Implementing global knowledge in local practice: a WHO lung health initiative in Nepal

Abstract

Clinical practice guidelines are used widely to improve the quality of primary health care in different health systems, including those of low-income countries. Often developed at international level and adapted to national contexts to increase the feasibility of effective uptake, guideline initiatives aim to transfer global scientific knowledge into local practice. The WHO's Practical Approach to Lung Health (PAL) is an example of such an initiative and is currently being developed to improve the quality of care for youths and adults with respiratory diseases.

We assessed ex-ante the feasibility of successful implementation of PAL in a pilot programme in rural Nepal, studying three components: the quality of the innovation (i.e. the guidelines), the effectiveness of the implementation strategy (i.e. training) and the receptiveness of the social system of health staff at all levels (i.e. social and organizational characteristics). We assessed the guideline innovation with the AGREE instrument for guidelines, the intended implementation strategy by critical comparison with literature on effective strategies, and the social system with both a stakeholder analysis and a descriptive analysis of the health care system at district level.

This ex-ante assessment of an adaptive local implementation of international WHO guidelines showed that in July 2002 the ‘implementability’ of the package was challenged on the three components studied. To increase the chances of successful implementation, the national guideline development process should be improved and the implementation strategy needs to be upgraded. In order to successfully transfer global knowledge into local practice, we need to develop additional multifactorial sustained interventions that tackle other culture-specific and health system-specific barriers as well. The primary health workers are key informants for these barriers.

Introduction

Multilateral efforts to improve the quality of primary health care in varying national systems often include the development and introduction of standard clinical practice guidelines. This process typically starts out with the development of generic tools at global level, as can be seen in a variety of programmes for essential drugs and disease control (Ross-Degnan et al. 1992; Gove 1997; WHO 2002b). The tools are not health system-specific and hence, the potential for instantaneous implementation in many health systems is limited. Consequently, generic tools are meant to be adapted to the specific local context. Overall, the ultimate challenge lies in transferring global scientific knowledge into local practices. In low-income countries in particular, these circumstances are frequently characterized by a chronic lack of sufficiently competent health workers and other limited resources.

Success or failure in the promotion of changes in clinical practice in developed countries with well-trained professionals depends on the characteristics of the implementation process (Wensing et al. 1998; Woolf et al. 1999; Grimshaw et al. 2001). Given the multitude of international initiatives and further globalization of efforts, there is a need for more systematic study of how generic guidelines are implemented through national health systems in developing countries.

This paper takes the example of the ‘Practical Approach to Lung Health’ (PAL),¹ a World Health Organization (WHO) initiative to improve the syndromic management of lung diseases in youths (over 5 years of age) and adults in middle- and low-income countries (WHO 2002a). PAL is presented as a package that consists of generic clinical practice guidelines and accompanying training materials. It targets the multi-purpose health worker at peripheral primary and secondary care facilities. In several countries, elements of PAL are currently being field tested and evaluated (WHO 2003).

In Nepal, the global, generic PAL package was adapted to the specific Nepalese health care context as part of the guideline development process and also to facilitate its further introduction (Box 1) (Bentley 1995; His Majesty's Government of Nepal 2001; Bishai et al. 2002; Adhikari and Maskay 2004). A pilot implementation of the adapted PAL package, called PAL-Nepal, started in governmental primary health care centres, health posts and sub-health posts of Nawalparasi, a rural lowland district (Figure 1). The pilot implementation of PAL-Nepal is subject to an evaluation of effects, including costs, organizational effects and health outcomes, using a cluster randomized trial design. Also, it is subject to a qualitative assessment of the development and implementation process itself.

This article presents the latter of the two studies and answers the research question: ‘What are the chances of successful implementation of PAL-Nepal given the characteristics of the guidelines, the planned implementation strategy and the social system?’.

The study was carried out by external researchers (AtA, LN, CvdH) who were observers during PAL-Nepal activities and had no active involvement in its development and implementation.

Methods

Adapted from Rogers (1995), we identified and assessed three components in the development and implementation process of PAL-Nepal: the innovation (the guidelines), the implementation strategy (training), and the social system (the social and organizational context of health workers at the various levels in the Nepalese health care system). We perceived that successful implementation depends on the quality of the innovation itself, the effectiveness of the implementation strategy and the receptiveness of the social system, in conjunction with each other. The first two components are elements of the new health policy and the latter determines the context for adoption and diffusion. For each component we selected specific assessment instruments.

