- Top of page
Visceral Leishmaniasis (VL) casts a heavy burden upon the Indian subcontinent (ISC), more specifically in the Gangetic plains of Bangladesh, India and Nepal. The annual VL incidence in this focus is estimated at 160 000–315 000 cases, that is, 80% of the world's VL burden (Alvar et al. 2012). VL generally touches the poorest of the poor (Boelaert et al. 2009) and is fatal if left untreated. In the last decade, the development of a rapid diagnostic test (rK39) and new treatment options such as Miltefosine oral treatment have allowed diagnosis and treatment of VL to shift from specialised centres to the primary healthcare (PHC) level (Bhattacharya et al. 2006). An ambitious regional elimination programme was launched in 2005 focusing on early diagnosis and Miltefosine-based treatment as one of its main control strategies (WHO 2005; Mondal et al. 2009). Very soon, concerns were raised that large-scale and unsupervised Miltefosine use might fuel drug resistance (Croft et al. 2006).
Unfortunately, VL control programmes give little attention to monitoring of patient adherence and clinical outcomes (Hasker et al. 2010; Malaviya et al. 2011). Current VL registers capture treatment initiation, but treatment completion or final outcome is rarely recorded. Monitoring of clinical outcomes is crucial for the patient though, as Miltefosine treatment needs to be maintained for 28 days to be effective, and has considerable, mainly gastrointestinal side effects (vomiting, diarrhoea), jeopardising correct intake and adherence. Treatment generally results in rapid clinical improvement with resolution of fever and improvement of appetite after a few days, also contributing to the risk of poorer adherence. But as VL is a chronic condition, the verdict ‘cure’ can only be pronounced some months after treatment, currently empirically set at 6 months. Secondly, monitoring of clinical outcomes on any VL drug is also crucial at programme level. Tertiary-level hospitals and specialised centres have long been the only source of research data on VL treatment and treatment efficacy, with the inherent risk of bias in two ways: possibly better outcomes because of dedicated staff and resources on the one hand, but a selected population of referred or more severe cases on the other. Data on treatment outcomes at primary healthcare level are essential to appreciate the quality and effectiveness of the control programme and the efficacy of a non-supervised, ambulatory treatment such as Miltefosine. Such data, if reliable, would be instrumental in the formulation of drug policy for the VL elimination programme.
Declining efficacy of a given treatment in VL is difficult to detect. Failure to treatment may appear through relapse after initial cure, rather than through immediate non-response to therapy during the drug course. Patients who relapse furthermore may not choose to return to the health-care provider or site where they were initially treated so the problem may be missed by the prescribing doctor (Hasker et al. 2010).
The gold standard for ascertaining cure in VL is a negative smear in direct microscopy on post-treatment bone marrow or spleen aspirate taken 6 months after last drug was taken, which is not feasible in the PHC setting. Diagnostic antibody tests such as DAT or rK39 RDT that have replaced direct microscopy as a diagnostic tool, remain positive for several years after treatment and thus cannot be used as a test of cure. Therefore, in practice, a patient is declared as definitively cured from VL when no clinical signs (fever or increase in spleen size since last visit) of VL have reappeared 6 months after completion of therapy (TDR 2010). In case of failure of treatment, presenting either as persistence of VL symptoms or return of symptoms of VL during or within 6 months after treatment, patients should thus be referred to a referral hospital for bone marrow of spleen aspiration and second-line treatment.
In most efficacy studies, outcome at 6 months is taken as end point (Sundar et al. 2002), but a recent cohort study in Nepal (Rijal et al. 2013) and surveys in India (S. Bursa, MSF, personal communication) indicate that relapses in the period 6–12 months post-treatment may be as frequent as before 6 months. A 6-month end point might therefore overestimate the true efficacy of certain drug regimens (Rijal et al. 2013).
