How Far Are We from Visceral Leishmaniasis Elimination in Bangladesh? An Assessment of Epidemiological Surveillance Data

Introduction In 2005, Bangladesh, India, and Nepal joined forces to eliminate Visceral Leishmaniasis (or kala-azar) from the region by 2015. In Bangladesh the elimination target is set at less than one new case per 10,000 population per year at upazila (sub-district) level. As the deadline approaches, we review the status of the elimination initiative in this country. Methods We collected all available disease surveillance data at the Disease Control Unit of the Directorate General of Health Services, Government of Bangladesh from 1994 to 2013. Additionally, we retrieved data from the Civil Surgeon Office from the Mymensingh district, one of the most heavily affected areas in Bangladesh. Results Between 1994 and 2013, 109,266 kala-azar cases causing 329 deaths were reported from 37 endemic districts in Bangladesh. Only 16 districts reported cases every year. The Mymensingh district was the most affected with 53,582 (49.04%) cases. Between 2008 and 2013 only 16 upazilas showed incidence rates above the elimination target in which they ranged from 1.06 to 18.25 per 10,000 people per year. Discussion While clear progress has been made towards eliminating VL, 16 upazilas in Bangladesh had not yet reached the target in 2013, based on official notification data that probably suffered from under-reporting bias. The elimination initiative urgently needs to establish methods to ascertain and monitor the elimination target.


Introduction
On the Indian subcontinent, Visceral Leishmaniasis (VL), or kala-azar, is caused by Leishmania donovani, which is transmitted from man to man by the sand fly Phlebotomus argentipes, the only known vector [1]. Of the 200,000 to 400,000 new cases of VL worldwide, more than 90% are reported from India, Bangladesh, Sudan, South Sudan, Ethiopia, and Brazil [2]. VL affects the poorest communities in these countries and is almost always fatal if not treated. The first report of VL ever came from Jessore, currently located in southwestern Bangladesh, where an epidemic outbreak killed an estimated 75,000 people between 1824 and 1827 [3]. Over the next decades, kala-azar became endemic in the region and spread slowly through Bengal, where it devastated the population of Burdwan and other areas. Subsequently, the disease spread eastward into Assam. Between 1892 and 1898, one-third of the population of the Nowgong district in Assam, India died [4]. Another epidemic of VL in conjunction with the Spanish Influenza epidemic claimed a further 200,000 lives in Assam and in the Brahmaputra valley between 1918 and 1923 [5]. Up to 1940, more than 1,000,000 VL cases were reported in former Bengal where the first mass treatment measures were undertaken [6,7]. The incidence finally declined because of the dichlorodiphenyltrichloroethane (DDT) spraying by the Malaria Eradication Programs in the 1950s, and VL was thought to be eliminated by 1970. Between 1968 and 1980, only 59 cases were reported in Bangladesh [8]. But since the 1980s, after the interruption of DDT spraying, there has been a dramatic resurgence of VL, with 73,467 cases reported from Bangladesh between 1994 and 2004, and many more reported in India [3].
In 2005, Bangladesh, India, and Nepal joined efforts to eliminate kala-azar. Elimination was thought feasible in this region because (i) human beings are considered the only reservoir host of L. donovani, (ii) Phlebotomus argentipes is the only vector, (iii) the disease is confined to a limited number of districts, (iv) a rapid diagnostic test allows easy diagnosis, and (v) effective oral treatment was available [9]. The respective Health Ministers of the three countries signed a Memorandum of Understanding (MOU) with the aim to reduce the annual incidence rate of VL to less than one per 10,000 inhabitants in the endemic communities by 2015, an elimination goal endorsed by the World Health Organization (WHO) [9]. The five strategies adopted in the VL elimination initiative were (i) early diagnosis and treatment, (ii) strengthened epidemiological surveillance, (iii) integrated vector management, (iv) social mobilization, and (v) operational research [9].
For any disease control program, proper epidemiological surveillance is a key issue. It allows for the establishment of the past and present disease burden, and will guide the program to take timely and appropriate action on case detection, patient management, vector control, and community awareness. In an elimination initiative, routine surveillance data are essential to keep track of the elimination target, though additional measures are required to ascertain the elimination status of given areas. As the set VL elimination target of 2015 is fast approaching, we have analyzed the available epidemiological information in Bangladesh to advise the national and regional disease control policy.

