Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
How to Mon�tor for Populat�on Health Outcomes: Gu�del�nes for develop�ng a mon�tor�ng framework
Public Health Intelligence Occasional Bulletin No. 44
��
–
How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
AuthorDr John Wren from the Public Health Intelligence (PHI) Unit
in the Ministry of Health wrote this guide.
Citation: Ministry of Health. 2007. How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework. Occasional Bulletin No. 44.
Wellington: Ministry of Health.
Published in July 2007 by the Ministry of Health
PO Box 5013, Wellington, New Zealand
ISBN 978-0-478-19150-9 (print) ISBN 978-0-478-19153-0 (Internet)
HP 4414
This document is available on the Ministry of Health website: http://www.moh.govt.nz
��� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
ForewordPublic Health Intelligence (PHI) is the epidemiology group of the Ministry of Health. PHI carries out the Ministry’s statutory responsibility to monitor the health of the New Zealand population by analysing population health outcomes and risks and determinants, and examining inequalities across regional boundaries and between population groups. An important role for PHI is the delivery and dissemination of epidemiology evidence for the development of policy and decision-making in the health sector.
How to Monitor for Population Health Outcomes presents guidance to public and population health programme managers, and interested others, on how to develop indicators to monitor progress on achieving population health outcomes sought from their programmes. This guideline has been designed to complement the information and guidance presented in the Ministry’s Guide to Developing Public Health Programmes: A generic programme logic model (Ministry of Health 2006). The need to develop a robust outcomes monitoring framework is driven by an increased expectation from central government that there will be a focus on results in the design and delivery of publicly funded services. This expectation is highlighted by the requirements of the Public Finance Act 2004 and the Crown Entities Act 20041 (The Treasury and State Services Commission 2007a; b).
We welcome your comments and suggestions about the contents, and any additions or clarifications you might have.
Barry BormanManager (Epidemiologist)Public Health Intelligence
1 Crown entities includes District Health Boards.
�v How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
AcknowledgementsThe author wishes to particularly thank:
• Geoff Stone, Senior Advisor, Centre for Social Research and Evaluation, Ministry of Social Development, for his contribution to the development of definitions in the Glossary and insights into outcomes monitoring in social settings other than public health
• Craig Wright, Senior Advisor (Statistics), Public Health Intelligence (PHI), Ministry of Health, for his significant preparation in Part C of the discussion on the small numbers problem and the statistical ability to detect change in health outcomes of interest; and
• Dyfed Thomas and Sarah Gerritsen, (PHI), for their preparation of Part D.
The author also wishes to thank the following external peer reviewers for their insightful comments:
• Anne Dowden, Director – Evaluation, Research New Zealand
• Rob Smith and Tony Walzl, Allen and Clarke Policy and Regulatory Specialists
• Ray Prebble, Editor, Macmillan and Prebble.
The following people provided valuable internal peer review:
• Dr Chris Wong, Public Health Physician, Ministry of Health
• Sarah Gerritsen, Senior Advisor (Population Health Research), PHI
• Dr Kirstin Lindberg, Senior Advisor (Public Health Medicine), PHI.
v How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
ContentsForeword ................................................................................................................................���
Acknowledgements.................................................................................................................�v
Part A: Introduct�on ................................................................................................................ 1
How to use this guide ........................................................................................................... 1
Why measure performance? ................................................................................................. 2
Part B: What Object�ves and Assoc�ated Outcomes Need to be Mon�tored? .............................. 3
Introduction: outcomes monitoring programmes .................................................................. 4
What should you measure to demonstrate the success of the programme? ........................... 4
What makes good outcome statements that are measurable? ............................................... 6
Have you used action words in active statements? ........................................................... 6
Are they SMART objectives and/or outcomes? .................................................................. 7
Monitoring double-barrelled objectives ............................................................................ 8
Prioritising objectives/outcomes for measurement .............................................................. 8
Management selection criteria for prioritising what is essential to monitor in public health programmes ............................................................................................. 10
Instrument selection criteria for prioritising what is possible to monitor appropriately in public health programmes ......................................................................................... 12
Additional questions and guidance for selecting outcome measures and indicators ........... 13
Do you need a new outcome indicator or instrument? .................................................... 13
Match the measurement instrument to the objective ...................................................... 14
Cost-effectiveness measures .......................................................................................... 15
Decide on the type of instrument ................................................................................... 15
Collecting and analysing information .................................................................................. 16
Part C: Issues to Cons�der �n Outcomes Mon�tor�ng ............................................................... 17
Outcomes monitoring as a tool for ‘continuous programme improvement’: the problem of accountability, attribution and performance management ................................................. 17
Outcomes monitoring versus evaluation ............................................................................ 19
The problem of small numbers ........................................................................................... 19
Detecting change in small numbers ............................................................................... 20
v� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Part D: Ava�lable Data and Informat�on from Publ�c Health Intell�gence .................................. 24
Survey data ........................................................................................................................ 24
Administrative data ............................................................................................................ 26
PHIOnline – www.phionline.moh.govt.nz ............................................................................ 27
Glossary ............................................................................................................................... 30
References ........................................................................................................................... 44
Further reading ................................................................................................................... 45
F�gures:Figure 1: A framework for prioritising what outcomes to monitor ................................................ 9
Figure 2: Simulated effect of 15% reduction in violent offences ............................................... 22
Figure 3: Power to detect change due to an intervention, for specified incidence, population size and 10% incidence reduction ......................................................................... 23
Figure 4: The PHIOnline home page ........................................................................................ 27
Figure 5: PHIOnline single map interface ................................................................................. 28
Figure 6: PHIOnline double map interface ............................................................................... 29
Tables:Table 1: Distinguishing objectives, outcomes and outputs ....................................................... 5
Table 2: List of action words for writing performance objectives................................................. 6
Table 3: Defining SMART objectives ........................................................................................... 7
Table 4: Checklist and score card for prioritising outcomes and indicators for monitoring ........ 13
Table 5: Summary of New Zealand Health Monitor surveys, 2002–2012 .................................. 25
Table 6: Summary of administrative data sources .................................................................... 26
1 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Part A: Introduct�on
Key Po�nts• This How to aims to help public and population health programme managers to develop
timely, appropriate and cost-effective outcome measures for programmes funded by government.
• Measuring programme performance is essential for maintaining and enhancing support for public health programmes.
• The process of developing a logic model and an outcomes monitoring framework is intended to facilitate a process of ‘continuous programme improvement’.
• This How to should be read in conjunction with A Guide to Developing Public Health Programmes: A generic programme logic model (Ministry of Health 2006).
How to use th�s gu�deThis How to aims to provide guidance to public and population health programme managers and planners – whether they be in District Health Boards, non-governmental organisations or government departments – on how to select timely, appropriate and cost-effective outcome measures for public health programmes.
Throughout this How to we aim to provide plain English advice, checklists and examples to guide you through the process of selecting which outcomes to measure and how they should be measured. We provide information about some issues that are likely to arise from adopting a more rigorous outcomes-focused framework to monitoring public health programmes. We also outline the types of data held by the Ministry of Health that may be freely used to help monitor the performance of your public health programmes.
The guide is structured into five parts.
• Part A: Introduct�on outlines the purpose of this document and where it fits in relation to other Public Health Intelligence (PHI) publications.
• Part B: What Object�ves and Assoc�ated Outcomes Need to be Mon�tored? describes a process for translating policy goals and objectives into measurable outcome statements, and a process for selecting and prioritising the outcomes to be monitored.
• Part C: Issues to Cons�der �n Outcomes Mon�tor�ng discusses a number of key issues arising from the application of an outcomes monitoring approach to publicly funded programmes and small populations.
• Part D: Ava�lable Data and Informat�on from Publ�c Health Intell�gence introduces PHIOnline, a free public access internet-based service to a wide range of government statistics that may be useful in monitoring a population health programme.
• A Glossary provides an alphabetical list of terms, definitions and explanatory text relating to outcomes monitoring language.
2 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
This How to builds on the material presented in A Guide to Developing Public Health Programmes: A generic programme logic model (Ministry of Health 2006). The Guide provides a general introduction on how to design and implement comprehensive and measurable public health programmes in New Zealand, and includes some material on designing outcome measures (Ministry of Health 2006). This How to significantly expands on that material, and material presented at a PHI analytical workshop on Monitoring Public Health Outcomes held in October 2006. Ideally the Guide should be read before this How to.
The Ministry of Health’s Leading for Outcomes website http://www.leadingforoutcomes.org.nz/ is another Ministry-led initiative that promotes and uses an outcomes framework to improve how we think about and improve health in New Zealand, starting with cardiovascular disease (CVD) and diabetes. The website sets out a model of risk factor and disease progression, an outcomes hierarchy and an indicators framework for measuring progress towards achieving better CVD and diabetes health outcomes in New Zealand, with a view to extension of the approach to other health and disability areas.
Why measure performance?The main motivation for introducing management tools such as programme logic models, outcomes monitoring and managing for results is an increased expectation from central government that the design and delivery of publicly funded services will have a strong results focus. This expectation is highlighted by the requirements of the Public Finance Act 2004 and the Crown Entities Act 20042 and associated guidance documents (The Treasury and State Services Commission 2007a; b) for the development of statements of intent by departments and Crown agencies.
To sum up, measuring performance:
• facilitates change and improvement
• is a mechanism for accountability
• supports planning and decision-making relating to resources
• can highlight areas requiring further work.
Ideally, performance measures for public health programmes should provide information about:
• a change in health status and health determinants achieved in priority population groups, including changes in inequalities
• resource and service utilisation
• the programme’s responsiveness to the target population.
2 Crown entities includes District Health Boards.
3 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Part B: What Object�ves and Assoc�ated Outcomes Need to be Mon�tored?
Key Po�nts1. It is not always possible to monitor every objective and outcome, so you will need to
establish a list of prioritised objectives/outcomes for monitoring.
2. The process of planning an outcomes monitoring framework begins with translating policy goals and objectives into SMART objectives:
• Specific
• Measurable
• Achievable (sometimes ‘Accurate’ or ‘Action-oriented’ are used)
• Relevant (sometimes ‘Realistic’ is used)
• Time-based.
3. Management and instrument criteria can be used to assess whether it is possible and essential to monitor the objective or outcome of interest. Management selection criteria are:
• attribution (accountability)
• centrality
• cost-benefit
• robustness to withstand public scrutiny
• timing.
Instrument selection criteria are:
• availability
• reliability
• sensitivity
• validity.
4. Planning for outcomes monitoring requires decisions about:
• What information is required? What is the data going to be used for?
• What data is already available? Is new data really needed?
• Who is going to collect the data?
• What type of measurement instrument could or should be used?
• How many measurement instruments are to be used?
• Who is going to do the analysis?
• When is the data required by to inform timely decision-making?
5. Table 4 provides a checklist and scorecard to help prioritise which objectives and outcomes are essential to monitor.
6. See the Glossary for a full list of terms and associated definitions used throughout this How to, and extra explanatory comment and examples illustrating the use of the terms.
4 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Introduct�on: outcomes mon�tor�ng programmesEstablishing which outcomes need to be – and can be – successfully monitored requires careful thought and planning. Preparing an outcomes monitoring plan will help you to establish a successful monitoring programme.
An outcomes monitoring plan is a process for the routine systematic collection and recording of timely information about aspects of a programme to assess whether progress is being made on achieving the programme’s objectives, and how the programme could be improved. Information is timely when it contains data directly relevant to the programme and – importantly – is available in a timeframe where it can usefully inform decisions about whether the programme is performing as planned, and what things need to be changed to improve the performance of the programme if required.
Collecting baseline data at the start of a programme is vital to be able to assess what change has occurred over time, particularly when you are trying to attribute a change to an effect of the intervention.
In preparing the outcomes monitoring plan you will need to assess:
• what needs to be measured to demonstrate success
• how things should be measured and by whom (deciding what indicators are going to be used, after considering issues such as validity, reliability, sensitivity, attribution, availability of suitable instruments, cost-benefit)
• the timeframes for delivering information about the performance of the programme against the stated objectives to decision-makers and key stakeholders
• whether a formal evaluation process is required.
A good outcomes monitoring process:
• identifies and prioritises the outcomes that are essential to monitor and that can be monitored appropriately
• provides timely information to key decision-makers and stakeholders about the progress made on achieving the desired outcomes
• uses outcome measures robust enough to withstand public scrutiny
• is cost effective
• includes baseline data relevant to the intervention.
What should you measure to demonstrate the success of the programme?Critical to any successful outcomes monitoring plan is to identify what could and should be measured in order to show that the programme is being implemented as planned, and that progress is being made to achieve the desired health outcome(s). The steps involved are:
1. identify what could be measured
2. prioritise these into those that should be measured if resources allow
3. identify the essential few that must be monitored.
� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Underpinning this approach is the assumption that it is not always feasible – or necessary – to measure everything.
The first place to look to determine what could and should be monitored is to identify what the key programme objectives, outcomes and outputs are. You should be able to identify these from the ‘logic model’ developed for the programme. A logic model is a planning tool that describes the purpose, what, where, when, and how of the programme being implemented.
Ideally, programme objectives, outcomes and outputs should be clearly distinguished from each other (see Table 1 for definitions of these terms) and expressed in action words. Expressing outcomes statements clearly makes it easier to identify what actions need to be undertaken, what changes need to take place, what the desired end result is, and consequently what needs to be monitored.
Table 1: D�st�ngu�sh�ng object�ves, outcomes and outputs3
Object�ves are statements about the results a programme seeks to achieve. Any programme must have at least one objective.
Objectives may form a hierarchy that moves from a limited set of high-order objectives that are synonymous with aims or goals to be achieved in the long term (five to seven years). Underneath high-level objectives are more intermediate-level objectives that are to be achieved in a three- to five-year time frame, and which must be achieved in order to attain the high-level objective. The lowest levels of objectives are immediate or operational objectives that must be achieved first – typically in one or two years.
Objectives may be translated directly into ‘outcomes’ if they deal with only one issue. However, double-barrelled objectives will require multiple outcome measures to be developed.
Outcomes are specific statements about the intended change in public health-related attitudes, knowledge, behaviours, or physical health status in the target population(s) sought by undertaking the planned public health activity. In some situations ‘process’ outcomes may be desirable.
Process outcomes typically measure the amount of effort put into a programme and the quality of the service provided. They can also be appropriate where it is important to monitor community support for a programme. Process outcomes that measure effort can be expressed as ‘outputs’.
Outputs are things (such as goods) produced, services delivered, events held, or participation generated resulting from the activities undertaken.
