Comparative analysis of local strategies to tackle health inequalities.
A research project of the School of Applied Social Sciences.
There are wide variations across Primary Care Trust (PCT) areas in England with making progress towards the Government’s 2010 national health inequality targets. These targets are focused on the Spearhead Group of 70 local authority areas in the bottom fifth nationally for three or more of: male and female life expectancy, premature mortality for cancers and for cardiovascular disease, and the Index of Multiple Deprivation. The Spearhead areas map onto 62 PCTs following the recent NHS reconfiguration. For life expectancy, only eight areas are on track to meet both the male and female targets of a 10% narrowing of the gap with England by 2010. Twenty-eight areas are off-track for both the male and female targets. Underlying these differences in life expectancy trends are striking contrasts in progress with reducing premature mortality due to cancers and circulatory diseases, major contributors to the life expectancy gap. For example, in one ex-coalfield Spearhead area in County Durham, premature mortality from cancers declined by 25% between 1995/97 and 2003/05, while in another ex-coalfield Spearhead area in the same county the decline was 10% (www.fti.neighbourhod.gov.uk). The effect is that in the former the absolute and relative gaps with the national average have narrowed while in the latter they have widened.
The Department of Health’s response to the evidence about the variation in progress towards the 2010 targets is to encourage Spearhead PCTs and their partners to focus particularly on preventing deaths from early middle age due to cardiovascular disease, cancer and respiratory disease. High impact interventions are recommended, especially increasing smoking cessation services and proactively case-finding and treating high blood pressure and cholersterol among at-risk groups. All the Spearhead PCTs have been issued with the trajectories in all-age all-cause mortality necessary for them to meet the 2010 targets, with the expectation that they will plan their interventions to meet these trajectories. However, there is little evidence about what might be behind the differential performance of Spearhead areas because no systematic research has been undertaken that compares each area as a ‘case’. It is not clear, for example, whether the Spearheads missing the targets have been slow to build capacity in smoking cessation clinics, whether their primary care organisations have been reactive rather than proactive with secondary prevention, or whether other factors to do with ways of working, leadership or local context are behind these differences.
The project is funded by the following grant.
- Comparative Analysis Of How Local System (£198656.80 from NHS Service Delivery & Organisation R&D Programme)
The aim of this project is to establish what attributes of Spearhead areas are associated singly or in combination with better or worse outcomes regarding progress with narrowing health inequalities. There are seven objectives:
1. To establish the recent trend in the relative gap between each Spearhead area and the national trends for premature mortality from circulatory diseases and cancers and for teenage conceptions. These are the three outcomes with which the project will be concerned.
2. To derive a theoretically, empirically and practically informed list of attributes and descriptors regarded as likely to be associated with these outcomes.
3. To survey every Spearhead PCT in England to gather data on these attributes.
4. To analyse these data using Qualitative Comparative Analysis (QCA), a technique that enables patterns of association between attributes and outcomes to be identified.
5. To undertake a longitudinal analysis using existing QCA data for North West England to explore causal attribution in more depth.
6. To explore causation further by engaging practitioners in workshops to discuss results from the analysis.
7. To enable PCTs and their partners to use the results to identify how they might improve and from which PCTs they might learn in order to make more progress with tackling health inequalities.
This project applies Charles Ragin’s method of Qualitative Comparative Analysis (QCA) to resolve this by treating outcomes as the product of causal combinations of attributes. QCA has been developed with policy applications in mind. It enables necessary conditions to be identified from logical statements describing the different combinations of conditions that are sufficient for a given outcome. Software such as fsQCA and Tosmana is available to analyse data constructed as a series of attributes and their states for each case, together with a value for an outcome for each case. Contrary to many conventional approaches to multivariate analysis, QCA does not regard the impact of a causal condition on an outcome to be the same regardless of the state or level of other causal conditions, a simplifying assumption unlikely to reflect the realities of how outcomes from policy programmes are produced.