Assessment of innovation

We used the generic and standardized Appraisal of Guidelines Research & Evaluation (AGREE) Instrument (The AGREE Collaboration 2001) to assess whether the guidelines could be expected to be effective in producing the desired changes in outcomes. The AGREE instrument appraises the methods used for developing the guidelines, the content of the final recommendations, and the factors linked to their uptake.

The material for this assessment consisted of the background documents for the generic PAL clinical practice guidelines (Pio and Chaulet 1997; Scherpbier et al. 1998), the minutes of the Adaptation Working Group (AWG) meetings held in Nepal (Mattousch 2001a,b), personal observation reports and the context-specific PAL-Nepal algorithms. Three independent reviewers familiar with developing and assessing clinical practice guidelines assessed the PAL-Nepal documents with the AGREE instrument. For multi-rater Kappa analysis, we used SPSS statistical software and additional macros (SPSS inc. 1997, 2000).

Assessment of implementation strategy

For the implementation strategy, we established whether, given the available evidence from international literature, we expected the strategy offered to be effective. To assess this, we critically analyzed the planned implementation strategy using the framework suggested by Hulscher et al. (1999), which is based on evaluation research literature and theories on implementation and behaviour change, and on the data collection checklist developed by the Cochrane collaboration (Hulscher et al. 2001). Additionally, we compared the planned implementation strategy with characteristics of strategies that enhance effective implementation, as summarized by Grol and Grimshaw (1999, 2003), Grimshaw et al. (2001) and Hulscher et al. (2000).

Input for this analysis was derived from the minutes of the AWG meetings (Mattousch 2001a,b), training plans and training manuals.

Assessment of the social system

Finally, we assessed the receptiveness of the social and organizational context of health workers at the various levels in the Nepalese health care system. We perceived the social system at national level foremost as a political system and carried out a stakeholder analysis to assess the feasibility of the political acceptance of PAL-Nepal (Reich 1995; Reich and Cooper 1997). For this analysis, three respondents (senior health officials) rated the stakeholders' position towards the PAL-Nepal goals and mechanisms, and the anticipated power of the stakeholders to influence the achievement of these goals.

In addition, we described factors at district and local level that influence the system's ability to take on central guidance. Input came from the literature, discussions with health care workers and personal observations during meetings at national and district/local level in 2001 and 2002 (AtA, LN, RS), the researcher's diary (AtA) and e-mail communication during the research period 2000–2002 (LN, AtA), and field visits totalling 7 months over a period of 2 years.

Time frame of assessment

The first phase in the development and implementation of PAL-Nepal started in November 1999, building on the work of the international expert meetings that started the construction of the generic PAL package. It ended in July 2002, just before the actual introduction of the PAL-Nepal guideline package to the targeted health workers, which marked the start of the implementation phase. The time locus for this paper is 1 July 2002.

Results

Innovation

The results of the assessment of the PAL-Nepal guidelines with the AGREE instrument are shown in Table 1. The ‘Comments of the reviewers’ reflect the specific findings per item of the AGREE instrument. The inter-rater agreement was low: multi-rater Kappa was 0.29. We hypothesized that part of the variation could be explained by variation in ‘agree’ and ‘strongly agree’ or ‘disagree’ and ‘strongly disagree’. The multi-rater Kappa for clustered answering categories ‘agree’ and ‘disagree’ was 0.65, confirming our hypothesis.

Implementation strategy

Generic PAL training materials for a 5-day training course were developed at international level (by WHO) and adapted at national level (by WHO and the National Tuberculosis Centre). The National Tuberculosis Centre coordinated the adaptation and translation of all materials, and would conduct the training. The implementation strategy in PAL-Nepal consisted mainly of training of primary care health workers using a mix of classroom teaching and interactive elements including clinical practice and exercises. Participants received training documents, a copy of the algorithms – including an A4-sized decision-support tool – as well as a large, poster-sized copy of the algorithms. A critical analysis of the intervention, using the format proposed by Hulscher et al. (1999), is presented in Box 2.

Box 3 shows the comparison of the planned implementation strategy with four characteristics of strategies that are likely to enhance effective implementation.

The social system: stakeholder analysis at national level

The stakeholder analysis revealed that the initiators of PAL-Nepal, the National Tuberculosis Centre and WHO, were the most supportive and powerful stakeholders in the pilot implementation in Nawalparasi. The complete list of potential stakeholders included 15 other international and national health care organizations and groups of professionals in the field of health care in Nepal. The respondents considered all these to be supportive of the implementation of PAL-Nepal but without having an interest or power to influence its implementation.

