third of the suspected and reported cases actually had the disease and it was significantly lower than in other areas. The reported data available in the surveillance system seems to be an overestimation of leptospirosis incidence in Matale and we suspect that this had affected the Robertson study in which the authors discussed Matale specifically. To further explain this diversity of clinical judgement, we looked at the cases reported from the three medical units in Kandy hospital. While one unit reported 54 clinically suspected cases during our study period of 4 months, the other two units each reported fewer than 10 suspected cases. This difference is highly unlikely, given that the same number of admission days are allocated to each unit. The most plausible explanation is that clinical suspicion varied widely according to treating physicians. While the 2008 outbreak of leptospirosis undoubtedly existed, exact case load and geographical distribution are questionable due to the lack of point-of-care diagnostic facilities and gross under-/overreporting of cases, based on the treating physician.

We have another major concern about the use of 2002 data for agriculture/paddy field distribution. In 2007, the government launched an island-wide programme (‘Api Vavamu, Rata Nagamu’), which made it mandatory to cultivate all abandoned paddy fields. This was seen especially in 2008, where a large number of people who were not traditional farmers got involved in paddy farming activities. This programme would have changed the paddy field distribution in 2002 considerably, and it may be a reason for not showing a significant association with the 2008 outbreak.

We would like to point out some other inaccuracies in the paper: (1) Sri Lanka does have a seasonal pattern for rainfall, but being a tropical country, we do not have four seasons as described by the authors. (2) The authors referred to our study published in 2008 [Reference Agampodi4] as evidence for diversity of serovars during the 2008 outbreak. This report was for the 2002–2003 period, not for the 2008 outbreak. (3) The authors referred to the report on interim analysis of the 2008 outbreak and mentioned that nine serovars were isolated. There has been no published literature on serovar isolation from Sri Lanka recently. The citation in the paper was based on results of the microscopic agglutination test. (4) In the paper, the authors used MOH areas as the unit of analysis, and MOH was defined as ‘Ministry of Health’. This is incorrect – MOH areas are ‘Medical Officer of Health’ areas, which are divisional-level health administrative units in Sri Lanka.

We also were very interested as to why authors reported the ‘prevalence’ of leptospirosis. Conventionally, we express leptospirosis disease as incidence because it is an acute condition.