Visceral leishmaniasis in southeastern Nepal: A cross-sectional survey on Leishmania donovani infection and its risk factors

Authors


Corresponding Author Karl Schenkel, Robert Koch Institute, Seestrasse 10, 13353 Berlin, Germany. E-mail: schenkelk@rki.de

Summary

Objective  To document the frequency of Leishmania donovani infection at community level in a highly endemic region in southeastern Nepal, and to assess socioeconomic and environmental risk factors.

Methods  A random cross-sectional population survey was held in two visceral leishmaniasis (VL) foci in Morang District in April to May 2003, enrolling individuals 2 years or older and residing in the endemic area for at least 12 months. Leishmania infection was defined as a direct agglutination test (DAT) titre equal to or higher than 1:3200. Risk factors were identified by logistic regression.

Results  The direct agglutination test was positive in 7.5% (95% CI: 5.1–10.8) and the leishmanin skin test (LST) in 13.2% (95% CI: 9.9–17.2) of the 373 study participants. No case of current kala-azar was found, but 5.1 % (95% CI: 3.1–7.8) reported having suffered from VL. Independent risk factors for Leishmania infection were proximity of the house to ponds [odds ratio (OR) 3.7, 95% CI: 1.6–8.5], family size (OR 4.4, 95% CI: 1.6–12.6), age geqslant R: gt-or-equal, slanted15 years (OR 5.5, 95% CI: 1.2–25.0) and house constructed in mud (OR 3.0, 95% CI: 1.1–7.6). Bednets, not impregnated and in poor condition, were used by 95.2% (95% CI: 92.3–97.0) of the population, but did not show any protective effect.

Conclusion  This study shows that there is a serious problem of transmission of VL in this area of Nepal. The risk factors identified are linked with the socioeconomic level and the environment. The population would benefit from a community intervention to improve the environmental and housing conditions in the villages.

Abstract

Objectif  Documenter la fréquence des infections àLeishmania donovani au niveau de la communauté, dans une région fortement endémique dans le sud-est du Népal et évaluer les facteurs de risque socio-économiques et environnementaux.

Méthodes  Une étude transversale aléatoire de population a été menée dans deux foyers de leishmaniose viscérale dans le district de Morang dans la période avril-mai 2003. Les individus de 2 ans et plus, résidant dans la zone endémique depuis au moins 12 mois ont été recrutés. L'infection à la leishmaniose a été définie sur base du test d'agglutination direct (DAT) pour un titre ÿ 1/3200. Les facteurs de risque ont été identifiés par la régression logistique.

Résultats  Le test DAT était positif chez 7,5% (IC: 5,1–10,8) et le test cutanée à la leishmanine (LST) chez 13,2% (IC95%: 9,9–17,2) des 373 participants. Aucun cas de Kala-azar n'a été trouvé. Mais 5,1% (IC95%: 3,1–7,8) des participants ont rapporté avoir souffert de leishmaniose viscérale. Les facteurs de risque indépendants pour l'infection à la Leishmaniose étaient les suivants: la proximité des habitations aux mares (OR: 3,7, IC95%: 1,6–8,5), la taille de la famille (OR: 4,4, IC95%: 1,6–12,6), l’âge ÿ 15 ans (OR: 5,5, IC95%: 1,2–25,0) et l'habitation dans des maisons construites en terre (OR: 3,0, IC95%: 1,1–7,6). Des moustiquaires non imprégnées et en mauvais état étaient utilisées par 95,2% (IC95%: 92,3–97,0) de la population, mais cela n'a révélé aucun effet protecteur.

Conclusion  Cette étude démontre l'existence d'un sérieux problème de transmission de la leishmaniose viscérale dans la région étudiée du Népal. Les facteurs de risques identifiés sont associés à des aspects socio-économiques et environnementaux. Cette population bénéficierait bien d'une intervention de communauté dirigée vers l'amélioration des conditions environnementales et des habitations dans les villages.

Abstract

Objetivo  Documentar la frecuencia de infección a nivel comunitario por Leishmania donovani en un área altamente endémica del sureste de Nepal, y evaluar los factores de riesgo socioeconómicos y ambientales asociados.