3 See the Glossary for a full list of other terms and associated definitions used throughout this How to, and extra explanatory comment and examples illustrating the use of the terms.
� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
What makes good outcome statements that are measurable?Have you used act�on words �n act�ve statements?In some cases, even though a programme’s objectives and outcomes can be identified, they may be stated in a way that makes them difficult to measure or unsuitable for programme planning or contracting purposes. This can happen because the language used is too passive in tone, or is too complex or abstract. Consequently, in some situations it may be necessary to translate the identified objectives into language that is more suitable for monitoring. Ensuring the language used is active rather than passive will help to do this.
Active statements involve ‘action’ words, and a list of action words identified as suitable for public health is provided in Table 2.
Table 2: L�st of act�on words for wr�t�ng performance object�ves
Accept Adopt Advocate Analyse Arrange
Approve Appraise Bargain Calculate Care
Change Choose Classify Categorise Challenge
Chart Compare Conduct Construct Contrast
Co-operate Check Defend Define Demonstrate
Describe Develop Differentiate Discriminate Draw
Evaluate Execute Explain Express Fill out
Forecast Formulate Generate Identify Inform
Instal Interview Judge Justify Label
List Locate Manipulate Modify Name
Operate Organise Outline Persuade Plan
Prepare Prescribe Produce Purchase Question
Rank Recall Recognise Reflect Remove
Research Resolve Review Select Sort
Specify State Study Take Tell
Translate Use Write
Source: Bartholomew et al 2006
Take, for example, the following objective:
Communication plans will be developed to ensure stakeholders are kept informed.
This sentence contains the idea of what needs to be done, but responsibility for doing it and therefore an emphasis on the action needed is lost because of the passive form. Compare this with the alternative wording, using an active form and action words:
The committee will develop communication plans to ensure stakeholders are kept informed.
7 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Using a bullet list for objectives lets you start each objective with an action word, which can be a forceful way of conveying the objectives; for example:
The committee will:
• develop communication plans to ensure stakeholders are kept informed
• prepare a report on the nutrition in schools open day
• inform parents about the progress their children have made.
Are they SMART object�ves and/or outcomes?Another good approach to identifying which objectives and/or outcomes are able to be measured, or to reconfigure them so that they are measurable, is to ask yourself, are they SMART objectives (Iverson 2003)? SMART objectives/outcomes are defined in Table 3.
Table 3: Defin�ng SMART object�ves
Item Definition
Specific An objective should address a specific target or accomplishment. Specific implies that
an observable action, behaviour or achievement is described, which is also typically
linked to an identifiable change in rate, number, percentage or frequency.
Measurable A method should be established to indicate that an objective has been met. That is,
there should be a system, method or procedure for tracking and recording the change in
behaviour or action towards which the objective is directed.
Achievable* Though not necessarily easy or simple, the objective should be feasible – that is,
capable of being achieved. Objectives should be limited to what can realistically be
done with available resources, and ideally the resulting change should be ‘attributable’
to the action undertaken.
Relevant* An objective should be significant to the people involved in the programme (from
beneficiaries to the programme’s sponsoring organisation), and the objectives should
be capable of having an impact or making a change.
Time-based An objective should be achievable within a specific timeframe. Generally this takes the
form of a start and end date. The time may be short (two or three months, up to two
years), medium (three to five years), or long term (five years to seven years).
Source: Iverson A. 2003. Preparing Program Objectives: Theory and Practice. The International Development Research Centre: Evaluation Unit, Toronto
* What A and R stand for is inconsistent in the literature. A is sometimes given as ‘accurate’, ‘action-oriented’, ‘accountable’ or ‘attributable’. R is sometimes given as ‘relevant’ or ‘realistic’. In some situations, these alternative phrases may be more useful, in which case use the term that is most appropriate or that most clearly helps to describe the intent of the objective.
� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Mon�tor�ng double-barrelled object�vesDouble-barrelled objectives contain multiple components, and if well written have the advantage of being able to convey multiple objectives in one single sentence rather than using several sentences. However, they are also more complex to monitor. Using the above criteria, we need to ask the question: Are double-barrelled objectives SMART? Here are two examples of double-barrelled objectives.
• Objective 1: Enable people with chronic conditions to improve their health, slow progress of their condition(s), and maintain independence wherever possible by aligning community and hospital services across [ ] DHB.
• Objective 2: Reduce the incidence of cancer, diabetes, and cardiovascular disease by 20% over the next five years.
The first objective has multiple components in that it refers to chronic conditions, and talks about slowing the progress of the conditions, maintaining independence and aligning services. It could be argued that this objective, as written, is a good high-level objective in that it uses some action words that provide an overall general direction for the DHB, and it complies adequately with the SMART criteria in Table 3 in terms of a high level objective statement.
However, for operational and monitoring purposes it is does not adequately comply because it does not specify the chronic conditions, time frame and services. All of these issues would have to be addressed when developing the logic model and outcomes monitoring plan to achieve this high-order objective. Multiple outcome measures would have to be used to assess whether this objective was achieved, because different measures would need to be used for each component.
The second objective also has multiple components, but it clearly specifies the chronic conditions of interest, the size of change sought, and the timeframe for achieving the changes. The objective complies with the criteria in Table 3, although the objective would also require the use of multiple outcome measures – a different measure for each chronic condition – for progress on its achievement to be monitored appropriately.
In answer to the question, are double-barrelled objectives SMART? the best advice is generally to avoid double-barrelled objectives when writing low-order (ie, operational) objectives. This will make planning clearer and the objectives easier to monitor. Where double-barrelled objectives may be desirable (eg, when writing high-level objectives), make sure they comply as closely as possible with the criteria in Table 3. Also, remember that where double-barrelled objectives are used, multiple outcome measures will generally have to be developed to monitor each component of the objective statement.
Pr�or�t�s�ng object�ves/outcomes for measurement After identifying and clarifying the range of objectives and/or outcomes that could be monitored in an ideal world, it is highly likely that it will not be feasible – or necessary – to monitor all the objectives and outcomes planned for. Consequently, a process of prioritisation will have to take place to select the objectives and outcomes that are both essential and possible to monitor.
� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
We suggest two types of criterion to use as aids in making decisions about which outcomes should be and can be monitored:
• management considerations
• instrument selection considerations.
Management considerations are concerned with promoting the development of an appropriate and robust monitoring system. Instrument selection considerations are concerned with establishing scientific credibility for the monitoring system. Figure 1 illustrates how these two types of criterion relate to each other. They are also further discussed in the following pages. Although the criteria can be defined separately, in practice they are interrelated and a judgement will have to be made as to where the balance lies between the merits of each.
F�gure 1: A framework for pr�or�t�s�ng what outcomes to mon�tor
By examining each objective and indicator using the criteria set out below, you will be able to develop a prioritised selection of the outcomes and associated indicators that can and should be measured appropriately. To help with this, it may be useful to use a scorecard such as that provided in Table 4. The scorecard provides a method for rating each outcome and indicator against the criteria. Note that more than one indicator can be used to monitor an outcome, but it may not be necessary or cost effective to do so.
The scorecard is simple to use: just assign a ‘1’ score to each criterion that has been successfully met. All the scores are added up, and the outcomes and associated indicators with the highest score should receive the highest priority for inclusion in the outcomes monitoring plan. A slightly more complex approach, which would provide better differentiation between possible indicators, would be to use a scoring approach that allows for decimal points; for example, instead of scores of 1 or 2, a score of 1.5 could be allocated. In some situations a more sophisticated approach may be desirable, in which case a ‘weighted’ system could be applied to the scorecard. In a ‘weighted’ approach, some criteria would be judged as more or less important than others, and consequently a higher or lower range of possible scores could be allocated to the selected criteria. For example, because of the size and nature of the programme it may be decided that the instrument selection criteria of validity and reliability are so important that they are worth double points compared to the others (ie, the results are going to be weighted (biased) towards indicators that score well on those factors).
Management criteriafor selecting outcomesthat should bemonitored:• attribution (accountability)• centrality• cost–benefit• robustness to withstand scrutiny• timing.
Instrument selectioncriteria for selectingoutcomes that can bemonitored robustly:• availability• reliability• sensitivity• validity.
Outcomesprioritised
formonitoring
10 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Management select�on cr�ter�a for pr�or�t�s�ng what �s essent�al to mon�tor �n publ�c health programmesAttr�but�on (accountab�l�ty)Attribution is the extent to which change in the outcome of interest is associated with the type of activity undertaken. Attribution is an important issue to consider because it has implications for how well a programme’s activities can be said to have resulted in any changes observed, and for assigning the level of accountability for the success or failure of a programme or its components.
Strong attribution requires being able to establish a clear and unambiguous causal link between what you do and what happens – something that is often problematic for many public health programmes. When assigning attribution, you need to beware of any ‘attributional bias’ that results from over-attributing a change to any particular activity.
Central�tyHow important is the outcome of interest to establishing the success of the programme? The more central the outcome, the more important it is to measure it appropriately. Where data is not available, think carefully about developing a new measure, or using an appropriate ‘proxy’ measurement (see ‘Availability’ below).
Whether a programme is deemed successful or not is not always related to whether the central health outcome of interest has been achieved. A programme may fail in one aspect but still be successful if other outcomes of importance have been achieved.
Cost–benefit‘Cost–benefit’ refers to the balance between the cost of using and/or developing a measure and the benefit that will be gained from implementing it. For many programmes, you will need to make a judgement about whether the cost of developing a monitoring regime or measure is worth the benefit to be gained from the information provided. Where the cost outweighs the benefit, then consider using an appropriate proxy indicator, or information from a pre-existing monitoring regime.
In general, the proportion of the total budget spent on monitoring should be in the region of 5% to 20%, depending on the type and size of the programme. If the programme is a pilot initiative, the results of which could be used to significantly influence whether the programme is expanded into a major effort, then 20% of the total budget could be appropriate. If the programme has a large budget and is well established, applied in standard ways and supported by evidence for its effectiveness, then a budget allocation of 5% may be more appropriate for monitoring purposes.
Cost per output is probably the best method to use when calculating the cost of a monitoring programme or a component of it. For example, the cost of purchasing information from an existing data set may be cheaper than doing it yourself. Or, it may be cheaper to use a proxy measure than to use a direct measure (assuming the proxy measure is robust enough to withstand public scrutiny and the information trade-off is acceptable). Note that cost effectiveness is just one of a number of criteria that should be used in planning an outcomes monitoring regime.
11 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Robustness to w�thstand scrut�nyRobustness will be provided by ensuring that the instrument selection criteria (presented below) are observed. Be very clear about why a particular set of outcomes has been selected for monitoring and not others, and why particular measurement instruments have been selected over others. Ask yourself the following.
• What is being monitored?
• Why and how is it going to help to assess the effectiveness or efficiency of the programme as a whole?
• Why is it essential?
• When must the information be available?
• How is the data to be collected?
• Who is going to do the collection and analysis?
• Is the proposed approach cost effective?
• Will the information be provided in a timely manner?
• Will the monitoring system withstand scrutiny by stakeholders?
For public health programmes funded by government money, it is important that assessments of the effectiveness of the programme be able to withstand public scrutiny. In this context, we suggest it is usually advisable not to attribute a programme’s success to a single criterion: robustness is provided by the strength of logic of the total outcomes monitoring regime.
T�m�ngTiming has two aspects to it. The first concerns establishing ‘attribution’ (see above). In this case, timing refers to establishing the time sequence between when an activity took place and when a change in the desired outcome was observed. The second aspect relates to being able to report to stakeholders on the performance of a programme in a timely manner, which also means the proposed indicator must be able to be produced within an appropriate timeframe.
As a general rule, the timeframe for short-term objectives can be as short as two to three months or up to two years. The medium term is defined as three to five years, and a timeframe of five to seven years is usual for achieving long-term objectives (The Treasury and State Services Commission 2007a; b).
In practical terms, it may not be feasible to monitor/measure progress on attaining short-term, or even medium-term, objectives using national data. For example, information from national survey and administrative data sets is typically not available anywhere from 6 to 36 months after the data was originally collected due to data checking and quality control processes. This may mean that it is not feasible to use information from these data sources as indicators because the data is not available soon enough.
12 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Instrument select�on cr�ter�a for pr�or�t�s�ng what �s poss�ble to mon�tor appropr�ately �n publ�c health programmesAva�lab�l�tyThis refers to whether data, or a measuring instrument, already exists at the local or national level that could be used to help monitor the outcome of interest. Where possible, use existing data that may be sourced from within your organisation or a range of government agencies or organisations such as public health organisations in your area. A good place to start seeing what data is already available to help monitor your programme is at PHIOnline, www.phionline.moh.govt.nz, which includes data at national, District Health Board and Territorial Local Authority level (see Part D for an introduction to the range of information available to you on PHIOnline). Using existing information sources can save a considerable amount of time and effort.
Where such data exists, efforts should be made to use that data or instrument rather than invest in developing new data sources or instruments. If issues such as cost, complexity and/or timeliness prevent direct measurement of the outcome of interest, consider using a proxy indicator. If achieving the central outcome is vital, and data is not readily available, careful thought needs to be given to spending extra effort on developing a new measure, or on identifying a suitable proxy measure.
Rel�ab�l�tyReliability is the extent to which a measure, when used repeatedly in the same way, will produce the same or a similar result. For example, if you were interested in monitoring a person’s weight and used a weighing machine that produced a different reading each time it was used (when all other factors were the same), then the machine would not be a reliable instrument to use.
Sens�t�v�tySensitivity refers to how well a measure is able to detect when a change has occurred in the outcome being monitored. For example, a weighing machine that was able to differentiate a change in weight by 100 gram graduations is more sensitive than a machine that can only detect changes in 500 gram amounts.
Specificity is another criterion that is particularly appropriate in clinical settings, where it is used to assess how accurate a clinical test is in identifying people at risk.
Val�d�ty Validity refers to how well a measurement, index or indicator reflects the outcome it is intended to measure. For an outcome measure to be valid, it should be both sensitive and specific.
Often there is more than one way of measuring an outcome, and the most valid measure may not be practical in all situations. For example, the best way to measure body composition (particularly lean and fat mass) is under-water densitometry (weighing), but this is not practical outside small clinical studies. Therefore, measures of weight adjusted for height (eg, body mass index, BMI) are often used to reflect body composition, particularly body fat mass. Although BMI is correlated with body fat mass, this relationship varies according to body build, age and ethnicity. Therefore, BMI is not recommended for assessing excess body fat mass (obesity) at an individual level.
