Monitoring drug effectiveness in kala-azar in Bihar, India: cost and feasibility of periodic random surveys vs. a health service-based reporting system

Authors


Corresponding Author Shyam Sundar, Department of Medicine, Institute of Medical Sciences, Banaras Hindu University, Varanasi, India. Tel.: +91 542 2369632; Fax: +91 542 2367568; E-mail: drshyamsundar@hotmail.com

Summary

Objective  In 2009, a random survey was conducted in Muzaffarpur district to document the clinical outcomes of visceral leishmaniasis patients (VL) treated by the public health care system in 2008, to assess the effectiveness of miltefosine against VL. We analysed the operational feasibility and cost of such periodic random surveys as compared with health facility-based routine monitoring.

Methods  A random sample of 150 patients was drawn from registers kept at Primary Health Care centres. Patient records were examined, and the patients were located at their residence. Patients and physicians were interviewed with the help of two specifically designed questionnaires by a team of one supervisor, one physician and one field worker. Costs incurred during this survey were properly documented, and vehicle log books maintained for analysis.

Results  Hundred and 39 (76.7%) of the patients could be located. Eleven patients were not traceable. Per patient, follow-up cost was US$ 15.51 and on average 2.27 patients could be visited per team-day. Human resource involvement constituted 75% of the total cost whereas involvement of physician costs 51% of the total cost.

Conclusion  A random survey to document clinical outcomes is costly and labour intensive but gives probably the most accurate information on drug effectiveness. A health service-based retrospective cohort reporting system modelled on the monitoring system developed by tuberculosis programmes could be a better alternative. Involvement of community health workers in such monitoring would offer the additional advantage of treatment supervision and support.

Abstract

Objetivo:  En el 2009 se realizó un estudio aleatorio en el distrito de Muzaffarpur para documentar los resultados clínicos de pacientes con leishmaniasis visceral (LV) tratados dentro del sistema de salud pública durante el 2008, con el fin de evaluar la efectividad de miltefosina vs LV. Hemos analizado la idoneidad operativa y el coste de estos estudios aleatorios periódicos en comparación con las monitorizaciones rutinarias realizadas en los centros sanitarios.

Métodos:  Se tomó una muestra aleatoria de 150 pacientes de los registros guardados en los Centros Sanitarios de Atención Primaria (CAPs). Se examinaron las historias clínicas y los pacientes fueron localizados en sus residencias. Se entrevistaron los pacientes y los médicos con la ayuda de dos cuestionarios específicamente diseñados por un supervisor, un médico y un trabajador de campo. Los costes del estudio se documentaron y se llevaron los libros de registro de los coches para realizar el análisis.

Resultados:  Se localizaron 139 (76.7%) pacientes, mientras que a 11 no se les encontró. El coste del seguimiento por paciente fue de US$ 15.51 y en promedio se visitaban 2.27 pacientes por equipo y día. Los recursos humanos constituían un 75% del coste total, mientras que el coste del médico correspondía a un 51% del coste total.

Conclusión:  Un estudio aleatorio para documentar los resultados clínicos es caro y requiere de mano de obra intensiva, pero probablemente da la información más precisa sobre la efectividad de medicamentos. Un sistema de informes, basado en una cohorte retrospectiva del servicio sanitario y modelado según el sistema desarrollado por los programas de tuberculosis, podría ser una mejor alternativa. Involucrar a trabajadores sanitarios comunitarios en la monitorización ofrecería una ventaja adicional a la supervisión y al apoyo en el tratamiento.

Introduction

Visceral leishmaniasis (VL, kala-azar) is a chronic infectious disease, caused by a parasite and transmitted through an arthropod vector. In the Indian subcontinent, Leishmania donovani is the causative parasite and Phlebotomus argentipes, a phlebotomine sandfly, is the vector. More than 90% of the world’s VL burden occur in the Indian subcontinent, Sudan and Brazil. In India, more than 90% of VL cases are reported from a single state, Bihar. A centrally organised and sponsored VL elimination initiative has been launched in endemic areas since 2005. The Government of India provides anti-leishmanial drugs, insecticides and technical support, and the state government implements the programme through the Primary Health Care system and district hospital in integration with other vector born diseases. Financial assistance is being provided to endemic states since December 2003 to facilitate effective strategy implementation by States (National Vector Born Disease Control Program, NVBDCP Available from: http://nvbdcp.gov.in/kala-new.html) . The goal of the VL elimination initiative is to reduce the annual incidence rate of kala-azar and post-kala-azar dermal leishmaniasis (PKDL) to less than one per 10 000 population at sub-district level by the end of 2015 (World Health Organization 2005). Several authors have pointed to the need to strengthen the surveillance system in the VL elimination initiative (Chappuis et al. 2007).

