Using Routinely Collected Administrative Data in Public Health Research: Geocoding Alcohol Outlet Data

Abstract

We describe our process of geocoding alcohol outlets to create a national longitudinal exposure dataset for Wales, United Kingdom from 2006 to 2011. We investigated variation in the availability of data items and the quality of alcohol outlet addresses held within unitary authorities. We used a standard geocoding method augmented with a manual matching procedure to achieve a fully spatially referenced dataset. We found higher quality addresses are held for outlets based in urban areas, resulting in the automatic geocoding of 68 % of urban outlets, compared to 48 % in rural areas. Missing postcodes and a lack of address structure contributed to a lower geocoding proportion. An urban rural bias was removed with the development of a manual matching procedure. Only one-half of the unitary authorities provided data on on/off sales and opening times, which are important availability factors. The resulting outlet dataset is suitable for contributing to the evidence-base of alcohol availability and alcohol-related harm. Local government should be encouraged to use standardised data fields, including addresses, to enable accurate geocoding of alcohol outlets and facilitate research that aims to prevent alcohol-related harm. Standardising data collection would enable efficient secondary data reuse using record linkage techniques, allowing Appl. Spatial Analysis (2017) 10:301-315

A best practice framework to measure spatial variation in alcohol availability

“…The small-area geography used in CHALICE was the lower layer super output area defined by the 2001 Census (n ¼ 1896) (Office for National Statistics, 2001). We first compiled a data set of alcohol outlets from Local Authority licensing records, described in detail elsewhere (Fone et al, 2016;Fry et al, 2016). Briefly, we located every alcohol outlet in Wales to address-level precision for each of the 24 yearly quarters during the study period from 2006 to 2011.…”

“…In the UK, these data are contained in Ordnance Survey (OS) MasterMap (Ordnance Survey, 2017) and AddressBase Premium (ABP) (Ordnance Survey, 2014) data sets respectively that are available to researchers and government organisations under license. Details on the precise locations and type of outlets can be a difficult and time-consuming process to collate (Fry et al, 2016) and may not always be available at the level of detail required. The final requirement relates to the interpretability of the AOD measure.…”

“…The data requirements reflect the availability of topographic, outlet and population data. (Fry et al, 2016) and may not always be available at the level of detail required. The final requirement relates to the interpretability of the AOD measure.…”

“…Other outlet attribute data (e.g. type, opening times and trading dates) are also important to collect as detailed in Fry et al (2016). These attributes are not considered in this paper, as the aim is to focus on the spatial measurement of AOD.…”

Mapping the interview transcript: Identifying spatial policy areas from daily working practices

“…The process was complicated by the bilingual nature of Welsh place names, which resulted in some inconsistencies in spellings and usage within and between the transcripts and the gazetteer. This was exacerbated by inaccuracies introduced during transcription where the (typically non‐Welsh‐speaking) transcriber looked‐up or sometimes guessed at the spelling of less common place names, resulting in spelling errors (for a recent commentary on the issues of matching Welsh place names to OS address products, see Fry et al, ). However, there did not seem to be any evidence of conflict in the use of Welsh place names in the daily working practices of interviewees, even in English‐speaking areas of the localities, possibly because of the institutionalised nature of the Welsh language in the organisations in which the practitioners worked.…”