13 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
However, BMI can be useful for assessing obesity at a population level if used and interpreted appropriately. BMI based on direct measurement of height and weight is more valid than BMI based on self-reported height and weight, particularly if measurements are taken by trained observers using appropriate equipment and standardised procedures.
Ultimately, the choice of measure for a particular programme will depend on a number of factors, including validity and practicality.
Table 4: Checkl�st and score card for pr�or�t�s�ng outcomes and �nd�cators for mon�tor�ng
Outcome of �nterest
Ind�cator name
Cr�ter�a Total score
Pr�or�ty
At Av Ce Cb Re Rb S T V
Outcome 1
Output 1
Objective 1
Cr�ter�a abbrev�at�ons
At Attribution (Accountability) Av Availability Ce Centrality
Cb Cost–benefit Re Reliability Rb Robustness to withstand public scrutiny
S Sensitivity T Timing V Validity
Source: Adapted from Table 1 ‘How to select indicators’ (page 68) in United Nations Development Programme Evaluation Office 2002
Add�t�onal quest�ons and gu�dance for select�ng outcome measures and �nd�cators Do you need a new outcome �nd�cator or �nstrument? Think carefully before developing a new outcome indicator or instrument, ask yourself the following.
• Why is a new indicator/instrument needed?
• Is it because there is a problem with an existing indicator/instrument, or is it because it really is a new outcome?
• Can we use existing data creatively to answer the question?
Every instrument has its own strengths, weaknesses and peculiarities – the perfect instrument does not exist. If you are unfamiliar with the instrument that best suits your plans, seek help from someone who knows about it already. Public Health Intelligence at the Ministry of Health have extensive expertise in measuring public health outcomes. If in doubt, seek their advice.
14 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Match the measurement �nstrument to the object�veChoose the level of observat�onWhat is the focus of interest? Is it an individual or a group of people? If it is an individual, perhaps the measurement is of an individual’s change in knowledge and behaviour due to the intervention, or maybe it involves observing a change in their health status over time. Is the focus on groups of patients (eg, patients from a particular age group, with a specific disease, or submitted to a certain intervention), or their ability to access services? If the intervention’s utility or the general quality and cost-effectiveness of different care systems is the main interest, compare the quality of care between different systems; say, between primary and secondary care or between geographic locations.
Formulate and descr�be the measure’s a�msWhat is your aim for the measure? Do you want to describe, compare or evaluate health outcomes? The selection of your instruments is highly related to the endpoints of your project. What do you want to use the instrument for?
The principal uses for a health measure are as follows.
• A health status measure can be used as an indicator, measuring the current health condition/state of a person or a population group at a point in time. In addition to validity, both reproducibility and specificity to the chosen health condition are important. Reproducibility is particularly important when undertaking a robust pre- and post-assessment of change following the intervention.
• A health outcomes measure can be used as a comparison, relating differences at different points (eg, before and after intervention). For this type of action, sensitivity and responsiveness to change are important. Put simply, the measure must be able to register small changes in people’s health.
• A health outcomes assessment implies that, apart from being an outcome measure, it is an attempt to use the information through feedback to the users of the information, including government funders of the programme. Apart from achieving the outcomes sought, government funders may also be interested in cost-effectiveness measures.
• Process outcome mon�tor�ng can use both qualitative and quantitative measures to report on process evaluation issues. This type of information helps to provide the ‘story behind the statistics’, and is useful for reporting on the following aspects that may be critical to the monitoring and success of the programme:
– community capacity
– service/agency capacity
– compliance with good practice
– programme integrity.
Examples of these types of measure are goal attainment scaling, global assessment scales and other rubric-based scoring.
1� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Cost-effect�veness measuresCost effectiveness refers to the balance between the cost of implementing the intervention and the benefit or effectiveness gained from programme. There are many ways of measuring cost effectiveness. The three approaches suggested by (The Treasury and State Services Commission 2007a; b) as being appropriate for Crown entities are:
• cost-benefit analys�s, which is seen as the ‘gold standard’ for establishing cost effectiveness
• cost per un�t of �mpact/outcome, which is a simpler approach (assuming it is feasible to quantify the impact/outcome gained)
• cost effic�ency, which is a method by which the cost per output is measured. This approach is likely to be particularly appropriate where the service provided (eg, a public health information service) is remote from any health outcome.
Cost per output is probably the most useful method when calculating the cost of a monitoring programme or a component of it. For example, the cost of purchasing information from an existing data set may be cheaper than doing it yourself. Note that cost-effectiveness is just one of a number of criteria that should be used when planning an outcomes monitoring regime.
Dec�de on the type of �nstrumentIt is important to note that the ‘psychometric’ qualities of the instrument you choose must be able to support your goals and objectives. This means it is essential that the instruments are valid and reliable (see above).
• In general, a condition-specific measure will have a narrow focus but will contain considerable detail in the area of interest. If you are interested, say, in one disease condition, and the assessment is mainly of symptoms and function, then use a condition-specific measure.
• If a specific domain, such as daily functioning or mental wellbeing in different populations, is your interest, use a dimension-specific instrument.
• If you are interested in general health or in the interaction between different conditions, or if you are interested in populations that may include healthy people, then you should use generic instruments.
• If you think the influence of other diseases or conditions that you have not measured may influence the results of the problem or the disease of interest, combine disease-specific and generic instruments.
No one instrument will prove satisfactory for all purposes. You may need to combine instruments because a reasonable instrument does not exist. But beware: when possible, use the instruments in their original form. Do not change them or use only parts of them. Validation covers only the complete instruments that were tested (see above on reliability and validity).
Be careful with instruments that are taken from one country to be used in another. Cross-cultural validation needs to follow strict rules. Even a survey that has been validated in Canada, the UK or the US may not be valid in New Zealand because of different cultural dynamics and language uses. The formal validation of an instrument is a costly and time-consuming process. How much of this work you do depends on your resources. And don’t forget the practicalities: the necessary time to fill in questionnaires and the costs of mailing and analysis. Think about your target group: for example, not every instrument suits children or older people.
1� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
The measures selected must align with the programme’s goals and objectives. It is therefore important to understand the programme’s goals and objectives so that appropriate selections can be made from the many measurement instruments available. In particular, you will need to determine whether an intervention’s effect is to be measured or descriptively assessed.
You will need broadly validated instruments that have been used in other studies if you want the intervention to describe the health status of a defined population or a specific disease category. Short, feasible and reliable instruments are recommended if care providers are to use them in their clinical work.
Collect�ng and analys�ng �nformat�onSelecting the right measures is only part of the process of establishing a monitoring system. You will also need to think about how the information will be collected and processed. There is no point choosing the perfect set of measures if there is no feasible way to collect and use the information.
Ask the following questions.
• What is the cost of collecting the required information?
• How easy is it to access the information?
• Are there administrative, privacy and ethical issues?
• Has the cost of data analysis been allowed for?
• Who is going to do the analysis?
• Who is the audience for or user of the information?
It takes time and skill to correctly analyse and interpret data, and this needs to be allowed for when planning a monitoring regime. The amount of time to allow depends on the complexity of the analysis and the type of information collected. Ask the following types of questions.
• What type of analysis is required? That is, what type of questions have to be answered: descriptive or explanatory? (Generally, explanatory analysis will require more work than descriptive analysis to prepare.)
• Who is going to use the information?
• What decisions will be made with the findings?
• When does it need to be done by?
• Who is going to do it? Do they have the skills?
In general, where the users of the information are senior decision-makers and/or the funding for the programme is significant, the more robust the analysis needs to be in order to withstand scrutiny, and consequently the more time should be allocated to this task. For major regional and national programmes involving the analysis of multiple indicators and the preparation of complex tables and report writing, a time period of several weeks or months is not unreasonable.
17 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Part C: Issues to Cons�der �n Outcomes Mon�tor�ng
Key po�nts1. Outcomes monitoring is intended to help:
a. facilitate a process of ‘continuous programme improvement’
b. build confidence in, and support for, public health programmes.
2. Outcomes monitoring is not a cheap replacement for a traditional evaluation process.
3. Outcomes monitoring may, or may not, include a traditional formative process and impact evaluation exercise.
4. Beware of the problem of small numbers.
Outcomes mon�tor�ng as a tool for ‘cont�nuous programme �mprovement’: the problem of accountab�l�ty, attr�but�on and performance management The main motivation for introducing management tools such as programme logic models, outcomes monitoring and managing for results within the state sector is an increased expectation from central government that there will be a focus on results in the design and delivery of publicly funded services.
This expectation is highlighted by the requirements of the Public Finance Act 2004 and the Crown Entities Act 20044 (The Treasury and State Services Commission 2007a; b). This legislation means that there is now an increased emphasis on ‘maximising results for the available resources’, and managers must report on the impacts, outcomes, or objectives that a programme seeks to achieve or contribute to, and how it will contribute to implementing government policy directions (The Treasury and State Services Commission 2007a; b). Increasingly, these expectations are now being transmitted into the services contracted for by government agencies.
The reporting requirements in the acts include stating the:
• rationale for the main types of interventions planned and the results expected
• risks identified and how they will be managed
• main measures that will be used to monitor progress and performance – including the cost effectiveness of the activities
• how the programme will link to other government agency programmes to support them, including those activities undertaken by non-government organisations (The Treasury and State Services Commission 2007a; b).
4 Crown entities includes District Health Boards.
1� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Implicit in this approach is the idea of ‘accountability’ for achieving the objectives set and outcomes sought. As a general principle, the degree to which accountability can be maintained depends on the ability of the person or organisation to meaningfully influence the changes sought. Strong accountability can only be maintained where there is a clear and unambiguous causal link (ie, attribution) between cause and effect.
For many government activities, including public health programmes, establishing clear causal links between an intervention and change in the outcome sought – in this case population health – is difficult. A number of practical difficulties must be overcome (Nutbeam 1998). Consequently, when assigning attribution in public health programmes, you should beware of any ‘attributional bias’ that would result in over-attributing a change to a single programme activity.
It is likely that success will not be achieved through a single activity, but by undertaking a range of activities and working with other programmes and across-government agencies to achieve the objectives. This means that using outcomes monitoring as a strict contract performance management tool is problematic. This does not mean that non-government organisations and providers of government services should not be accountable for the choice of methodology and quality of implementation undertaken to achieve the health objectives.
The use of logic models and outcomes monitoring should be seen as a way of instituting a ‘continuous improvement cycle’ in the design and delivery of publicly funded programmes (The Treasury et al 2005), rather than as a contract performance management tool to be feared. With the continuous improvement approach, the performance of organisations is focused on their ‘understanding, reviewing, and learning from the efficiency and effectiveness of their operations’ (The Treasury et al 2005). The approach provides a tool for helping programmes to evolve over time in response to changes in the environment and as programme objectives are achieved.
This approach is a recognition that being able to attribute outcome changes to particular interventions or outputs ‘won’t always be feasible’ (The Treasury and State Services Commission 2007b). This does not absolve service providers from any responsibility and accountability for designing, implementing, monitoring and reporting on the effectiveness and efficiency of their programmes. It does mean that where establishing attribution is difficult, it is even more important that a clear logic (rationale) for the intervention be stated, and that where a number of similar but small programmes are being implemented, all the programmes use the same reporting framework and attribution takes place at the group level rather than at the individual programme provider level.
In terms of contract performance management, providers should be reassured that the approach recommended here includes considering the full range of reasons for meeting or not meeting a target. The important thing is that there is a robust programme planning and monitoring plan that will enable us to answer questions such as:
• What impact have our interventions had?
• What else was going on at the same time that also had an impact?
• Can we learn from our performance monitoring, and how can we improve in response to the lessons?
1� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Outcomes mon�tor�ng versus evaluat�on Along with the new emphasis on using logic models for planning and monitoring for health outcomes there has been an implicit shift towards a monitoring process that relies on quantitative methods using epidemiological and biostatistical methods, rather than traditional formative and process evaluations. This shift in emphasis should not be seen as something particularly new, but rather as a reflecting the ongoing evolution in public management thinking and practice. In this case, the use of logic models can be seen as an evolving from ‘Managing by Objectives’ to ‘Managing for Outcomes’. Similarly, ‘outcomes monitoring’ reflects an evolution in the practice of ‘evaluation’.
• Traditional evaluation of public health programmes tended to focus on qualitative methods and formative and process questions, although current practice is to use multiple methods.
• Monitoring for outcomes tends to use quantitative methodologies and focuses on endpoints.
However, depending on what the programme is supposed to achieve, formative and process evaluation may still form a legitimate part of an outcome monitoring framework. For example, where community acceptance of a new type of intervention is critical to achieving the ultimate health outcome sought, then a formative and process evaluation would be appropriate.
A full programme evaluation that includes formative, process and outcomes evaluation is likely to be appropriate for major national-level campaigns. Such evaluation examines long-term changes in health status and the determinants of health. These include changes in knowledge, awareness and behaviour; shifts in social, economic and environmental conditions; as well as changes to public policy and health infrastructure.
Outcome or impact evaluation also seeks to measure the reduction in health status inequities between population subgroups. In this approach, it is important to identify and measure short-, medium- and long-term outcomes to ensure the ongoing support and relevance of the activity for those whose agendas are shorter term. Outcomes evaluation also uses indicators as benchmarks, or proxy measures, to assess the extent to which objectives have been met. Matching objectives to associated indicators in a logic model helps to ensure the availability of relevant data sources for programme evaluation.
Note that a monitoring regime using quantitative methodologies should not be seen as a cheap replacement for evaluation. Both approaches require skilled staff to advise on the most robust methods and measures to use and to undertake the data analysis, and both rely on robust administrative systems.
The problem of small numbersMonitoring the outcome of an intervention requires a health indicator that can, among other things:
• detect the outcome sought from the intervention
• be used to provide a baseline measure against which future measurements, which commence after the intervention has started, are compared
• consistently be used over the time period of interest.
20 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
The first is usually the hardest to determine, but statistical power analysis calculations and the explanations that form the logic behind the use of the indicator can help decide whether you should even begin a monitoring process using the indicator. The second is usually easy to determine, because it simply requires data before and after the intervention. The third is more difficult than it appears because it can be easy to introduce changes in the way the data is collected over time. Assessing whether the indicator has been consistently used requires someone with a good knowledge of the process.
A good health status indicator measures either:
• prevalence (the number people with the condition in the population, divided by the total number of people in the population), which can be expressed as a ratio; or
• �nc�dence (the number of new cases of the condition in the population divided by the total number of people in the population) of a disease, condition or risk factor, which can be expressed as a ratio. Generally, incidence is for a specified period of follow-up of the population of concern (eg, one year).
