1. Performance of kala-azar surveillance in Gaffargaon subdistrict of
Mymensingh, Bangladesh
AUTHORS AND AFFILIATIONS
Kazi Mizanur Rahman1,2
, Indira Samarawickrema2
, David Harley2
, Anna Olsen3
, Colin D
Butler2,4
, Dinesh Mondal1
, Shariful Amin Sumon1
, Subrata Kumar Biswas1
, Stephen P
Luby1,5
, Adrian C Sleigh2
1
International Centre for Diarrhoeal Disease Research, Bangladesh, Dhaka, Bangladesh
2
Australian National University, Canberra, ACT, Australia
3
University of New South Wales, Sydney, NSW, Australia
4
University of Canberra, Canberra, ACT, Australia
5
Stanford University, Stanford, CA, USA
ABSTRACT
Background: Elimination of kala-azar is planned for South Asia and this will require good
surveillance along with other strategies. We assessed surveillance in Gaffargaon upazila (a
subdistrict of 13 unions) of Mymensingh district, an area of Bangladesh highly endemic for
kala-azar.
Methods: In 4703 randomly sampled households within nine randomly sampled villages
drawn from three randomly sampled unions, we actively searched for kala-azar cases with
onset between January 2010 and December 2011. We then searched for medical records of
these cases in the patient registers of Gaffargaon upazila health complex (UHC). These
records form the data source for monthly reporting. We also investigated factors associated
with the medical recording by interviewing the cases and their families. We also did a general
1

2. observation of UHC recording systems and interviewed health staff responsible for the
monthly reports of kala-azar cases.
Results: Our active case finding detected 58 cases but 29 were not recorded in the
Gaffargaon UHC patient and laboratory registers. Thus, only 50% of cases were reportable
via the government surveillance system. Interviews with health staff reveal the heavy
reporting burden for multiple diseases, variation in staff experience and high demands on the
staff time. Recording was more likely for those age 18 years or more, males, with an
educated father, palpable spleens and abdominal pain. The odds of being recorded in the
UHC registers were 18 times higher for patients treated with miltefosine than sodium
antimony gluconate (95% CI = 3.4 – 100.3). Conclusions: Fifty percent of kala-azar cases
occurring in one highly endemic area of Bangladesh were recorded in registers that were the
source for monthly reports to the national surveillance system. Recording was influenced by
patient, treatment, and staff factors. Our findings have policy implications for the national
surveillance system.
Keywords: Visceral leishmaniasis, kala-azar, elimination, surveillance, Bangladesh
2

3. INTRODUCTION
Visceral leishmaniasis (VL) is caused by Leishmania protozoan parasites transmitted by
Phlebotomine sandflies[1]. VL is a systemic illness with significant morbidity and mortality
and is characterized by fever and splenomegaly. The disease is endemic in impoverished
tropical areas globally [2, 3]. In South Asia the disease is known as kala-azar meaning ‘black-
fever’ and only affects humans, but elsewhere VL is also enzootic.
A memorandum of understanding signed by the governments of India, Bangladesh and Nepal
in 2005 states that kala-azar will be eliminated by 2015. Elimination is defined as annual
incidence of less than 1 per 10,000 population at the district or sub-district level [4]. The
elimination programme has four phases – preparatory, attack, consolidation, and
maintenance. By late 2013 the elimination programme had reached the end of the preparatory
phase (establishment of strategies including pilot insecticide spraying, facilities for diagnosis,
treatment, and surveillance, passive and active case detection and vector monitoring). The
attack phase is beginning in 2014 and will include implementation and monitoring of all
elimination strategies[5], and additional vector control activities. Accordingly the focus is on
vector control (indoor spraying, treated bed nets, environmental management and community
participation) as well as access to early diagnosis and complete case management, and
continued transmission surveillance.
Surveillance is crucial in monitoring disease and is identified as one of the main elimination
strategies for kala-azar [5]. India’s surveillance system reports 12 – 20 percent of incident
cases with improving ascertainment noted by 2010 [6-8]. In Bangladesh, the surveillance
provides estimates and trends of the nationwide occurrence of kala-azar [9]. Kala-azar is
reported every month starting from the upazila or subdistrict level. There are also research
3

