Demographic and Socioeconomic Conditions (Screening Survey)
Site-specific demographic and socioeconomic conditions are shown in . A total of 18,933 households with 106,425 inhabitants were screened in the three countries (). The proportion of males was higher in Vaishali and Muzzafarpur (India), and Mahottari (Nepal) compared to Rajshahi (Bangladesh) (). The high level of poverty at all four study sites can be seen in the following indicators: (a) Very young population; 34.05% (95% CI, 33.8%–34.3%) of the study population were children under 15 y. (b) Crowded living conditions with 6.8±3.3 persons per household and an average of 2.7±2.2 rooms per house. (c) Inadequate housing, with 79.1% (95% CI, 77.8%–80.4%) living in thatched houses with mud plaster or made only with bamboo sticks. (d) High illiteracy rates; with 47.4% (95%CI, 45.8%–49.0%) of household heads were illiterate. (e) Low-paid work; 26.8% (95% CI, 25.4%–28.2%) of heads of households were landless labourers or unskilled workers. Looking at site-specific variation, there was little difference in the large proportion of children in the total population (reflecting an almost triangle-shaped population structure), but illiteracy and crowding were particularly high in Mahottari (56.4% and 7.4 persons per HH respectively), inadequate housing was frequent in Rajshahi (95.3%), and landless labourers and unskilled workers predominated in Vaishali (40%). In Muzaffarpur the indicators showed a slightly more favourable situation.
Demographic and socio-economic conditions.
VL epidemiological findings in study areas.
Estimated VL Case Load (Screening Survey)
The screening survey () identified 364 persons in the total study population with fever of more than 2 wk; 293 of these had a negative rK39 and were referred to the hospital for further diagnosis. Seventy-one of them (71/364, 19.5%) had splenomegaly and a positive rK39 test and were therefore classified as “VL case,” strictly speaking as “probable VL case” to be treated according to national guidelines. Additionally 166 VL cases were reported in the interviews who had been diagnosed within 12 mo preceding the interview and where the reported date of diagnosis was within that 12 mo period. Thus the average estimated annual VL incidence rate in all study sites was 22.3 per 10,000 (237/106,425; 95% CI 20.0–26.0). The proportion of cases detected through screening, out of all cases identified in the 12 mo period (i.e., reported cases within preceding 12 mo based on passive case detection [PCD], plus newly detected cases through active case finding [ACF]) was 30.0%; this proportion can be used as an indicator of the weakness of the passive case finding system across study sites.
There were important site-specific differences (): The highest estimated annual VL incidence rate was registered in India (Vaishali) and Bangladesh (Rajshahi), and the lowest in Nepal (Mahottari). The proportion of probable (rK39 positive) VL cases out of patients with fever was high in Vaishali, India (55.1%) but only between 10% and 15% at the other three study sites.
The proportion of cases found through ACF out of all cases identified was high in the more neglected districts such as Rajshahi, Bangladesh (where 49.3% of cases were detected only through ACF) and low in Muzaffarpur, India where case detection and treatment through NGOs has a long tradition (only 5.8% case detection through ACF); in districts with a reasonable PCD system including a reasonable level of VL awareness in the population (see below) less than half of all cases identified had been detected through ACF: 48.2% in Vaishali, India and 37.5% in Mahottari, Nepal. All rK39-positive febrile patients with enlarged spleens were parasitologically confirmed in Nepal, but only half of them in India, because many patients preferred the private sector for treatment. Parasitological diagnosis was not available in Bangladesh at the subdistrict level.
Effort to Detect New Cases through Active Case Finding (Screening Survey)
On average 267 houses (18,933 houses/71 new VL cases) had to be visited in VL endemic villages in order to identify one probable case with prolonged fever, splenomegaly, and positive rK39 test, which corresponded to at least three to four working days of a trained interviewer or village health worker. As can be expected, the lower the disease prevalence rate, the higher the screening effort to find a new case. The number of houses to be screened was highest in Muzaffarpur, India (1,432) and lowest in Rajshahi, Bangladesh (166, ).