In both measures, the numerator or population from which one identifies the cases must match the denominator population.
The problem with these types of indicators in New Zealand is that either or both the numerator (the number of people with the health status of interest) or the denominator (the number of people in the population as a whole) may be very small. When the numbers are small, then statistically it can be difficult to measure the prevalence or incidence, and it also becomes very difficult to detect whether a change has occurred, let alone to decide whether any detectable change is related to any effect of the intervention.
Example 1: Illustrat�ng mortal�ty �nc�dence: New Zealand’s crude mortal�ty rate �n 200�In 2006, 28,390 deaths were registered in New Zealand and, as of 30 June 2006, 4,127,000 people were estimated to be living in New Zealand. So the crude mortality rate is 28,390 divided by 4,127,000, which equals 0.0069. This translates to 1 death per 145 people, or 688 deaths per 100,000 people in the denominator population.
This is an incidence rate. Mortality rates are always incidence rates due to the fact that there is no prevalent pool, because the cases are all deaths in New Zealand and so they match the same population as the denominator.
Detect�ng change �n small numbersThe ability of an indicator to detect the effect of an intervention depends on a combination of factors, including:
1. sens�t�v�ty (the ability to pick up accurately a change in the outcome measured)
2. spec�fic�ty (how accurate a test is at identifying people at risk – this is particularly important in clinical testing situations)
3. power (the probability of detecting an effect, or degree of change, of a specified size).
21 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Determining the sensitivity and specificity of an indicator is often complicated and requires specialist knowledge beyond the scope of this How to. The power of your indicator to detect changes in the outcome of interest depends on the actual size of the effect (degree of change) taking place, or that is expected to take place based on prior experience or the literature. The larger the change, the easier it is to detect.
The power to detect also depends on the background prevalence or incidence of the indicator of interest. Finally, the size of the population of interest and the variability in the indicator go hand in hand to determine how likely it is your indicator will detect the intervention.
Following are two examples showing how the problem of small numbers in the New Zealand population inhibits our ability to statistically identify change in the health status of the population due to the effects of an intervention.
Example 2: Power to detect a change �n the number of v�olent offences �n a small commun�tyA small community has implemented a programme to reduce interpersonal violence. This community collects data on violent offences reported to the local police from one year before the programme was initiated. The community has a population of 1000 people and there are 25 violent offences reported to the police in the year before the programme. Overseas evidence from similar programmes indicates a reduction in reported offences of 15%. This translates to an annual incidence of 25/1000 or 0.025 offences per person per year. A reduction of 15% would be just under four fewer cases a year in this community.
Under this scenario, and if the reduction is actually 15%, the power of this indicator to statistically detect the reduction in the following year would be 2.7%. That means if we were to run the programme for another 37 years we would on average detect the difference once in that time. We would conclude that an indicator with higher incidence would be needed to monitor the change made due to the programme.
Figure 2 illustrates this scenario. It simulates a 10-year period for our small community: five years before and five years after the violence prevention intervention was initiated at the beginning of year six. The pattern highlights the usual statistical variation that one could expect to see. Note that in this example, because of the small numbers, it would be very hard to argue statistically that the programme has been successful – even though it might have been.
22 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
F�gure 2: S�mulated effect of 1�% reduct�on �n v�olent offences
Source: Craig Wright, Public Health Intelligence
Example 3: Power to detect the effects of a health �ntervent�onThe larger the population, or the larger the effect change that we want to detect, the easier it is to detect a change. The interaction between the size of the underlying incidence of the health issue, a desire to detect a 10% change, and population size is illustrated in Figure 3. Here, the three lines indicate the statistical power to detect a 10% change in three populations comprising 1000 people, 10,000 people, and 100,000 people, for a health problem that has an incidence of 1 to 20% in the population.
It can be seen that the population of size 1000 never has a power over 20% for the whole incidence range of 0–20%, making it unlikely to detect interventions in a population of this size. Population 10,000 reaches a reasonable level of power (ie, 80%) at about 9% incidence, so it would be unlikely that one would want to use indicators with under a 10% incidence in communities of size 10,000 when expecting a reduction of 10%. Finally, indicators in the large 100,000 size population are likely to detect a 10% decrease in the indicator above 2% incidence (with 80% confidence).
Number of offences reported
InterventionPeriod
35
30
25
20
15
10
5
01 2 3 4 5 6 7 8 9 10
Year
23 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Power100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10% 11% 12% 13% 14% 15% 16% 17% 18% 19% 20%
Incidence
Population100010,000100,000
F�gure 3: Power to detect change due to an �ntervent�on, for spec�fied �nc�dence, populat�on s�ze and 10% �nc�dence reduct�on
Source: Craig Wright, Public Health Intelligence
24 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Part D: Ava�lable Data and Informat�on from Publ�c Health Intell�gence
Key po�nts1. Public Health Intelligence has data you can use:
a. New Zealand Health Monitor Survey programme population survey data
b. Administrative data, which includes hospital records, disease registrations, and mortality data.
2. Go to PHIOnline (www.phionline.moh.govt.nz) to access data.
Public Health Intelligence (PHI) has two main types of data you can utilise as indicators for outcomes monitoring: survey data collected from population surveys, and administrative data, which includes hospital records, disease registrations, and mortality data.
Survey dataPHI is responsible for the New Zealand Health Monitor (NZHM) survey programme, which is a population-based, integrated, ongoing survey programme. Data is collected for the following three major health information domains:
• health outcomes (health status, disease states)
• causes of these outcomes (social and environmental determinants, risk and protective factors)
• health services (access, utilisation, need, coverage, quality, responsiveness, cost).
There are two main types of vehicles for collecting information in the NZHM programme: serial cross-sectional surveys and serial cohort studies. Table 5 presents a summary of the NZHM surveys.
2� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Table �: Summary of New Zealand Health Mon�tor surveys, 2002–2012
Survey Top�c / data areasFrame (target populat�on)
Sample Mode Frequency
New Zealand Health Survey
Chronic diseases, biological and behavioural risk factors, reported health status, health service utilisation, sociodemographics.
All New Zealanders
12,500 adults (15+ years) and 5000 children (birth to 14 years)
Face-to-face, computer-assisted (CAPI) questionnaire plus anthropometric measurements in respondent’s home.
Previously 1992/93, 1996/97, and 2002/03. Now every three years (2006/07, 2009/10 etc).
Te Rau Hinengaro New Zealand Mental Health Survey
Prevalence, severity, impairment and treatment of major mental health disorders.
New Zealand adults (16+ years)
Approx. 13,000
Face-to-face CAPI questionnaire in respondent’s home.
Approximately every 10 years. Previously 2004, next planned for 2014.
New Zealand Nutrition Surveys
Food and nutrient intake, factors influencing dietary intake, nutritional status and nutrition-related status.
New Zealand adults (15+ years) or New Zealand children (5–14 years)
Approx. 4000– 5000
24-hour dietary recall and food frequency questionnaire, self-administered questionnaire, plus examination, in respondent’s home or at school for children.
Every five years, alternating between adult and child (adult 1997 next 2007/08; child 2002, next 2012).
New Zealand Oral Health Survey
Oral health status, oral health beliefs, attitudes, knowledge and practices.
All New Zealanders
Approx. 6000 to 8000
Face-to-face CAPI questionnaire and oral examination.
Every 10 years from 2008.
New Zealand Tobacco Use Survey
Tobacco use and the psychosocial correlates of smoking behaviours. Prevalence and consumption data available from the NZ Health Survey in 3rd year.
New Zealand adults (15 to 64 years)
Approx. 4000 to 6000
Face-to-face CAPI questionnaire in respondent’s home.
Two out of every three years (2005, 2006, 2008, 2009, etc).
New Zealand Alcohol and Drug Use Survey
Alcohol and illicit drug use, and the behaviours associated with alcohol and drug use.
New Zealand adults (16 to 64 years)
Approx. 8000
Face-to-face CAPI questionnaire with audio-assisted self-complete section (A-CASI) in respondent’s home.
Every two years from 2007.
Source: Ministry of Health 2005.
2� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
In addition to the cross-sectional surveys listed above, PHI is also involved in serial cohort studies (often referred to as record linkage studies). These include the New Zealand Census – Mortality Study and Cancer Trends.
Descriptive reports on the results of the above surveys can be found on PHI’s website. Access to the unit record survey data sets is available on application; see: www.moh.govt.nz/phi/surveys.
Adm�n�strat�ve dataPHI often utilises administrative data collected by the New Zealand Health Information Service (NZHIS). These data sources include hospital discharges, cancer registrations and mortality data. Table 6 describes these data sources in more detail.
Table �: Summary of adm�n�strat�ve data sources
Short name of data source
Descr�pt�on/keywordsSource of data
Per�od covered Delay
Mortality data Mortality data from the Mortality Data Collection
NZHIS 1970–2003 3 years
Foetal mortality Foetal and infant mortality data from the Mortality Data Collection
NZHIS 1988–2003 3 years
Cancer registrations All cancers NZHIS 1950–2004 3 years
Priority sites (lung, female breast, cervix, prostate and colorectal)
NZHIS 1950–2004 2 years
Notifiable diseases Notifiable diseases from Environmental Science and Research’s (ESR’s) schedule
ESR 1997–2006 18 months
Sexually transmitted infections
Sexually transmitted infections from sexual health clinics
ESR 1997–2003 18 months
Public hospital discharges
Public hospital discharge from the National Minimum Data Set (NMDS)
NZHIS via DHBSPF
1971–2006 6 months
Filtered public hospital discharges
Filtered public hospital discharge from the NMDS
NZHIS via DHBSPF
1988–2006 6 months
Private hospital discharges
Private hospital discharge from the NMDS
NZHIS 1980–1995, 2001–2003
Irregularly
Birth registrations Live and still birth registrations from the BDM registrations
BDM via SNZ
1980–2006 18 months
Mental health services Mental health services from the Mental Health Information National Collection (MHINC)
NZHIS 2001–2006 6 months
Notes: DHBSPF = District Health Board Service Provision Framework; BDM = Births, Deaths and Marriages; SNZ = Statistics New Zealand.
27 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
All of the above data sources are:
• in SAS version 8 file format
• at unit record level
• licensed to Public Health Intelligence, Ministry of Health.
PHIOnl�ne – www.ph�onl�ne.moh.govt.nzNational administrative data and survey data can often be disaggregated to the regional level, which may be more useful for your monitoring purposes. The main access portal for this information is PHIOnline. It is a powerful visualisation tool and provides an alternative way to access health information through a mapping interface rather than traditional text-based documents. Data is displayed for different DHB as well as at the Territorial Local Authority level, allowing you to visualise variability within your DHB.
F�gure 4: The PHIOnl�ne home page
The home page provides information for users on accessing the visualisation tool. Each visualisation is thought of as an ‘atlas’, and the information has been grouped in accordance to the health priorities found in the New Zealand Health Strategy. To view the atlases, just click one of the links listed on the left-hand side of the home page shown in Figure 4.
PHIOnline incorporates a number of formats to view health information on the one web page. The interface consists of an interactive map (or maps) with linked tables, charts and graphs. The standard view of the website shows the one map as the centrepiece of the interface (Figure 5).
2� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
The map is thematically drawn in relation to the range and classification of the particular data set.
PHIOnline offers a number of classifications, which the user can experiment with. To complement the map, a table of the data is also displayed, as well as a time series of the data set. You can rank the data in the table by value, highest-to-lowest or lowest-to-highest, as well as by alphabetical order. Data is displayed by region as well as providing a national figure. You can download all data displayed on the site. The site also contains extensive metadata for each data set.
F�gure �: PHIOnl�ne s�ngle map �nterface
Another interface allows you to compare two different health data sets, displaying two maps on the one web page (see Figure 6). A scatter plot for both data sets is also displayed.
In addition to survey and administrative data, PHIOnline contains data on:
• oral health
• Get Checked diabetes
• water quality
• elected services
• problem gambling.
2� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
F�gure �: PHIOnl�ne double map �nterface
30 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
GlossaryTe
rmD
efin�
t�on
Expl
anat
ory
com
men
t and
exa
mpl
e
Attr
�but
�on
Att
ribu
tion
refe
rs to
the
exte
nt to
whi
ch c
hang
e in
the
outc
ome
of
inte
rest
is a
ssoc
iate
d w
ith
the
type
of a
ctiv
ity
unde
rtak
en. S
tron
g at
trib
utio
n re
quir
es b
eing
abl
e to
est
ablis
h a
clea
r and
una
mbi
guou
s ca
usal
link
bet
wee
n ca
use
and
effe
ct. T
his
is o
ften
diffi
cult
for m
any
publ
ic h
ealt
h pr
ogra
mm
es.
Test
ing
for a
ttri
buti
on in
volv
es id
enti
fyin
g a
caus
al th
eory
and
th
en c
onfir
min
g th
e ca
usal
mec
hani
sm, u
sing
at l
east
thre
e of
the
follo
win
g w
ays.
Che
ck:
• w
ith
othe
r obs
erve
rs –
did
they
not
ice
the
sam
e th
ings
for t
he
sam
e hy
poth
esis
ed re
ason
s?
• th
at p
rogr
amm
e re
sult
s re
late
to th
e pr
ogra
mm
e co
nten
t
• th
at c
lient
impa
ct (r
esul
ts) o
ccur
red
clos
e in
tim
e to
del
iver
y of
th
e ac
tivi
ty
• th
at th
ere
are
no o
ther
suf
ficie
ntly
pla
usib
le e
xpla
nati
ons
• fo
r a p
atte
rn o
f tel
ltale
out
com
es s
ugge
stin
g on
e ca
use
or
anot
her
• th
at in
crea
sed
dose
(mor
e ac
tivi
ty) l
eads
to in
crea
sed
effe
ct (o
r gr
eate
r res
ults
)
• by
usi
ng a
com
paris
on g
roup
.
(For
mor
e, s
ee D
avid
son
2005
: 70–
84.)
Att
ribu
tion
is a
n im
port
ant i
ssue
to c
onsi
der b
ecau
se it
has
im
plic
atio
ns fo
r how
wel
l a p
rogr
amm
e’s
acti
viti
es c
an b
e sa
id
to h
ave
resu
lted
in a
ny c
hang
es o
bser
ved,
and
for a
ssig
ning
the
leve
l of a
ccou
ntab
ility
for t
he s
ucce
ss o
r fai
lure
of a
pro
gram
me
or
its
com
pone
nts.