Métodos  Estudio croseccional, aleatorizado, realizado en dos focos de Leishmaniasis Visceral (LV) en el distrito de Morang, entre Abril y Mayo del 2003. Se incluyeron individuos mayores de dos años que llevasen residiendo al menos 12 meses en el área endémica. La infección por Leishmania se definió como un título igual o mayor de 1:3200 para la Prueba de Aglutinación Directa (PAD). Los factores de riesgo se identificaron mediante regresión logística.

Resultados  De los 373 participantes, un 7.5% (IC 5.1–10.8) obtuvo una PAD positiva y un 13.2% (95% IC 9.9–17.2) dio positivo en la Prueba Cutánea con Leishmanina (PCL). No se halló ningún caso activo de kala-azar, pero un 5.1 % (IC: 3.1–7.8) reportó haber sufrido de LV. Los factores de riesgo independientes para infección por Leishmania fueron: proximidad de la vivienda a estanques (OR 3.7 95% IC: 1.6–8.5), tamaño del núcleo familiar (OR 4.4 95% IC 1.6–12.6), edad geqslant R: gt-or-equal, slanted 15 años (OR 5.5 95% IC 1.2–25.0) y construcción de la vivienda con barro (OR 3.0 95% IC 1.1–7.6). Un 95.2% (95% IC 92.3–97.0) de la población usaba redes mosquiteras sin impregnar y en malas condiciones, que demostraron no tener efecto protector.

Conclusión  Este estudio muestra que existe un problema serio de transmisión de LV en esta área de Nepal. Los factores de riesgo identificados están relacionados con el nivel socio-económico y el medio ambiente. La población se beneficiaría de una intervención comunitaria que mejorase las condiciones medioambientales y de las viviendas en los poblados.

Introduction

Visceral leishmaniasis (VL), a vector-borne disease caused by Leishmania donovani, is one of the major public health threats in the Terai (lowlands) in southeast Nepal and the bordering districts of the Indian state of Bihar. The first officially recorded case of VL in Nepal was in 1980 from Dhanusha District (Bista 1998). Since then, there has been a steady increase in the reported cases, and presently 12 districts are identified to be endemic with more than 5.5 million people estimated to be at risk of the disease (unpublished report, MOH Nepal, 2002). From 1980 to 1989, the incidence rate per 100 000 person-years remained below 10. Since then, it has been increasing steadily and in the last few years it has been reported to have increased from 43 to 55 per 100 000 person-years. The case fatality rate for VL has been reported to be around 1% in the last few years. However, these surveillance data from the Ministry of Health are considered to be underestimates as they do not include cases treated outside the public health facilities (Bista 1998). In 2004, a seroprevalence of 6% was found in two endemic communities (Kasaini and Gidhaniya) in the Morang District in eastern Terai (Koirala et al. 2004).

Koirala et al. (1998) have described factors that contributed to the rising trend of kala-azar (KA) in Nepal: lack of surveillance, cross-border migration from Bihar, inadequate treatment of cases and an increasing vector density are contributing to an increasing burden of infection and disease.

B.P. Koirala Institute of Health Sciences (BPKIHS) is a tertiary care centre in Dharan, southeast Nepal. According to BPKIHS officials, over the past 10 years, an increasing number of KA patients were admitted to the tropical medicine ward. Many of those patients were from the Dulari village development committee (VDC), a rural village community in the neighbouring Morang District, about 20 miles to the southeast of Dharan. To give recommendations for control at community level, and to identify the socioeconomic and environmental risk factors for L. donovani infection, a random population survey was undertaken in Dulari.

Material and methods

Study area and population

In March and April 2003, a random cross-sectional survey was undertaken in wards 1 and 9 of the Dulari VDC, an area where VL is endemic. The biological tests performed are explained later in this section. VDCs are the fourth administrative division in Nepal, corresponding to the territory of a village community. Each VDC is composed of approximately 10 wards (of approximately 1000 inhabitants each). Wards 1 and 9 of Dulari VDC were known to be the most affected areas within the VDC on the basis of case notification data in the district epidemiological surveillance system. Figure 1 gives the geographical location of the Dulari VDC within Morang District.

Figure 1.