4. data for selected endemic sub-districts extending back up to eight years and they usually
provide kala-azar rates higher than the government surveillance system [3, 10-12]. But a
variety of methods were used and results are not comparable with surveillance data.
Furthermore, they cannot be repeated regularly nationwide. Thus, the government led passive
surveillance system provides the only ongoing nationwide data and hence the only
nationwide indicator of progress towards the elimination goal.
Here we report a community-based study of kala-azar surveillance in a highly endemic sub-
district of Bangladesh. We estimate the proportion of incident kala-azar cases actually
recorded in the medical registers of the sub-district health facility, the source of data for the
report of monthly case counts. We analyse factors influencing case recording and make
recommendations to improve surveillance.
METHODS
Study setting
The study was conducted in Gaffargaon sub-district of Mymensingh district in Bangladesh.
Gaffargaon is located 80km north-east of Dhaka, the capital of Bangladesh. The sub-district
occupies 398 sq. km. with a population of 430,000 [13]. Gaffargaon is divided into 15 unions
and 214 villages. Around half of the population of Gaffargaon are male and half are literate.
The majority of the working population are farmers. Among the sub-districts , Gaffargaon
reported the third highest number of cases of kala-azar in 2011 [14].
Study sample and case detection
4

5. The sampling strategy here reported was developed for a larger study identifying barriers to
early diagnosis and case management of kala-azar in Fulbaria and Gaffargaon sub-districts of
Mymensingh district. In the barriers study we compared the two sub-districts measuring
diagnostic delay of incident cases arising over a two year period (Jan 1, 2010 – Dec 31, 2011)
as an indicator of the success of the two different elimination programmes - international
non-governmental organization (Fulbaria) and Bangladesh government (Gaffargaon). We
calculated the sample size of incident kala-azar cases using an estimate of 20 days difference
(i.e. 60 vs 40) in mean delay to diagnosis between the two sub-districts, 80% power, 95%
confidence and a design effect of 1.3 (clustering of unions and villages in a multistage
sampling). These statistical assumptions were based on findings from a survey conducted by
KMR in 2009 in Fulbaria (unpublished data). We thus concluded that 75 kala-azar cases were
required from each of the two sub-districts for the barriers study.
For the surveillance study reported here, we initially calculated sample size needed in each of
the two sub-districts to detect 50% under-reporting with 10% precision and 90% confidence
level. A sample of 68 kala-azar cases – a smaller number than that needed for the overarching
barriers study – was indicated on the basis of the power calculation using Epi Info StatCalc to
calculate the sample size needed to measure a single proportion. Therefore, the sample size of
75 needed for each of the two sub-districts for the barriers study also suffices for the
surveillance study. We focused on Gaffargaon because passive reporting via the government
system operates here. We also cross-checked in Fulbaria by noting medical recording of kala-
azar cases by the treating clinic operated by Médecins Sans Frontières (MSF) [15].
We expected our house to house case finding in Gaffargaon would yield 75 kala-azar cases
from 5400 households based on cumulative incidence of 1.4 cases per 100 households over
5

6. two years in Fulbaria (data not published). We did multistage cluster sampling in Gaffargaon
(Figure 1). First, we excluded two of 15 unions as they were part of the Gaffargaon central
administrative area where the upazila health complex (UHC) is located and we expected that
careseeking would be different from the other 13 unions. Then, from the remaining 13 unions
we randomly sampled three unions and in each of these unions we randomly selected three
villages. We then sampled households in each village, aiming for about 600 households per
village. If the village was large, we sub-divided it into different para or clusters, and
randomly sampled clusters to get 600 households. The resulting sample included a total of
4703 households and we sought current or past kala-azar cases in each of these sampled
households.
Going from house to house in the period December 2011 to May 2012, we sought cases that
had occurred between January 2010 and December 2011. In each household the informant
was a case or a senior family member. Our case definition required diagnosis by a qualified
health care provider based on clinical presentation and a positive confirmatory diagnostic
test.
Data collection: interviews, observations and UHC records
If there was more than one kala-azar case in a household, we studied the earliest case. Kala-
azar cases and their families were interviewed using a structured questionnaire. For cases less
than 18 years of age, a parent or carer was interviewed. Surviving family members of
deceased cases were interviewed. We collected demographic and economic data, date of
symptom onset, signs and symptoms, care seeking experiences and diagnosis.
We also studied recording and reporting of kala-azar cases making general observations at
the Gaffargaon UHC. As well, we interviewed three key health service staff from the
6