There are some important reasons for monitoring not just the reported incidence of new VL cases but also treatment compliance and its outcome. The causative parasite, L. donovani, has a human reservoir; any patient not fully cured remains infectious to the vector (World Health Organization 1990). Moreover, as Styblo realised when expanding tuberculosis control programmes in East Africa in the late 1970s, a disease control programme that fails to cure a high enough proportion of patients enroled may do more harm than good by promoting drug resistance (Styblo 1976; Styblo & Bumgarner 1991). Drug resistance to antimonials is already widespread in the VL endemic region, and miltefosine resistance might easily develop if the drug is used on a large scale as monotherapy in an unregulated manner (Sundar et al. 2001; van Griensven et al. 2010).

Failure of VL treatment can present as initial failure or as relapse after treatment is completed, or as PKDL, a dermatological complication of leishmanial infection. Apart from the clinical problems they pose, these increase the reservoir of leishmanial infection and thereby its transmission (Addy & Nandy 1992).

The Kala-azar elimination initiative has recognised that case detection rates, compliance rates and cure rates are key indicators to monitor the effectiveness of the initiative undertaken. Nonetheless, the recording and reporting system currently in use lacks adequate provisions for monitoring VL patients during treatment and for their final clinical outcome. Monitoring clinical outcomes is complex in kala-azar, as it conventionally requires an assessment by a clinician 6 months after treatment completion; whilst the drug treatment itself usually lasts for less than a month.

We conducted a cross-sectional random survey of all kala-azar patients treated in 2008 in all the Community Health Centres (CHC) of Muzaffarpur district, Bihar State, to document treatment effectiveness (Hasker et al. 2010). We noticed that about 40% of patients were still treated with pentavalent antimonials, despite the programme recommendation dated 2005 to use miltefosine as first-line treatment. Twenty-eight per cent of VL patients in Muzaffarpur in 2008 (95% CI 20–35%) needed retreatment. Especially patients treated with antimonials (sodium stibogluconate) had a poor treatment outcome: 40% (95% CI 28–52%) of these required a second course of treatment, whereas only 15% of the 40 patients treated with miltefosine (95% CI 3–27%) did. There were some discrepancies between the results obtained through review of the health centre records and those obtained through interviews with patients.

The recording and reporting system currently in use in the health facilities was unable to generate this essential information on treatment outcomes. To better monitor treatment effectiveness, periodic random surveys for all the patients such as the one described here could be an option. Whilst the periodic random survey method is probably more accurate in assessing individual treatment outcomes, it will require a substantial effort from the control programme. Another option could be to improve the routine recording and reporting system with involvement of grass root level health workers. Such a surveillance method depends on the existing networks of the health system; it might be more appropriate and sustainable in the given context than periodic random surveys. In this study, we analysed the operational and economical feasibility of a periodic random survey and documented its strengths and its constraints. We also identified bottlenecks in the current recording and reporting system at Primary Health Care level and discuss a possible alternative system with the involvement of grass root level health workers who could also supervise treatment and follow up the VL cases.

Method

Description of the study area

Muzaffarpur District in Bihar state has been highly endemic for kala-azar since 1972. Districts in this state typically have a population of 2–3 million inhabitants and one district hospital that provides all specialised services. Primary Health Care in rural areas of India is organised as a 3-tier system: At the top of the pyramid is the CHC, in theory catering for a population of 120 000, which corresponds to the administrative division, called ‘Block’. The CHC is staffed with physicians and has a limited number of beds. Each CHI supports four Primary Health Centres that is the lowest level of the system at which physicians are available. A Primary Health Care centre (PHC) in turn should support 6 Health Sub-centres run by auxiliary nurses/midwives (ANMs). At village level, there are volunteer outreach workers known as accredited social health activists (ASHAs). They are supervised by the ANMs and are the link between the community and the health system. The system is well organised with all ASHAs attending meetings at the CHI once every 4 weeks and ANMs meeting twice a month. Although according to the norms, there should be one CHI per 120 000 inhabitants and one Primary Health Centre per 30 000, the actual population covered is much higher. The rural areas of Muzaffarpur in which we conducted our research have a population of almost 4 million but altogether there are only 14 CHIs and 46 additional Primary Health Centres. The establishment of additional Primary Health Centres is in progress (http://statehealthsocietybihar.org).

