First steps in demonstrating an impact on health service use of the Welsh Government Supporting People homelessness prevention programme using linked administrative data

ABSTRACTObjectivesTo assess the feasibility of using linked administrative data to deliver a quantitative component to the evaluation of the national homelessness prevention programme in Wales, Supporting People. Specifically, to establish whether any impact of Supporting People on the use of health services could be demonstrated. Approach: Data on the Supporting People programme was acquired from two pathfinder Local Authorities in Wales and anonymously linked to the routine health records held in the SAIL Databank at Swansea University. Any bias in linking rates was examined and health events analysed for the period 12 months before and 12 months after the point when homelessness prevention support began (the ‘crisis point’). The study evaluated a range of possible control groups that could be created using administrative data and undertook some preliminary comparative analysis. ResultsThe results of the study are due to be published on 10th March 2016 so detailed results will be available for August. We will briefly explore the unique contribution of data linking to researching the hard-to-reach and/or vulnerable groups supported by the intervention (e.g. people with substance misuse problems, women experiencing domestic violence and people with a criminal offending history) and the challenges of delivering a quantitative evaluation using linked administrative data. Findings will be reported on linking rates including any potential areas where particular service groups were relatively less well represented. Patterns of the following health events will be reported around the ‘crisis point’: i) number of days when GP events occurred; ii) number of Au0026E visits and iii) number of emergency hospital admissions. Interpretations are made of consistent patterns of health events across different service user subgroups. The GP diagnosis codes, GP prescribing codes, reasons for visiting Au0026E and reasons for emergency hospital admissions that showed the greatest decrease after the intervention will be reported. A brief discussion will be presented of the suitability of different control groups and the findings of the preliminary comparative analysis. ConclusionsUsing primary research methods, demonstrating the impact of a homelessness prevention programme on outcomes like health service use represents an insuperable challenge. The use of data linking data delivers a challenging but potentially robust method to demonstrate impact and the feasibility study was able to take the first steps towards demonstrating improved health outcomes for individuals.