Introduction Visceral leishmaniasis (VL), also known as Kala-azar, is a protozoan parasitic disease transmitted by phlebotomine sandflies. It is estimated that in 2004 a total of 1,071,743 disability-adjusted life years were lost to VL in South East Asia alone [1]; if left untreated the disease is fatal. India contributes the highest number of VL cases worldwide and in 2011 reported 33,187 cases, of which 76% originated in Bihar state [2]., Despite a fall in reported cases in 2012 in India, there is evidence of substantial under-reporting of VL [3], [4] and the estimated annual incidence lies between 146,700–282,800 cases [5]. In 2005, India, Nepal, and Bangladesh signed a tripartite memorandum of understanding committing to the elimination of VL by 2015. In 2007, Médecins Sans Frontières (MSF), in collaboration with the Rajendra Memorial Research Institute of Medical Science (RMRI; Patna, Bihar, India), and the National Vector Borne Disease Control Programme of India, carried out an observational cohort study in Vaishali district, Bihar of 251 VL patients treated using 20 mg/kg liposomal amphotericin B (Ambisome; Gilead Pharmaceuticals, Foster City, CA, USA) as first-line treatment. All patients were admitted to the district hospital for the duration of treatment, and given four doses of Ambisome 5 mg/kg over a 10-day period. The intent-to-treat analysis yielded a treatment effectiveness rate of 98.8% at 6 months, with no relapses and an excellent safety profile [6]. In coordination with the RMRI, MSF shortly thereafter signed an agreement with the State Health Society, Bihar to reinforce the existing programme by identifying VL patients in Vaishali district and treating with Ambisome 20 mg/kg. Ambisome is a brand name for Liposomal Amphotericin B. There are a number of preparations of Liposomal Amphotericin B available on the market; however due to the lack of standard and widely applicable regulations or guidance for liposomal technology, it is important that this specific preparation be named. At time of publication, none of the rival preparations have undergone peer reviewed non-inferiority studies against Ambisome nor received stringent regulatory approval for use in VL. It is for this reason that MSF and the WHO currently only use Ambisome rather than other preparations. However it is urgent that clear regulatory guidelines for endemic countries be established by a normative setting organisation like the WHO and other existing formulations be formally evaluated [7]. This observational retrospective cohort study describes the field outcomes and experiences of the programme over the subsequent 5 years, discussing the challenges of implementing Ambisome treatment at the field level, and describes the long-term outcomes of treatment after 5 years of routine operational use. Methods MSF developed an integrated programme within the existing healthcare facilities of Vaishali district that utilised the district hospital for inpatient care and five rural primary healthcare centers (PHCs) for ambulatory treatment in the community. Vaishali district, with a population of 3.5 million, lies at the center of Bihar's VL belt, and is surrounded by three of the highest endemic districts for VL in the state. A comprehensive strategy of information, education, and communication (IEC) was implemented within five administrative blocks within the district where MSF supported government facilities. The intention was to treat patients with complicated VL as inpatients at the district hospital, and other VL patients in the PHCs, however the widespread lack of community belief in the PHC system [8] meant that it was understood from the outset that the majority of patients who could have been treated in the PHCs would seek care directly at the district hospital. Patients diagnosed with VL at PHC level aged 55 years, pregnant, lactating, severely malnourished, had a hemoglobin (Hb) level of 2 weeks duration and splenomegaly) were confirmed using rK39 rapid diagnostic tests (DiaMed-IT LEISH; DiaMed AG, Cressier, Switzerland). Patients presenting with relapse, or in whom there was continued suspicion of VL despite negative diagnostic tests, were referred to a tertiary center (RMRI) for parasitological confirmation through splenic or bone marrow aspiration. The RMRI is a tertiary research institution that specializes in all aspects of VL research and treatment. General demographic data were recorded for all patients diagnosed with VL, in addition to clinical history, Hb level, height, weight and malaria rapid diagnostic test result. Also recorded was ‘caste’, a form of social stratification used in India, and the categories used in the study were: scheduled caste, other backward class, scheduled tribe, and general category. Other backward class is a collective term used by the government of India for castes that are educationally and socially disadvantaged. Scheduled caste and scheduled tribe are terms used for two groups of historically disadvantaged people recognized in the Constitution of India. These three groups combined account for approximately 60% of India's population. General category comprises those who do not fit within the other categories and are not considered to be disadvantaged. Initially only inpatients deemed to be at high risk of HIV (e.g. those experiencing a relapse of VL or with a history suggestive of higher risk, such as migrant workers) were offered an HIV test, however this policy was changed in March 2011 so all patients treated at the district hospital