The social system: characteristics at district and local level

Table 2 summarizes the potential barriers at district and local levels that were identified by the AWG and by WHO and PAL-Nepal partners, as well as suggestions made to address these barriers.

Discussion

Our three-pronged assessment of the PAL-Nepal guidelines enabled a systematic study of the factors that are seen as important for a successful implementation. In the pilot stage of programme development, it identified potential complementary improvements in the guidelines formulation, a broader and more multifactorial implementation strategy, and some obstacles at district and facility levels.

The innovation

The AGREE instrument indicated areas to improve the present version of the guidelines. First, although the scope and purpose of the guidelines was clearly stated, it could be more specific about demographic and geographic characteristics of the target population. Secondly, clarity and presentation were generally of high quality. In addition, the reviewers recommended an easily accessible, comprehensive guideline document that integrates the generic and Nepal-specific recommendations.

Thirdly, in the area of stakeholder involvement, it was recognized that the target users – health workers at health posts and sub-health posts – had not been involved in the development of the generic guidelines, nor in the adaptation process for PAL-Nepal. Likewise, patient views had not been incorporated in the development process of the guidelines. Their perspectives – for example, on the accessibility and use of facilities, or the cultural acceptability of treatments – could have identified potential barriers.

Fourthly, the low scores for applicability can partly be explained by the fact that the cost implications and organizational barriers have yet to be evaluated for the pilot implementation of PAL-Nepal. That does not explain, however, why PAL-Nepal lacked recommendations for monitoring and auditing. Some indicators for monitoring and audit of tuberculosis and asthma management were mentioned in the background document, but not for chronic obstructive pulmonary disease and pneumonia. Including such recommendations in the PAL-Nepal guidelines is helpful for health workers, supervisors and mid-level managers.

Fifthly, the score for rigour of development will increase when the link between the evidence base and the recommendations is clearly shown. The generic PAL guidelines referred to different national guidelines and international scientific papers. Most of these documents were made available to the AWG for PAL-Nepal. It is not mentioned to which recommendations the evidence was linked and what the evidence was for the adaptations made in Nepal. The recommendations made in the generic guidelines were validated a posteriori (Jefferson 2001). Unfortunately, the validation document was not available during the adaptation process in Nepal, or to the AGREE reviewers. There is substantial evidence in support of the generic PAL recommendations. Most evidence is from contexts in developed countries; hence, it is still informative to confirm this evidence in the context of developing countries. Evaluation of costs, organizational effects and health outcomes of PAL-Nepal – currently being carried out by an international consortium of researchers – is expected to reveal the health benefits and related cost of the PAL-Nepal guidelines.

Finally, background documents and the PAL-Nepal guidelines lacked a statement of editorial independence. It was not clearly stated whether the guideline developers might receive personal or institutional benefits from the recommendations.

The validity of the AGREE instrument in the Nepal context needs to be dealt with. The low inter-rater agreement might be related to characteristics of the AGREE instrument itself, PAL-Nepal or the Nepal context. AGREE is an appropriate instrument in the European context, where guidelines target health workers with a high level of education and address a limited area of clinical practice. In addition, algorithms can be used as an additional decision support tool that summarizes the main actions and recommendations. In PAL-Nepal, a set of algorithms is the core element and replaces a textual guideline document. In the case of less skilled health workers and the absence of specific reference manuals on respiratory diseases, the logical choice of the PAL developers has been the introduction of algorithms based on a pragmatic symptomatic approach. The generic background document (Scherpbier et al. 1998) is an extensive book containing a great deal of information that, according to one reviewer, ‘one expects in a medical textbook rather than in a guideline’. The size of this document is not only an important obstacle in the implementation of PAL-Nepal, but it also might challenge the validity of AGREE.

The instrument does not assess the medical quality of the recommendations in the guidelines, nor does it assess whether basic conditions are met for introducing the guidelines, such as educational level of the health workers, drug availability and limited complexity of the algorithm. Also, the relative importance of the domains in AGREE might be unequal. For example, at sub-health post level a notification that the authors were not paid by pharmaceutical companies (editorial independence domain) seems a lesser priority than, say, a visually well-depicted algorithm with clearly stated recommendations (clarity and presentation domain). To better understand the applicability of the AGREE instrument in the context of developing countries, more case studies are needed.