VL diagnosis and treatment are currently available only at CHIs. VL diagnosis in these centres is based on presenting signs and symptoms, and the results of an rK39 rapid immunochromatographic test. If required, e.g. in case of presumed recurrence, parasitological examination is performed at higher health facilities such as a district hospital or medical college. Monotherapy with 28 days of oral miltefosine is the recommended first-line regimen in the elimination initiative. Six months after treatment completion patients are invited for a follow-up visit to determine the final treatment outcome. If the patient has completed treatment and is in good clinical condition at the end of the follow-up period, he or she is considered cured.

Patients and data collection

The lists of all VL cases treated in 2008 were collected from all the 14 CHCs totalling 1879 cases. A random sample of 150 cases was selected as described elsewhere [Hasker et al. (2010)]. Locations of the patients are shown in Figure 1 and were representative of all patients in the district. We checked the records of the patients kept at their respective CHC and noted the information related to their date of reporting, date of diagnosis and initiation of treatment, type of drug given and final outcome of the treatment. Two interview questionnaires were used: one to interview the physicians in the CHCs and a second for patients. We prepared a visit plan to minimise the travel costs and time. We allowed a maximum of three visits to locate and interview a patient; if contact was not possible within those three visits, the patient was considered as untraceable. For locating patients, help from health worker like ASHA, ANMs and local community leaders were sought. In case of a minor, patient’s guardian was interviewed. If a patient was not available for interview or had died, adult family members or nearest relatives were interviewed. The flow chart of the entire procedure for interviewing patients is given in Figure 2. Medical records of patients were verified by the physician to confirm the diagnosis. The research team comprised an experienced physician and a well-trained field worker with sociology background. The research team used its own vehicle for transportation. On an average, the team worked for 8 h per day including transportation and refreshment time in between.

Figure 1.

 Distribution of 139 randomly selected VL cases during 2008 in Muzaffarpur district, Bihar, India.

Figure 2.

 Flow chart showing the follow up procedure of Kala-azar patients at their homes.

Cost estimation

For calculation purposes, we recorded the time necessary in ‘Team days’ and maintained a vehicle log book, account book and other financial documents. We calculated various costs and the total distance travelled from Kala-azar Medical Research Centre (KAMRC) in Muzaffarpur city, the nodal centre for operating the study. Time (team days) and cost of transportation for obtaining information from CHCs and patients were calculated separately. In our calculation, we did not include the costs of investments and amortisation, in particular the costs of a car and GPS devices for recording the Lat/Long points of the patients’ homes.

Ethical aspects

Ethical clearance was obtained from the Ethical committee of Banaras Hindu University in Varanasi, India, as well as from the ethics committee of the University of Antwerp, Belgium. Any person identified with (suspected) VL in the process of the study, whether because of treatment failure/relapse or a first episode, was eligible for free diagnosis and treatment at the KAMRC.

Results

Of 150 randomly selected patients, households of only 139 could be located, and of those 139, one patient had died. We could locate 115 cases on the first visit, and 24 cases required a second visit. Eleven patients could not be found even after three attempts and were considered as ‘untraceable’. Of 115 cases that could be located in the first visit, we could interview only 101 cases on the same day; 14 cases required a second visit to be interviewed. The remaining 24 cases located at the second visit could all be interviewed then (Table 1). After each unsuccessful attempt to locate a patient, before making the next attempt, we explored the reasons in consultation with ANMs and ASHAs. The most common reasons were (i) incorrect registration of name and incomplete addresses of the patients recorded in the CHC register; (ii) false identification given by patients as they might be originating from a CHC area other than the one they sought treatment in; (iii) the patient was a visiting relative of a household in the CHC area and returned to his/her original permanent residence; or (iv) the patient originated from the same CHC area but migrated elsewhere.

Table 1.   Community health centre (CHC) wise detail of total registered visceral leishmaniasis (VL) cases, no. randomly selected and no of visits required for effective follow up of 150 patients
CHC nameArea (Sq km)Total population (Dec. 2008)Total VL cases in 2008Number randomly selectedNo. of visits required for effective follow upUnsuccessful after three attempts
OneTwoThree
Aurai21927760042202  
Bochahan2212201391631173 1
Giaghat23324262497853  
Kanti31140303014322165 1
Katra1932283124233   
Kurhani30141119711514112 1
Minapur2503094059155   
Motipur33336490919019125 2
Muraul180216999811   
Musahari17029859290422  
Paroo34732692442028204 4
Sahebganj2122185732031174  
Sakara2052885826564  2
Saraiya2732970291671688  
Total34484103915183615010138 11

Manpower required and estimation of cost

To collect the patients’ information from all the 14 CHCs, the team had to spend 10 team days an average of 0.7 team days per CHC and had to travel a total of 700 km. To follow up 150 patients, it took a total of 210 visits in 66 team days with an average of 3.18 visits per team-day. The ‘effective coverage’ of patients (meaning a patient could be located and interviewed or finally could not be located) was 2.27 patients per team-day. The average distance travelled for effective coverage per patient was 35.2 km (Table 2). The total cost incurred for obtaining information from all 14 CHCs was US$ 275.05 (Table 3). Per patient, average follow-up cost was US$ 15.51 with a total of US$ 2328.09 (Tables 2 and 3). Costs on human resources made up a major portion of the total costs at each stage of the study. It constituted 70% and 75%, respectively, for collecting information from 14 CHCs and for follow up of patients (Table 3).