Knowledge about VL (In-Depth Survey)
People's knowledge about VL was poor in the neglected districts, particularly Rajshahi, Bangladesh, compared to the better served districts in India and Nepal: In India almost all interviewees (98%; 873/891 combining Muzaffarpur and Vaishali) were aware of kala-azar, the local term for VL. This was less in Bangladesh (91%; 507/556) and Nepal (82%;1,915/2,336)). Fever as the leading symptom was identified by 92% of interviewees in India, but only by 72% in Nepal and 30% in Bangladesh. Ninety-eight percent of the Indian and 97% of Nepali respondents knew that VL is curable, but only 64% in Bangladesh were aware of this (denominators as above). Likewise the knowledge about sand flies (local term in questionnaire) transmitting the disease was frequent in India (71%) and Nepal (88%) but rare in Bangladesh (21%).
Health-Seeking Behaviour (In-Depth Survey)
Across all sites local unqualified village health workers were preferred as first-choice health care providers. This choice was associated with their excellent accessibility (on average 15 min travel time). Choosing health care beyond the community, the Indian respondents preferred private providers over governmental ones while in Bangladesh and Nepal public services were preferred over the private ones (). The choice between private or public health care professionals was not dependent on travel times and transport costs to private and public practitioners, because these were similar in the study sites or, in Bangladesh, even longer/more expensive to reach the preferred government doctors. Respondents in India would use for the treatment of VL mainly the private sector (50%) and less the public sector (30%), while in Nepal and particularly in Bangladesh the pattern was the reverse: mainly use of public sector (Nepal 45%; Bangladesh 52%) and less of private practitioners (11% and 13%, respectively). Additionally, in India people resorted to local unqualified village health workers for VL treatment (12%), in Nepal to indigenous healers (23%), and in Bangladesh to local chemists (28%). Main reasons for choice of health care provider for VL treatment were: geographical accessibility for village health workers, indigenous healers, and local chemists; but for selecting between private and public sector the most frequently mentioned factors were “faith” (belief that VL can be treated adequately) and “good interpersonal communication.”
Choice of health care provider beyond community level by 113 VL patients.
The selection pattern between private and public health sector for treatment of kala-azar was roughly confirmed by interview answers of 113 current and past (<12 mo) VL patients in the in-depth study (; not all patients identified answered): In India 80% (53/70) had used the private sector and 20% the public sector. In Bangladesh the use of public sector was 81.2% (28/35). Of the eight patients interviewed in Nepal, six (75.0%) went to the public sector and two (25%) went to local chemists.
Delay in Diagnosis and Treatment (In-Depth Study of VL Patients)
Delays of more than 2 wk () between onset of symptoms (“feeling ill”) and seeking care was frequent (57.8% of the 113 VL patients who responded) with the highest proportion in Rajshahi, Bangladesh (65.7%). The delay between resorting to the health care provider and receiving the diagnosis was 1 wk or less in most cases (58.4%), but particularly long in Vaishali, India, because of the outsourcing of diagnostic services from public hospitals and health centres to private laboratories (58.8% with more than 4 wk delay). There was significant delay between diagnosis and start of treatment by more than 2 wk particularly in the neglected districts of Vaishali and Rajshahi.
Reported treatment delay of more than 2 wk from onset of symptoms to diagnosis and start of treatment as a proportion of 113 VL patients in India, Bangladesh, and Nepal.
Treatment Interruption and Reasons (In-Depth Interview)
Of 113 VL patients, 22.8% interviewed reported having interrupted the treatment at one stage. In most cases (85.7%) the reported interruption was 2 wk or less; this was particularly the case in Muzaffarpur, India and Mahottari, Nepal; however, in Rajshahi, Bangladesh half of VL patients and in Vaishali, India, one-third of patients had interrupted their treatment for more than 3 wk. Main reported reasons for treatment interruption were lack of money for treatment (68.7%) and side effects (15.7%).