Serious underreporting of visceral leishmaniasis through passive case reporting in Bihar, India

Authors


Corresponding Author Dr Shyam Sundar, Department of Medicine, Institute of Medical Sciences, Banaras Hindu University, Varanasi 221 005, India. Tel.: +91 542 2367795, Fax: +91 542 2367568, E-mail: shyam_vns@sify.com

Summary

Objectives  Visceral leishmaniasis (VL) is a major public health problem in Bihar, India. Unfortunately, accurate data on the incidence or prevalence of the disease are not available. This longitudinal study was undertaken to determine the incidence of VL in a Community Development Block area of the state of Bihar. Survey results were compared with official reports of the disease to assess the extent of underreporting by the Government health system.

Methods  Three health subcentre areas in Kanti Block, consisting of 14 villages with a total population of 26 444, were selected. Active surveillance was performed every month from January 2001 to December 2003 by house to house survey to detect cases of fever for more than 15 days. Patients clinically suspected of suffering from VL were subjected to parasitological examination for confirmation. Analysis of records of the reporting agencies in the district was undertaken to compare and assess the extent of underreporting.

Results  A total of 202 cases of VL were identified in 3 years giving an average annual incidence rate of 2.49/1000 population (95% CI = 2.15–2.83). As identification data of patients was not available with the official reporting agencies for 2001 and 2002, extent of underreporting could be assessed for 2003 only. In the study population, 65 cases of VL were detected during 2003 providing an annual incidence rate of 2.36/1000 population. Only eight (12.30%) cases were reported officially, resulting in underreporting by a factor of 8.13. In 2003, the official incidence rate of VL for Kanti Block was 0.31/1000 against the actual rate of 2.36/1000. As the constraints for official reporting at the block and the district levels are similar, the underreporting at district level was also assumed to be similar. This finding has significance in the preparation for elimination programme.

Abstract

Objectifs  La leishmaniose viscérale (LV) est un problème majeur de santé publique à Bihar en Inde. Malheureusement, des données précises sur l'incidence ou la prévalence de la maladie ne sont pas disponibles. Cette étude longitudinale a été menée afin de déterminer l'incidence de la LV dans une zone formée de bloc de développement d'habitations communautaires dans l’état de Bihar. Les résultats ont été comparés aux reports officiels sur la maladie afin de mesurer l'ampleur du sous report par les systèmes de santé gouvernementaux.

Méthode  Trois zones de santé hors du centre dans le bloc de Kanti, comprenant 14 villages, avec une population de 26444 habitants ont été sélectionnées. Une surveillance active a été menée mensuellement, de janvier 2001 à décembre 2003 avec visite de chaque maison pour détecter les cas de fièvre de plus de 15 jours. Les patients suspects de LV ont été examinés pour confirmation parasitologique. L'analyse des agences de reports du district a été menée pour comparer et évaluer l'ampleur du sous report.

Résultats  Au total 202 cas de LV ont été identifiés sur 3 ans, ce qui correspond à un taux d'incidence annuelle moyenne de 2,49/1000 (IC95%: 2,15–2,83). L'ampleur du sous report des cas a seulement étéévaluée pour 2003, car les données d'identification des patients n’étaient pas disponibles pour 2001 et 2002. Dans la population étudiée, 65 cas de LV ont été détectés en 2003, ce qui donne un taux d'incidence de 2.36/1000. Seuls 8 de ces cas (12,3%) avaient été rapportés officiellement, ce qui correspond à un sous report 8,13 fois plus bas. En 2003, le taux d'incidence officielle de la LV pour le bloc de Kanti était de 0,3/1000 par rapport à un taux actuel de 2,36/1000. Comme les contraintes pour le report officiel au niveau du district et du bloc sont similaires, le sous report au niveau du district est supposé aussi être similaire. Cette observation a une signification importante dans les concepts de programme d’élimination de la maladie.

Abstract

Objetivos  La leishmaniasis visceral (LV) es un problema importante de salud pública en Bihar, India. Desafortunadamente, no existen datos disponibles sobre la incidencia o prevalencia de la enfermedad. Este estudio longitudinal se realizó con el fin de determinar la incidencia de LV en un área del Bloque para el Desarrollo Comunitario (Community Development Block) del estado de Bihar. Los resultados de la encuesta se compararon con los reportes oficiales de enfermedad para evaluar el nivel de subreporte del sistema de salud gubernamental.