Whe
n as
sign
ing
attr
ibut
ion,
bew
are
of a
ny ‘a
ttri
buti
onal
bia
s’ th
at
wou
ld re
sult
in o
ver-
attr
ibut
ing
a ch
ange
to a
sin
gle
prog
ram
me
acti
vity
.
For m
any
publ
ic h
ealt
h pr
ogra
mm
es, e
stab
lishi
ng c
lear
cau
sal
links
bet
wee
n an
inte
rven
tion
and
cha
nge
in a
hea
lth
outc
ome
is
diffi
cult
, and
ther
e ar
e a
num
ber o
f pra
ctic
al d
ifficu
ltie
s th
at m
ust
be o
verc
ome.
Whe
re e
stab
lishi
ng a
ttri
buti
on is
pro
blem
atic
, it
is im
port
ant t
hat a
cle
ar lo
gic
(rat
iona
le) f
or th
e in
terv
enti
on
be s
tate
d, a
nd th
at w
here
a n
umbe
r of s
imila
r pro
gram
mes
are
be
ing
impl
emen
ted
all t
he p
rogr
amm
es u
se th
e sa
me
repo
rtin
g fr
amew
ork,
and
att
ribu
tion
take
s pl
ace
at th
e gr
oup
leve
l rat
her
than
at t
he in
divi
dual
pro
gram
me
prov
ider
leve
l.
Exam
ple
1. I
f a s
ervi
ce p
rovi
der i
s co
ntra
cted
to p
rovi
de a
n in
form
atio
n se
rvic
e, th
en in
form
atio
n on
out
puts
(eg,
num
ber o
f mat
eria
ls
dist
ribu
ted,
num
ber o
f enq
uiri
es re
spon
ded
to in
a ti
me
fram
e,
leng
th o
f tim
e to
resp
ond
to re
ques
ts) i
s ap
prop
riat
e. T
his
sort
of
info
rmat
ion
will
hel
p to
pro
vide
cos
t effi
cien
cy in
form
atio
n th
at w
ill g
ive
info
rmat
ion
abou
t the
leve
l of p
erfo
rman
ce
prov
ided
by
the
serv
ice.
How
ever
, the
just
ifica
tion
for t
his
type
of i
nter
vent
ion
relie
s on
the
stre
ngth
of t
he lo
gic
for t
he
supp
ly o
f the
ser
vice
.
31 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
2. I
f a s
ervi
ce p
rovi
der i
s co
ntra
cted
to u
nder
take
wor
kfor
ce
deve
lopm
ent,
a tr
aini
ng e
valu
atio
n m
ay in
clud
e fo
llow
-up
surv
eys
to fi
nd o
ut th
e ex
tent
to w
hich
trai
ning
con
tent
has
be
en u
sed
to re
spon
d to
and
/or r
esol
ve a
clie
nt’s
nee
ds.
3. W
here
cha
nge
in th
e he
alth
out
com
e of
inte
rest
is h
ard
to
iden
tify
(eg,
sui
cide
pre
vent
ion
in a
sm
all c
omm
unit
y), t
hen
the
effe
ctiv
enes
s of
the
inte
rven
tion
mig
ht b
e as
sess
ed b
y gr
oupi
ng s
tand
ardi
sed
data
from
all
the
prog
ram
me
prov
ider
s in
to o
ne c
olle
ctiv
e re
sult
for t
he ty
pe o
f int
erve
ntio
n as
a
who
le.
Acco
unta
b�l�t
yA
ccou
ntab
ility
refe
rs to
hol
ding
a p
erso
n or
an
orga
nisa
tion
re
spon
sibl
e fo
r obt
aini
ng th
e ob
ject
ive
set t
hrou
gh im
plem
enta
tion
of
the
inte
rven
tion
.
Bala
nced
sco
reca
rdTh
e ba
lanc
ed s
core
car
d m
odel
exp
licit
ly li
nks
orga
nisa
tion
al
stra
tegy
, org
anis
atio
nal c
apac
ity,
ser
vice
del
iver
y, s
ervi
ce
deve
lopm
ent a
nd c
lient
out
com
es. I
t is
char
acte
rised
by
feed
back
lo
ops
usin
g le
ad (m
easu
ring
eff
ort)
and
lag
(mea
suri
ng re
sult
s)
indi
cato
rs.
Caus
al o
r ex
plan
ator
y th
eory
Goo
d pr
ogra
mm
e im
plem
enta
tion
relie
s on
a fe
w c
riti
cal p
rinc
iple
s,
fore
mos
t of w
hich
is w
ell-f
ound
ed a
nd c
lear
ly s
tate
d ca
usal
(o
r exp
lana
tory
) the
ory.
Thi
s in
volv
es d
efini
ng a
cor
e pr
oble
m,
iden
tify
ing
fact
ors
that
are
pro
ven
to (o
r pot
entia
lly) c
ontr
ibut
e to
th
e pr
oble
m, a
nd id
enti
fyin
g th
e do
wns
trea
m e
ffec
ts th
is p
robl
em
crea
tes.
Cle
ar a
nd re
leva
nt p
rogr
amm
e ob
ject
ives
(see
Obj
ect�v
es)
seek
to c
hang
e ke
y co
ntri
buti
ng fa
ctor
s. S
uita
ble
indi
cato
rs (s
ee
Ind�
cato
rs) c
an th
en b
e de
velo
ped
that
che
ck w
heth
er a
ny c
hang
es
a pr
ogra
mm
e m
akes
to c
ontr
ibut
ing
fact
ors
actu
ally
pro
duce
a
redu
ctio
n in
pro
blem
eff
ects
.
In th
e fa
mily
vio
lenc
e ar
ea o
ne p
robl
em m
ight
be
defin
ed a
s ch
ildre
n’s
expo
sure
to p
aren
tal f
amily
vio
lenc
e. C
ontr
ibut
ing
fact
ors
can
rang
e ac
ross
:
• lim
ited
care
-giv
er a
war
enes
s ab
out t
he n
egat
ive
shor
t- a
nd
long
-ter
m e
ffec
ts o
f fam
ily v
iole
nce
on c
hild
ren
• ge
nera
l vol
atili
ty in
one
or b
oth
pare
nts
• in
suffi
cien
t rel
atio
nshi
p an
d pr
oble
m-s
olvi
ng s
kills
wit
h on
e or
bo
th c
areg
iver
s
32 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
• su
bsta
nce
abus
e an
d/or
men
tal i
llnes
s
• en
viro
nmen
tal s
tres
sors
suc
h as
une
mpl
oym
ent/
low
inco
me,
ov
ercr
owde
d ho
usin
g, s
ocia
l iso
lati
on, a
larg
er fa
mily
, ch
ildre
n’s
beha
viou
r pro
blem
s, la
ck o
f soc
ial s
uppo
rt s
ervi
ces
Prob
lem
eff
ects
may
incl
ude:
• ch
ildre
n’s
safe
ty (p
hysi
cal i
njur
y, m
enta
l hea
lth)
• ch
ildre
n’s
beha
viou
r (fr
etti
ng, a
ctin
g ou
t, a
ggre
ssio
n)
• ch
ildre
n’s
deve
lopm
ent (
enga
gem
ent w
ith
scho
ol, p
osit
ive
soci
alis
atio
n).
Usi
ng a
‘pro
blem
tree
’ app
roac
h al
low
s ap
prop
riat
e so
luti
ons
to
be id
enti
fied
and
a th
eory
of c
hang
e or
act
ion
to b
e fo
rmul
ated
(s
ee T
heor
y of
cha
nge/
act�o
n).
For m
ore,
see
:
Prob
lem
Tre
e An
alys
is (O
DI)
htt
p://
tiny
url.c
om/2
eg9g
j
Prob
lem
, Alte
rnat
ives
, Obj
ectiv
es T
ree
(Par
tici
pati
on)
http
://t
inyu
rl.c
om/2
f551
s
Cost
-eff
ect�v
enes
s
Cost
-ben
efit
Cost
per
un�
t of
�mpa
ct/o
utco
me/
outp
utCo
st-e
ffic�
ency
Cost
eff
ecti
vene
ss is
the
bala
nce
betw
een
the
cost
of d
evel
opin
g a
mea
sure
or i
ntro
duci
ng a
new
mon
itori
ng re
gim
e, a
nd th
e be
nefit
th
at w
ill b
e ga
ined
from
impl
emen
ting
it.
Ther
e ar
e m
any
way
s of
m
easu
ring
cos
t eff
ecti
vene
ss. T
he th
ree
appr
oach
es s
ugge
sted
by
(The
Tre
asur
y an
d S
tate
Ser
vice
s Co
mm
issi
on 2
007a
; b) a
s ap
prop
riat
e fo
r Cro
wn
agen
cies
and
ent
itie
s ar
e:
• co
st-b
enefi
t ana
lysi
s, w
hich
is s
een
as th
e ‘g
old
stan
dard
’ for
es
tabl
ishi
ng c
ost e
ffec
tive
ness
• co
st p
er u
nit o
f im
pact
/out
com
e, w
hich
is a
sim
pler
app
roac
h to
use
(ass
umin
g it
is fe
asib
le to
qua
ntif
y th
e im
pact
/ o
utco
me
gain
ed)
For m
any
prog
ram
mes
, a ju
dgem
ent w
ill h
ave
to b
e m
ade
as to
w
heth
er th
e co
st o
f dev
elop
ing
a m
onito
ring
regi
me
or m
easu
re
is w
orth
the
bene
fit to
be
gain
ed fr
om th
e us
e of
the
info
rmat
ion
prov
ided
. Whe
re th
e co
st o
utw
eigh
s th
e be
nefit
, con
side
r usi
ng
an a
ppro
pria
te p
roxy
indi
cato
r, o
r whe
ther
info
rmat
ion
from
an
alre
ady
exis
ting
mon
itori
ng re
gim
e co
uld
be u
sed.
In g
ener
al, t
he p
ropo
rtio
n of
the
tota
l bud
get s
pent
on
mon
itori
ng
shou
ld b
e in
the
regi
on o
f 5%
to 2
0%, d
epen
ding
on
the
type
and
si
ze o
f the
pro
gram
me.
Whe
re th
e pr
ogra
mm
e is
a p
ilot i
nitia
tive
, th
e re
sult
s of
whi
ch c
ould
be
used
to s
igni
fican
tly
influ
ence
w
heth
er th
e pr
ogra
mm
e is
exp
ande
d in
to a
maj
or e
ffor
t, th
en
20%
of t
he to
tal b
udge
t cou
ld b
e ap
prop
riat
e.
33 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
• co
st-e
ffici
ency
, whi
ch is
a m
etho
d by
whi
ch th
e co
st p
er o
utpu
t is
mea
sure
d. T
his
appr
oach
is li
kely
to b
e pa
rtic
ular
ly a
ppro
pria
te
whe
re th
e se
rvic
e pr
ovid
ed (e
g, a
pub
lic h
ealt
h in
form
atio
n se
rvic
e) is
rem
ote
from
any
hea
lth
outc
ome.
How
ever
, if t
he p
rogr
amm
e ha
s a
larg
e bu
dget
and
the
prog
ram
me
is w
ell e
stab
lishe
d, a
pplie
d in
sta
ndar
d w
ays
and
supp
orte
d by
evi
denc
e fo
r its
eff
ecti
vene
ss, t
hen
a bu
dget
al
loca
tion
of 5
% m
ay b
e ap
prop
riat
e fo
r mon
itori
ng p
urpo
ses.
Cost
per
out
put i
s pr
obab
ly th
e be
st m
etho
d to
use
whe
n ca
lcul
atin
g th
e co
st o
f a m
onito
ring
pro
gram
me
or a
com
pone
nt
of it
. Fo
r exa
mpl
e, th
e co
st o
f pur
chas
ing
info
rmat
ion
from
an
exis
ting
dat
a se
t may
be
chea
per t
han
doin
g it
your
self.
Or,
it m
ay
be c
heap
er to
use
a p
roxy
mea
sure
than
to u
se a
dir
ect m
easu
re
– a
ssum
ing
the
prox
y m
easu
re is
robu
st e
noug
h to
wit
hsta
nd
publ
ic s
crut
iny
and
the
info
rmat
ion
trad
e-of
f is
acce
ptab
le. N
ote
that
cos
t eff
ecti
vene
ss is
just
one
of a
num
ber o
f cri
teri
a th
at
shou
ld b
e us
ed in
pla
nnin
g an
out
com
es m
onito
ring
regi
me.
Dat
a
D
ata
sets
Dat
a is
fact
ual i
nfor
mat
ion
that
aft
er a
naly
sis
is u
sed
to re
ason
or
mak
e de
cisi
ons.
The
wor
d is
com
mon
ly u
sed
to re
fer t
o ei
ther
sin
gle
or m
ulti
ple
piec
es o
f inf
orm
atio
n, a
nd in
tech
nica
l rep
orts
is o
ften
us
ed in
the
plur
al fo
rm (a
s in
‘dat
a ar
e co
llect
ed’)
.
Dat
a se
ts re
fers
to c
olle
ctio
ns o
f pie
ces
of in
form
atio
n th
at a
re
colle
cted
for p
arti
cula
r pur
pose
s. ‘D
ata
sets
’ is
syno
nym
ous
wit
h da
taba
ses.
For e
xam
ple,
the
Phar
mho
use
data
set
col
lect
s pi
eces
of
info
rmat
ion
abou
t the
pre
scri
bing
of d
rugs
sub
sidi
sed
in N
ew
Zeal
and
by P
harm
ac.
The
New
Zea
land
Hea
lth
Info
rmat
ion
Ser
vice
is re
spon
sibl
e fo
r the
co
llect
ion
and
mai
nten
ance
of m
any
of th
e la
rge
nati
onal
hea
lth-
rela
ted
data
set
s, s
uch
as m
orta
lity
and
mor
bidi
ty d
ata.
Oth
er
gove
rnm
ent a
genc
ies
are
resp
onsi
ble
for o
ther
dat
a se
ts th
at
can
be u
sefu
l for
mon
itori
ng p
ublic
hea
lth
prog
ram
mes
. A la
rge
num
ber o
f the
se d
ata
sets
are
hos
ted
on th
e PH
IOnl
ine
web
site
(w
ww
.phi
onlin
e.m
oh.g
ovt.
nz).