7. Gaffargaon UHC who were involved in the kala-azar reporting system. We asked them to
detail their role in kala-azar reporting. We asked about obstacles to reporting and how the
reporting system could be improved. These semi-structured interviews were audio-recorded.
We examined Gaffargaon UHC records to determine if cases we detected had been entered in
the registers. Inpatient, outpatient and laboratory registers were reviewed. We sought record
entries in the 2010-2011 period for all kala-azar cases we detected in the community. We
used the name and address to confirm identification of the patients in the register books.
Data management and analysis
Quantitative data were digitized using Epi Info (version 3.5.3). Information from the patients
and families was recorded in MS Excel in a file which also noted whether the patient was
recorded in the UHC registers. Qualitative data included extensive notes on general
observations and interview recordings. The recordings were transcribed in the Bengali
language for the three UHC staff.
Quantitative data were analysed using STATA (version 8). We calculated the proportion of
our study cases found in the Gaffargaon UHC records and compared frequencies of patient
and health system factors for those found and not found in the hospital records. The factors
were age, sex, education, symptoms, companion to clinic, place of diagnosis, involvement of
private providers, treatment location, drug and year.
Interview transcripts were compared and contrasted for recurring themes and informative
quotations related to the research questions. Quotations used in this publication were
7

8. translated into English by author KMR. All analyses were supported by the general
observations.
Ethics
Informed written consent was obtained from participants. For illiterate interviewees, the
consent form was read out loud and the participants’ fingerprints were obtained on the
consent forms. Before conducting the study of the kala-azar reporting system and
interviewing staff of Gaffargaon UHC, written permission was obtained from the relevant
authority based centrally in Dhaka and at Gaffargaon. Individual consent was also obtained
from interviewed staff who were advised that they would not be identified. Ethical approvals
were provided by the Human Research Ethics Committee of the Australian National
University and the Ethical Review Committee of the International Centre for Diarrhoeal
Disease Research, Bangladesh (icddr,b).
RESULTS
Study participants
We screened 4703 households and identified 89 people fulfilling our case definition of kala-
azar. After exclusions (refused to participate, migrated out, extra cases in a family and
special care), 58 cases remained for analysis in the study (Figure 2).
Kala-azar patient flow and monthly reporting
Our general observations revealed the flow of kala-azar patients and related information at
Gaffargaon UHC (Figure 3). We identified three registers (laboratory, kala-azar and
admission) where staff recorded the patient name and other details. Each month, using the
8

9. kala-azar register as the final source of the information, the responsible nurse reports to the
UHC statistician the consolidated data on the number of kala-azar cases (by age, sex,
treatment status). Then the statistician compiles a monthly kala-azar tally along with the
routine ‘disease profile’ report and sends this information to the Civil Surgeon’s office which
is responsible for that district (Figure 4).
Surveillance of kala-azar in Gaffargaon: active vs passive case detection
After reviewing the UHC registers we concluded that there were no records for 29 of our 58
actively detected cases. Thus, overall, only 50% were detectable by the government passive
surveillance system - 44% in 2010 and 59% in 2011.
Factors influencing kala-azar reporting
Using the survey data we explored patient (Table 1) and health system factors (Table 2) as
determinants of reporting. Age 18 years or more, male sex, educated father, palpable spleen
and abdominal pain increased the likelihood of being recorded by the UHC; these effects
were quite substantial but (constrained by the small sample size) were not statistically
significant (Table 1). We did not find any association between wealth and reporting (data not
shown).
Almost all (55 of 58) the study patients visited the Gaffargaon UHC in seeking care,
indicating that non-attendance was not associated with non-recording. Twenty four patients
(41%) were diagnosed at the UHC and 31 (54%) were diagnosed by a private health care
provider. The odds of recording were lower for those diagnosed at public, relative to private,
facilities (OR = 0.7; 95% CI = 0.2-1.9) (Table 2).Thirty received oral miltefosine tablet and
28 received injectable sodium antimony gluoconate (SAG). The odds of being recorded in the
UHC registers were 18 times higher for patients treated with miltefosine than SAG (95% CI
9