The basic analytical unit in QCA is not the ‘variable’ but the ‘configuration’ and the cases conforming to each configuration. A simple example is as follows. There are three cases, the local areas, in two of which, cases 1 and 2, the gap in premature deaths from circulatory diseases relative to the national average has narrowed and in one of which, case 3, the gap has widened. Thus, the outcome has two states – narrowed or widened. Three attributes are then considered. Attribute ‘A’ is a comprehensive programme of secondary prevention (present=A/absent=a); attribute ‘B’ is robust estimates of people with symptomatic CHD in the local population (achieved=B/not achieved=b); and attribute ‘C’ is deprivation (high=C/not high=c). The attributes configure for cases 1 and 2 as A, B, C and A,B,c respectively. For case 3 they configure as a,b,c. From this we can deduce that the gap narrows with a combination of comprehensive secondary prevention and robust estimates of CHD cases, regardless of whether deprivation is high or not high. Comprehensive secondary prevention is necessary but not sufficient for the gap to narrow; it needs to be combined with good estimates of CHD cases. In reality we would have more cases and more attributes, and it is possible to work with attributes that have a range of states (such as very high, high, medium, low and very low, either using Tosmana, which works with ordinal variables, or fuzzy sets in fsQCA, where attribute membership is based on the extent to which a case is in or out of a category). With a larger number of cases patterns of necessary and sufficient causation can be discovered with more confidence, although in probabilistic terms. Given that these are associations, there is a need to have good theoretical grounds for regarding attributes as causal conditions. This proposal, however, also has a longitudinal dimension because QCA data already exist for North West England, enabling a time series to be constructed for this region that can be used to explore causal effects more confidently.
All Spearhead PCTs in England and their partner organisations represented on local health partnerships were invited to participate in the study, with initial contact being organised through SHAs and Government Offices. Meetings were held with regional Directors of Public Health to brief them on the project and engage their assistance with approaching the PCTs and health partnerships. An analysis has been undertaken of the most recent health inequality trends, based on three year rolling averages for each Spearhead local authority area over the past 5 years in the relative gap for under-75 deaths in circulatory diseases and cancers and for teenage conceptions. These variables have been coded categorically as narrowing or not narrowing, forming the outcomes for the QCA exercise.
Four sources were used to identify attributes considered relevant to the outcomes. Firstly, a review was undertaken of the research evidence on what impacts at population level on the three outcomes in terms of organisational attributes, programme attributes and contextual attributes, using the NICE web site, Medline, renewal.net and other sources identified during the course of this stage of the project. Useful reviews already exist, such as in the case of teenage conceptions the national evaluation of the Teenage Pregnancy Strategy (Parker et al., 2005). Also relevant were reviews of other national programmes, such as Health Action Zones (Barnes et al., 2005) and the Neighbourhood Renewal Strategy (GFA Consulting, 2005). Secondly, practitioner stakeholders were consulted in three workshops held in Leeds, Birmingham and London using a mini-Delphi technique to identify key attributes. Thirdly, the current ESRC project led by Blackman investigating how health inequalities are being tackled in local areas of England, Scotland and Wales provides a source of interview data relevant to this proposal (http://www.publicservices.ac.uk/). Over 200 interviews have been completed with practitioners working locally at a strategic level; these have been coded using NVivo, enabling them to be searched for possible attributes. Fourthly, some of the attributes identified through a more rapid process for Blackman’s GO-NW study were included.
Some attributes were assessed for each area from secondary data sources, such as the level of deprivation. Most have been self-assessed by the local lead for the outcome in question, using a confidential questionnaire and scales of achievement normally categorised as less than basic, basic, basic/good, good, good/exemplary and exemplary. These categories were presented with descriptors to guide the self-assessment, and each self-assessment will be evidenced by requesting the respondent to provide an example to support it. Candid responses were encouraged by respondents not knowing how their area has been categorised for the outcome, a guarantee of confidentiality, a guarantee of anonymity until voluntarily surrendered if the PCT and partners subsequently agree to be grouped into learning partnerships based on the results, and emphasising the mutual learning that will be possible from having as complete and accurate a set of data as possible.
The questionnaires were sent to named contacts for each local area, with a separate questionnaire for circulatory diseases, cancers and teenage conceptions. There was a telephone helpline for any queries arising from the questionnaire and regular follow up and reminders over the four month period that was allowed for questionnaires to be completed and returned.
On return, the questionnaires were checked and any queries followed up by telephone with respondents. The data has been coded for entry to SPSS. The sub-set of attributes selected for QCA have been based on an initial exploration of the data using crosstabulations of attribute values and outcomes.
Respondents were asked to assess attributes as the situation existed three years ago as well as for the present day. The assessment for three years ago is likely to be a more valid indication of how the attributes affect outcomes, given the latter will be measured using mortality and conception data trends where the most recent data point will be up to two years in the past.
The analysis will result in tables showing how the attributes combine for each (anonymised) case and with what effect on outcomes. The patterns will be considered in terms of the influence of both local contextual attributes and process attributes, as well as their combinations. The will include considering how assessments of the attributes differ for the present compared to three years ago, and the implications of this for reaching the 2010 health inequality targets and progress beyond this date. The analysis will also provide an evidence-base to inform the development of focused learning partnerships with PCTs.
Not yet available