Context
Bangladesh is administratively divided into six divisions, namely: Chittagong, Barishal, Dhaka, Khulna, Rajshahi, and Sylhet. All of the divisions except Sylhet are reporting VL cases, but not to the same extent. Each division is further subdivided into districts (a total number of 64), sub-districts, called upazilas or thanas (a total of 482), Union Parishads (UPs), and wards. The governmental health care system is structured along the same administrative divisions, with a national, divisional, district, upazila (sub-district), union, and ward level. Three levels in this system deliver VL treatment: (i) the Upazila Health Complex (UHC) (the lowest level, offering indoor facilities with 31 to 50 beds) (ii) the District Sadar Hospital (DSH) (a 200-to 500-bed hospital), and (iii) Medical Colleges. A UHC caters to a population of 200,000 to 300,000, while a DSH covers approximately 2 to 5 million inhabitants.

Epidemiological surveillance system
Disease surveillance is organized as follows. At the end of each month, the UHC and DSH send their morbidity and mortality reports to the Civil Surgeon Office at the district headquarters. The District Civil Surgeon transmits these reports to the Director of Disease Control at the Directorate General of Health Services (DGHS), who will notify country data to the World Health Organization. As is commonly the case in routine surveillance, these figures are an underestimation of the true number of VL cases. Medical colleges do not report although some VL patients get reported when they are referred back to the UHC or DHS after some days of treatment. This happens quite often, as VL drugs are provided for free in the public health services but are not always available in the medical colleges. Another important factor to consider when analyzing trends in the surveillance data is the recent change in drug policy. While the injectable Sodium Stibogluconate (SSG) was the only drug of choice for many years, since the middle of 2009, the oral drug Miltefosin was introduced as first-line drug (except for women of childbearing age and pregnant women). As this drug is not available in the local market, patients are now more motivated to attend a UHC to get this oral medication. The recent introduction of single-dose AmBisome therapy also attracts VL patients to the UHC, as does the recent involvement of one international Non-governmental organization (NGO), Médecins Sans Frontières (MSF), in VL care.

Data collection and analysis
We have collected all data on VL available at the central level (DGHS) for the years 1994 to 2013. We also collected district-level data for the Mymensingh district from the Civil Surgeon's Office in Mymensingh for the same period. As the VL elimination target is set at the upazila level ''to reduce VL incidence rate below one per 10,000 population per year at upazila level,'' [9] we calculated the Incidence Rates (IR) for each upazila. We used the total population of the upazila in the corresponding year in the denominator. The data were analyzed by using SPSS, Minitab, and Microsoft Excel.

Results
From 1994 to 2013, the DGHS of Bangladesh reported 109,266 cases of VL and 329 VL-related deaths from 37 endemic districts. During this 20-year period, there were three years (1997, 2002, and 2006) with more than 8,000 reported cases. The highest number of cases was reported in 2006, and the annual number of cases diminished after that peak year ( Figure 1). Altogether, 37 districts reported VL cases at some point during this period. The number of districts reporting VL cases increased from 21 districts in 1994 to 29 districts in 1998 and 2008. Sixteen districts reported VL cases continuously throughout the 1994-2013 period and accounted for 96.82% of the total case number reported during this period ( Figure 2). The highest number of VL cases was reported from the Mymensingh district, accounting for 49.04% (53,582) of the total number of cases ( Figure 2). The Mymensingh district reported 110 deaths due to VL, or 33.85% of the country's total. The second highest number of cases occurred in the Pabna district (12,067 or 11.04%) ( Figure 2) with 11 deaths, followed by the Tangail and Jamalpur districts where 10,170 (9.31%) and 6,965 (6.37%) cases, respectively, were reported ( Figure 2), including 45 (13.85%) and eight (2.46%) deaths. Twelve districts reported more than 1,000 cases in the study period, six between 500 and 999, and 19 reported between one and 499 cases.    Table 2 shows that Mymensingh is the most affected district. The Incidence Rate (IR) per 10,000 people per year varies from 0.1 to 16.5 in the seven affected upazilas ( Table 3). As Table 3 shows, almost all cases are concentrated in five of the seven upazilas. The Tangail district has three endemic upazilas, there are two affected upazilas in the Pabna district, and the other six districts have six endemic upazilas (one in each) ( Table 2). Among the five most endemic upazilas in the Mymensingh district, the Fulbaria upazila diagnosed 4,085 cases from 2008 to 2013. The second highest case number diagnosed is in the Trishal upazila, where 4,020 cases were diagnosed, followed by Bhaluka (1,501), Muktagacha (1,310), Gafargaon (1,053 cases), Haluaghat (120 cases), and Nandail (21 cases). Other upazilas diagnosed few cases from 2008 to 2013 (Table 3). The IR in the five highly endemic upazilas, Trishal, Fulbaria, Muktagacha, Gafargaon, and Bhaluka was 16.5, 15.7, 5.5, 5.2, and 4.5 per 10,000, respectively, from 2008 to 2013 (Table 3).
From 2008 to 2013, there were 16 upazilas in nine districts with an average IR exceeding the elimination target, ranging between 1.06 to 18.25 per 10,000 inhabitants (Table 2).