10. = 3.4-100.3). The proportion supplied with any kala-azar treatment drug from the UHC was
100% and 78% among those recorded and not recorded, respectively. Receiving kala-azar
treatment at the UHC, relative to elsewhere, was associated with 80% lower odds of being
recorded.
Interviews with staff indicated a number of factors influencing reporting including a lack of
confidence in their knowledge of kala-azar. There were other issues such as the burden of
recording numerous notifiable diseases on the monthly ‘disease profile’ report.
“It is not just one report that we need to prepare from the hospital every month … All
diarrhoea, ARI, kala-azar (cases) …(plus)… reports from the field (administrative
level lower than the subdistrict), we send all these consolidated figures.”
As well, the reporting system itself has changed, leading to new routines along with a new
kala-azar reporting form.
“We used to send the report on kala-azar with other diseases in the monthly ‘disease
profile’. In that form there was a row for kala-azar. Later a separate form for kala-azar
was introduced. We now send reports on kala-azar both through the ‘disease profile’
as well as in the separate form.”
The informant indicated that reporting channels were sometimes complex and duplicated.
This reflected the need to get the information through quickly but this can be difficult in hard
copy due to the need for a signature by the Upazila Health and Family Planning Officer
(otherwise known as the Thana Health Administrator or THA).
10

11. “We send the reports to the office of Civil Surgeon. We often send them through
email. Sometimes we send the numbers over telephone if there is any hurry. Even
through mobile phone. …We send the hard copies after getting it signed by the THA.”
In addition, different nursing staff recorded the patient data at different times. This may cause
the quality or completeness of recording to vary. However, the system used was unlikely to
duplicate reporting.
“We put a serial number ... this serial number is the registration number. …
We show the patient only once.”
DISCUSSION
Only half of the kala-azar cases arising in Gaffargaon sub-district, a highly endemic area
within Mymensingh district, were recorded in the upazila health records. Consequently the
monthly tallies of kala-azar cases reported to the government of Bangladesh represent only
about half of the actual cases that occur. Investigation of socio-demographic and health
system factors associated with non-recording was generally uninformative – most factors
were not significantly associated and effect estimates had wide confidence limits. Limits
were exceptionally wide for the odds ratio linking miltefosine to recording (3.4 – 100.3), but
the point estimate indicated a strong association for this health system factor (18.4).
Previous population based studies in Bangladesh have not reported surveillance performance
in any way comparable to our study [3, 10-12]. In India, Singh et al (2006) assessed kala-azar
surveillance in Bihar in 2003. They searched households for people with fever for over 15
days duration and confirmed kala-azar with microscopic parasite identification in spleen or
bone marrow aspirates. Sixty-five cases were detected through their active case search and
11

12. only 8 (12.3%) were reported [7]. The same group in a later study found that only 17% of the
cases were reported [8]. Those over 30 years of age were significantly less likely to be
reported, but no other patient, family or health system factor was shown to be important.
Our study was done on a two-year sample of kala-azar patients in a geo-demographically
defined segment of the Bangladesh population in a highly endemic area. We achieved our
principal aim of estimating with reasonable accuracy the proportion of incident kala-azar
cases being recorded by the health system (i.e. 50%). But our sample size did not have
statistical power to enable conclusive sub-analyses of factors related to surveillance.
However, the qualitative data we collected were able to provide some perspective on the
issue of preparing tallies and reporting to the government. The system in use involves
recording in a kala-azar register and an admission register which are sources for reporting
kala-azar to the statistician to enable the monthly tally. The staff we interviewed made
comments about the data flow and forms, and the figure we produced showed a rather
complex system that could be simplified.
We were not able to determine the actual proportion reported because we could not separate
the contribution expected from our sample from the contribution expected from the rest of the
population served by the same UHC. If we could have searched for cases in the entire
population of Gaffargaon (around 430,000) for a particular month then we would be able to
compare the number of cases detected in the community and number of cases reported from
the UHC. But this would require substantial resources.
We infer from our study that around 50% of kala-azar cases in a highly endemic area of
Bangladesh are not detectable by the national surveillance system. Obstacles to reporting
were related to the reporting system and the reporting burden for multiple diseases.
12

