**3.4.4 Distractions**

Similarly, the specific and uncoordinated reporting requirements of vertical programs/donors can lead to several forms being filled by a sole health worker for the same problem, distracting them from more productive uses of their time.

Although the Travis 2004 overview provided limited evidence, it concluded that:

*"Disease or service-specific strategies to strengthen health systems on their own are unlikely to bring about the improvements in health systems needed to achieve the MDGs. …Such an approach must be complemented by a substantial additional body of knowledge and action that takes the functioning of the health system as its core concern".* 

#### **3.5 Can actions to strengthen disease-specific programs that are effective in one area be successfully spread within a country?**

There is some moderate evidence from research that such programmes can be successfully "scaled up". This means that more studies have found successful scale up than those which have found less successful scale up. However, there is a publication bias towards reporting the "scale ups" that were successful, and not reporting the unsuccessful ones.

One example is the strengthening strategy used to scale up NGO-CHW projects across one set of districts in Zambia. The original model was refined, and then a pilot scale up programme was made, which was then itself developed to allow spread in other regions. The HIV/AIDS CHW model was extended to provide programmes including malaria control, and immunization (Øvretveit, 2008) .

There is evidence that success appears to depend on how the implementation is carried out, and on certain enabling factors in the environment, as well as on the type of disease specific programmes: more complex multiple-component programmes appear to be less successful in scale up, but this may be because the capacity was not there to ensure continual coordination in some cases. Research based guidance for scale-up is given in Øvretveit , 2006.

#### **3.6 Are many strengthening strategies together more successful?**

The evidence (WB, 2005) is that they can be, but often are not because of a lack of resources and management capacity at different levels to coordinate and implement the different strategies. There is some evidence that multiple compatible strategies, where different changes reinforce each other, are likely to have a more significant, long-term effect than single-action strategies alone (e.g. integrated delivery of health service, multiple component healthcare reforms). But the risks of failed implementation are higher because:

Strengthening Health Systems in Yemen:

circumstances.

than salaried employees.

Review of Evidence and Implications for Effective Actions for the Poor 291

There is also research into strategies used for scale-up which have proven successful. There are useful frameworks for scale up of successful pilot strengthening approaches in Yemen including one by Cooley & Kohl 2005. Their tested framework gives a three-step process to carry out ten key tasks which their study suggests were needed for effective scaling up. Key choices to be made in deciding how to apply a strengthening intervention more widely (e.g. in scale-up) are: the sequencing of elements of the strengthening programme; and the pace

of spread (e.g. rapid or phased); the areas to spread to and the sequence of areas.

(Nyonator et al., 2006) gives some possible lessons for how to carry out a strengthening strategy in Yemen. The model aimed to reorient primary health care from clinics to communities, by relocated nurses to live and work in community-constructed clinics and using volunteers to mobilizing traditional social institutions to get community support. The scale up strategy

The study notes actions which helped to overcome constraints to scale-up by comparing slow and faster implementing districts. One was to use "peer exchange" to discuss the details of practical changes which would be needed and to use the original pilot as a demonstration model for visits. This is combined with training for upgrading clinical skills, new referral arrangements, quality

The study notes that once the initiative gets started in one or two zones there is spread of the new approach within districts, but spread is slow across district boundaries because of staff exchanges. So within- and across- district involvement of leaders from neighbouring communities was necessary. It also notes a resource constraints problem to scale up where often fewer resources are available than

The action taken to address some of the problems of nurses working in areas they were not from, was a "community engaged" approach to decentralized training. Communities select nurse trainees, who are sent to a local training centre where fees are paid by the districts and communities to be served by the trainees. On graduation, nurses return home, rather than to a post in a distant location.

