- Open Access
SUPPORT Tools for Evidence-informed Policymaking in health 18: Planning monitoring and evaluation of policies
Health Research Policy and Systemsvolume 7, Article number: S18 (2009)
This article is part of a series written for people responsible for making decisions about health policies and programmes and for those who support these decision makers.
The term monitoring is commonly used to describe the process of systematically collecting data to inform policymakers, managers and other stakeholders whether a new policy or programme is being implemented in accordance with their expectations. Indicators are used for monitoring purposes to judge, for example, if objectives are being achieved, or if allocated funds are being spent appropriately. Sometimes the term evaluation is used interchangeably with the term monitoring, but the former usually suggests a stronger focus on the achievement of results. When the term impact evaluation is used, this usually implies that there is a specific attempt to try to determine whether the observed changes in outcomes can be attributed to a particular policy or programme. In this article, we suggest four questions that can be used to guide the monitoring and evaluation of policy or programme options. These are: 1. Is monitoring necessary? 2. What should be measured? 3. Should an impact evaluation be conducted? 4. How should the impact evaluation be done?
This article is part of a series written for people responsible for making decisions about health policies and programmes and for those who support these policymakers. The series is intended to help such people ensure that their decisions are well-informed by the best available research evidence. The SUPPORT tools and the ways in which they can be used are described in more detail in the Introduction to this series . A glossary for the entire series is attached to each article (see Additional File 1). Links to Spanish, Portuguese, French and Chinese translations of this series can be found on the SUPPORT website http://www.support-collaboration.org. Feedback about how to improve the tools in this series is welcome and should be sent to: STP@nokc.no.
Scenario 1: You are a senior civil servant with overall responsibility for several healthcare programmes. You wish to ensure that you have the information necessary to assess how various programmes are performing and the impact they are having.
Scenario 2: You work in the Ministry of Health and have been instructed to prepare a memo on various issues that should be taken into consideration when the national vaccination programme is evaluated.
Scenario 3: You work in a unit supporting the government in its use of evidence in policymaking. You are preparing a monitoring and evaluation plan for the national tuberculosis control programme.
For policymakers (Scenario 1), this article suggests a number of questions that their staff might be asked when planning the monitoring and evaluation of a new policy.
For those who support policymakers (Scenarios 2 and 3), this article suggests a number of questions to consider when planning how to monitor the implementation of policies and programmes, and the evaluation of their impacts.
Policymakers and other stakeholders will often need to know whether a new policy or programme has been implemented in accordance with their expectations. Is the programme rollout progressing as planned? Are the objectives being achieved, and are the allocated funds being spent appropriately? Monitoring is the term commonly used to describe the process of systematically collecting data to provide answers to such questions . The term performance monitoring is often used when the main focus of an evaluation is comparing "how well a project, program, or policy is being implemented against expected results" .
Indicators are frequently used as part of the monitoring process. An indicator has been defined as a "quantitative or qualitative factor or variable that provides a simple and reliable means to measure achievement, to reflect the changes connected to an intervention, or to help assess the performance" . An indicator can be a simple count of events, e.g. the number of vaccinations conducted within a set period of time, or a construct based on various data sources, e.g. the proportion of all children being fully immunised before their first birthday.
The term evaluation is sometimes used interchangeably with monitoring, but the former usually suggests a stronger focus on the achievement of results. These terms are not used consistently and may mean different things to different people. The term impact evaluation is frequently used when an attempt is made to evaluate whether observed changes in outcomes (or 'impacts') can be attributed to a particular policy or programme.
Questions to consider
Is monitoring necessary?
What should be measured?
Should an impact evaluation be conducted?
How should the impact evaluation be done?
1. Is monitoring necessary?
The importance of monitoring depends on the perceived need among relevant stakeholders to know more about what is happening 'on the ground'.
Determining whether a system for monitoring a policy or programme should be established may depend on several factors, including:
Whether a monitoring system is already in place that includes the desired indicators, or if a new set of indicators is required
The likely costs of establishing the system required. For example, could a few new items be added to data collection procedures already in place, or is it necessary to conduct additional large-scale household surveys or to develop a completely new tool?