The implementation strategy

The discussion about effective implementations is ongoing and lively (Grol and Grimshaw 2003; Grol et al. 2003; Jamtvedt et al. 2003; Grimshaw and Eccles 2004; Grimshaw et al. 2004). We used this discussion to assess the effectiveness of an implementation strategy prospectively in a developing country. Our findings give some support to the hypothesis that the PAL-Nepal implementation strategy needs upgrading to become effective. In our critical analysis of this strategy as well as its comparison with the international literature, we identified potential areas for improvement.

Firstly, there is a body of evidence that training alone is not effective in changing clinical practice (Thomson O'Brien et al. 2001). The PAL mono-event strategy therefore needs expansion. It is likely that a multi-faceted and multi-events strategy will be more effective, for example, organizing multiple opportunities to learn and practice working with the guidelines. Recent studies, however, show that it might be more complicated than just adding more and different approaches (Jamtvedt et al. 2003; Grimshaw and Eccles 2004; Grimshaw et al. 2004). Studies in Nepal on effective strategies for improvement of primary health care will also provide useful suggestions (Kafle et al. 1997, 2001).

Secondly, analysis of implementation barriers was only documented as part of the guideline adaptation process, and more specifically the treatment recommendations in the generic guideline. The AWG members studied the applicability of these recommendations in the Nepalese context before including these in the PAL-Nepal guidelines. They identified several barriers related to drug treatment and human resources, although these have not been addressed systematically. The adaptation process can benefit from using checklists for barriers and facilitators as used, for example, by Flottorp and Oxman (2003). These tools can also be integrated in the generic PAL package, as is the case in the adaptation of PAL in South Africa.

The social system: stakeholder analysis at national level

The assessment of stakeholders at national level has been informative in that it has shown the real status of the policy process: as long as PAL-Nepal is dealt with as a pilot implementation project of the WHO and the Ministry of Health – and by other potential future stakeholders – it will not mobilize opposition or support from others. At this stage, the receptiveness for PAL-Nepal was high. Respondents added that if PAL-Nepal is to be implemented nationwide, less tolerant stakeholders might be identified due to the financial implications. So far, the financial input has been relatively low and also has been covered by an international donor. The respondents commented that positive results from the effect evaluation were necessary to mobilize political will and more financial resources. These critical factors have not yet been addressed explicitly.

The social system at district and local levels

During the development of PAL-Nepal, several aspects of the social and organizational context at district and local levels were identified that potentially obstruct its effective implementation. Some were addressed effectively (inapplicable treatments), others were only temporarily solved (staff transfers) or not at all (drug availability, absenteeism of PAL-trained staff in sub-health posts). In the literature, other obstructing factors are reported. General factors affecting health service delivery in Nepal are poor human resources, difficult geography and poor general infrastructure (Bentley 1995; Bishai et al. 2002; Osrin et al. 2003; Pokhrel and Sauerborn 2004). Others address factors that more specifically influence the district and local services' ability to take central guidance. Campbell et al. (2003) mention a lack of human and financial resources and motivation. These two factors hamper the utilization of clinical protocols and operational guidelines, although these tools are considered to be important and widespread in Nepal (Campbell et al. 2003). In addition, high absenteeism and communication gaps between villagers and rural health workers, on the one hand, and the higher health authorities, on the other, play an important role (Justice 1983, 1984, 1986).

Aitken (1994) presents an explanatory model for the functioning of district health services that shows ‘two value systems or theories with entirely different aims and expectations… Officially, the organization's value system emphasizes the services delivered: their quality and the number delivered.’ The implicit theory is:

‘that the organization exists in order to distribute and account for funds and to provide the staff … with income. The duty of staff is therefore the provision of reports showing how these funds have been distributed and justifying their expenditure in terms of “showing progress” towards government targets… The actual services provided are not seen to be very important. The staff are aware of both theories, often translating the demands of one into the language of the other.’ (Aitken 1994)

It is crucial to understand these local circumstances in which changes are meant to take place.