The study provides an analysis of issues and principles some of which may apply in other settings and for other strengthening-strategies. The first highlighted was the role of research: not just evidence from a district which replicated the pilot, both of which convinced policy-makers and others that the pilot would work elsewhere, but to identify problems and guide the scale-up. The second was the need for specific guidelines about parts of the programme that needed to be changed, the steps needed to get the operational change, and for monitoring whether change was taking place. The third observation was that the pilot was useful as a demonstration of the model and as a training centre. There was a need to resource the pilot founding implementation team to

communicating the evidence and progress as well as sustaining the effort: newsletters documented community and worker experience with the programme and conferences, demonstration exchanges, and staff meetings. The report notes that "CHPS is thus a complex story. Its core strategy is based on a complex experiment, multiple replication efforts, and diverse sources of evidence. But, its core

Fig. 1. Example of lessons from research from a health service strengthening strategy

pass on their experience and motivation. A fourth item was the value of many ways of

agenda is quite simple for stakeholders to understand and embrace".

were used in the pilot. The "faster" districts had found additional funds, usually not from government, for example "private practitioners" –paramedics who are community financed rather

A Community-based Health Planning and Services (CHPS) initiative in Ghana

used decentralized planning to adapt the operational details to local

assurance, and community-based health management.

#### **3.6.1 Consensus and support**

It is more difficult to obtain consensus and support for each component of a multiple-action strategy than for a single approach.

### **3.6.2 Management and oversight**

More complex multiple-action strategies demand greater management capacity if the actions are to be mutually-reinforcing. Persistent oversight is needed for effective consensus building, planning, coordination, review, and readjustment. Management capacity may not be able to provide these.

### **3.6.3 Timing**

Because of limited resources and capacity, the specific actions for multiple-action strategies will need to be phased-in at different times so that these resources and capacity are not overwhelmed by the demands of many actions at one time.

## **3.6.4 Implementation of non-mutually reinforcing actions**

There is a possibility that specific actions will not be implemented, or may be implemented in ways that undermine other components of the strategy. For example, incentives to provide specific services (e.g. special payments for immunization), can reduce incentives to provide other services for which there is greater need (Øvretveit, 2006).

Overall, the evidence shows that, the more complex the strengthening strategy (e.g. many changes, phased changes, with a large overall change), the more support is required (expert facilitators, external training and supervision). Multiple component and sophisticated strengthening strategies can be more effective only if properly implemented – it is costly to provide this support nationally and some level of ongoing support or supervision is often required (Øvretveit, 2006).

#### **3.7 Is how the strategy implemented more important than the type of strategy?**

One conclusion from this review of research is that that almost any strategy might be possible to implement, if certain conditions and implementation methods are present. There is positive evidence for this from successful implementation, as well as negative evidence from the failed implementations which did not have supportive conditions or were not wellmanaged.

A systematic review of 150 studies using high quality experimental designs (Øvretveit et al., 2008) noted that many of the strategies studied had significant amounts and types of resources to ensure full implementation: similar results could only be expected if the resources or conditions were repeated. One common element in the few studies with many positive outcomes was efforts to assess needs and constraints. In these studies 'constraint reduction plans' were found in 66% of the randomized interventions. However, many interventions that did not use this approach also had positive outcomes. Also, the research often did not describe to what extent these constraint reduction plans were implemented.

Overall there is evidence, that, of all the fully implemented strategies, some were effective in strengthening service delivery for poor people. What appears to be important is targeting poor people, ensuring regular measurement of impact, and oversight to ensure the poor benefit.

It is more difficult to obtain consensus and support for each component of a multiple-action

More complex multiple-action strategies demand greater management capacity if the actions are to be mutually-reinforcing. Persistent oversight is needed for effective consensus building, planning, coordination, review, and readjustment. Management capacity may not

Because of limited resources and capacity, the specific actions for multiple-action strategies will need to be phased-in at different times so that these resources and capacity are not

There is a possibility that specific actions will not be implemented, or may be implemented in ways that undermine other components of the strategy. For example, incentives to provide specific services (e.g. special payments for immunization), can reduce incentives to

Overall, the evidence shows that, the more complex the strengthening strategy (e.g. many changes, phased changes, with a large overall change), the more support is required (expert facilitators, external training and supervision). Multiple component and sophisticated strengthening strategies can be more effective only if properly implemented – it is costly to provide this support nationally and some level of ongoing support or supervision is often

**3.7 Is how the strategy implemented more important than the type of strategy?** 

One conclusion from this review of research is that that almost any strategy might be possible to implement, if certain conditions and implementation methods are present. There is positive evidence for this from successful implementation, as well as negative evidence from the failed implementations which did not have supportive conditions or were not well-