Whether the findings are likely to be useful. What actions should be taken if monitoring reveals that things are not going as planned?
Monitoring is not worthwhile if data remain unused. Data are particularly useful if corrective action is undertaken when a gap is identified between expected and actual results. Such findings may result in expectations being reconsidered. This may take the form of assessments, for example, of whether the initial plans were too ambitious, or whether a new policy has failed to work as effectively as expected.
2. What should be measured?
Indicators that focus on various parts of the 'results chain' (i.e. on inputs, activities, outputs, outcomes or impacts - see Figure 1) are typically used to monitor the implementation of a programme or policy option. In some circumstances it may be seen as sufficient to monitor inputs (i.e. the provision of resources such as personnel and equipment). In others it may be important to monitor the activities of the programme or its outcomes (such as the number of children fully immunised).
Validity: the extent to which the indicator accurately measures what it purports to measure
Acceptability: the extent to which the indicator is acceptable to those who are being assessed and those undertaking the assessment
Feasibility: the extent to which valid, reliable and consistent data are available for collection
Reliability: the extent to which there is minimal measurement error, or the extent to which findings are reproducible should they be collected again by another organisation
Sensitivity to change: the extent to which the indicator has the ability to detect changes in the unit of measurement
Predictive validity: the extent to which the indicator has the ability to accurately predict relevant outcomes
Costs related to data collection and the capacity to analyse and feed back data to managers and providers may also limit the choice of indicators. In settings where analytical resources are scarce, it may be preferable to select a simple indicator even if it does not have the best predictive validity, rather than an indicator that requires statistical manipulation.
A trade-off is often apparent between, on one hand, wanting to use desired and optimal indicators and on the other hand, having to use those indicators which are based on existing data. There are good reasons not to select more indicators than are absolutely essential. These reasons include the need to limit the burden of data collection within a health system, avoid the collection of data that are not utilised, and focus on collecting data of higher quality, even if this means collecting less data overall .
Routinely collected information from health systems may provide valuable data that can be used as a data source for monitoring purposes. Data can also be collected specifically for the purpose of monitoring, e.g. through surveys or interviews. Consideration should be given to the level of motivation among those expected to collect data. In many instances, health personnel will need to integrate data collection into a busy daily schedule. Therefore if the information being collected has little or no local obvious value to them, their motivation for undertaking such tasks may be low. Similarly, if incentives or penalties are associated with the findings from the monitoring process (e.g. where the payment of providers is linked to performance indicators), the risk of data manipulation or system gaming should be considered.
3. Should an impact evaluation be conducted?
One of the limitations of monitoring activities, as described above, is the fact that such activities do not necessarily indicate whether a policy or programme has had an impact on the indicators that have been measured. This is because indicators used for monitoring will almost always be influenced by factors other than those related to particular interventions. This makes it extremely difficult to determine which factors caused the observed changes. If monitoring reveals that performance is improving, this does not necessarily mean that the intervention is the (only) causal factor. It is conceivable that the indicators would have improved anyway even in the absence of the intervention (see Figure 2).
The establishment of a causal relationship between a programme or policy and changes in outcomes is at the core of what impact evaluation is about. What would have happened to those receiving an intervention if they had not in fact received it, is the central impact evaluation question, according to the World Bank .
There may be strong reasons to expect positive results based on solid documentation from, for example, previous evaluations. However, very often such evidence is lacking. Or the evidence available may not be applicable to the current setting. Thus, there is a real risk that a new programme may be ineffective or, even worse, cause more harm than good. This issue is important for policymakers to clarify when implementing new programmes. It is also important because of the benefit that such knowledge could bring to future health policymaking both in the programme setting and other jurisdictions.