Método  Se seleccionaron tres áreas con subcentros de salud en el bloque de Kanti, compuestas por 14 pueblos y una población total de 26,444 habitantes. Con el fin de detectar casos de fiebre con una duración mayor a los quince días, se mantuvo una vigilancia activa mensual entre Enero 2001 y Diciembre 2003, realizada mediante encuestas casa por casa. Los pacientes con sospecha clínica de LV fueron sometidos a pruebas parasitológicas. También se analizaron los archivos de agencias de información del distrito, para comparar y evaluar la extensión del subreporte.

Resultados  Se identificaron 202 casos de LV en tres años, con una tasa de incidencia anual promedio de 2.49/1000 habitantes (95% CI = 2.15–2.83). Puesto que las agencias oficiales no tenían datos de pacientes para los años 2001 y 2002, el nivel de subreporte solo pudo evaluarse para el 2003. En la población de estudio se detectaron 65 casos de LV durante el 2003, con una tasa de incidencia anual de 2.36/1000 habitantes. Solo 8 del total de casos (12.30%) fueron reportados oficialmente, resultando en un subreporte de 8.13 veces. En el 2003, la tasa de incidencia oficial de LV para Kanti Block fue de 0.31/1000 habitantes, mientras que la tasa real era de 2.36/1000 habitantes. Puesto que las dificultades para el reporte oficial en el block y el distrito son similares, el subreporte a nivel de distrito se asume ha de ser similar. Este hallazgo tiene un especial significado para la preparación del programa de eliminación.

Introduction

Visceral leishmaniasis (VL, kala-azar) is a dreaded disease, caused by over 20 species of Leishmania and transmitted by the bites of female sandfly (Desjeux 1992). It is endemic in 62 countries with 200 million people at risk (Guerin et al. 2002). The global annual incidence of VL is estimated to be 500 000 (Desjeux 1996; Guerin et al. 2002); its prevalence, 2.5 million cases (WHO 1998). Over 90% of VL cases occur in five countries: India, Bangladesh, Nepal, Sudan and Brazil (Desjeux 1996; Murray 2002). If left untreated, clinical VL is usually fatal. The worldwide deaths caused by VL in the year 2003 are estimated to be 59 000 (WHO 2003).

The VL is a major public health problem in the state of Bihar, and adjoining areas of states such as West Bengal, Jharkhand and Uttar Pradesh in India (Bora 1999). The disease is also endemic in neighbouring Bangladesh and Nepal. The state of Bihar accounts for more than 90% of the cases in the country (Lal et al. 1996; Bora 1999; Sundar et al. 2000). The Government of India aspires to eliminate VL by 2010 from India. For prevention and control of any disease, health policy planners need to know the extent of the problem in a specified area, besides other epidemiological parameters, before launching any intervention program. For this an effective community-based surveillance system is required to find out the cases. Unfortunately, in the developing countries like India, a proper surveillance system is lacking for most of the diseases.

There is no doubt that the number of cases occurring in India and around the world is considerably greater than the officially reported figures and the number of infections greater still (Badaro et al. 1986; WHO 1990; Ashford et al. 1992; Seaman et al. 1992) as there are subclinical cases. The disease shows an iceberg phenomenon (WHO 1988, 1994; Seaman et al. 1992). Most of the official data are obtained exclusively through passive case detection (Desjeux 1996). There are no studies to determine the incidence and prevalence of the disease in India. In the state of Bihar, there is no active surveillance system, and similar situation exists in other states endemic for the disease. In Bihar, only those VL cases are being reported by the Government who are registered for treatment with the public health care delivery system i.e. Primary Health Centres (PHC), District Hospitals and Government Medical colleges. The objective of the present study was to determine the incidence of VL in a pre-determined area of the state of Bihar and to estimate the extent of underreporting through passive case reporting by the Government health system.

Materials and methods

This study was conducted in the district of Muzaffarpur, one of the districts with highest VL endemicity in the state. The district's total population is 3.743 million, of which 3.40 million live in rural areas. There are 14 Block PHCs in the district. Kanti Block with an estimated population of 352 000 in 2003 (considering the 2001 census population of 337 670 and a growth rate of 2.1% in the state) was selected for the present study (2001 census of India). There are 48 subcentres in the Kanti Block PHC, each catering to a population ranging from 6000 to 10 000. These subcentres were arranged into 16 clusters of three geographically contiguous subcentres in each cluster and one cluster was selected randomly. This approach was adopted to improve the operational feasibility. The three subcentres in the selected group had 14 villages and every household in these villages was covered (Figure 1). A house-to-house survey was performed from September 2000 to March 2001 to get information about socio-demographic and environmental characteristics. The total population of the study area was 26 444 distributed in 4083 households in 2001. Field investigators were trained in data collection methods to minimise inter-investigator variation. The data was collected using pre-tested, semi-structured schedules. Registers and reports of all the Block PHCs, district hospital and medical college located in the district Muzaffarpur were undertaken for the years corresponding to the study period (2001–2003).