34 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
Eval
uat�o
n
F
orm
at�v
e
P
roce
ss
I
mpa
ct
Eval
uati
on is
a re
sear
ch p
roce
ss th
at a
ims
to a
sses
s sy
stem
atic
ally
an
d ob
ject
ivel
y th
e re
leva
nce,
per
form
ance
and
suc
cess
of o
ngoi
ng
or c
ompl
eted
pro
gram
mes
. Ev
alua
tion
aim
s to
pro
vide
info
rmat
ion
abou
t whe
ther
und
erly
ing
assu
mpt
ions
or t
heor
ies
used
in th
e pr
ogra
mm
e de
velo
pmen
t wer
e va
lid, w
hat w
orke
d an
d w
hat d
id n
ot
wor
k, a
nd w
hy. T
ypic
ally
eva
luat
ion
aim
s to
det
erm
ine
the
rele
vanc
e,
effic
ienc
y, e
ffec
tive
ness
, im
pact
and
sus
tain
abili
ty o
f a p
rogr
amm
e.
Eval
uati
on s
houl
d al
low
for l
esso
ns le
arnt
to b
e tr
ansf
erre
d in
to
othe
r pro
gram
mes
.
A c
ompr
ehen
sive
eva
luat
ion
prog
ram
me
usua
lly in
volv
es th
ree
part
s:
• fo
rmat
ive
eval
uati
on
• pr
oces
s ev
alua
tion
• im
pact
/out
com
e ev
alua
tion
.
Form
ativ
e ev
alua
tion
focu
ses
on th
e es
tabl
ishm
ent o
f the
pr
ogra
mm
e. I
t is
part
icul
arly
use
ful w
hen
the
prog
ram
me
is
new
, unt
este
d an
d re
quir
es c
omm
unit
y ac
cept
ance
if it
is to
be
suc
cess
ful.
It se
eks
to u
nder
stan
d w
hat t
he c
ondi
tion
s fo
r pr
ogra
mm
e su
cces
s ar
e su
ppos
ed to
be.
Proc
ess
eval
uatio
n fo
cuse
s on
how
wel
l the
pro
gram
me
func
tion
s.
Are
the
cond
itio
ns fo
r suc
cess
pre
sent
in th
e w
ay th
e pr
ogra
mm
e is
be
ing
deliv
ered
?
Impa
ct /
outc
ome
eval
uatio
n as
sess
es h
ow s
ucce
ssfu
l the
pr
ogra
mm
e ha
s be
en in
del
iver
ing
the
desi
red
outc
omes
. Im
pact
ev
alua
tion
sho
uld
alw
ays
try
to id
enti
fy u
nint
ende
d be
nefit
s an
d an
y ne
gati
ve p
rogr
amm
e co
nseq
uenc
es.
3� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
Oth
er te
rms
that
may
be
asso
ciat
ed w
ith
eval
uati
on in
clud
e:
• ou
tput
• ou
tcom
e
• im
pact
.
Ev�d
ence
d-ba
sed
publ
�c h
ealth
(E
BPH
)
Defi
niti
ons
of w
hat ‘
evid
ence
d ba
sed
publ
ic h
ealt
h’ m
eans
are
ev
olvi
ng (K
ohat
su e
t al 2
004)
. In
the
cont
ext o
f the
Gui
de to
D
evel
opin
g Pu
blic
Hea
lth P
rogr
amm
es: A
gen
eric
pro
gram
me
logi
c m
odel
(Min
istr
y of
Hea
lth
2006
) and
this
How
to, t
he d
efini
tion
by
Bro
wns
on (2
003)
is p
refe
rred
:
EB
PH is
the
deve
lopm
ent,
impl
emen
tati
on, a
nd e
valu
atio
n of
eff
ecti
ve p
rogr
ams
and
polic
ies
in p
ublic
hea
lth
thro
ugh
appl
icat
ion
of p
rinc
iple
s of
sci
enti
fic re
ason
ing,
incl
udin
g sy
stem
atic
use
s of
dat
a an
d in
form
atio
n sy
stem
s an
d ap
prop
riat
e us
e of
pro
gram
pla
nnin
g m
odel
s.
See
als
o Ko
hats
u (2
004)
and
Jeni
cek
(199
7) fo
r alte
rnat
ive
form
ulat
ions
defi
ning
EB
PH. K
ohat
su’s
(200
4) fo
rmul
atio
n is
ap
plic
able
to th
e po
licy
deve
lopm
ent p
roce
ss, w
hile
Jeni
cek’
s (1
997)
defi
niti
on is
an
appl
icat
ion
of e
vide
nce-
base
d m
edic
ine
conc
epts
to p
ublic
hea
lth
prac
tice
.
(See
als
o Pu
blic
hea
lth.
)
Impa
cts
Impa
cts
are
the
cont
ribu
tion
mad
e to
the
achi
evem
ent o
f an
outc
ome
by a
spe
cifie
d se
t of a
ctio
ns (T
he T
reas
ury
and
Sta
te
Ser
vice
s Co
mm
issi
on 2
007b
).
Impa
cts
shou
ld b
e po
siti
ve, a
ltho
ugh
unin
tend
ed c
onse
quen
ces
from
an
acti
on m
ay o
ccur
that
resu
lt in
eit
her p
osit
ive
or n
egat
ive
cont
ribu
tion
s to
the
achi
evem
ent o
f the
des
ired
out
com
e.
Not
e: th
is d
efini
tion
and
use
of ‘
impa
ct’ a
re d
iffer
ent t
o ho
w
it is
defi
ned
and
used
in e
valu
atio
n m
etho
dolo
gy (s
ee a
lso
Eval
uati
on).
3� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
Ind�
cato
rs
Q
uant
�tat�v
e
Q
ual�t
at�v
e
Pr
oxy
Indi
cato
rs a
re e
ithe
r qua
ntita
tive
or q
ualit
ativ
e m
easu
res
that
as
sess
the
dire
ctio
n an
d si
ze o
f cha
nge
in th
e th
ing
bein
g m
easu
red.
Qua
ntita
tive
indi
cato
rs a
re n
umer
ical
mea
sure
men
ts th
at le
nd
them
selv
es to
sta
tisti
cal a
naly
sis.
(Thi
s H
ow to
focu
ses
on th
is
type
of i
ndic
ator
). Q
ualit
ativ
e in
dica
tors
are
pri
mar
ily m
easu
res
take
n in
a te
xtua
l (na
rrat
ive)
form
(ie,
non
-num
eric
al) a
nd m
ay o
r m
ay n
ot le
nd th
emse
lves
to q
uant
ifica
tion
. Ana
lysi
s of
qua
litat
ive
indi
cato
rs re
quir
es th
e us
e of
qua
litat
ive
anal
ytic
al te
chni
ques
. Pr
oxy
indi
cato
rs a
re m
easu
res
that
pro
vide
an
appr
oxim
ate
esti
mat
e of
cha
nge
in th
e ou
tcom
e of
inte
rest
. Pro
xy m
easu
res
may
be
appr
opri
ate
to u
se w
hen
dire
ct m
easu
res
cann
ot b
e us
ed b
ecau
se
of c
once
rns
abou
t eth
ics,
cos
t, c
ompl
exit
y of
mea
sure
men
t or
tim
elin
ess.
For
exa
mpl
e, h
ospi
talis
atio
ns fo
r int
enti
onal
sel
f-ha
rm
are
used
as
an a
ccep
tabl
e pr
oxy
mea
sure
of s
uici
de a
ttem
pt, a
nd
BM
I is
usua
lly a
ccep
ted
as a
pro
xy fo
r obe
sity
.
(See
als
o In
stru
men
t.)
Tim
e is
an
impo
rtan
t fac
tor i
n as
sess
ing
wha
t cha
nge
has
occu
rred
. In
dica
tors
may
mea
sure
cha
nge
over
the
shor
t,
med
ium
or l
ong
term
. Wha
t con
stitu
tes
shor
t, m
ediu
m o
r lon
g te
rm d
epen
ds o
n th
e co
ntex
t and
the
issu
e(s)
the
prog
ram
me
is
tack
ling.
Hea
lth
indi
cato
rs d
irec
tly
or in
dire
ctly
(ie,
pro
xy) m
easu
re a
he
alth
-rel
ated
cha
ract
eris
tic
of a
n in
divi
dual
, pop
ulat
ion
or th
e en
viro
nmen
t. T
he in
dica
tor m
ay m
easu
re o
ne o
r mor
e he
alth
as
pect
s (q
ualit
y, q
uant
ity
and
tim
e) (N
utbe
am 1
998)
.
Inpu
tsRe
sour
ces
put i
nto
a pr
ogra
mm
e to
car
ry o
ut a
n ac
tivi
ty. A
noth
er
term
for i
nput
is ‘e
ffor
t’.
Inpu
ts c
an b
e hu
man
, mat
eria
l, fin
anci
al o
r exp
ress
ed a
s ti
me.
Ef
fort
des
crib
es e
very
thin
g th
at ‘w
e’ u
se a
nd d
o fo
r ‘th
em’ –
the
prog
ram
me
reci
pien
ts.
Inst
rum
ent
An
inst
rum
ent i
s a
num
eric
al o
r tex
t-ba
sed
mea
suri
ng m
etho
d th
at
show
s th
e ex
tent
or a
mou
nt o
r qua
ntit
y or
deg
ree
of s
omet
hing
of
inte
rest
.
The
inst
rum
ent i
s th
e pr
oces
s us
ed to
pro
duce
the
desi
red
publ
ic
heal
th in
dica
tor.
(See
Ind�
cato
rs.)
Mea
sure
(s)
See
Ind�
cato
rs a
nd In
stru
men
t.
Met
r�c(
s)S
ee In
d�ca
tors
and
Inst
rum
ent.
Mon
�tor
See
Out
com
es m
on�to
r�ng
.
37 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
Obj
ect�v
es H
�gh-
orde
r/ge
nera
l S
pec�
fic/
�nt
erm
ed�a
te L
ow-o
rder
/
ope
rat�o
nal
SM
ART
Obj
ecti
ves
are
stat
emen
ts a
bout
the
resu
lts
a pr
ogra
mm
e se
eks
to a
chie
ve. A
ny p
rogr
amm
e m
ust h
ave
at le
ast o
ne o
bjec
tive
. O
bjec
tive
s m
ay b
e tr
ansl
ated
dir
ectl
y in
to ‘o
utco
mes
’ whe
re th
ey
deal
wit
h on
ly o
ne is
sue.
How
ever
, dou
ble-
barr
elle
d ob
ject
ives
will
re
quir
e m
ulti
ple
outc
ome
mea
sure
s to
be
deve
lope
d.
Obj
ecti
ves
may
form
a h
iera
rchy
that
mov
es fr
om a
lim
ited
set o
f hi
gh-o
rder
obj
ecti
ves
that
are
syn
onym
ous
wit
h ai
ms
or g
oals
, to
low
er-o
rder
or m
ore
spec
ific
or in
term
edia
te o
bjec
tive
s, a
s fo
llow
s.
Hig
h-or
der o
r gen
eral
obj
ectiv
es a
re s
tate
men
ts a
bout
the
over
-all
or lo
ng-t
erm
eff
ects
or u
ltim
ate
outc
omes
that
are
exp
ecte
d to
be
obta
ined
as
a re
sult
of th
e pr
ogra
mm
e, o
r to
whi
ch th
e pr
ogra
mm
e co
ntri
bute
s to
as
part
of a
who
le-o
f-go
vern
men
t ini
tiati
ve.
Att
ainm
ent o
f hig
h-or
der o
bjec
tive
s m
ay o
r may
not
be
dire
ctly
at
trib
utab
le to
any
one
pro
gram
me.
Lon
g-te
rm o
bjec
tive
s ar
e ge
nera
lly th
ose
to b
e ac
hiev
ed in
5 to
10
year
s.
Spec
ific
or in
term
edia
te-l
evel
obj
ectiv
es a
re s
tate
men
ts a
bout
th
e in
term
edia
te o
utco
mes
that
are
exp
ecte
d to
be
achi
eved
by
the
prog
ram
me,
and
that
are
mor
e di
rect
ly a
ttri
buta
ble
to th
e pr
ogra
mm
e’s
acti
viti
es. T
hese
obj
ecti
ves
are
gene
rally
thos
e to
be
achi
eved
in 3
to 5
yea
rs.
Low
-ord
er o
r ope
ratio
nal o
bjec
tives
are
sta
tem
ents
abo
ut th
e im
med
iate
out
com
es (w
hich
may
be
expr
esse
d as
out
puts
in
som
e ci
rcum
stan
ces)
exp
ecte
d to
be
achi
eved
or a
ccom
plis
hed
from
the
inte
rven
tion
that
are
dir
ectl
y at
trib
utab
le to
the
acti
viti
es
unde
rtak
en. T
hese
obj
ecti
ves
are
gene
rally
thos
e to
be
achi
eved
in
1 to
2 y
ears
.
In te
rms
of th
e Pu
blic
Fin
ance
Act
200
4, o
bjec
tive
s ar
e no
t ex
pres
sly
defin
ed, a
nd c
onse
quen
tly
can
be s
een
as m
eani
ng a
go
al o
r aim
. How
ever
, bec
ause
som
e co
re d
epar
tmen
tal a
ctiv
itie
s th
at s
ervi
ce th
e op
erat
ion
of g
over
nmen
t do
not fi
t wit
hin
the
defin
itio
n of
an
outc
ome
give
n in
the
Act
, obj
ecti
ves
are
also
see
n as
bei
ng s
imila
r to
impa
cts
and
outc
omes
. (S
ee O
utco
mes
.) (T
he
Trea
sury
and
Sta
te S
ervi
ces
Com
mis
sion
200
7a.)
In h
ealt
h te
rms,
hea
lth
goal
s (o
bjec
tive
s/ai
ms)
are
gen
eral
st
atem
ents
of i
nten
t and
dir
ecti
on s
et fo
r a h
ealt
h pr
ogra
mm
e,
whi
ch m
ay in
clud
e a
heal
th ta
rget
to a
chie
ve (N
utbe
am 1
998)
. In
this
doc
umen
t, th
e te
rm ‘o
bjec
tive
’ is
used
to id
enti
fy
part
icul
ar re
sult
s th
e pr
ogra
mm
e se
eks
to a
chie
ve. F
or m
onito
ring
pu
rpos
es, t
he re
sult
may
be
expr
esse
d in
term
s of
an
outc
ome
or
an o
utpu
t. (S
ee O
utco
mes
and
Out
puts
.)
Follo
win
g is
an
exam
ple
of a
hie
rarc
hy o
f obj
ecti
ves.