Conducting impact evaluations can be costly. Whether such studies represent good value for money can be ascertained by comparing the consequences of undertaking an evaluation with the consequences of not undertaking an evaluation. For example, is it likely that a programme would be stopped or modified if the results proved to be negative? If the answer is 'no', the value of undertaking an impact evaluation is clearly limited.
An impact evaluation is generally more likely to represent value for money when results can be obtained as the intervention is being rolled out. In such circumstances there is an opportunity to improve or stop the rollout based on the results of an impact evaluation conducted in the early stages of implementation. This would provide value for money in two instances: firstly, when a pilot study is not possible and, secondly, when it would be possible and practical to modify or stop the rollout (if needed) based on the results.
The Mexican government's health insurance scheme, Seguro Popular, is an example of an impact evaluation embedded in a programme rollout [9–11]. Implemented in 2001, the scheme was established in order to extend health insurance coverage to the almost 50 million Mexicans not yet covered by existing programmes. Taking advantage of the timetable of the progressive rollout, the government set up an evaluation comparing the outcomes for those communities receiving the scheme with those still waiting for it. In addition to evaluating whether the reform achieved the outcomes intended and did not have unintended adverse effects, the evaluation also provides for shared learning.
An impact evaluation may also be useful after a programme has been fully implemented, e.g. when there is uncertainty about continuing a programme. For example, the conditional cash transfer scheme, Progresa (later known as Oportunidades), which was introduced in the mid-1990s provided cash "on the condition that families fulfil particular elements of co-responsibility, such as sending children to school rather than work, providing them with a specially formulated nutritional supplement, and attending a clinic to receive a specified package of interventions for health promotion and disease prevention." . For evaluation purposes, 506 communities were randomly assigned to either enter the programme immediately or 2 years later . The findings from this impact evaluation directly informed policy decisions in Mexico, persuading the government "not only to continue with the programme, but also to expand it" .
4. How should the impact evaluation be done?
Attributing an observed change to a programme or policy requires a comparison between the individuals or groups exposed to it, and others who are not. It is also important that the compared groups are as similar as possible in order to rule out influences other than the programme itself. This can effectively be done by randomly allocating individuals or groups of people (e.g. within geographic areas) to either receive the programme or not to receive it, in what is called a randomised trial. Usually such trials are conducted as pilot projects before a programme is introduced at a national level. But they can also be undertaken in parallel with full scale implementation, as illustrated by the Mexican examples given above.
Randomised trials may, however, not always be feasible. Alternative approaches include the comparison of changes before and after programme implementation, with observed changes during the same time period in areas where the programme was not implemented (e.g. in neighbouring districts or countries). This is called a controlled before-after evaluation. Alternatively, an interrupted time-series may be used in which data are collected from multiple time points before, during, and after programme implementation.
Simply comparing the value of an indicator before and after programme implementation is not generally recommended since the risk of misleading findings is high - observed changes, e.g. HIV-incidence may be caused by known and unknown factors other than those related to the programme itself (see Figure 2) [14, 15]. An overview of a number of evaluation designs is provided in Additional File 2. The weaknesses and strengths of each method described in Additional file 2 are outlined in Additional file 3.
Impact evaluations should be planned well ahead of programme implementation in conjunction with relevant stakeholders, including policymakers. After a programme has been rolled out widely it is usually too late to carry out baseline measurements or to establish appropriate comparison groups. For example, using random assignments to decide whether communities will be included in a programme or not, cannot be done after the programme has been implemented nationally. Impact evaluations that are built into a programme from the start are thus more likely to yield valid findings than those evaluations conducted as an afterthought. Furthermore, if impact evaluations are seen as an integrated part of programme implementation, policymakers and others may be more committed to taking the findings into account.
The number of individuals or communities required for an impact evaluation should also be estimated at an early stage. This will ensure that there is sample size large enough for meaningful conclusions to be drawn from the evaluation findings.
In healthcare, as in most other areas, programmes need to be both effective and cost-effective. To assess the economic aspects of a programme, resource use and costs must be estimated, preferably based on data collected from real-life implementation . Decisions on what economic data to collect should therefore also be made at an early stage, before the evaluation starts.
