The PPUL in northwest Bangladesh in the population of 5 years and older, as found by means of a random cluster survey, is 15.1 per 10,000. This study, which included about 0.5% of the total population of the area, was based on established multistage cluster sampling techniques. We believe that the results give a reliable picture of the leprosy situation in northwest Bangladesh, in an area where an extensive leprosy control programme has been implemented for more than 10 years.
Potential sources for selection and information bias were considered, especially as only those present during the survey were included. Selection bias on cluster level is not likely, but on individual level selection bias is possible as the survey is announced in advance and those afraid of the diagnosis may go into hiding. Males are less likely to be at home during the day and indeed only 42% of those examined are males. In our data, however, the PPUL among males and females is the same. It is possible that, due to stigma, those with leprosy have a higher chance of being unemployed or rejected at school, so they could be over-represented at the survey, but as all patients found were in the early stage of the disease, this does not seem to be a likely reason for the high number of cases found in our study. We conclude that the possible sources of bias probably have had no effect.
In the past, over-diagnosis has not been a problem in this particular field programme, as was confirmed by an independent evaluator in 2001
, but to avoid possible over-diagnosis in this study, all suspected cases were seen by senior leprosy control officers with more than 5 years experience in the diagnosis of leprosy at referral centre level, and confirmed by a medical doctor.
We found that the PPUL (including children under five) found by active screening was nearly 6 times higher than the registered prevalence (13.4 vs. 2.31/10,000). Registered prevalence is largely based on passive case detection. A large difference between the official new case detection (NCD) or prevalence, based on passive case detection, and the NCD or prevalence found by door-to-door surveys has been described before. For example, Schreuder et al. found by a rapid village survey in Java, Indonesia, two and a half times the number of known cases
, and Bakker et al. found during a survey on a few small Indonesian islands 96 cases of leprosy of whom only 11 were previously known.
Different sample surveys in India have also revealed sample prevalences 4–5 times the recorded prevalence.
Self-healing of leprosy contributes to the difference between active and passive case-finding. In South India Ekambaram et al. found that the percentage of self-healing among non-lepromatous patients was around 74%.
In Africa Browne found that 34% of non-treated patients healed spontaneously.
shows the PPUL in the general population sample as described in this paper, together with the PPUL in the subgroups of contacts of leprosy patients as found during the intake of the COLEP trial.
These subgroups were defined by their physical distance to the index patient. The age distribution in the general population examined is similar to the distribution in the contact group, so this is not a major cause for bias. In the contact group of the COLEP study as a whole, the PPUL rate was 73/10,000, compared to 15.1/10,000 in the population sample.
With regard to the different categories in the contact group, we conclude that even in the most distant category (the neighbours of the neighbours and social contacts) the PPUL rate (49/10,000) does not come down to the same level as that of the general population. It may therefore be preferable under such high-endemic circumstances to conduct full village or neighbourhood surveys instead of (close) contact surveys.
Table 3 Prevalence of previously undiagnosed leprosy (PPUL) per 10,000 in the subgroups of the contact population of the COLEP study  and in the general population sample.
There is a marked variance in PPUL among the different clusters. A gradient along geographical lines was not found. The clusters with a low number of newly found cases are scattered over both districts, as are the clusters with the highest numbers. In the three urban clusters however, relative high numbers of cases were found. This is in contrast to the findings of Kumar et al. in Agra, India, where the prevalence of leprosy in the urban areas was about one third lower than in the rural areas.
Sterne et al. observed a lower incidence of leprosy in the semi-urban district capital of the Karonga District in Malawi
, while Lapa et al. report that in the State of Pernambuco, Brazil, leprosy is mainly an urban disease.
In conclusion, our data show that the PPUL in the general population is six times higher than the registered prevalence, but three times lower than that in the most distant subgroup of contacts of leprosy patients in the same area. It has to be kept in mind however, that most new cases in populations where leprosy is relatively highly endemic come from the non-close contact group. Hence full village or neighbourhood surveys might be preferable to contact surveys under such circumstances.
There are indications that in lower endemic areas the incidence of leprosy among contacts declines faster as the physical distance to the patient increases.
If that is indeed the case, screening of contacts further removed from the patient might not be as useful in lower endemic areas.