13. CONCLUSIONS
The kala-azar passive surveillance system in Bangladesh needs to improve in order to record
the 50% of cases currently missed. The reporting system could be simplified. For example, if
treatment was restricted to those with a positive result, the only record used for reporting
could be the rK-39 diagnostic test result. An improved surveillance system would help
programme managers deploy resources more effectively and monitor progress towards
elimination.
ACKNOWLEDGEMENT
The study was funded by the Core Research Fund of the International Centre for Diarrhoeal
Disease Research, Bangladesh (icddr,b) which supported all the data collection related
expenditure. Icddr,b is thankful to the Governments of Australia, Bangladesh, Canada,
Sweden and the UK for providing core/unrestricted support. The PhD research fund of the
Australian National University provided additional support. KMR also received support
through the Endeavour Postgraduate Award of the Australian Government.
We are grateful to Communicable Disease Control, Directorate General of Health Services,
Ministry of Health and Family Welfare, Bangladesh and Upazila Health Complex of
Gaffargaon for providing all sorts of supports needed for the study.
Staff involved in data collection for the study worked to their utmost sincerity, hard-work and
honesty. This research work would not be possible without the spontaneous participation of
the community members of Gaffargaon.
13

14. REFERENCES
1. Control of the leishmaniasis: report of a meeting of the WHO Expert Committee on
the Control ofLeishmaniases, Geneva, 22-26 March 2010., in WHO technical report
series; no.949. 2010, WHO: Geneva.
2. Boelaert, M., et al., The poorest of the poor: a poverty appraisal of households
affected by visceral leishmaniasis in Bihar, India. Trop Med Int Health, 2009. 14(6):
p. 639-44.
3. Bern, C., et al., Risk factors for kala-azar in Bangladesh. Emerg Infect Dis, 2005.
11(5): p. 655-62.
4. SEARO. Kala-azar Disease Burden in SEA Region. [cited 18/07/2013]; Available
from: http://www.searo.who.int/entity/vector_borne_tropical_diseases/data/Kala-
azar/en/index.html.
5. SEARO, Regional strategic framework for elimination of kala-azar from the South-
East Asia Region (2011-2015). 2012.
6. Mubayi, A., et al., Transmission dynamics and underreporting of Kala-azar in the
Indian state of Bihar. J Theor Biol, 2009. 262(1): p. 177-85.
7. Singh, S.P., et al., Serious underreporting of visceral leishmaniasis through passive
case reporting in Bihar, India. Trop Med Int Health, 2006. 11(6): p. 899-905.
8. Singh, V.P., et al., Estimation of under-reporting of visceral leishmaniasis cases in
Bihar, India. Am J Trop Med Hyg, 2010. 82(1): p. 9-11.
9. Training Module on Kala-azar Elimination Program, D.G.H.S. Disease Control Unit,
Ministry of Health and Family Welfare, Government of Bangladesh, Editor. 2010.
10. Bern, C. and R. Chowdhury, The epidemiology of visceral leishmaniasis in
Bangladesh: prospects for improved control. Indian J Med Res, 2006. 123(3): p. 275-
88.
11. Ferdousi, F., et al., Visceral Leishmaniasis Eradication is a Reality: Data from a
Community-based Active Surveillance in Bangladesh.Trop Med Health. 40(4): p. 133-
9.
12. Rahman, K.M., et al., Increasing incidence of post-kala-azar dermal leishmaniasis in
a population-based study in Bangladesh. Clin Infect Dis, 2010. 50(1): p. 73-6.
13. BBS, Community Report: Mymensigh Zilla. Population and Housing Census 2011. ,
in Population and Housing Census 2011. 2012, Bangladesh Bureau of Statistics,
Ministry of Planning.
14