Figure 1.

 Map of the blocks of Muzaffarpur district, study area is shown by an arrow in the Kanti block.

Sample size

In the year 2000, a retrospective survey was undertaken in the study area and collected information about cases of VL that occurred during past 10 years using a significant political event as a cut off point. From that data, we obtained an annualised VL incidence rate of approximately 2.5/1000 population. Based on this rate and using a permissible error of 15%, a sample size of 70 933 was required for the study to estimate the incidence of the disease (using the formula 4 pq/L2, where L is permissible error). In order to make the study operationally feasible we decided to conduct the survey in a population of 26 444 over 3 years. As we used peripheral healthcare units (subcentres) as sampling units for selection, we had to cover this population yielding a cumulative figure of 81 009 person years over 3 years (Table 1). To detect sizeable difference between cases reporting at private and public health facilities, the sample size requirement was calculated using the same formula. From our initial data we estimated that 80% of the VL patients were contacting private facilities. Taking permissible error as 15%, a total of 45 VL patients were required.

Table 1.   Annual and cumulative average annual incidence rate of visceral leishmaniasis (per 1000 population) in the study area from 2001 to 2003
Year of occurrenceTotal population*Number of patientsIncidence rates
  1. * By population enumeration in 2001 and for subsequent years calculated by taking the exponential growth rate of 2.1% (census 2001).

200126 444652.46 (1.86–3.06)
200226 999722.67 (2.05–3.29)
200327 566652.36 (1.79–2.93)
Total81 0092022.49 (2.15–2.83)

Diagnosis of visceral leishmaniasis cases

Any individual with fever for more than 15 days with or without splenic enlargement was considered to be a ‘suspected’ case of VL. For those opting for treatment at Kala-azar Medical Research Centre (KAMRC), a non-governmental organisation (NGO) in Muzaffarpur, all the patients were subjected to parasitological examination and a patient was considered to be a ‘confirmed’ case of VL only when the diagnosis was made by the identification of amastigotes in the splenic or bone marrow aspirate. Patients seeking care at other facilities (private doctors, private hospitals and public health institutions) were considered as positive based only on positive parasitological reporting.

Surveillance for VL cases

In this longitudinal study all the houses in the study area were given a house number and that was displayed at a suitable visible place with an indelible marker. For the identification of VL, a three pronged strategy was adopted to minimise the missed cases. (i) The field workers visited every household at monthly intervals and enquired for any case currently treated for VL either in public or private sector and for cases having fever for more than 15 days. (ii) Opinion leaders in the villages were motivated to participate voluntarily to give information about such cases. At least one such leader was identified in each hamlet. (iii) Pamphlets were distributed and pasted at important public places to give information about the signs and symptoms of the disease and to report at a designated place. The suspected cases were followed regularly to know the final diagnosis. These suspected and any other partially treated or relapse cases were offered diagnosis and treatment free of cost to KAMRC at Muzaffarpur.

Informed consent

Households were briefed about the nature of the study and the consent of the subjects/head of the household was taken on a written consent form. The study was approved by the Ethical Committee of the Institute of Medical sciences, Banaras Hindu University, Varanasi.

Calculation of incidence rate

For the calculation of incidence rates all the confirmed cases of VL occurring from January 2001 to December 2003 were considered. The date of diagnosis was considered as the period of occurrence of the disease. Our enumerated population was considered as the denominator for 2001 and for subsequent years extrapolated population using a growth rate of 2.1% (2001 census of India) was used.

Existing system of case reporting

Visceral leishmaniasis case reporting in the state is based on the cases seeking treatment from public health care facilities (Block PHCs, District hospitals and Medical colleges). These cases are reported every month to District Malaria Control Unit for onward transmission to state capital. There is no active surveillance system in place. Therefore, a large proportion of VL patients (approximately 80%) who seek treatment from non-governmental sector do not get reported at all. Non-availability of regimens other than sodium antimony gluconate (SAG) at Block PHC contributed to the preference of patients for non-governmental health facilities.