Our observations and experiences confirm the comments made by Campbell, Justice and Aitken. We observed a multitude of separate activities in the district capital to support and improve the quality of health services, sponsored by national and international organizations, including the PAL-activities. These activities generated needed additional staff income, i.e. daily allowances and travel allowances. Also, they gave health workers a welcome change from work in remote stations and an opportunity to meet up with colleagues. This resulted in significant absenteeism of health workers from clinical work. Additional earnings through drug retail or private practice were the rule rather than the exception. Drugs that were not available at government facilities could often be purchased from the private businesses of health workers. Health workers spent their working hours combining private and public jobs, favouring the more profitable private jobs, while making sure that reports were presented on time. The centrally run staff appointment system also contributed to absenteeism, as health workers were appointed to unfamiliar localities. Health workers could spend a lot of time in administrative offices negotiating transfers to a more suitable location. Health workers and senior supervisors reported that supervision of peripheral staff in their own duty stations was rare. Supervising tasks had to compete with providing clinical health services at district level, facilitating training courses and workshops, and travelling to regional and national administrative offices. Although the current mechanism for introduction of PAL-Nepal seems appropriate in this stage of PAL development (i.e. through the additional financial incentives provided by a donor in a small pilot), we anticipate that in future, alternatives may be required linking implementation to adaptation of existing systems, for example through the integration of PAL-guidelines into existing training curricula.

We observed that despite these difficult circumstances, health staff were very positive and willing to contribute to the implementation of PAL-Nepal, for example, by facilitating meetings or participating in content discussions. PAL-Nepal is endorsed by the Department of Health Services and facilitated and promoted by the National Tuberculosis Centre, which established operational success through charismatic leadership, staff motivation and strong communication lines (Hamlet and Baral 2002). This programme setting ensured attention and cooperation from the health workers.

Conclusion

Our ex-ante assessment of the adaptive local implementation of international WHO guidelines showed that the feasibility to implement PAL-Nepal could be improved, given the characteristics of the guideline package, the implementation strategy and the social system (in July 2002). The innovation has potential for improvement, the implementation strategy is limited and the social system has several cultural and manpower problems at district and local levels. Besides the technical and organizational challenges posed by the assessment of the innovation and the implementation strategy, we argue that the social and organizational reality of the district health services in Nepal needs a more prominent role in the efforts to change clinical practice. Borrowing the approach – and the reputation as well – from the successful tuberculosis control programme might prove to be an important step. It can be explored further whether its organizational structure (including reporting, supervision and public relations) can be copied to the PAL-Nepal network of involved primary care health workers in rural Nepal.

We emphasize that both the context-specific PAL-Nepal and the global, generic PAL package are in the development stage. Our findings contribute to the maturation of the PAL programme, as do the findings from colleagues who evaluate PAL elsewhere in the world.

This paper is one in a series of assessment reports supporting PAL's further implementation. Developing good generic innovative interventions globally is just the first step in a long process towards successful implementation in local, specific contexts. Assessment of the feasibility of implementation can contribute to efficient allocation of scarce resources. The selected instruments allow for specific recommendations for further development of both the generic as well as the adapted guideline package in Nepal. In order to successfully transfer global knowledge into local practice, we need to support the development of additional sustained interventions that tackle other, culture- and health system-specific barriers as well. Health workers are key informants for these barriers.

Biographies

Guus ten Asbroek, MSc, is a public health researcher in the Department of Social Medicine of the Academic Medical Centre, University of Amsterdam, The Netherlands. He studied health sciences and epidemiology, and has 10 years of professional experience from work in The Netherlands, Africa and Asia. His research topics are in the domain of health systems research, especially process and policy analysis of quality improvement and health system change. His most recent work includes an evaluation of a primary care programme implementation in Nepal and the development of a national performance indicator framework for the Dutch health system.

Diana Delnoij, MA, Ph.D., is a senior scientist at the Netherlands Institute of Health Services Research (NIVEL) and has a broad experience in health system analysis and health services research. Her fields of expertise are health systems, primary care, health insurance, provider payment systems, consumer satisfaction and patient participation. [Address: NIVEL, PO Box 1568, 3500 BN Utrecht, The Netherlands. Email: d.delnoij@nivel.nl]

Louis Niessen, MD, Reg.PH, Ph.D., is a medical epidemiologist and a senior staff member at the Institute for Health Care Policy and Management/Institute for Medical Technology Assessment, Erasmus MC, Erasmus University, The Netherlands. He is involved in national and international projects on health policy, disease burden and economic evaluation in European, African and Asian settings. His field experiences include Nepal, Tanzania and Peru. He is a principal investigator in the economic evaluation of the Practical Approach to Lung Health. [Address: Institute for Health Care Policy and Management/Institute for Medical Technology Assessment, Erasmus University, PO Box 1738, 3000 DR, Rotterdam, the Netherlands. Email: niessen@bmg.eur.nl]