A systematic review of 150 studies using high quality experimental designs (Øvretveit et al., 2008) noted that many of the strategies studied had significant amounts and types of resources to ensure full implementation: similar results could only be expected if the resources or conditions were repeated. One common element in the few studies with many positive outcomes was efforts to assess needs and constraints. In these studies 'constraint reduction plans' were found in 66% of the randomized interventions. However, many interventions that did not use this approach also had positive outcomes. Also, the research often did not describe to what extent these constraint reduction plans

Overall there is evidence, that, of all the fully implemented strategies, some were effective in strengthening service delivery for poor people. What appears to be important is targeting poor people, ensuring regular measurement of impact, and oversight to ensure the poor

**3.6.1 Consensus and support** 

strategy than for a single approach.

**3.6.2 Management and oversight** 

overwhelmed by the demands of many actions at one time.

**3.6.4 Implementation of non-mutually reinforcing actions** 

provide other services for which there is greater need (Øvretveit, 2006).

be able to provide these.

required (Øvretveit, 2006).

managed.

were implemented.

benefit.

**3.6.3 Timing** 

There is also research into strategies used for scale-up which have proven successful. There are useful frameworks for scale up of successful pilot strengthening approaches in Yemen including one by Cooley & Kohl 2005. Their tested framework gives a three-step process to carry out ten key tasks which their study suggests were needed for effective scaling up. Key choices to be made in deciding how to apply a strengthening intervention more widely (e.g. in scale-up) are: the sequencing of elements of the strengthening programme; and the pace of spread (e.g. rapid or phased); the areas to spread to and the sequence of areas.

A Community-based Health Planning and Services (CHPS) initiative in Ghana (Nyonator et al., 2006) gives some possible lessons for how to carry out a strengthening strategy in Yemen. The model aimed to reorient primary health care from clinics to communities, by relocated nurses to live and work in community-constructed clinics and using volunteers to mobilizing traditional social institutions to get community support. The scale up strategy used decentralized planning to adapt the operational details to local circumstances.

The study notes actions which helped to overcome constraints to scale-up by comparing slow and faster implementing districts. One was to use "peer exchange" to discuss the details of practical changes which would be needed and to use the original pilot as a demonstration model for visits. This is combined with training for upgrading clinical skills, new referral arrangements, quality assurance, and community-based health management.

The study notes that once the initiative gets started in one or two zones there is spread of the new approach within districts, but spread is slow across district boundaries because of staff exchanges. So within- and across- district involvement of leaders from neighbouring communities was necessary. It also notes a resource constraints problem to scale up where often fewer resources are available than were used in the pilot. The "faster" districts had found additional funds, usually not from government, for example "private practitioners" –paramedics who are community financed rather than salaried employees.

The action taken to address some of the problems of nurses working in areas they were not from, was a "community engaged" approach to decentralized training. Communities select nurse trainees, who are sent to a local training centre where fees are paid by the districts and communities to be served by the trainees. On graduation, nurses return home, rather than to a post in a distant location.

The study provides an analysis of issues and principles some of which may apply in other settings and for other strengthening-strategies. The first highlighted was the role of research: not just evidence from a district which replicated the pilot, both of which convinced policy-makers and others that the pilot would work elsewhere, but to identify problems and guide the scale-up. The second was the need for specific guidelines about parts of the programme that needed to be changed, the steps needed to get the operational change, and for monitoring whether change was taking place. The third observation was that the pilot was useful as a demonstration of the model and as a training centre. There was a need to resource the pilot founding implementation team to pass on their experience and motivation. A fourth item was the value of many ways of communicating the evidence and progress as well as sustaining the effort: newsletters documented community and worker experience with the programme and conferences, demonstration exchanges, and staff meetings. The report notes that "CHPS is thus a complex story. Its core strategy is based on a complex experiment, multiple replication efforts, and diverse sources of evidence. But, its core agenda is quite simple for stakeholders to understand and embrace".

Fig. 1. Example of lessons from research from a health service strengthening strategy

Strengthening Health Systems in Yemen:

likelihood of implementation of the strategy (Travis, 2004).

health workers and capable managers (WB, 2005).

and persistence to make the changes needed.

motivation and retention in most situations).

including volunteer services.