In summary, the protracted nature of VL and the need for final outcome evaluation long after the end of the drug course pose a challenge to the primary health-care system, a situation very similar to other chronic infectious diseases as tuberculosis or HIV: In the directly observed treatment (DOTS) Strategy for control of tuberculosis, where treatment is given for 6 or 8 months, patients are registered systematically in a register that keeps track of periodic evaluations (sputum smears) and final treatment outcome in one single line. Treatment outcome is noted using well-defined, mutually exclusive categories (cure, treatment completed, treatment failure, death, default or transfer) (WHO 2001, 2006). Analysis of the treatment results is made through Retrospective Quarterly Cohort Analysis, calculating the proportions of the different treatment outcomes with as denominator the total number of patients reported for the quarter concerned one year earlier. In antiretroviral treatment (ART) care, there are three types of cohort analysis: (i) a quarterly analysis of new patients started on ART in the latest three-month period, (ii) a cumulative analysis of all patients ever started on ART and (iii) a group cohort treatment outcome analysis related to set time periods (WHO 2001, 2003; Ormerod et al. 2002; Chemtob et al. 2001; Veen et al. 1998). We thus adapted the retrospective cohort monitoring methodology to the specific case of VL treatment. The objective of this study was to field test this method in a few selected health facilities in VL endemic zones in India and Nepal.
- Top of page
The total number of patients treated in the three PHCs in India was 310; in the two district hospitals in Nepal 289, and at BPKIHS in Nepal 232. For KAMRC in Muzaffarpur, we only report on the outcomes of the 566 patients treated on Miltefosine only. Treatments provided at the PHCs in India were either Miltefosine (68.7%) or Sodium Stibo-Gluconate (SSG, 31.3%), the latter despite the fact that it had officially been abandoned for its known lack of efficacy. In the two Nepalese district hospitals, the treatment of choice was Miltefosine (69.2%) with Amphotericin B as alternative in case of contraindication or Serious Adverse Events (30.8%). At BPKIHS, 58.6% received Miltefosine, 39.6% Amphotericin B and 1.7% liposomal amphotericin B (four patients).
Overall, early and late (6M) treatment outcomes in the different clinical settings and with the different drugs prescribed, on a cumulative cohort up to May 2011, are given in Table 2 and 3: Treatment completion rate (which corresponds with early cure rates because we define cure as treatment completion with clinical improvement) with Miltefosine in the two district hospitals in Nepal was 87.0%, and in the three PHCs in India 84.0%. Defaulter rates for Miltefosine-treated patients were 2.5% in Nepal and 14% in India. In Nepal, three patients were switched to second-line treatment because of non-response to Miltefosine treatment on clinical grounds. No aspiration and Giemsa staining was performed to formally confirm this as treatment failures.
Table 2. Early treatment outcomes in cumulative cohort from 2009 to May 2011 early treatment outcomes
|Setting||Early cure||Defaulter||Death||Treatment switch for non-response||Referral||Treatment switch for SAE||Grand total||Treatment completion rate %||Defaulter rate %|
| District hospitals||Nepal||MILTEFOSINE||174|| 5||0||3||10||8||200||87.0|| 2.5|
|Amphotericin B|| 81|| 0||1||0|| 7||0|| 89||91.0|| 0.0|
| PHCs||India||MILTEFOSINE||179||30||2||0|| 2||0||213||84.0||14.1|
| || ||SSG (SAG)|| 72||21||1||1|| 2||0|| 97||74.2||21.6|
| BPKIHS||Nepal||MILTEFOSINE||123||10||0||0||0||3||136|| 90.4||7.4|
|Amphotericin B|| 89||3||0||0||0||0|| 92|| 96.7||3.3|
|AmBisome|| 4||–||–||–||–||–|| 4||100.0||–|
| KAMRC (DOTS)||India||MILTEFOSINE||555||4||2||0||0||5||566|| 98.1||0.7|