level aged >14 years were offered testing. All women ≥14 were offered a pregnancy test. Treatment regimen Patients received four doses of 5 mg/kg Ambisome over 4–10 days depending on the clinical severity of their illness. Initially all patients were treated on days 0, 1, 4, and 9. However, once the safety of the treatment was established, and because of increasing patient numbers and the hospital's limited capacity, the duration of treatment for all clinically stable inpatients was reduced to 4 consecutive days. The 10-day regimen was maintained for severely ill inpatients and for ambulatory patients treated at the PHCs. Initial cure was defined as improvement of symptoms, cessation of fever, and reduction of spleen size at time of discharge. Considering the risks of splenic puncture and in light of a previous study showing >98% cure rate at 6 months using the same regimen [6], test-of-cure was planned only on those patients with suspicion of treatment failure, of whom there were none. Patients received health education regarding VL, and advice given to return or be actively contacted at 3, 6, and 12 months for follow-up. Additionally, all patients received health education regarding post Kala-azar dermal leishmaniasis (PKDL) and the possibility of relapse of VL, and were advised to return to the district hospital if either situation occurred. Follow-up Towards the end of the 5-year period of analysis, an active follow-up survey was conducted to determine the long-term relapse rates for a sub-cohort of VL patients. All patients not known to be HIV-positive residing within 8 of the 16 administrative blocks constituting Vaishali district, and who had completed VL treatment between two reference dates (September 2010 to December 2011) spanning 18 months prior to the survey date (March 2012) were traced. Any history of mortality, relapse, or retreatment was recorded. The eight administrative blocks, with an average of 215,000 residents per block, were selected for ease of access and highest density of patients treated within the programme. Statistical analysis All data were entered into a standard Microsoft Excel database; double data-entry was not done. Regular database cleaning comprised checks for inconsistencies with reference to source documents where necessary. An epidemiologist ensured the database was well maintained and regularly audited the quality of data transfer. Nutritional status (Body Mass Index) was assessed using weight and height data, whilst World Health Organization Anthro and Anthro Plus software (Geneva, Switzerland) was used to calculate a weight-for-height Z-score for children aged 4 weeks of illness prior to treatment). Ethics statement This analysis met the Médecins Sans Frontières Institutional Ethics Review Committee criteria for a study involving the analysis of routinely collected program data. Although a new treatment in the Indian setting, the programme utilised a recognised treatment for VL and was run in coordination with the State Health Society through a memorandum of understanding, which is the usual procedure for NGOs operating in this context. All electronic data were analysed anonymously. Results A total of 8749 patients were treated for VL during the 5-year MSF-supported program (July 2007 and August 2012), with admissions following a similar seasonal pattern each year (Figure 1). 10.1371/journal.pntd.0002603.g001 Figure 1 Admissions by month between 2007 and 2012. Patient characteristics Of the 8749 patients, 42.9% of patients were female, a proportion substantially lower than the 47.12% of females in the background population sex distribution [10]. The mean ±SD age was 22.7±17.1 years (range 0.5–90); 44.5% of patients were aged 6 cm. Both these indicators directly correlated with duration of illness prior to treatment (Pearsons Correlation r = 0.215 and r = −0.108 for Hb and Spleen size respectively, p 8 1503 (17.2) – 6.4±6.1 (0–159) Spleen size, cm [ = 8744] 6 3019 (34.5) – 6.1±3.8 (0–32) Hemoglobin, g/dL [ = 8723] 8 4689 (53.6) – 8.4±2.2 (2–18) History of relapse [ = 8741] No relapse 8364 (95.6) – Single relapse 352 (4.0) – Multiple relapses 33 (0.4) – History of previous treatment for VL ( = 8741) No treatment 8364 (95.6) – Sodium stilbogluconate (SSG) 135 (1.5) – Amphotericin B deoxycholate 104 (1.2) – Miltefosine 128 (1.5) – Other 18 (0.2) – Nutritional status ( = 7254)*** Normal 4292 (59.2) – Moderate acute malnutrition 1656 (22.8) – Severe acute malnutrition 1306 (18.0) – Co-infections HIV 161 (1.8) – Tuberculosis 51 (0.6) – Malaria 11 (0.1) – Pregnant 49 (0.6) – Where n 19 years age group (p 8 weeks, respectively. The median duration of illness prior to admission was 4 weeks (IQR 3–8), whilst the mean duration was 6.4 weeks (SD 6.1). The odds of late presentation (defined as presenting >4 weeks after developing symptoms) were significantly higher in females (OR 1.2; 95% CI 1.1–1.3; p = 0.001), those from a scheduled caste (OR 1.2; 95% CI 1.0–1.3; p = 0.03), and age ≥15 years (OR 1.4; 95% CI 1.3–1.6; p 90% for liposomal amphotericin B doses ranging from 5–20 mg/kg. A key phase III study by Sundar et al. demonstrated the safety and >95% efficacy of a single-dose regimen of liposomal amphotericin B 10 mg/kg at 6 months [42]. The study was pivotal in the adoption of this regimen as first-line treatment for VL in South East Asia by the World Health Organization Expert Committee [15]. However, common limitations of these studies include the small cohort sizes and the lack of validation of the results under field