**most countries?** 

**resources?** 

**strategies** 

Review of Evidence and Implications for Effective Actions for the Poor 293

**3.10 Are there strategies which are less dependent on the environment, which work in** 

There was no strong evidence from the review of research that some strategies were more "robust" than others, and less influenced by some of the conditions which appeared to affect implementation of all the strategies. A number of studies reported success where implementation had included constraints assessment and actions to reduce constraints (table 1 at end of this review). The implications are that, if decision makers take action to ensure that as many of these conditions as possible are met, this would increase the

The research reviewed shows that one condition profoundly influencing all strengthening strategies are whether there are adequate resources for the change, for example as indicated by average per-capita income of the country, health care expenditure, and availability of

Findings from the WB, 2005 HSS case studies show that some strategies were not implemented because of lack of resources initially, or a reduction in available resources later, typically when donor or project finance ceased. There was evidence from the cases that availability of finance was a necessary but not sufficient condition for health service strengthening: some human resource strategies had financial resources, but a shortage of

**4. Conclusions from the review of research into health service strengthening** 

There is evidence of high levels of unmet health needs in Yemen, and of the potential for health services to prevent and alleviate suffering, especially of poor people. There is evidence of a number of deficiencies in the allocation of services, their performance and accessibility. This evidence suggests that changing the allocation of resources and increasing efficiency could do much to meet existing needs. However, the changes will not be easy to make, will take time, and will need capable management and incentives for change at all levels. Central and local government will need support to build the capacity, commitment

Research in Yemen has found vertical programmes have been effective, but possibly at the expense of generic primary health care and district services. Research found that PHC could provide EPI successfully on an outreach basis, and this "integration" model has recently been used for Malaria, TB, IMCI, nutrition, and bilharzias in a GAVI programme. Evidence shows that some PHC and districts can successfully provide prevention and care for these diseases and clients, but others need additional strengthening so as to be able to do so. Research from other lower income countries shows the strength of service delivery (amount, accessibility for those most in need and quality) is most strongly influenced by the resources available for the service. This, in turn, depends on the amount of finance from government and private (individual and other), the number and skills of health workers, the facilities and supplies (especially drugs), and participation of the community in different ways

Other specific factors have a greater or lesser influence on the strength of service delivery in different situations (e.g. pay and conditions of government workers can be critical for

**3.11 Should we implement a strategy if we are uncertain if we have the right** 

health workers prevented full implementation (e.g. in Ethiopia, Afghanistan).

#### **3.8 How important is it to adapt the strategy to fit the situation, and to continue to adapt it?**

There was some evidence from the review that adaption – taking a strategy and adapting it to the country and local conditions - was associated with fuller strategy implementation. In addition, that continuous strategy adaption in scale up led to fuller implementation and that adaptation was easier in small-scale interventions. In Ghana, a scale-up of a child and maternal health service strengthening pilot was made using an approach adapted for the situation using peer demonstration, diffusion, and teamwork (Phillips et al, 2006).

There are different examples of intervention approaches which decisions makers can use. A two-phased approach involves a pilot, then feedback, and then further modification of the intervention, followed by regional or national dissemination. A three-phased approach may prove to be even more effective. If time and resources allow, the initial pilot may be followed by additional pilots at the same time within different country contexts, allowing for more detailed guidance for decision-makers. There is evidence that implementation effectiveness is increased by providing continuous feedback to the strategy team and leaders about health service needs, constraints, implementation progress, and health service impact. This can be done by independent researchers.

Data from the WB, 2005 HSS case studies show that the planning of all the strategies in each country included some type of assessment of constraints and adaptation of ideas to the country situation. There was great variation after initial national planning in, continual adaptation (e.g. whether annual reviews and re-planning were carried out to adjust the strategy to changing circumstances), and also in adaptation by lower levels to the situation and needs of local areas.

There is evidence that strategies with not only initial, but also continual and local adaptation were more successful from the examples of decentralization in Ethiopia, Ghana Uganda, and Zambia (Øvretveit et al., 2008).