Table 3. Late (6 m) treatment outcomes in cumulative cohort from 2009 to May 2011 early treatment outcomes
|Setting||Final cure||Defaulter||Death||Treatment switch for non-response||Referral||Treatment switch for SAE||Relapse||Lost to follow||Grand total||Cure rate %||Treatment failure rate %|
|District Hospitals||Nepal||MILTEFOSINE|| 76|| 5||0||3||12||8|| 0||96||200|| 38.0|| 1.5|
|Amphotericin B|| 5|| 0||1||0|| 7||0|| 3||73|| 89|| 5.6a|| 4.5|
|PHCs||India||MILTEFOSINE||162||30||2||0|| 2||0||10|| 7||213|| 76.1|| 5.6|
|SSG (SAG)|| 69||21||2||1|| 2||0|| 2|| 0|| 97|| 71.1|| 5.2|
| BPKIHS||Nepal||MILTEFOSINE||106||10||3||0|| 0||3||12|| 2||136|| 77.9||11.8|
|Amphotericin B|| 64|| 2||0||0|| 0||0|| 0||26|| 92|| 69.6|| 0.0|
|AmBisome|| 4||–||–||–||–||–||–||–|| 4||100.0||–|
| KAMRC (DOTS)||India||MILTEFOSINE||513|| 4||4||0|| 0||5||38|| 2||566|| 90.6|| 7.4|
For the late treatment outcomes, routine data collection could not be installed as planned as hardly any patient returned for the follow-up visit at six months post-treatment. Active tracing through telephone contacts and home visits was organised by the research teams, with good results in India. In Nepal, however, this was particularly time intensive because of the large distance between the team's headquarter and the two hospitals. Moreover, in the border district Mahottari, more than 50% of the VL patients were Indian citizens seeking treatment across the border, whereby often only a temporary address in Nepal was recorded in the register, resulting in high loss-to-follow-up rates. The final cure rate obtained despite active tracing therefore was 38.0% for patients treated with Miltefosine, against 76.1% in India. Treatment failure rates were 1.5% and 5.6%, respectively.
In contrast, the RQCM approach yielded more complete information in the research centres BPKIHS for Nepal and KAMRC for Bihar. End-of-treatment cure rates with Miltefosine in BPKIHS were 90.4% and 98.1% under DOT in KAMRC; defaulter rates were resp. 7.4% and 0.7%. Late treatment outcomes here were much more complete with minimal loss-to-follow-up rates, and showed 77.9% cure and 8.8% relapse in BPKIHS Dharan and 90.6% cure and 6.7% relapse at KAMRC Muzaffarpur (Sundar et al. 2012).
In Figure 3, we plotted out the quarterly outcomes by cohort in Miltefosine-treated patients from BPKIHS, where systematic follow-up was done up to 12 months post-treatment, showing trends in cure rates over time, and revealing an increase in the relapse rate from 8.8% at 6M to 16.2% at 12 months.
Figure 3. Quarterly cohort outcomes of Miltefosine-treated VL patients at end of treatment, at six months and at twelve months after treatment in BPKIHS, Dharan, 2009–2011.
Download figure to PowerPoint
- Top of page
The main objective of this study was to develop and optimise new tools for monitoring VL treatment outcomes and test them in the primary healthcare setting. Retrospective cohort monitoring is considered a good tool for monitoring care outcomes in long treatments and chronic diseases within the health system (Khader et al. 2012a,b; UNRWA 2009). As the data are collected by the providers and can be analysed at an intermediate (district) level, immediate feedback to providers and remedial action is possible, for example, in case of a high defaulter rate. A potential benefit of the monitoring is that it stimulates the PHC staff to improve counselling on follow-up visits–and consequently on treatment adherence, as shown in TB and HIV/AIDS programmes (Lowrance et al. 2007; Volmink & Garner 2007).
With the proposed tools initial and final cure rates, defaulter rates and failure rates, as done in TB programmes, can be calculated. Their implementation, however, requires commitment and strategies to systematically monitor and register early and late treatment outcomes of all patients diagnosed and treated.