 Geographical location of study area: wards 1 and 9 of Dulari village development centre (VDC) and their location in Morang District, southeast Nepal, 2003.

The sample size for our survey was calculated assuming an expected prevalence of recent infection of 10%, a required precision of 3% and an alpha level of 5%. During a 6-week period, Wards 1 and 9 were systematically searched in a north-to-south approach to visit randomized households by a team of local VDC workers. Households were randomly selected from an updated VDC census list in a simple random approach. Inclusion took place during a 6-week period. Survey teams composed of local VDC workers systematically searched Wards 1 and 9 in a north-to-south approach to retrieve all the households that had been selected as eligible for inclusion in the sample. All persons aged 2 years or older, living in the endemic area for at least 12 months, were eligible for inclusion. A person was considered to be part of a particular household if he or she shared meals and sleeping facilities with the other members of that household.

Our case definition for a clinical suspect of KA was any person with a reported fever of more than 38.5 °C for 2 weeks, after exclusion of malaria by OptiMAL® test (DiaMED, Switzerland) along with clinical signs of either spleen or liver enlargement or lymphadenopathy on palpation. We did not include children <2 years of age for ethical reasons, as parents would object to taking venous blood samples and conducting skin tests with small children. Furthermore, not every health worker in the field team was able to puncture veins in very small children in the field.

Persons reporting to have been diagnosed by a professional caregiver with KA within the last 5 years were considered as previous cases of VL. Persons agreeing to participate were informed about the aims of the study and written informed consent was obtained. Ethical approval had been obtained in February 2003 by the ethical committee of BPKIHS.

Markers of infection

Consenting persons were subjected to a leishmanin skin test (LST) and 2 ml venous blood was collected on a filter paper for the direct agglutination test (DAT). The leishmanin used during this study, a 5 × 106 promastigotes/ml suspension of Leishmania infantum MHOM/TN/80/IPT1, obtained from Istituto Superiore di Sanità in Rome, Italy, was injected intradermally into the ventral side of the forearm. Between 48 h and a maximum of 72 h after inoculation, results of the LST were read with the ‘ball-pen technique’ described by Sokal (1975). A positive LST was defined as a papular skin reaction of at least 5 mm diameter at the leishmanin injection site. The LST reflects T-cell-mediated hypersensitivity and is therefore a marker of past infection, whereas the DAT is based on humoral immunity and reflects more recent infection. As the BCG vaccination leads to false-negative results for LST and the vaccination coverage in the study population is unknown, we used LST for describing the burden of infection only but did not include it in the risk factor analysis. DAT as a marker of more recent infection was included, as it reflects infection during the past 2–5 years, a more reasonable time span to take into account household conditions, which may change over a lifetime.

The DAT was executed in the laboratory of the Institute of Tropical Medicine, Antwerp, as described by Boelaert et al. (1999). In brief, promastigotes of L. donovani MHOM/SD/00/1S, stabilized suspension that had been formaldehyde fixed, Coomassie Brilliant Blue stained and dispensed into 5 ml vials, were reconstituted with a mix containing sodium chloride, gelatine, sodium azide and distilled water. Five millimetre punches of dried filter paper blood spots were eluted overnight and later pipetted into a microtitration plate with the DAT-antigen preparation. A positive DAT was defined as a DAT titre of at least 1:3200.

Risk-factor analysis

Data analysis

Socioeconomic risk factors for infection were assessed in a bivariate and multivariate analysis for the whole survey and separately by wards. For this purpose, an infected person was defined as a resident with a DAT titre of at least 1:3200. Variables included in the bivariate analysis were sex, age (dichotomized geqslant R: gt-or-equal, slanted15 years), family size, house type (mud, brick, wood, cement), ownership of cows/buffaloes or small animals such as fowl; environmental factors, such as the proximity of the household to stagnant water (within a radius of 50 m); number of rooms per house; electricity and yearly family income as proxies for socioeconomic status; localization of the sleeping site and use of bednets. Variables associated with the outcome ‘infection’ at the P < 0.20 level in the bivariate analysis (Pearson's chi-square tests) were included in a logistic regression procedure. The final model was obtained by a backward selection strategy recommended by Kleinbaum et al. (1998). The analysis was carried out using the statistical software spss 10.0 for Windows (SPSS, Chicago, IL, USA).