#### **3.9 Is stakeholder involvement and consultation necessary to effective implementation?**

Overall there is some limited evidence from the research that consulting or involving those who make the change, or who can stop it, is necessary for implementation. But there are also counter examples from authoritarian governance situations such as China where success was due to strong implementation structures without consultation (Kaufman, et al 2006). The research can help decision makers be more aware of the different approaches to consultation and involvement of different parties and levels of the health system, and of examples where this has been done. Research does not show if involvement and consultation is always necessary or which type is most effective in which situations.

The research reports adaptation by decision makers or implementers alone, after consultation. It also reports consultation and stakeholder involvement with little adaptation, and as pre-implementation preparation or as a form of education (Fajans, 2006)

The evidence suggests that some pre-implementation consultation can increase the speed and depth to which a strategy is implemented, but much depends on the country's history and culture. Many studies refer to lack of stakeholder consultation, or of lack of involvement and commitment as one explanation for less successful implementation. Where there has been successful implementation, widespread involvement in a process for agreeing the strategy is often reported as building commitment and as a key factor explaining successful implementation.

There was some evidence from the review that adaption – taking a strategy and adapting it to the country and local conditions - was associated with fuller strategy implementation. In addition, that continuous strategy adaption in scale up led to fuller implementation and that adaptation was easier in small-scale interventions. In Ghana, a scale-up of a child and maternal health service strengthening pilot was made using an approach adapted for the

There are different examples of intervention approaches which decisions makers can use. A two-phased approach involves a pilot, then feedback, and then further modification of the intervention, followed by regional or national dissemination. A three-phased approach may prove to be even more effective. If time and resources allow, the initial pilot may be followed by additional pilots at the same time within different country contexts, allowing for more detailed guidance for decision-makers. There is evidence that implementation effectiveness is increased by providing continuous feedback to the strategy team and leaders about health service needs, constraints, implementation progress, and health service impact.

Data from the WB, 2005 HSS case studies show that the planning of all the strategies in each country included some type of assessment of constraints and adaptation of ideas to the country situation. There was great variation after initial national planning in, continual adaptation (e.g. whether annual reviews and re-planning were carried out to adjust the strategy to changing circumstances), and also in adaptation by lower levels to the situation

There is evidence that strategies with not only initial, but also continual and local adaptation were more successful from the examples of decentralization in Ethiopia, Ghana Uganda,

Overall there is some limited evidence from the research that consulting or involving those who make the change, or who can stop it, is necessary for implementation. But there are also counter examples from authoritarian governance situations such as China where success was due to strong implementation structures without consultation (Kaufman, et al 2006). The research can help decision makers be more aware of the different approaches to consultation and involvement of different parties and levels of the health system, and of examples where this has been done. Research does not show if involvement and

The research reports adaptation by decision makers or implementers alone, after consultation. It also reports consultation and stakeholder involvement with little adaptation,

The evidence suggests that some pre-implementation consultation can increase the speed and depth to which a strategy is implemented, but much depends on the country's history and culture. Many studies refer to lack of stakeholder consultation, or of lack of involvement and commitment as one explanation for less successful implementation. Where there has been successful implementation, widespread involvement in a process for agreeing the strategy is often reported as building commitment and as a key factor explaining successful

consultation is always necessary or which type is most effective in which situations.

and as pre-implementation preparation or as a form of education (Fajans, 2006)

**3.9 Is stakeholder involvement and consultation necessary to effective** 

**3.8 How important is it to adapt the strategy to fit the situation, and to continue to** 

situation using peer demonstration, diffusion, and teamwork (Phillips et al, 2006).

This can be done by independent researchers.

and needs of local areas.

**implementation?** 

implementation.

and Zambia (Øvretveit et al., 2008).

**adapt it?** 

#### **3.10 Are there strategies which are less dependent on the environment, which work in most countries?**

There was no strong evidence from the review of research that some strategies were more "robust" than others, and less influenced by some of the conditions which appeared to affect implementation of all the strategies. A number of studies reported success where implementation had included constraints assessment and actions to reduce constraints (table 1 at end of this review). The implications are that, if decision makers take action to ensure that as many of these conditions as possible are met, this would increase the likelihood of implementation of the strategy (Travis, 2004).