Hig
h-or
der o
bjec
tive
(syn
onym
ous
wit
h vi
sion
, aim
or g
oal)
:
• En
able
peo
ple
wit
h ch
roni
c co
ndit
ions
to im
prov
e th
eir
heal
th, s
low
pro
gres
s of
thei
r con
diti
on(s
), a
nd m
aint
ain
inde
pend
ence
whe
reve
r pos
sibl
e by
alig
ning
com
mun
ity
and
hosp
ital s
ervi
ces
acro
ss [
] D
HB
.
Inte
rmed
iate
-lev
el o
bjec
tive
asso
ciat
ed w
ith a
chie
ving
the
high
-le
vel o
bjec
tive:
• Re
duce
the
inci
denc
e of
can
cer,
dia
bete
s, a
nd c
ardi
ovas
cula
r di
seas
e by
20%
resp
ecti
vely
ove
r the
nex
t 5 y
ears
.
3� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
Obj
ecti
ves
shou
ld b
e w
ritt
en a
s SM
ART
obje
ctiv
es:
• Sp
ecifi
c
• M
easu
rabl
e
• Ac
hiev
able
(som
etim
es re
ferr
ed to
as
‘acc
urat
e’ o
r ‘ac
tion
-or
ient
ed’)
• Re
leva
nt (s
omet
imes
refe
rred
to a
s ‘r
ealis
tic’
)
• Ti
me-
base
d.
Low
-ord
er o
r ope
rati
onal
obj
ecti
ves
asso
ciat
ed w
ith
achi
evin
g th
e in
term
edia
te le
vel:
• In
crea
se th
e le
vel o
f kno
wle
dge
by 5
0% in
the
targ
et
popu
lati
on g
roup
s ab
out t
he c
ause
s an
d pr
even
tion
of b
reas
t an
d pr
ostr
ate
canc
er, d
iabe
tes
and
card
iova
scul
ar d
isea
se.
• In
crea
se th
e up
take
of c
ance
r scr
eeni
ng s
ervi
ces
by 5
0% in
the
high
-ris
k po
pula
tion
s in
the
next
3 y
ears
.
• In
crea
se a
cces
s by
50%
to e
arly
dia
gnos
is s
ervi
ces
in ru
ral
com
mun
itie
s in
the
next
2 y
ears
.
Out
com
es
In
term
ed�a
te
Cr
oss-
agen
cy
Pr
oces
s
Out
com
es a
re s
peci
fic s
tate
men
ts a
bout
the
inte
nded
cha
nge
in
publ
ic h
ealt
h-re
late
d at
titu
des,
kno
wle
dge,
beh
avio
urs,
or p
hysi
cal
(inc
ludi
ng m
enta
l) h
ealt
h st
atus
in th
e ta
rget
pop
ulat
ion(
s) s
ough
t by
und
erta
king
the
plan
ned
publ
ic h
ealt
h ac
tivi
ty.
Inte
rmed
iate
out
com
es (a
lso
calle
d in
teri
m o
utco
mes
) are
ste
ps
alon
g th
e w
ay to
the
desi
red
end
outc
ome.
The
y ar
e of
ten
smal
ler
chan
ges
that
nee
d to
hap
pen
befo
re th
e fin
al d
esir
ed o
utco
me
can
be re
ache
d.
Cros
s-ag
ency
out
com
es a
re o
utco
mes
whe
re th
e co
llect
ive
effo
rt o
f m
ore
than
one
gov
ernm
ent a
genc
y or
pro
gram
me
is re
quir
ed if
the
outc
ome
is to
be
achi
eved
.
Proc
ess
outc
omes
typi
cally
mea
sure
the
amou
nt o
f eff
ort p
ut in
to a
pr
ogra
mm
e an
d th
e qu
alit
y of
the
serv
ice
prov
ided
. The
y ca
n al
so
be a
ppro
pria
te w
here
it is
impo
rtan
t to
mon
itor c
omm
unit
y su
ppor
t fo
r a p
rogr
amm
e. P
roce
ss o
utco
mes
that
mea
sure
eff
ort c
an b
e ex
pres
sed
as o
utpu
ts.
(See
Pro
cess
out
com
e m
on�to
r�ng
.)
Hea
lth
outc
omes
sho
uld
be e
xpre
ssed
in s
uch
a w
ay th
at
indi
cate
s th
e ty
pe, d
irec
tion
and
ext
ent o
f cha
nge
in h
ealt
h kn
owle
dge,
beh
avio
ur o
r sta
tus
soug
ht in
an
indi
vidu
al o
r po
pula
tion
gro
up.
Exam
ple
• Ra
tes
of s
mok
ing
cess
atio
n in
you
ng w
omen
und
er 2
0 ye
ars
of
age
incr
ease
by
20%
at t
he e
nd o
f the
pro
gram
me.
Att
ainm
ent o
f the
out
com
es is
mea
sure
d th
roug
h in
dica
tors
(see
In
d�ca
tors
). Id
eally
, the
hea
lth
outc
ome
soug
ht s
houl
d be
dir
ectl
y at
trib
utab
le to
the
inte
rven
tion
, but
this
is u
sual
ly p
robl
emat
ic in
he
alth
pro
mot
ion
acti
viti
es (N
utbe
am 1
998)
.
This
defi
niti
on c
an b
e se
en a
s a
spec
ific
appl
icat
ion
to p
ublic
he
alth
of t
he d
efini
tion
of a
n ‘o
utco
me’
pro
vide
d in
the
Publ
ic
Fina
nce
Act
200
4 (s
2(1)
): ‘A
sta
te o
r con
diti
on o
f soc
iety
, the
ec
onom
y or
the
envi
ronm
ent a
nd in
clud
es a
cha
nge
in th
at s
tate
or
con
diti
on.’
3� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
Cros
s-ag
ency
out
com
es s
houl
d pr
obab
ly b
e th
e ru
le ra
ther
than
th
e ex
cept
ion
in c
ompr
ehen
sive
pub
lic h
ealt
h pr
ogra
mm
es. I
n su
ch s
ituat
ions
, it i
s im
port
ant t
hat t
he ro
les
and
resp
onsi
bilit
ies
of th
e pa
rtic
ipat
ing
agen
cies
are
agr
eed
– in
clud
ing
who
is
resp
onsi
ble
for m
onito
ring
out
com
es –
bef
ore
the
prog
ram
me
begi
ns.
Out
com
es
mon
�tor�
ngA
pla
n fo
r the
rout
ine,
sys
tem
atic
col
lect
ion
and
reco
rdin
g of
in
form
atio
n ab
out a
spec
ts o
f a p
rogr
amm
e ov
er ti
me.
The
pur
pose
is
to a
sses
s w
heth
er p
rogr
ess
is b
eing
mad
e on
ach
ievi
ng th
e pr
ogra
mm
e ob
ject
ives
.
Prep
arin
g th
e ou
tcom
es m
onito
ring
pla
n re
quir
es a
sses
sing
:
• w
hat n
eeds
to b
e m
easu
red
to d
emon
stra
te s
ucce
ss
• ho
w th
ings
are
goi
ng to
be
mea
sure
d an
d by
who
m (d
ecid
ing
wha
t ind
icat
ors
are
goin
g to
be
used
, aft
er c
onsi
deri
ng
issu
es s
uch
as v
alid
ity,
relia
bilit
y, s
ensi
tivi
ty, a
ttri
buti
on a
nd
feas
ibili
ty) (
see
Ind�
cato
rs.)
•
the
tim
efra
mes
for d
eliv
erin
g in
form
atio
n ab
out t
he
perf
orm
ance
of t
he p
rogr
amm
e ag
ains
t the
sta
ted
obje
ctiv
es
to d
ecis
ion-
mak
ers
and
key
stak
ehol
ders
in a
tim
ely
man
ner.
Out
com
es m
onito
ring
may
or m
ay n
ot in
clud
e a
form
al e
valu
atio
n pr
oces
s de
pend
ing
on th
e ob
ject
ives
of t
he p
rogr
amm
e. (S
ee
Eval
uat�o
n.)
Out
puts
Out
puts
are
thin
gs (e
g, g
oods
) pro
duce
d, s
ervi
ces
deliv
ered
, ev
ents
hel
d, o
r par
tici
pati
on g
ener
ated
resu
ltin
g fr
om th
e ac
tivi
ties
un
dert
aken
. (S
ee O
utco
mes
and
Pro
cess
out
com
e m
on�to
r�ng
).
This
defi
niti
on is
con
sist
ent w
ith
that
pro
vide
d in
the
Publ
ic
Fina
nce
Act
200
4. C
omm
on p
ublic
hea
lth
outp
uts
are
thin
gs s
uch
as in
form
atio
n se
rvic
es p
rovi
ded,
mee
ting
s he
ld o
r att
ende
d,
and
educ
atio
nal/
trai
ning
or s
ocia
l mar
keti
ng p
rodu
cts
or s
ervi
ces
deliv
ered
. Suc
h ou
tput
s ar
e in
tend
ed to
hav
e an
impa
ct th
at
assi
sts
or re
sult
s in
ach
ievi
ng th
e de
sire
d pu
blic
hea
lth
outc
ome,
su
ch a
s a
chan
ge in
hea
lth
know
ledg
e an
d be
havi
our.
(See
Out
com
es a
nd P
roce
ss o
utco
me
mon
�tor�
ng.)
40 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
Proc
ess
outc
ome
mon
�tor�
ngTh
is is
like
out
com
e m
onito
ring
but
incl
udes
kee
ping
trac
k of
how
w
ell a
pro
gram
me
is b
eing
impl
emen
ted
in it
s di
ffer
ent a
spec
ts.
Proc
ess
outc
ome
mon
itori
ng p
rovi
des
the
basi
s fo
r pro
gram
me
staf
f to
ass
ure
them
selv
es th
at g
ood
outc
omes
are
ass
ocia
ted
wit
h go
od
prog
ram
me
deliv
ery.
If t
he d
esir
ed o
utco
mes
are
not
ach
ieve
d, th
en
the
agen
cy w
ill h
ave
suffi
cien
t inf
orm
atio
n to
ass
ess
whi
ch a
spec
ts
of p
rogr
amm
e de
liver
y ne
ed to
be
deve
lope
d or
cha
nged
.
(See
als
o Ba
lanc
ed s
core
card
.)
A p
rogr
amm
e th
eory
of c
hang
e w
ill in
dica
te w
hat k
inds
of i
nput
s an
d pr
oces
ses
are
likel
y to
pro
duce
a p
reve
ntio
n ef
fect
. For
ex
ampl
e, a
hom
e-vi
siti
ng p
rogr
amm
e m
ight
det
ail t
he fo
llow
ing
as v
ital t
o go
od c
lient
out
com
es:
• ap
prop
riat
ely
expe
rien
ced,
ski
lled
and
supe
rvis
ed s
taff
• st
aff w
ho a
re e
mpa
thet
ic a
nd w
arm
in th
eir i
nter
acti
ons
• a
fam
ily/w
hana
u as
sess
men
t tha
t ide
ntifi
es s
tren
gths
and
ch
alle
nges
like
ly to
hel
p or
hin
der c
hild
-rea
ring
(eg,
incl
udes
a
scre
en fo
r fam
ily v
iole
nce,
sub
stan
ce a
buse
, dep
ress
ion)
• th
e as
sess
men
t pro
cess
enc
oura
ges
co-o
pera
tion
and
redu
ces
defe
nsiv
enes
s
• fa
mili
es a
re v
isite
d at
suf
ficie
ntly
regu
lar i
nter
vals
• id
eally
the
ethn
icit
y of
wor
kers
and
pri
mar
y ca
regi
vers
are
m
atch
ed to
max
imis
e en
gage
men
t
• th
ere
is a
dev
elop
men
tal m
odel
to c
ompa
re c
hild
ren’
s pr
ogre
ss a
gain
st
• th
e de
velo
pmen
tal m
odel
is e
asy
to c
omm
unic
ate
to
care
give
rs a
nd th
ey ‘b
uy’ i
t
• de
velo
pmen
tal o
utco
mes
for c
hild
ren
and
cond
itio
ns th
at
hind
er n
orm
al d
evel
opm
ent a
re s
yste
mat
ical
ly tr
acke
d
• gu
idan
ce is
pro
vide
d in
a re
spon
sive
way
(ie,
is ti
mel
y,
solu
tion
focu
sed,
real
isti
c an
d ch
ecke
d fo
r und
erst
andi
ng)
• go
al-s
etti
ng is
use
d to
enh
ance
car
egiv
er m
otiv
atio
n an
d to
pr
iori
tise
acti
ons.
Regu
lar s
uper
visi
on m
eeti
ngs
and/
or p
rogr
amm
e re
view
s pr
ovid
e sp
ace
for c
lient
s’ p
rogr
ess
to b
e ch
ecke
d ag
ains
t the
qua
lity
of
prog
ram
me
deliv
ery.
Any
cha
nges
in c
lient
pla
ns, p
rogr
amm
e de
sign
or p
rofe
ssio
nal d
evel
opm
ent c
an b
e co
nsid
ered
a p
roce
ss
outc
ome.
41 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
Publ
�c h
ealth
Popu
lat�o
n-ba
sed
publ
�c h
ealth
Pu
bl�c
hea
lth
act�o
n
Ther
e is
a ra
nge
of te
rms
and
defin
itio
ns re
lati
ng to
pub
lic h
ealt
h,
incl
udin
g pu
blic
hea
lth
acti
on a
nd p
opul
atio
n-ba
sed
publ
ic h
ealt
h.
The
Wor
ld H
ealt
h O
rgan
izat
ion
(199
8) d
efine
s pu
blic
hea
lth
as:
Th
e sc
ienc
e an
d ar
t of p
rom
otin
g he
alth
, pre
vent
ing
dise
ase,
and
pr
olon
ging
life
thro
ugh
the
orga
nize
d ef
fort
s of
soc
iety
.
It ca
n be
arg
ued
that
this
defi
niti
on o
mit
s th
e im
port
ant p
oint
that
pu
blic
hea
lth
prim
arily
focu
ses
on th
e he
alth
of t
he p
opul
atio
n or
po
pula
tion
sub
-gro
ups
and
not t
he h
ealt
h of
spe
cific
indi
vidu
als.
Th
is fo
cus
on p
opul
atio
ns in
pub
lic h
ealt
h is
iden
tifie
d m
ore
expl
icit
ly in
mor
e re
cent
defi
niti
ons.
• ‘P
ublic
hea
lth
is c
once
rned
wit
h im
prov
ing
the
heal
th o
f the
po
pula
tion
, rat
her t
han
trea
ting
the
dise
ases
of i
ndiv
idua
l pa
tien
ts’
(Dep
artm
ent o
f Hea
lth
2004
cite
d in
Pub
lic H
ealt
h el
ectr
onic
Lib
rary
).