Results

The random sample included 373 persons: 196 from ward 1 and 177 from ward 9 (Figure 2). The mean age of study subjects was 32.2 years (SD = 19.1) and 59.8% were females. Table 1 shows the frequency of some potential socioenvironmental risk factors. The median number of family members was six persons per household with an interquartile range (IQR) from 5 to 7. The median yearly family income per capita was US$ 85.7 with an IQR of US$ 49.0–128.6.

Figure 2.

 Inclusions in sample.

Table 1.   Percentage of behavioural and socioeconomic characteristics by ward in Dulari VDC, Nepal, 2003
  Ward 1 (n = 196) Ward 9 (n = 177)Both wards combined (n = 373)
%95% CI%95% CI%95% CI
  1. †Chickens, ducks, goats.

Electricity available80.674.4–85.955.447.7–62.868.663.6–73.3
Cattle near household57.149.9–64.277.470.5–83.366.863.7–73.3
Small animals near household†59.252.0–66.170.162.7–76.764.359.2–69.2
Ponds in proximity19.414.1–25.629.422.8–36.724.119.9–28.9
House with >three rooms49.041.8–56.245.838.3–53.447.542.3–52.7
Sleeping on the ground77.671.1–83.252.044.4–59.565.460.3–70.2
Individual bednet use93.989.5–96.896.692.8–98.795.292.3–97.0
History of migration2.10.6–5.22.30.6–5.72.11.0–4.4

Markers of infection

A positive LST was found in 16/190 persons tested in ward 1 (8.4%, 95% CI: 4.9–13.3), of whom 10 were females, and 32/175 in ward 9 (18.3%, 95% CI: 12.9–24.8), of whom 16 were females. The mean age of LST positives in ward 1 was 28.8 years (SD = 16.7) and 32.6 years (SD = 19.5) in ward 9. In both wards taken together, cumulatively 48/365 (13.2%, 95% CI: 9.9–17.2) persons were found with a positive LST.

A DAT titre of 1:3200 or higher was found in 6/196 persons tested in ward 1 (3.1%, 95% CI: 1.1–6.5), of whom 3 were females, and 22/177 persons tested in ward 9 (12.4%, 95% CI: 8.0–18.2), of whom 12 were females. The mean age of persons with a positive DAT was 24.2 years in ward 1 (SD = 8.9) and 38.7 years in ward 9 (SD = 18.3). Cumulatively, 28 persons (7.5%, 95% CI: 5.1–10.8) were found with a positive DAT in both wards. Table 2 shows the data obtained on frequency of infection. The DAT titre was negative (<1:3200) in 190 persons in ward 1 (96.9%, 95% CI: 93.5–98.9) and in 155 persons in ward 9 (87.6%, 95% CI: 81.8–92.0).

Table 2.   Diagnostic findings (LST, DAT) and history of KA by ward, Dulary community, 2003
 Ward 1 (n = 196)†Ward 9 (n = 177)†Both wards combined
%95% CI%95% CI%95% CI
  1. n = 190, 175 for LST, respectively, in ward 1 and ward 9.

LST positive8.44.9–13.318.312.9–24.813.29.9–17.2
DAT geqslant R: gt-or-equal, slanted1:32003.11.1–6.512.48.0–18.27.55.1–10.8
Current kala-azar (KA)00–2.100–1.900–1.3
History of KA2.00.6–5.18.54.8–13.65.13.2–8.0
DAT geqslant R: gt-or-equal, slanted1:3200 and LST positive00–1.92.80.9–6.51.30.5–3.3
DAT geqslant R: gt-or-equal, slanted1:3200 and history KA2.00.6–5.17.94.4–12.94.83.0–7.7
LST positive and History KA00–1.92.30.6–5.71.10.3–2.9

We did not find any clinical suspect of KA during our field survey. However, a history of KA was reported by 4/196 persons in ward 1 (2.0%, 95% CI: 0.6–5.1) and 15/177 persons in ward 9 (8.5%, 95% CI: 4.8–13.6). Of 19 persons, 17 with a positive history of KA declared having suffered from the disease over the last 5 years prior to the interview (1998–2002). Only 4 of 19 individuals with a known history of KA had a positive LST (21.1%, 95% CI: 6.1–45.6). There were no statistically significant associations between DAT and LST results. Figure 3 gives percentages of positive LST and DAT results by age group. Table 3 shows adjusted odds ratios (OR) of the variables of the final logistic regression model. No association was found with the presence of cattle, the use of bednets or sleeping on the ground. The bednets, not impregnated and in bad condition, were used by 95.2% (95% CI: 92.3–97.0) of the population, but did not show any protective effect.