Estimation of underreporting

For the estimation of under-reporting of VL, only the figure of 2003 was considered, as the required information to ascertain the permanent residential addresses of the patients was not available in 2001 and 2002 with the PHCs and the district hospital. To ascertain the reported cases of the study area, the list of cases treated at the three public health facilities were matched with our list, and if needed, verified in the field. The list of an adjacent block PHC, where the patients from the study area could go, was also matched. Underreporting was calculated for the 14 study villages and the entire Kanti Block. However, it has two assumptions for these calculations. First, all the cases of VL from the study area were identified by our surveillance activities and second, the incidence of the disease in the block is similar to the study area. Under reporting was extrapolated to district level assuming that the treatment seeking behaviour and consequently level of passive detection were similar in the entire district.

Results

Incidence and profile of VL cases in the study area

In the study area, the total number of VL cases found by the surveillance was 65, 72 and 65 in 2001 2002 and 2003, respectively and the corresponding ‘incidence rates’ were 2.46, 2.67 and 2.36 per 1000 population. The average annual incidence rate was 2.49/1000/year (95% CI 2.15 and 2.83) (Table 1). Of the total 202 VL patients, 62.4% were males as compared with 53.5% in the total population (statistically significant: χ2 = 5.99; P < 0.014) and 47.0% were below 15 years of age as compared with 41.9% in the population (not significant: χ2 = 1.96; P > 1.64) were below 15 years of age (Table 2). The maximum numbers of patients (42.1%) were diagnosed from January to April. The average duration of illness was 3.18 months. Out of the total cases, 88.61% were either treated by private doctors or NGOs.

Table 2.   Profile of visceral leishmaniasis patients diagnosed from January 2001 to 2003 in the study population
CharacteristicsFrequency (n = 202)PercentTotal population (%)
  1. PHC, Primary Health Centres.

  2. * Not significant, † significant.

Age of the patients*
 0–14 years9547.041.9
 15+ years10752.958.1
Sex†
 Male12662.453.5
 Female7637.646.5
Month of diagnosis
 January–April8542.1
 May–August7034.6
 September–December4723.3
Duration of illness
 <2 months6632.7
 2–3 months8642.6
 3–4 months2612.9
 4 months and more2411.9
Place of treatment
 PHC94.46
 District hospital/ medical college146.93
 NGO/charitable hospital14270.30
 Private doctor3718.31

Reporting of VL cases from the study area and the Kanti block

The incidence rate according to the officially reported cases at the block level was calculated only for the year 2003, as the complete addresses of the patients were not available to ascertain their actual place of residence for the years 2001 and 2002. Of the 65 cases identified in the sampled villages during 2003, only eight cases (12.3%) were treated and reported by the public health facilities. Extrapolating the average incidence rate of the study area (2.49/1000) for the entire Kanti block we estimated a total of 876 VL cases in the block in 2003. After scrutinising the addresses of the patients in the records it was found that block PHC treated and reported 77 (8.77%) cases and an additional 32 cases (3.66%) were reported by district hospital, medical college and adjacent block PHCs (Table 3). Thus the total cases reported from the block were 109 (12.6%). As per the reported cases, the annual incidence rate for Kanti Block should have worked out as 0.31/1000 population (Table 4).

Table 3.   Percentage of the visceral leishmaniasis cases reported by public health care facilities from the study area and from the entire Kanti block during 2003
Place of reportingReported from the study areaReported from the entire Kanti block
Number of cases (n = 65)*Percent of the total casesNumber of estimated cases (n = 876)†Percent of the total estimated cases
  1. PHC, Primary Health Centres.

  2. * Total number of cases that occurred in the study area.

  3. † Total number of cases estimated for the entire Kanti block.

Kanti block PHC23.08778.78
District hospital23.08121.37
Medical college34.61161.83
Adjacent block PHCs11.5340.46
Total812.3010912.44
Table 4.   Incidence of visceral leishmaniasis (VL) in Kanti block and the entire Muzaffarpur district according to the number of officially reported cases by the public health care delivery system from 2001 to 2003 (incidence rate per thousand population)
YearKanti blockMuzaffarpur district
No. of VL cases*Incidence rate†No. of VL cases*Incidence rate**
  1. * Population is given in parentheses.