Impact evaluations are likely to be most informative if a process evaluation is included. A process evaluation may examine whether the programme or policy option was delivered as intended. It may also investigate the processes of implementation and change, explore responses to the programme, and explore reasons for the findings of the evaluation .
See Table 2 for examples of impact evaluations.
Budget, time or data constraints may act as disincentives to ensuring rigorous implementation of an evaluation. Such constraints can affect the reliability of impact evaluations in a number of ways:
By compromising the overall validity of the results, for example, due to insufficient planning or follow-up, or through a paucity of baseline data, a reliance on inadequate data sources, or the selection of inappropriate comparison groups
Through the use of inadequate samples, e.g. due to the selection of samples that are convenient to sample but may not be representative, as a result of sample sizes being too small, or by a lack of sufficient attention to contextual factors
Such constraints can be addressed by starting the planning process early or finding ways to reduce the costs of data collection. It is important to ensure, however, that neither the possible threats to the validity of the results, nor the limitations of the sample, are such that the results of the evaluation will be unable to provide reliable information. Before conducting an evaluation, an assessment should therefore be made as to whether an adequate evaluation is possible. If it is not, an assessment needs to be undertaken as to whether a programme should be implemented without prior evaluation, in the face of uncertainty about its potential impacts .
Impact evaluations are not worthwhile if the findings are not used. Results should be used to inform decisions about whether to continue, change or stop existing programmes. Clearly, other interests will also need to be taken into consideration. For instance, decision makers may elect not to emphasise particular findings from certain evaluations when such findings conflict with other interests that are perceived as more important . However, it is important to avoid the suppression of findings from impact evaluations, e.g. for political reasons. Failing to use evaluation findings contradicts one of the main objectives of conducting such evaluations: to learn from experience and share the knowledge that has been generated. Using independent parties to conduct impact evaluations may decrease the risk of having the findings manipulated or held back from the public.
A number of aspects related to monitoring and evaluation have been described in this article. At present, many programme monitoring and evaluation efforts are commonly done using methods that do not yield valid assessments of the implementation of a policy or programme or valid estimates of effects. Sometimes such evaluations are not done at all. By taking the issues described in this article into consideration, policymakers and those who support them should be able to develop plans that will generate new and directly useful knowledge.
Useful documents and further reading
Segone M (ed). Bridging the gap: The role of monitoring and evaluation in evidence-based policy making. UNICEF, the World Bank and the International Development Evaluation Association. http://www.unicef.org/ceecis/evidence_based_policy_making.pdf
MacKay K. How to Build M&E Systems to Support Better Government. 2007. Washington DC, The World Bank. http://www.worldbank.org/ieg/ecd/docs/How_to_build_ME_gov.pdf
Monitoring and Evaluation (M&E): Some Tools, Methods and Approaches. 2004. Washington DC. The World Bank. http://lnweb90.worldbank.org/oed/oeddoclib.nsf/24cc3bb1f94ae11c85256808006a0046/a5efbb5d776b67d285256b1e0079c9a3/$FILE/MandE_tools_methods_approaches.pdf
Framework for Managing Programme Performance Information. 2007. National Treasury of South Africa. http://www.treasury.gov.za/publications/guidelines/FMPI.pdf
Barber S. Health system strengthening interventions: Making the case for impact evaluation. 2007. Geneva, Alliance for Health Policy and Systems Research. http://www.who.int/alliance-hpsr/resources/Alliance%20%20HPSR%20-%20Briefing%20Note%202.pdf
Savedoff WD, Levine R, Birdsall N. When will we ever learn? Improving lives through impact evaluation. Report of the Evaluation Gap Working Group. 2006. Washington DC, Center for Global Development. http://www.cgdev.org/content/publications/detail/7973/
Grimshaw J, Campbell M, Eccles M and Steen N. Experimental and quasi-experimental designs for evaluating guideline implementation strategies. Family Practice 2000; 17: S11-S18. http://fampra.oxfordjournals.org/cgi/reprint/17/suppl_1/S11
Links to websites
Independent Evaluation Group (IEG) at the World Bank: http://www.worldbank.org/ieg - IEG is an independent unit within the World Bank. IEG assesses what is effective or not effective with regard to policy options, how a borrower plans to run and maintain a project, and the lasting contribution of the Bank to a country's overall development.