Table 2.   Effort and cost for collecting information from community health centre (CHC) and patients
Efforts and cost of information from 14 CHCsEfforts and cost of information from 150 patients
Total no of team days required10Total team days required66
Total no. of CHC visits required14Total no. of patient visits (attempts) required210
Average team days required per CHC0.7Average visit paid for ‘effective coverage’ of one case1.4
Average no. of CHC covered per team-day1.4Average no. of patient visits per team day3.2
Total distance travelled700 kmTotal distance travelled5280 km
Average distance travelled per CHC50 kmAverage distance for effective coverage of a case35.2
Total CostUS$275.05Total costUS$2328.09
Average cost per CHCUS$19.65Average cost per patientUS$15.52
Table 3.   Break-up cost for collecting information from (a) CHCs and (b) patients
HeadsSub-headsAmount (US$./team-day)Team days consumedTotal amount (US$)
  1. CHC, community health centre.

(a) CHCs
 Human resourceSupervisor10.8310 team-days 
Field worker5.76 
Driver2.88 
Sub-total19.47194.70
 TransportationFuel/lubricant5.3153.10
Maintenance 12.63
 Snacks 0.656.50
 Stationary/printing  6.50
 Communication  1.62
   Grand total275.05
(b) Patients
 Human resourcePhysician18.0466 team-days 
Field worker5.76 
Driver2.88 
Sub-total26.681760.88
 TransportationFuel/lubricant6.06399.96
Maintenance 95.67
 Snacks 0.6542.88
 Stationary/printing  21.66
 Communication  7.04
   Grand total2328.09

Discussion

We studied a random sample of 150 kala-azar patients treated in 2008 in the Muzaffarpur district by CHCs with the aim of assessing their long-term treatment outcome and efficacy of existing system and looking for alternatives. CHCs used to report the final monthly VL figures and not patients’ full details to District Headquarters. Hence we had to collect the available detail of patient’s characteristics (name, age, sex and head of households) and addresses from CHCs. We managed to conduct this survey on 150 patients in the sample at a total cost of US$ 2603.14 and spent a total of 76 team days. Often incomplete and at times erroneous patient characteristics and addresses recorded at CHC facilities were the major bottlenecks identified in the study. Because of this, 35 patients could not be located during the first visit, so the retrieval rate with a single visit was only 76.7% (115/150). With proper addresses and identification of patients, it would have been possible to locate all the patients in the first visit and thus reducing 25% (19 team days) time. This would have reduced the costs for the survey from US$ 15.51 per patient to US$ 11.05 per patient.

In this study, researchers made the follow-up visits from their nodal centre at Muzaffarpur city to all the CHC facilities and to the homes of patients. If the same follow up were performed by the CHC facilities for their respective patients using their own health workers, cost of collecting information about patients from CHCs would be curtailed and cost of follow up of patients could be reduced. However, recording of detailed addresses with good identification characteristics of patients would be essential to facilitate the location of a patient’s home. Involving a medical doctor in patient interviews constituted 51% of the total cost incurred at that stage. To reduce the cost of the survey, this part could be carried out by paramedical workers also; however, involvement of physicians would provide a better quality data. Government health workers like ASHA, ANMs and other community leaders were important key informants to locate the patients during our survey.

The survey was conducted on a simple random sample of patients, incorporating only 8% of the total VL cases treated in 2008. If such random surveys were adopted as a national strategy for monitoring treatment effectiveness, they would need to be carried out at periodic intervals. They could possibly be implemented by the CHCs under the direction and supervision of district headquarters at a lower per patient cost as the distances covered to locate a patient would be shorter and CHC personnel would know their target area better. Still, the question remains whether the teams at CHC level have the time, manpower, equipment and financial resources needed to conduct, analyse and systematically report such periodic random surveys. It seems also an extra-ordinary effort to ask from teams in routine health care practice, for one single disease, given they have multiple tasks and duties in the health system.