Figure 3.

 LST and DAT results by age group; number of positive diagnostic tests within age group; Dulari, wards 1 and 9, 2003.

Table 3.   Independent factors associated with leishmanial infection: crude OR with 95% Cl by ward as a result of bivariate analysis; adjusted OR with 95% CI in total sample as estimated by logistic regression). Dulari community, Nepal, 2003.
 Ward oneWard nineBoth wards combinedAdjusted OR95% CI
Crude OR95% CICrude OR95% CICrude OR95% CI
  1. * significant risk factor ( p value < 0.05), but unstable point estimate of OR.

  2. **positively associated factor in bivariate analysis (p = 0.06), but unstable point estimate of OR; significantly associated risk factor in separate logistic regression by ward (p value < 0.05).

Proximity of ponds4.40.9–22.93.51.4–8.64.21.9–9.13.71.6–8.5
Presence of small animals**0.20.1–0.60.60.3–1.30.40.2–1.1
Age ≥ 15 years0.250.0–2.21.20.5–2.91.00.5–2.15.51.2–25.0
Mud house****0.90.4–2.42.51.0–5.93.01.1–7.6
≥6 family members4.00.5–35.32.90.9–9.03.41.3–9.04.41.6–12.6

Independent risk factors for Leishmania infection (persons with DAT geqslant R: gt-or-equal, slanted 1:3200) found significant in multivariate analysis for the entire surveyed area were

  • the proximity of the house to ponds (OR 3.7, 95% CI: 1.6–8.5);
  • a family size of geqslant R: gt-or-equal, slanted6 members (OR 4.4, 95% CI: 1.6–12.6);
  • a house constructed in mud (OR 3.0, 95% CI: 1.1–7.6);
  • age geqslant R: gt-or-equal, slanted15 years (OR 5.5, 95% CI: 1.2–25.0).

However, in separate analysis by ward, because of lack of power, proximity of ponds, age geqslant R: gt-or-equal, slanted15 years and geqslant R: gt-or-equal, slantedsix family members were not significant in ward 1 alone, and mud house was not significant in ward 9 alone.

Discussion

We obtained a high estimate of past leishmanial infection based on LST (8.4% in ward 1 and 18.3% in ward 9) and a lower estimate based on DAT serology (3.1% in ward 1 and 12.4% in ward 9). Living in proximity to ponds, family size and mud-type housing were identified as independently associated factors for infection with L. donovani in this community.

Although we did not find any active case of VL in this cross-sectional approach, we found 2% of the ward 1 and 8.5% of the ward 9 sample reporting to have suffered from KA previously, based on a clinical case definition that is recommended by WHO (WHO 1999).

However, these figures cannot be extrapolated to the entire Morang District, as KA is known to be clustering within the wards 1 and 9 of Dulari VDC, and those wards were purposively selected for the survey. The data shown give an indication of the public health impact of VL in high-prevalence foci in Nepal. The difference of infection estimates between the two wards may be explained by the fact that in ward 1 infection was geographically spread in a rather circumscribed small area only, whereas in ward 9 we found it more evenly distributed throughout the entire ward.

Socioenvironmental conditions were quite different between the two wards. Stagnant water, especially, was more common in ward 9. Disease awareness in Dulari community is quite high, as the outbreak had been ongoing for more than 4 years. Today, people seek help for their symptoms quite rapidly, mainly at BPKIHS hospital, and this might explain why we did not find active VL cases in the field. Our seroepidemiological results correlate well with the results of a study in the Terai lowlands by Koirala et al. (2004), who found that 6% had a DAT titre >1:2000 (n = 1083).