#### **3.11 Should we implement a strategy if we are uncertain if we have the right resources?**

The research reviewed shows that one condition profoundly influencing all strengthening strategies are whether there are adequate resources for the change, for example as indicated by average per-capita income of the country, health care expenditure, and availability of health workers and capable managers (WB, 2005).

Findings from the WB, 2005 HSS case studies show that some strategies were not implemented because of lack of resources initially, or a reduction in available resources later, typically when donor or project finance ceased. There was evidence from the cases that availability of finance was a necessary but not sufficient condition for health service strengthening: some human resource strategies had financial resources, but a shortage of health workers prevented full implementation (e.g. in Ethiopia, Afghanistan).

### **4. Conclusions from the review of research into health service strengthening strategies**

There is evidence of high levels of unmet health needs in Yemen, and of the potential for health services to prevent and alleviate suffering, especially of poor people. There is evidence of a number of deficiencies in the allocation of services, their performance and accessibility. This evidence suggests that changing the allocation of resources and increasing efficiency could do much to meet existing needs. However, the changes will not be easy to make, will take time, and will need capable management and incentives for change at all levels. Central and local government will need support to build the capacity, commitment and persistence to make the changes needed.

Research in Yemen has found vertical programmes have been effective, but possibly at the expense of generic primary health care and district services. Research found that PHC could provide EPI successfully on an outreach basis, and this "integration" model has recently been used for Malaria, TB, IMCI, nutrition, and bilharzias in a GAVI programme. Evidence shows that some PHC and districts can successfully provide prevention and care for these diseases and clients, but others need additional strengthening so as to be able to do so.

Research from other lower income countries shows the strength of service delivery (amount, accessibility for those most in need and quality) is most strongly influenced by the resources available for the service. This, in turn, depends on the amount of finance from government and private (individual and other), the number and skills of health workers, the facilities and supplies (especially drugs), and participation of the community in different ways including volunteer services.

Other specific factors have a greater or lesser influence on the strength of service delivery in different situations (e.g. pay and conditions of government workers can be critical for motivation and retention in most situations).

Strengthening Health Systems in Yemen:

**5.3 Strengthening management and systems** 

"change champions," and implementation teams.

**5.4 Reduce constraints to health service strengthening** 

**Factor Description** 

with adequate management systems are:

performance and accountability

implementation of most types of strategies:

**Enabling/Hindering** 

Resources for the strengthening strategy

Management and governance capacity

Political stability and support

Source: Øvretveit, 2006

Table 1. Enabling/Hindering Factor

management information and human resource management.

Review of Evidence and Implications for Effective Actions for the Poor 295

For these health service strengthening actions to be carried out, actions to strengthen the health system will be needed. These are actions which increase the capability of managers at all levels and how they work together, and improve the management systems such as for

For most strategies, managers at all levels need to be developed and given time to plan and implement strengthening interventions (rather than solely manage routine operations), with some managers dedicated full-time to implementing the strengthening intervention. Aspects of leadership associated with successful strengthening changes include: a clearly communicated mandate from top management that gives authority, resources, and accountability to leaders and teams throughout the organization, as well as respected

The evidence suggests that some strategies which require stronger management capacity should not be pursued on more than a pilot basis until the capacity has been developed. Strategies which might be considered later, and which have proven to have some success

1. Performance based contracting or other ways of linking financing to measures of

2. Payment of incentives to health workers to increase the quantity and quality of services.

In planning and implementing a strategy, decision makers would be advised to assess and address the following factors which have been reported in research to enable/hinder

> Funding for the strengthening strategy Number of personnel engaged in carrying out the strengthening intervention Competence of managerial and front-line personnel (e.g. professionalism, skills, expertise in change processes), particularly their ability to adapt the intervention to local circumstances (i.e. through an assessment of constraints, opportunities, resources)

> Ability/power of each management level to prompt the level below to take action Ability of each management level to hold others accountable (i.e. impose rewards, sanctions) Degree of corruption Degree of local community participation and assistance in the implementation process

> Frequency of changes in government, or leadership implementing the strategy Degree and consistency of support by powerful interest groups Degree of consensus among powerful interest groups Degree and consistency of popular support