By piloting this method in the different settings and comparing outcome indicators, we were also able to make a series of observations: The data obtained in our study provided a snapshot picture of the status of implementation of the current treatment guidelines with Miltefosine as first-line drug regimen. In routine practice, the proportion of patients receiving recommended first-line treatment was around 60–70% in Nepal, related to the absence of paediatric formulae and availability of Amphotericin B. In Indian PHCs, the proportion fluctuated in relation to supply, with periods of SSG being used.
PHCs in India have only access to first-line treatment and therefore have a high referral rate, in contrast to the District hospitals in Nepal with access to second line and more lab facilities. Non-response and treatment switch for serious adverse events are rarely reported, because these patients are referred to second line, so they end up classified as referral, or they simply default and seek treatment elsewhere.
The cure rates of Miltefosine obtained in this piloting study give a first evaluation of the effectiveness of the current first-line policy with unsupervised ambulatory treatment (as applied in the three PHCs in India, and the two district hospitals as well as the reference hospital in Nepal): Early treatment outcomes show treatment completion rates from 84,0–90, 4% and call for efforts to reduce defaulter rates. The late (6 months) treatment outcomes that were complete for the PHCs in India and BPKIHS in Nepal were 76.1% and 76.5%, respectively, which is worryingly low, especially given the data from BPKIHS, indicating a considerable number of relapses beyond six months (Figure 3).
Given the relatively small number of patients, there will be a large variation in, for example, cure rates from one quarter to another in a given PHC, but calculating the outcomes on a quarterly basis forces to reflect on performance and leads to initiatives for improvement (Maher 2012).
The majority of patients adhered to the treatment follow-up including the EoT evaluation, probably encouraged by the financial incentive provided by the VL programme. The return visit scheduled at six months post-treatment was not respected, and treatment outcomes presented here were obtained through active tracing of the project staff and not by the clinic staff. Tracing patients who default during treatment or for the follow-up visits should be the role of the District Health Office (DHO) through the existing network of village health volunteers such as Accredited Social Health Activists (ASHAs) and Auxiliary Nurses/Midwives (ANMs), and the use of mobile phones to transfer the missing data on outcome (Malaviya et al. 2013). Involvement of community health workers in VL control activities has been advocated by others (Joshi et al. 2006; Malaviya et al. 2011; Singh et al. 2011; Malaviya et al. 2013), as they can help to identify suspected VL cases at community level, and refer or eventually even use a rapid diagnostic test for screening on the spot when adequately trained.
An alternative to routine collection of treatment outcomes at the primary healthcare level would be through spaced surveys. Hasker et al. (2010) performed such a survey, by tracing formerly treated patients and questioning them on clinical events since treatment. They found that 40% of SSG-treated patients and 15% of miltefosine-treated patients had taken a second VL treatment since. Compared to retrospective cohort monitoring, the survey method may have the advantage of being independent and punctual but it was costly, and may be biased by a high number of patients that cannot be retrieved if addresses are not recorded in detail (Malaviya et al. 2011).
Implications for public health and research
Based on a standard methodology applied in TB programs, we propose a treatment outcome monitoring tool for VL that can be routinely used at primary healthcare level. All the tools (register, report forms, manuals) can be downloaded from the website www.leishrisk.net/kaladrug. To implement the system requires (i) minimal training on treatment outcome definitions for PHC staff and (ii) a dedicated team at the district health office to organise the collection of late treatment outcomes with possibly the involvement of community health workers. District health office staff acquainted with the tuberculosis programme should have no problem in applying the same RQCM methodology on VL. The tools can be used by any VL treatment programme in the world, regardless of the treatment protocol(s) used. It should help in comparing the effectiveness in real-life conditions of different treatment options, as well as monitoring the effectiveness of current treatment strategies over time in the light of possible emergence of resistance.