Usually, leishmanial infection prevalence is thought to correlate with LST positivity. However, LST reflects the T-cell-mediated hypersensitivity and is therefore considered as a marker of past infection, whereas the DAT is considered as a marker of more recent infection. Furthermore, the low rate of LST responders among individuals with recent history of VL (21.1%) confirms previous observations on the long timespan (>1 year) required for LST conversion after parasitological cure (Ho et al. 1983). Moreover, in the area of observation, the coverage for BCG vaccination, which was shown to inhibit responsiveness to LST (Rab & Evans 1994), is unknown.

In analogy to the methods of Joshi et al. (1999), the DAT cut-off titre of 1:3200 or higher was used to determine infection with L. donovani. While some cases of infection may have been missed by this approach, we consider this DAT cut-off fairly specific for the purpose of documenting infection. Multiple epidemiological studies have shown that it readily discriminates KA patients from healthy endemic controls, with specificity in the range of 90–100% (Abdel-Hameed et al. 1989; Singla et al. 1993; Boelaert et al. 1999; Joshi et al. 1999).

Our findings concerning the positive association of living in mud-type houses with infection are supported by Bern's observations (2000) that cracked house walls were associated with an increased risk of VL infection. A recent study from India has also shown evidence for mud-plastered house walls as a risk factor for KA (Ranjan et al. 2005). Through drying and erosion, cracks and crevices appear in mud walls and may function as breeding sites for the female sand flies. This may happen in the mud walls of human habitats or animal shelters. As high soil humidity favours an ideal breeding habitat for sand flies, they may be more attracted to houses near ponds. In our study, most families with more than six members (60.5%, 95% CI: 52.5–68.1) did live in mud houses, a factor positively associated with infection. Nevertheless, in the final logistic regression model, both factors remained in the model, showing the independent effects of each variable on infection.

Our results did not support any association between the likelihood of living in a household with cattle and infection, perhaps because in Dulari VDC, the animal population is quite homogeneously and densely distributed within the human population.

Our results indicate that the ownership of small animals (mainly fowl) is negatively associated with infection: holding small animals protected individuals from infection (OR 0.4, 95% CI: 0.2–1.1). A possible explanation could be that sand flies are attracted by fowl more than by humans, and therefore their owners could be protected from transmission. Alexander (2002) discussed the interactions between chicken raising and the transmission of L. infantum by Lutzomyia longipalpis in rural areas of Brazil. According to the authors, chicken attract both male and female Lutzomyia and their presence may protect humans from infection (zooprophylactic effect). Finally, although an associated factor with infection, we were not able to elucidate the role of small animals in this study.

In contrast to the results of Bern et al. (2000, 2005) in our study, the use of bednets was not associated with the absence of L. donovani infection. However, we found bednets in poor condition and unimpregnated in the majority of the cases.

The results of this survey were presented to the Dulari community, and efforts to reduce the breeding sites of the vector population were recommended mainly by house wall spraying and eliminating cow dung and other animal debris from human habitats. As an alternative to house wall spraying, the more sustainable lime and mud plastering technique, developed by Kumar et al. (1995), was discussed. In this technique, the enviromentally friendly and affordable materials, lime and mud, are used for plastering house walls up to a height of approximately 1.50 m in order to smooth the wall surface. Cracks and crevices in walls disappear and peridomestic transmission is reduced.

The role of impregnated bednets in the prevention of VL has never been demonstrated at large-scale level under field conditions in contrast to the effect on the prevention of cutaneous leishmaniasis (Nadim et al. 2000). However, several authors mention that peak transmission of the parasite to humans occurs at night (Srinivasan & Panicker 1992; Dinesh et al. 2001) and intervention studies examining the protective efficacy of impregnated bednets are needed to draw any conclusions.

Acknowledgements

This study was conducted with the support of the Institute of Tropical Medicine (ITM) in Antwerp and the B.P. Koirala Institute of Health Sciences (BPKIHS) in Dharan, Nepal. Our special thanks go to D. Jacquet from ITM Applied Technologies & Production Unit for test execution and to P. Desjeux at WHO. Furthermore, we thank all medical staff members of the tropical ward at BPKIHS and the Dulari community health workers involved in the field work.

Ancillary