  2. † 95% Confidence interval is given in parentheses.

20011161 (3743 836)0.310 (0.292–0.328)
20021403 (3822 456)0.367 (0.348–0.386)
2003109 (352 000)0.31 (0.252–0.368)1270 (3902 727)0.325 (0.307–0.343)

Incidence of VL in Muzaffarpur district according to the official reporting

The incidence rates at the district level according to the officially reported cases were 0.310, 0.367 and 0.325 for the years 2001 2002 and 2003, respectively (Table 4) and the average annual incidence rate for the three years was 0.334/1000 population (95% CI = 0.323–0.345).

Discussion

This is the first report of incidence of VL based on active house-to-house survey from the endemic region of India. According to our estimates, the average annual incidence rate for VL was 2.49/1000 population (95% CI = 2.15–2.83) for the study area. More than two-thirds (70.3%) of the total 202 cases were treated by NGOs/charitable hospitals, because patients are offered free diagnosis and treatment at these places. Most of these patients are poor with very low income and cannot afford to pay for the treatment and therefore, prefer organisations offering free services. We took adequate precautions to identify patients of our study villages who had migrated out of the area to seek treatment for VL and to exclude the cases who were not the permanent residents of the study area.

Of the 202 VL cases identified in the study population, 201 were identified by our surveillance activities. Only one patient, who could not be detected, was identified while matching the cases treated at the adjacent block PHCs. He had moved to that area during his treatment. Missing a few more VL cases is possible because patients might have shifted to other places for their treatment and this information could not be elicited from the family members. Also, there is no established policy or documentation system for referring patients from block PHC to district hospital or medical college. Patients are free to seek treatment at any place and from any health system they want. Therefore, it is also possible that a patient is partially treated at one health facility and then switches to another facility or and his name is enlisted at both the places. It is also likely that a patient after treatment at one facility gets a relapse and goes to another facility and figures twice in the reporting system. Therefore, there could be a slight variation in the actual incidence on either side.

We found that during the year 2003 the proportion of patients seeking treatment from government health system (and therefore reported) in the study population and the whole block was 12.3% and 12.44%, respectively of the documented or estimated cases, giving corresponding underreporting figures of 8.13 and 8.04 times respectively. The fault lies in the reporting system which picks up only those VL cases which are registered for treatment with the public health care delivery system i.e. PHC, District Hospital and Government Medical colleges. The vast majority of patients opt for the private health care system and are thus left out of the reporting system. An attempt was made to find the reasons for poor utilisation of public health facility (data not given). Probably, distance of the facility, availability of drugs, type of drugs available and attitude of the healthcare providers are the important determinants for seeking treatment from government health facility. At the time of collecting data only SAG was available at Block PHC which is largely ineffective in the region under study (Sundar et al. 2000). Amphotericin B, the only effective drug available in the public health sector, was available only at the district hospital and the medical college, that too very irregularly. When effective drugs are not available at the public health outlets, people, who otherwise would have preferred public health facility, are forced to go to private doctors or NGOs. This was vindicated by the fact that in the study area only 11.39% of the 202 patients sought treatment from the government sector (Table 2) during 3 years of the study period (information based on interviewing the family members).

If we consider similar determinants for utilisation of government health facilities for treatment and official reporting of VL cases at the block and district level then reporting and incidence rate at the district level would also be approximately eight times less. Movement of patients from one district to another for treatment is possible and if immigrant and emigrant populations are similar, it is not going to affect the extent of the disease burden in the district. It was also noticed that in the medical college a few patients are also coming from the neighbouring country of Nepal and in our analysis we had to overlook this fact.

Recommendation

As there is no active surveillance system for the disease and only those cases are officially reported who seek treatment from the public health system; to improve the reporting system, Auxiliary Nurse Midwifes (now called Female Multipurpose Workers), working at the subcentre level, should be given the responsibility to identify the cases in the community during her home visits, irrespective of the place of treatment and then report these at the PHC level. An effort may also be made to collect information at least from reputable private practitioners and NGOs about the cases treated by them every month. For this, a person working at the district hospital could be given responsibility. It would minimise the extent of underreporting.

Acknowledgements

This work was supported by WHO/World Bank/UNDP Special Programme for Research and Training in Tropical Diseases (ID 99106). We are also grateful to Prof. R.N. Mishra, Biostatician, department of community medicine, IMS. BHU and the staff members of KAMRC for their valuable suggestions and cooperation.

Ancillary