• ‘[P
ublic
hea
lth
is] c
once
rned
wit
h th
e he
alth
of t
he c
omm
unit
y as
a w
hole
’ (Am
eric
an Jo
urna
l of P
ublic
Hea
lth 2
006,
web
site
(h
ttp:
//w
ww
.ajp
h.or
g/))
.
• Pu
blic
hea
lth
acti
on is
‘Co
llect
ive
acti
on fo
r sus
tain
ed
popu
lati
on-w
ide
heal
th im
prov
emen
t … a
nd to
redu
ce h
ealt
h in
equa
litie
s. R
espo
nsib
ility
for s
uch
acti
on is
not
con
fined
to
the
heal
th s
ecto
r but
sho
uld
incl
ude
all s
ecto
rs w
hose
ac
tion
s af
fect
the
heal
th o
f pop
ulat
ions
. Pop
ulat
ions
can
be
geo
grap
hic
and/
or d
efine
d by
fact
ors
such
as
ethn
icit
y,
gend
er, a
ge, s
exua
l ori
enta
tion
, inc
ome
etc.
’ (B
eagl
ehol
e et
al
2004
; Pub
lic H
ealt
h A
dvis
ory
Com
mit
tee
2006
).
• ‘P
ublic
hea
lth
is th
e or
gani
sed
resp
onse
by
soci
ety
to p
rote
ct
and
prom
ote
heal
th, a
nd to
pre
vent
illn
ess,
inju
ry a
nd
disa
bilit
y. T
he s
tart
ing
poin
t for
iden
tify
ing
publ
ic h
ealt
h is
sues
, pro
blem
s an
d pr
iori
ties
, and
for d
esig
ning
and
im
plem
enti
ng in
terv
enti
ons,
is th
e po
pula
tion
as
a w
hole
, or
popu
lati
on s
ub-g
roup
s.’ (
Nat
iona
l Pub
lic H
ealt
h Pa
rtne
rshi
p 20
06).
(See
als
o Ev
�den
ce-b
ased
pub
l�c h
ealth
.)
42 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
Rel�a
b�l�t
yRe
liabi
lity
is th
e ex
tent
to w
hich
a m
easu
re, w
hen
used
repe
ated
ly in
th
e sa
me
way
, will
pro
duce
the
sam
e or
a s
imila
r res
ult.
Fo
r exa
mpl
e, if
we
wer
e in
tere
sted
in m
onito
ring
a p
erso
n’s
wei
ght a
nd u
sed
a ch
eap
wei
ghin
g m
achi
ne th
at p
rodu
ced
a di
ffer
ent r
eadi
ng e
ach
tim
e it
was
use
d (w
hen
all o
ther
fact
ors
wer
e th
e sa
me)
, the
n th
e m
achi
ne w
ould
not
be
a re
liabl
e in
stru
men
t to
use.
Robu
stne
ss to
w
�thst
and
scru
t�ny
It is
impo
rtan
t tha
t ass
essm
ents
abo
ut th
e ef
fect
iven
ess
of p
ublic
he
alth
pro
gram
mes
fund
ed b
y go
vern
men
t mon
ey b
e ab
le to
w
iths
tand
scr
utin
y by
all
the
stak
ehol
ders
and
oth
er in
tere
sted
pa
rtie
s.
We
reco
mm
end
that
for m
ost p
ublic
hea
lth
prog
ram
mes
it is
ad
visa
ble
that
a p
rogr
amm
e’s
succ
ess
not b
e at
trib
uted
to a
si
ngle
cri
teri
on: r
obus
tnes
s is
pro
vide
d by
the
tota
lity
of th
e ou
tcom
e m
onito
ring
regi
me.
Sens
�t�v�
tyS
ensi
tivi
ty re
fers
to h
ow w
ell a
mea
sure
is a
ble
to a
ccur
atel
y de
tect
w
hen
a ch
ange
has
occ
urre
d in
the
outc
ome
bein
g m
onito
red.
For e
xam
ple,
a w
eigh
ing
mac
hine
that
was
abl
e to
diff
eren
tiate
a
chan
ge in
wei
ght b
y 10
0 gr
am g
radu
atio
ns is
mor
e se
nsit
ive
than
a
mac
hine
that
cou
ld o
nly
dete
ct c
hang
es in
500
gra
m g
rada
tion
s.
Targ
etA
targ
et is
the
amou
nt o
f cha
nge
desi
red,
or e
xpec
ted,
in a
spe
cific
ti
me
peri
od th
roug
h th
e im
plem
enta
tion
of t
he h
ealt
h pr
ogra
mm
e.
Targ
ets
shou
ld b
e ba
sed
on s
peci
fic a
nd m
easu
rabl
e ch
ange
s in
hea
lth
outc
omes
. Set
ting
targ
ets
help
s de
fine
the
proc
ess
tow
ards
ach
ievi
ng th
e he
alth
obj
ecti
ve s
et, a
ltho
ugh
targ
ets
requ
ire
the
exis
tenc
e of
an
appr
opri
ate
indi
cato
r (N
utbe
am 1
998)
.
(See
Ind�
cato
rs.)
Theo
ry o
f cha
nge/
act�o
nA
theo
ry o
f cha
nge/
acti
on is
a s
tate
men
t abo
ut h
ow p
ract
itio
ners
be
lieve
indi
vidu
al, i
nter
-gro
up a
nd s
ocia
l/ s
yste
mic
cha
nge
happ
ens,
and
how
the
prop
osed
act
ions
will
pro
duce
pos
itiv
e re
sult
s. A
goo
d th
eory
of c
hang
e ex
plic
itly
des
crib
es th
e as
sum
ptio
ns a
nd c
ausa
l lin
ks th
ough
t to
be im
port
ant f
or a
chie
ving
th
e he
alth
out
com
e so
ught
.
Whe
re in
itia
l evi
denc
e fo
r the
eff
ecti
vene
ss o
f an
inte
rven
tion
is
lack
ing,
a ro
bust
theo
ry o
f cha
nge
shou
ld b
e ar
ticu
late
d to
ju
stif
y th
e ch
oice
of i
nter
vent
ion.
Use
d in
this
way
, a th
eory
of
chan
ge p
rovi
des
the
‘logi
c m
odel
’ fra
mew
ork
to p
lan
and
just
ify
the
inte
rven
tion
sel
ecte
d. A
theo
ry o
f cha
nge
appr
oach
may
be
part
icul
arly
app
ropr
iate
whe
n in
terv
enti
ons
are
aim
ed a
t min
orit
y po
pula
tion
gro
ups.
43 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
Term
Defi
n�t�o
nEx
plan
ator
y co
mm
ent a
nd e
xam
ple
T�m
ely
data
/�n
form
at�o
nIn
form
atio
n is
tim
ely
whe
n it
cont
ains
dat
a di
rect
ly re
leva
nt to
the
prog
ram
me
and
– im
port
antl
y –
ava
ilabl
e in
a ti
me
fram
e w
here
it
can
usef
ully
info
rm d
ecis
ions
abo
ut w
heth
er th
e pr
ogra
mm
e is
pe
rfor
min
g as
pla
nned
and
wha
t thi
ngs
need
to b
e ch
ange
d to
im
prov
e th
e pe
rfor
man
ce o
f the
pro
gram
me,
if re
quir
ed.
Som
e na
tion
al d
ata
sets
are
sch
edul
ed to
be
colle
cted
on
5-ye
arly
dat
a. S
uch
data
is u
nlik
ely
to b
e ti
mel
y fo
r per
form
ance
m
onito
ring
pur
pose
s if
the
prog
ram
me
will
onl
y ru
n fo
r 3 y
ears
, sa
y. S
imila
rly,
nat
iona
l mor
talit
y an
d m
orbi
dity
dat
a is
typi
cally
on
ly a
vaila
ble
18 to
24
mon
ths
afte
r the
eve
nt. S
uch
info
rmat
ion
may
not
be
tim
ely
for s
ome
prog
ram
mes
, alt
houg
h lo
cal d
ata
may
be
ava
ilabl
e w
ithi
n a
mor
e us
eful
tim
e fr
ame.
T�m
�ng
Tim
ing
refe
rs to
the
tim
e pe
riod
bet
wee
n w
hen
an a
ctiv
ity
took
pla
ce
and
whe
n a
chan
ge in
the
desi
red
outc
ome
was
obs
erve
d.
Tim
ing
is a
n im
port
ant a
spec
t in
esta
blis
hing
‘att
ribu
tion
’, an
d be
ing
able
to m
onito
r whe
ther
a p
rogr
amm
e is
wor
king
.
Val�d
�tyVa
lidit
y re
fers
to h
ow w
ell a
pro
pose
d in
stru
men
t mea
sure
s th
e ou
tcom
e of
inte
rest
. Som
e ou
tcom
es m
ay b
e m
easu
red
dire
ctly
, w
hile
oth
ers
will
requ
ire
a pr
oxy
met
hod
that
resu
lts
in a
pro
xy
mea
sure
. Gen
eral
ly, t
he m
ore
dire
ct a
mea
sure
is, t
he m
ore
valid
it is
lik
ely
to b
e.
Ther
e is
oft
en m
ore
than
one
way
of m
easu
ring
an
outc
ome
of
inte
rest
, and
ther
e ca
n be
con
side
rabl
e de
bate
abo
ut h
ow ‘v
alid
’ a
mea
sure
act
ually
is.
For e
xam
ple,
the
best
way
to m
easu
re b
ody
com
posi
tion
(par
ticu
larl
y le
an a
nd fa
t mas
s) is
und
er-w
ater
den
sito
met
ry
(wei
ghin
g), b
ut th
is is
not
pra
ctic
al o
utsi
de s
mal
l clin
ical
stu
dies
.
Ther
efor
e, m
easu
res
of w
eigh
t adj
uste
d fo
r hei
ght (
eg, b
ody
mas
s in
dex,
BM
I) a
re o
ften
use
d to
refle
ct b
ody
com
posi
tion
,
part
icul
arly
bod
y fa
t mas
s. A
ltho
ugh
BM
I is
corr
elat
ed w
ith
body
fat m
ass,
this
rela
tion
ship
var
ies
acco
rdin
g to
bod
y bu
ild, a
ge
and
ethn
icit
y. T
here
fore
, BM
I is
not r
ecom
men
ded
for a
sses
sing
exce
ss b
ody
fat m
ass
(obe
sity
) at a
n in
divi
dual
leve
l.
How
ever
, BM
I can
be
usef
ul fo
r ass
essi
ng o
besi
ty a
t a p
opul
atio
n
leve
l if u
sed
and
inte
rpre
ted
appr
opri
atel
y. B
MI b
ased
on
dire
ct
mea
sure
men
t of h
eigh
t and
wei
ght i
s m
ore
valid
than
BM
I bas
ed
on s
elf-
repo
rted
hei
ght a
nd w
eigh
t, p
arti
cula
rly
if m
easu
rem
ents
are
take
n by
trai
ned
obse
rver
s us
ing
appr
opri
ate
equi
pmen
t and
stan
dard
ised
pro
cedu
res.
Ult
imat
ely,
the
choi
ce o
f mea
sure
for a
par
ticu
lar p
rogr
amm
e w
ill
depe
nd o
n a
num
ber o
f fac
tors
, inc
ludi
ng v
alid
ity
and
prac
tica
lity.
44 How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
References Bartholomew L, Parcel G, Kok G, et al. 2006. Planning Health Promotion Programmes: An intervention mapping approach. San Francisco. Jossey-Bass.
Beaglehole R, Bonita R, Horton R, et al. 2004. Public health in the new era: improving health through collective action. Lancet 363: 2084–6.
Brownson RC. 2003. Evidence-based Public Health. Oxford: Oxford University Press.
Davidson J. 2005. Evaluation Methodology Basics: Nuts and bolts of sound evaluation. Thousand Oaks. Sage.
Department of Health. 2004. Public Health Electronic Library Standards for Better Health. London: Department of Health. London.
Iverson A. 2003. Preparing Program Objectives: Theory and practice. Toronto: The International Development Research Centre Evaluation Unit.
Jenicek M. 1997. Epidemiology, evidenced-based medicine, and evidence-based public health. Journal of Epidemiology. Dec; 7(4): 187–97.
Kohatsu ND, Robinson JG, Torner JC. 2004. Evidence-based public health: an evolving concept. American Journal of Preventive Medicine 27: 417–21.
Ministry of Health. 2005. The New Zealand Health Monitor: Updated strategic plan. Wellington: Ministry of Health.
Ministry of Health. 2006. A Guide to Developing Public Health Programmes: A generic programme logic model. Wellington: Ministry of Health.
Ministry of Health. 2007. Leading for Outcomes. http://www.leadingforoutcomes.org.nz Wellington: Ministry of Health.
National Public Health Partnership. 2006. Public Health Classifications Project Phase One: Final report. Melbourne: National Public Health Partnership.
Nutbeam D. 1998. Health promotion glossary. Health Promotion International 13: 349–64.
Public Health Advisory Committee. 2006. Health is Everyone’s Business: Working together for health and wellbeing: A report to the Minister of Health on the implications of a changing context for public health in New Zealand. Wellington: Public Health Advisory Committee.
The Treasury, State Services Commission. 2007a. Guidance and Requirements for Crown Entities: Preparing the Statement of Intent. Wellington: The Treasury, State Services Commission.
The Treasury, State Services Commission. 2007b. Guidance and Requirements for Departments: Preparing the Statement of Intent. Wellington: The Treasury, State Services Commission.
4� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework
The Treasury, State Services Commission, Department of the Prime Minister and Cabinet. 2005. Guidance for Crown Entities: Planning and managing for results. Wellington: The Treasury, State Services Commission, Department of the Prime Minister and Cabinet.
United Nations Development Programme Evaluation Office. 2002. Handbook on Monitoring and Evaluating for Results. New York: United Nations Development Programme Evaluation Office.
World Health Organization. 1998. Health Promotion Glossary. Geneva: World Health Organization.
Further read�ngErnst K. 2004. The Canadian Outcomes Research Institute’s Program Logic Model. Calgary: The Canadian Outcomes Research Institute.
European Research Group Health Outcomes. 1996. Choosing a health outcomes measurement instrument: general advice proposed by the European Research Group on Health Outcomes Measures (ERGHO). Quality of Life Newsletter 15: 7–8. URL: http://www.mapi-research.fr/pdf/newsletter/qol15 0.PDF. Accessed 3 November 2005.
4� How to Monitor for Population Health Outcomes: Guidelines for developing a monitoring framework