International Initiative for Impact Evaluation (3ie): http://www.3ieimpact.org - 3ie seeks to improve the lives of poor people in low- and middle-income countries by providing and summarising evidence related to what policy options work, as well as when and why, and the costs involved.
Health Metrics Network: http://www.who.int/healthmetrics/en - The Health Metrics Network (HMN) has the strategic goal of increasing the availability and use of timely and accurate health information. To achieve this, HMN identifies strategies for Health Information System (HIS) development and strengthening, supports countries in implementing HIS reform, and increases knowledge about global public goods through research, technical innovation, and sharing lessons learned.
NorthStar: http://www.rebeqi.org/?pageID=34%26ItemID=35 - NorthStar is a tool for planning, conducting and evaluating quality improvement programmes.
Lavis JN, Oxman AD, Lewin S, Fretheim A: SUPPORT Tools for evidence-informed health Policymaking (STP). Introduction. Health Res Policy Syst. 2009, 7 (Suppl 1): I1-10.1186/1478-4505-7-S1-I1.
Development Assistance Committee Working Party on Aid Evaluation: Glossary of Key Terms in Evaluation and Results Based Management. 2002, Paris, OECD Publications, [http://www.oecd.org/dataoecd/29/21/2754804.pdf]
Harries AD, Gomani P, Teck R, de Teck OA, Bakali E, Zachariah R, Libamba E, Mwansambo A, Salaniponi F, Mpazanje R: Monitoring the response to antiretroviral therapy in resource-poor settings: the Malawi model. Trans R Soc Trop Med Hyg. 2004, 98: 695-701. 10.1016/j.trstmh.2004.05.002.
Jakobsen E, Palshof T, Osterlind K, Pilegaard H: Data from a national lung cancer registry contributes to improve outcome and quality of surgery: Danish results. Eur J Cardiothorac Surg. 2009, 35: 348-52. 10.1016/j.ejcts.2008.09.026.
Smith PC, Mossialos E, Papanicolas I: Performance measurement for health system improvement: experiences, challenges and prospects. Background Document for WHO European Ministerial Conference on Health Systems: "Health Systems, Health and Wealth". 2008, Copenhagen, World Health Organization, Europe, [http://www.euro.who.int/document/hsm/2_hsc08_ebd2.pdf]
Campbell SM, Braspenning J, Hutchinson A, Marshall M: Research methods used in developing and applying quality indicators in primary care. Qual Saf Health Care. 2002, 11: 358-64. 10.1136/qhc.11.4.358.
MacKay K: How to Build M&E Systems to Support Better Government. 2007, Washington DC, The World Bank, [http://www.worldbank.org/ieg/ecd/docs/How_to_build_ME_gov.pdf]
The World Bank: Impact evaluation: Overview. 2009, The World Bank, [http://go.worldbank.org/2DHMCRFFT2]
Moynihan R, Oxman A, Lavis JN, Paulsen E: Evidence-Informed Health Policy: Using Research to Make Health Systems Healthier. Rapport nr. 1-2008. 2008, Oslo, Nasjonalt kunnskapssenter for helsetjenesten, [http://www.kunnskapssenteret.no/Publikasjoner/469.cms]
Frenk J, Gonzalez-Pier E, Gomez-Dantes O, Lezana MA, Knaul FM: Comprehensive reform to improve health system performance in Mexico. Lancet. 2006, 368: 1524-34. 10.1016/S0140-6736(06)69564-0.
Gakidou E, Lozano R, Gonzalez-Pier E, Abbott-Klafter J, Barofsky JT, Bryson-Cahn C, Feehan DM, Lee DK, Hernandez-Llamas H, Murray CJ: Assessing the effect of the 2001-06 Mexican health reform: an interim report card. Lancet. 2006, 368: 1920-35. 10.1016/S0140-6736(06)69568-8.