An alternative to periodic random surveys (either organised by district headquarters or from the CHC) is to incorporate routine retrospective cohort reporting based on register keeping at CHC level – as practised already by the CHCs for tuberculosis control (Veen et al. 1998). In such a system, all newly registered VL cases by the CHC within a quarter would have to be reported at the end of that quarter to the district level with a breakdown by new cases and previously treated cases. For each new treatment episode, a new entry in the patient register would be made. A patient who had been treated before would thus be re-registered as ‘treatment after failure’, ‘treatment after default’, ‘relapse’, ‘transferred-in’ or ‘other’. This report would provide some indication of recent programme performance if the proportion of ‘treatment after failure’ or ‘relapse’ cases amongst patients registered for treatment is high. In addition, a specific report on final treatment outcome after 6 months would be made for all patients registered during a given quarter – provided that sufficient time has elapsed- matching the patients’ count with the numbers in the earlier case registration report. This treatment outcome report would provide the final treatment outcomes, i.e., ‘cured’, ‘treatment failure’, ‘defaulted’, ‘died’, ‘relapsed’, ‘transferred out’ or ‘other’ of all the patients grouped by ‘new cases’, ‘previously treated cases’ and also grouped by treatment regimen and thus would provide a detailed record on effectiveness of specific treatment regimen and current programme. At the end of any given quarter, e.g. end of the first quarter 2010, the CHC would have to file two reports, one on the new registered cases of that quarter, and another on the final treatment outcomes of all the patients registered in the first quarter 2009. This system would allow tracking the treatment effectiveness of VL drugs with a 1-year delay. This type of registration is currently being piloted in three sentinel CHCs in Bihar at the initiative of the Kaladrug-R research project.

The critical element in such a register-based system is to obtain the information on each treated patient 6 months after treatment. One way to obtain this information is to involve the peripheral health staff, ASHAs and ANMs, from the start of treatment. ASHA and ANM can be put in charge of supervising the miltefosine intake with proper recording and reporting at periphery. A model incorporating this strategy is presented in Figure 3. As miltefosine is an oral drug and now freely available under the VL elimination initiative, supervised treatment at patients’ door by ASHAs is possible. This would greatly improve drug compliance, and patients would not have to visit PHCs at frequent intervals, thus avoiding their time and wage losses. After confirmation of VL, patients could be given the drug for the first 3 days and called on the fourth day. If there were no adverse side effects, the patient would get drugs for the next 4 days. Drugs for the remaining 21 days could be handed over to the ANM of the respective sub-centre from where the local ASHA would collect it for supervised treatment. On 7th, 14th and 21st day, the ASHA would give (or patient may collect himself) drugs to the patient at home and would inquire about any adverse effect (Figure 3). Although this model advocates maintaining the referral system at each level, if patients by pass the channel they will be attended at higher level.

Figure 3.

 Model showing supervised treatment with miltefosine, recording/reporting and referral system involving peripheral workers.

Under such a model, post-treatment 6-month follow up by the ASHA is very easy, as she will have developed a good relation with the patient. The ASHA could make a home visit to check for persistence of symptoms, relapse or recurrence of symptoms of VL. If there were any warning symptom, she would report to CHC via ANM. With the availability of mobile phones, it would be easy to connect to the in charge at the CHC to report the final clinical outcome. If this system worked correctly, the ASHA could be further involved in active VL case detection. If she encountered any suspected VL case in the field, she would refer it to a sub-centre where ANM would perform an rk39 dip stick test and spleen palpation and would refer the patient to CHCs if suspected for VL. This would make early case detection possible and reduce the delay in reporting to CHCs.

Whilst periodic random surveys are costly and focus only on 6 month’s final outcome, the register-based retrospective cohort reporting, by its connection to community health workers and peripheral staff, provides opportunities for treatment support and active case detection. Information on non-compliance and reasons for drop outs would become available earlier and at a level which could intervene. With this approach, the recording and reporting system would be strengthened at the grass root level. By involving ASHAs, the quality of VL care would be strengthened through the concept of community participation in health care delivery. However, this model needs more evaluation in routine practice.

Conclusion

The recording and reporting system on clinical outcomes of VL at CHCs and district level needs to be improved. Treatment outcomes as well as side effects must be recorded in routine health care. Identification characteristics and addresses of patients should be recorded in such a format that the patient can be easily followed up later on. Of the two possible solutions, we explored here, periodic random surveys and health services-based retrospective cohort monitoring, the latter is more attractive, as it seems more feasible, cheaper and provides the additional benefit of strengthening the supervised treatment with miltefosine at patient’s doorstep.

Acknowledgement

This work was supported by NIAID, NIH TMRC Grant No. 1P50AI074321.

Ancillary