15. 14. DGHS, Health Bulletin 2012. 2012, Management Information System, DGHS.
Ministry of Health and Family Welfare, GOB.
15. [cited 2012 September 13 ]; Available from: http://www.msf.org.au/media-
room/press-releases/press-release/article/hope-for-kala-azar-sufferers-in-
bangladesh.html.
16. Rahman, M., et al. Healthcare-seeking from Village Doctor for Kala-azar in
Bangladesh. in 13th Annual Scietific Conference of ICDDR,B. 2011. Dhaka,
Bangladesh: ICDDR,B.
17. Mondal, D., et al., Visceral leishmaniasis elimination programme in India,
Bangladesh, and Nepal: reshaping the case finding/case management strategy. PLoS
neglected tropical diseases, 2009. 3(1): p. e355.
18. Zafar Ullah, A.N., et al., Effectiveness of involving the private medical sector in the
National TB Control Programme in Bangladesh: evidence from mixed methods. BMJ
Open. 2(6).
15

16. Figure 1: Study sampling
16

17. Figure 2: Study participants and exclusions
17
Households screened = 4703
Kala-azar cases identified = 89
Cases finally analysed = 58
Cases refused to participate = 2 Extra cases in a household = 15
Cases migrated out = 5
Cases receiving special clinical care
or care from outside Gaffargaon = 9

18. Figure 3: KA patient flow and reporting system observed in Gaffargaon UHC, 2012
Note: KA = kala-azar; UHC = upazila health complex; MO = medical officer
18

19. Figure 4: UHC kala-azar recording and reporting
19

20. Table 1: Patient factors influencing UHC recording of kala-azar patients
Factor Kala-azar cases Odds ratio
(OR)
95% CI
Recorded by
UHC n (%)
Not recorded by
UHC n (%)
Age in years
18 years or more 13 (45) 8 (28) 1.0
< 18 years 16 (55) 21 (72) 0.5 0.2-1.4
Sex
Female 8 (28) 12 (41) 1.0
Male 21 (72) 17 (59) 1.9 0.6-5.7
Education: >= 5 years completed
(reference: < 5 years)
Patient 11 (50) 5 (36) 1.8 0.4-7.4
Father 6 (38) 3 (15) 3.4 0.6-18.1
Mother 4 (25) 5 (25) 1.0 0.2-4.7
HH head 16 (72) 14 (58) 1.9 0.4-8.0
Patient accompanied by spouse 7 (24) 11 (38) 0.5 0.2-1.7
Patient accompanied by parents 14 (48) 13 (45) 1.2 0.4-3.3
Kala-azar symptoms/signs
Abdominal distension 17 (59) 17 (59) 1.0 0.4-2.9
Abdominal pain 19 (66) 15 (52) 1.8 0.6-5.2
Palpable spleen reported by a
health care provider
11 (38) 8 (28)
1.6 0.5-4.9
Skin darkening 21 (72) 20 (69) 1.2 0.4-3.7
20

21. Table 2: Health system factors influencing UHC recording of kala-azar patients
Factor Kala-azar cases Odds ratio
(OR)
95% CI
Recorded by
UHC n (%)
Not recorded by
UHC n (%)
Facility/ provider confirmed
kala-azar for the first time
Private 18 (62) 13 (44) 1.0
Public 11 (38) 13 (45) 0.7 0.2-1.9
KA drug administered
Not at UHC
At UHC 5 (17) 16 (55)
1.0
0.2 0.04-0.6
KA drug received for treatment
SAG 5 (17) 23 (79) 1.0
Miltefosine 24 (83) 6 (21) 18.4 3.4-100.3
KA treatment year
2010 16 (55.2)) 20 (69.0) 1.0
2011 13 (44.8) 9 (31.0) 1.8 0.60-5.40
21