# Methodology

This page gives links to some useful short reports on methodological topics. The page does not attempt to be comprehensive but gives a selection of relevant publications. Further useful publications can be found in the reference sections of the documents below. Please note that following the reorganisation within England, the regional public health observatories (PHO) and the Association of PHOs (APHO) are all now part of Public Health England (PHE). We welcome comments and ideas about what topics to include.

### Terminology

See our Guide to terminology (49Kb) for guidance on some of the terms used in the ScotPHO webpages.

### Standardisation

ISD has produced a guide to standardisation (196Kb) and an Excel tool (65Kb) which allows you to standardise your own data.

The English Eastern Region Public Health Observatory (now part of PHE) report INphoRM 6: Standardisation (196Kb) also describes and compares the different methods of standardisation.

Information on calculating confidence intervals for standardised rates can be found in the Association of Public Health Observatories (APHO; now part of PHE) technical briefing 3: Commonly used public health statistics and their confidence intervals. The accompanying excel spreadsheet tool allows you to calculate standardised rates and confidence intervals on your own data.

**Please note** that for the calculation of European age-sex standardised rates (EASRs), the original European Standard Population (ESP) introduced in 1976 has been replaced by the ESP 2013 which has an older population structure. Further details are given on the ONS website. Statistics providers across the UK, including ScotPHO, started to use ESP 2013 from January 2014. The effect of this is that for events that occur primarily amongst older people (eg all-cause deaths), the calculated EASR *increases* by changing to using the ESP 2013. Conversely, for events that happen mainly amongst younger people (eg hospital admissions for drugs misuse), the calculated EASR is likely to *decrease* by changing to using the ESP 2013. All EASRs should be clearly annotated to explain which version of the ESP has been used (but some older analyses of EASR based on ESP 1976 may not yet be annotated). When looking at time trends in EASRs, it is important to ensure that these are based on only one release of the ESP, as it is meaningless to compare an EASR calculated using ESP 1976 with one calculated using ESP 2013. Note that the numbers of events and crude rates are unaffected by the choice of standard population used.

### Confidence intervals

The APHO (now part of PHE) Technical Briefing 3: Commonly used public health statistics and their confidence intervals includes how to calculate confidence intervals for rates, proportions, means and age-standardised rates and ratios. It also describes the use of confidence intervals i.e. what they are and when they are used, and presents the PHE preferred methods for calculating the intervals for the type of statistics described. An excel spreadsheet tool allows you to calculate confidence intervals on your own data.

### Process control charts

The APHO (now part of PHE) Technical Briefing 2: Statistical process control methods in public health intelligence (1.3Mb) gives a good overview of this topic. Excel spreadsheet tools are available to download which allow you to produce funnel plots using your own data (using counts, crude rates, proportions and percentages, or directly and indirectly standardised rates).

### Health inequalities

The ScotPHO report on measurement techniques to help interpret health inequalities covers different methods for measuring inequalities in health, and has an accompanying downloadable excel spreadsheet tool which allows you to perform the different tests on your own data.

The ISD Geography, Population and Deprivation Analytical Support Team has produced a comprehensive document Deprivation measures: Guidance for analysts (82Kb). This offers background information, outlines the deprivation indices used by ISD, and recommendations are made on the choice of deprivation index to use in health analyses.

The Public Health Observatory Handbook of Health Inequalities Measurement focuses on the measurement and interpretation of health inequalities.

The Scottish Executive Measuring Inequalities Working Group report (2005) sets out recommendations made by the Working Group on Measuring Inequalities in Health in Scotland.

The Scottish Government's annual Long-term monitoring of health inequalities report includes the most up-to-date government recommendations for measuring and reporting inequalities in Scotland.

### Life expectancy and healthy life expectancy

See the references in our Healthy life expectancy section.

The English Eastern Region Public Health Observatory (now part of PHE) report INphoRM 3: Life expectancy sets out some of the issues relating to the construction, interpretation and uses of life tables and life expectancy estimates.

### Developing indicators

For a good overview see The Good Indicators Guide: Understanding how to use and choose indicators (535Kb) produced by APHO (now part of PHE).

### Target setting

The APHO (now part of PHE) technical briefing 4: Target setting in a multi-agency environment looks at key issues to consider when setting targets in a multi-agency environment, including the choice of appropriate methodologies, indicators and statistics, and consideration of the wider political and ethical context.

### Health Impact Assessment (HIA)

Information about HIA and a large number of HIA reports are available at the PHE HIA gateway.

### Small area modelled (or synthetic) estimates

See APHO (now part of PHE) technical briefing 8: Prevalence Modelling for details around the need for prevalence modelling, and an overview of various methods of generating estimates of the prevalence of diseases or risk factors in local populations.

Local prevalence of health behaviours can be modelled mathematically from national surveys. This has been done in England for a range of behaviours in their local health profiles, and in Scotland for tobacco smoking. Modelling is a low-cost way of producing local estimates for small areas across the whole country since it makes use of existing survey data and requires no new data collection. However, there are substantial limitations which need to be considered when interpreting the estimates, and which mean that modelled/synthetic estimates can't be used for monitoring the effectiveness of interventions.

### Other methodological resources that may be of interest:

Further technical briefings from Public Health England on:

- Measuring sustainable development
- Geodemographic segmentation
- Using small area data in public health intelligence
- Measuring smoking prevalence in local populations

The ScotStat Analytical Network electronic library of analytical guidance material.

The BMJ have published on:

The Evidence-Based Medicine "What is....?" series provides short bulletins explaining the key concepts behind evidence-based medicine.

### Page update:

- The last update of this page was completed in
**December 2014**. - The next review is due to be carried out
**by end December 2015**.