Robert Scherpbier, MD, MPH, has extensive working experience in international health. He specialized in tropical medicine and holds public health degrees from the Institute of Tropical Medicine, Antwerp, Belgium and the Johns Hopkins University, Baltimore, USA. He worked in the Ministry of Health in Ghana and joined the World Health Organization in 1997. He coordinated the development of an integrated case management strategy for respiratory diseases, WHO's Practical Approach to Lung health (PAL). Since 2002, he has been involved in evidence-based policy development for child health, including the Integrated Management of Childhood Illness (IMCI), child survival programme management, and technical support to countries in the Western Pacific Region. [Address: Family and Community Health Cluster, World Health Organization, Avenue Appia 20, 1211 Geneva 27, Switzerland. Email: scherpbierr@who.int]

Naveen Shrestha, MSc, is a lecturer in the Department of Community Medicine and Family Health, Institute of Medicine, Tribhuvan University, Nepal. His research activities are in the field of public health and health services research. His most recent work focuses on the rational use of drugs and the evaluation of the pilot implementation of PAL-Nepal. [Address: Department of Community Medicine and Family Health, Institute of Medicine, Tribhuvan University, GPBox: 8975 EPC 7200, Kathmandu, Nepal. Email: naveenshrestha@hotmail.com]

Dirga Sing Bam, MD, is Director of Nepal's National Tuberculosis Centre and Programme, a Board Member of the International Union Against Tuberculosis and Lung Disease (IUATLD), and was Director of the SAARC Tuberculosis Centre, Kathmandu until October 2004.

Christian Gunneberg, MB ChB, MFPHM (UK), is a public health physician and currently WHO Medical Officer (Tuberculosis) in Nepal.

Carin van der Hor, MA, is a social scientist with extensive experience in general management, especially in management of health care institutes and in training of managers in the field of health care in The Netherlands. She worked in Nepal as the general manager of the PAL-Nepal evaluation research project from January 2002 till January 2004. [Address: iMTA (institute for Medical Technology Assessment), Erasmus University Rotterdam, PO Box 1738, 3000 DR Rotterdam, The Netherlands. Email: carinvanderhor@yahoo.com]

Niek Klazinga, MD, Ph.D., has been professor of social medicine at the University of Amsterdam and chair of the Department of Social Medicine at the Academic Medical Centre, University of Amsterdam since 1999. He holds an MD from the State University of Groningen (1984) and a Ph.D. from the Erasmus University Rotterdam. Between 1985–1999 he was (chief) scientific officer at CBO, the Dutch Institute for Health Care Improvement. His present research activities concentrate on public health and health services research, more specifically on the development and use of quality indicators and quality improvement mechanisms in the health systems context. [Address: Department of Social Medicine J3 301, Academic Medical Centre/University of Amsterdam, PO Box 22700, 1100 DE Amsterdam, The Netherlands. Email: n.s.klazinga@amc.uva.nl]

We are grateful to the Nepalese respondents for providing detailed information for the stakeholders' analysis at national level and the social and organizational context of health workers in Nepal. We thank P ten Have, J de Koning and P Rosier of The Dutch Institute for Healthcare Improvement CBO for assessment of PAL-Nepal guidelines with the AGREE instrument.

This assessment would not have been possible without the kind cooperation of the Department of Health Services, Ministry of Health of His Majesty's Government of Nepal; the District Health Office and the health workers in Nawalparasi; the Department of Community Medicine and Family Health of Tribhuvan University, Kathmandu; and the National Tuberculosis Centre.

Funding for the evaluation of PAL-Nepal was provided by WOTRO, Netherlands Foundation for the Advancement of Tropical Research. This study was co-funded by the department of Social Medicine of the Academic Medical Centre for the University of Amsterdam.

We are grateful to the two anonymous reviewers for their valuable comments on earlier drafts of this article.

References

Author notes

1Institute of Medical Technology Assessment/Institute of Health Policy and Management, Erasmus Medical Centre, Erasmus University Rotterdam, the Netherlands, 2Department of Social Medicine, Academic Medical Center – University of Amsterdam, the Netherlands, 3previously: Stop TB, Communicable Diseases Cluster, World Health Organization, Geneva, Switzerland; currently: Family and Community Health Cluster, World Health Organization, Geneva, Switzerland, 4Department of Community Media and Family Health, Institute of Medicine, Tribhuvan University, Kathmandu, Nepal, 5National Tuberculosis Centre, Thimi, Nepal, 6World Health Organization, Nepal and 7Netherlands Institute of Health Services Research, Utrecht, The Netherlands