Discussion
The VL elimination program was launched in 2005, and its target was set to reduce the number of cases at upazila level below one case per 10,000 people by 2015 [9]. However, as the initial situation in 2005 and the current epidemiological situation are not well captured, it is extremely difficult to say in 2014 how far we are from the elimination target. Surveillance data from the Disease Control Department of DGHS show that cases were reported from 37 districts from 1994 to 2013, whereas previously, 45 districts were considered endemic [10]. Unfortunately, upazila-level data on VL are not available at DGHS before 2007, making it difficult to properly assess the trends at this level. Moreover, the level of underestimation of VL cases has probably reduced significantly during recent years, as was observed also in India, for at least two reasons: (1) the introduction of a rapid diagnostic test, which has improved access to diagnosis, and (2) the recent policy making free VL treatment available in the public services with drugs that are not available in the private market. This changing, under-reporting ratio does not facilitate the interpretation of trends.
About half of the total numbers of VL cases were reported from a single district (Mymensingh), where five upazilas are highly endemic. Therefore, the national program should intensify its efforts in these areas with high priority. Due to VL, there is a huge     economic loss for the affected families in an endemic community [11]. Currently, the access to diagnosis and treatment for VL is improved, as VL care is available free of charge at the UHC or district-level hospital in all endemic districts, which is clear progress. However, to reduce the VL incidence rates, a complete package of activities (prompt case detection, proper treatment and case management, and effective vector control) should be deployed on an urgent basis. Unfortunately, no vector control activities were carried out between 1999 and 2012 in many areas due to financial constraints, lack of trained human resources, and unavailability of insecticides for indoor residual spraying (IRS) because of procurement problems [12]-except for limited IRS in a few pilot projects in 2011 in the Mymensingh district. In 2012, IRS was implemented in eight highly endemic upazilas (Fulbaria, Trishal, Bhaluka, Gaforgaon, and Muktagacha in Mymensingh, Terokhada in Khulna, Madarganj in Jamalpur, and Nagorpur in Tangail) during the pre-and post-monsoon periods. Focal spraying was conducted in moderate-and low-endemic areas from January 2013. Recently insecticide-treated bed nets were distributed to all patients treated in UHCs in the highly endemic communities in the past three years. In 2011, the Fulbaria upazila of the Mymensingh district reported 1,608 cases. The most likely explanation for this high figure is that in mid-2010 an international NGO (MSF) started a VL control program in this upazila through active case search and treatment with AmBisome. They also provided food during hospitalization, and compensation for transport and wage losses. This intervention attracted a high number of patients, including some living in nearby upazilas, who used the address of their relatives or friends in Fulbaria as their own address in order to get access to the best facilities.
Passive surveillance records show 329 patients died in 20 years (1994-2013), which is probably an under-estimation, as there was drug scarcity before the introduction of Miltefosine as a first-line drug. Poor quality SSG was used during periods of drug scarcity, and there was a major incidence of the use of substandard Miltefosine for a certain period, which added to mortality [13]. The data on VL deaths were mostly recorded in patients who died in the government hospitals during drug administration, but those who died at home or in a private clinic due to VL or its complications were not recorded by passive surveillance. A community-based study showed that the VL case fatality rate in Bangladesh was 5.3% in males and about three times higher in females [14].
The reporting of PKDL cases should be strengthened, as country-wide data are lacking up to 2013. A recent study from Bangladesh shows that the cumulative incidence of PKDL can be up to 17% within five years of being treated for kala-azar [15]. Another study, performed in a less VL-endemic area, found a PKDL prevalence of about 6 per 10,000 people [16]. These PKDL cases are important for transmission dynamics as they are supposedly highly infectious.
Effective progress towards VL elimination requires continuous surveillance (Box 1) of mortality and morbidity as well as of the populations at risk. While clear progress has been made towards VL elimination in Bangladesh, 16 of 140 endemic upazilas had not yet reached the target in 2013, based on official notification data that suffer from underreporting bias. The elimination initiative urgently needs to establish methods to ascertain and monitor the elimination target. Box 1. Importance of Proper Disease Surveillance for VL Elimination Program N Measure the past and present disease burden. N Guide national program to take timely and appropriate action on: # Patient management # Vector control # Community awareness through IEC N Prevent unnecessary program expenditure, i.e., accurate surveillance record will provide guidance on specific areas in which to take necessary actions instead of in the whole upazila or union.