Frenk J: Bridging the divide: global lessons from evidence-based health policy in Mexico. Lancet. 2006, 368: 954-61. 10.1016/S0140-6736(06)69376-8.
Rivera JA, Sotres-Alvarez D, Habicht JP, Shamah T, Villalpando S: Impact of the Mexican program for education, health, and nutrition (Progresa) on rates of growth and anemia in infants and young children: a randomized effectiveness study. JAMA. 2004, 291: 2563-70. 10.1001/jama.291.21.2563.
Savedoff WD, Levine R, Birdsall N: When will we ever learn? Improving lives through impact evaluation. 2006, Washington DC, Center for Global Development, [http://www.cgdev.org/files/7973_file_WillWeEverLearn.pdf]
Shadish WR, Cook TD, Campbell DT: Experimental and Quasi-Experimental Desings for Generalized Causal Inference. 2002, Houghton Mifflin
Oxman AD, Fretheim A, Lavis JN, Lewin S: SUPPORT Tools for evidence-informed health Policymaking (STP). 12. Finding and using research evidence about resource use and costs. Health Res Policy Syst. 2009, 7 (Suppl 1): S12-10.1186/1478-4505-7-S1-S12.
Lewin S, Glenton C, Oxman AD: Use of qualitative methods alongside randomised controlled trials of complex healthcare interventions: methodological study. BMJ. 2009, 339: b3496-10.1136/bmj.b3496.
Oxman AD, Lavis JN, Fretheim A, Lewin S: SUPPORT Tools for evidence-informed health Policymaking (STP). 17. Dealing with insufficient research evidence. Health Res Policy Syst. 2009, 7 (Suppl 1): S17-10.1186/1478-4505-7-S1-S17.
Scheel IB, Hagen KB, Oxman AD: The unbearable lightness of healthcare policy making: a description of a process aimed at giving it some weight. J Epidemiol Community Health. 2003, 57: 483-7. 10.1136/jech.57.7.483.
Amuron B, Coutinho A, Grosskurth H, Nabiryo C, Birungi J, Namara G, Lewin J, Smith PG, Jaffar S: A cluster-randomised trial to compare home-based with health facility-based antiretroviral treatment in Uganda: study design and baseline findings. Open AIDS J. 2007, 1: 21-7. 10.2174/1874613600701010021.
Jaffar S, Amuron B, Birungi J, Namara G, Nabiryo C, Coutinho A, Grosskurth H: Integrating research into routine service delivery in an antiretroviral treatment programme: lessons learnt from a cluster randomized trial comparing strategies of HIV care in Jinja, Uganda. Trop Med Int Health. 2008, 13: 795-800. 10.1111/j.1365-3156.2008.02131.x.
London School of Hygiene and Tropical Evidence: Home-based HIV care just as effective as clinic-based care in Sub-saharan Africa. 2009, London School of Hygiene and Tropical Evidence, University of London, [http://www.lshtm.ac.uk/news/2009/homeHIVcare.html]
Fretheim A, Havelsrud K, MacLennan G, Kristoffersen DT, Oxman AD: The effects of mandatory prescribing of thiazides for newly treated, uncomplicated hypertension: interrupted time-series analysis. PLoS Med. 2007, 4: e232-10.1371/journal.pmed.0040232.
Please see the Introduction to this series for acknowledgements of funders and contributors. In addition, we would like to acknowledge Ruth Levine for helpful comments on an earlier version of this article.
This article has been published as part of Health Research Policy and Systems Volume 7 Supplement 1, 2009: SUPPORT Tools for evidence-informed health Policymaking (STP). The full contents of the supplement are available online at http://www.health-policy-systems.com/content/7/S1.
The authors declare that they have no competing interests.
AF prepared the first draft of this article. ADO, JNL and SL contributed to drafting and revising it.