conditions. There are several concerns regarding the introduction of liposomal amphotericin B into national programs, particularly in India. Firstly, the cost of the ‘gold standard’ preparation (Ambisome) is significantly higher than that of other treatment options, despite the agreed company ‘access price’ of USD 18 per 50 mg vial for use in developing countries [29]. However, a cost-effectiveness analysis comparing 10 different treatment modalities showed that if this price was reduced to below USD 9.80 per vial, single dose 10 mg/kg Ambisome would become the most cost-effective treatment [43]. A second concern is the capacity of the healthcare systems of VL endemic countries to maintain the necessary cold chain if Ambisome is to be used in rural areas. Thirdly, because Ambisome is given as an intravenous infusion, it could be considered technically challenging to correctly prepare and administer in those areas where there are limited numbers of qualified nurses and doctors. The outcomes of this MSF-supported programme provides strong evidence that many of these challenges can be overcome. The successful use of Ambisome to treat 1397 patients in rural PHCs with existing government staff suggests that, with appropriate support and training, it is possible to provide high-quality care for VL patients in such rural settings. A functional cold chain ILR requiring a range of 2–8°C to store vaccines for the Expanded Programme of Immunization in India already exists in the majority of rural PHCs, and vaccination in even the most rural areas is already well established in India. The present program has shown that providing an additional ILR for the storage of Ambisome is a pragmatic solution to the cold chain issue, especially considering the limited number of districts in India where VL is endemic. However, careful monitoring and maintenance of the ILRs remains essential. Increased availability of electronic thermometers to improve the identification of deviations outside set temperature ranges is recommended, as is the inclusion by the manufacturer of a visual vial indicator that will make it easier to identify and discard vials that have been exposed to high temperatures. At the district hospital level ILRs remain an option, but as the required storage temperature of Ambisome ranges between freezing and 25°C, storage in air-conditioned rooms is also a possibility, provided 24-hour generator back-up is available. The need for task shifting This program has also demonstrated that a lack of highly-skilled clinic staff is not a barrier to using ‘complex’ treatments such as Ambisome at the rural PHC level. After appropriate training, lesser-skilled health workers (e.g. dressing nurses) were able to independently prepare, administer, and monitor Ambisome treatment. As suggested by the exceptionally low default rate from ambulatory care, the delegation of responsibility and empowerment (or task shifting) for the management of VL patients to such individuals, following diagnosis by the PHC doctor, also meant that more thorough patient counseling and health education was possible at the point-of-care since the care provider had more opportunity to spend time with patients. This is important when considering the time constraints and heavy patient loads that are a daily reality for the more skilled health workers in the PHCs of India, which is reflected in the generally poor community perception of the quality of PHC care in India [8]. The low mortality rate at the PHC level also suggests that the health workers were able to identify and refer 22.9% of VL patients defined as higher-risk by the MSF protocol, who may have been better served by further assessment and treatment at higher centers of care. However, this conclusion would be better supported by an audit of the appropriateness of the referrals. The present study has also shown that patients accessing care at the PHC level present earlier than those seeking care at the hospital level. Ambulatory treatment at the community level has clear benefits from both societal and healthcare provider perspectives, and is a policy that should be encouraged. However, it is clear that without a change in the community perception of the PHC system in this context, which itself must be based on a good quality of care provision, the full potential of community/rural based management of VL will never be realised. In conclusion, although expensive, the 20 mg/kg Ambisome regimen is a safe, effective, and feasible treatment for VL patients in Bihar, India. With few severe adverse events, it was well accepted by both patients and medical staff within this MSF-supported program. The active follow-up survey performed after 4-years of routine use indicated that the VL relapse rate remains exceptionally low within 6 months of treatment; however, there is a substantial number of relapses at 6–12 months post-treatment. With a move towards a 10 mg/kg single-dose liposomal amphotericin B regimen and shorter-course combination therapies, and the target of disease elimination in mind, we suggest that 1-year follow-up monitoring be recommended for VL patients in these newer treatment modalities whose longer-term efficacy has yet to be established. Supporting Information Checklist S1 STROBE checklist. (DOC) Click here for additional data file. Table S1 Hb values at admission for VL patients by age and sex. (DOCX) Click here for additional data file. Table S2 Nutritional status at admission of VL patients by age. (DOCX) Click here for additional data file.