The DIAL study: Deaths Involving Alcohol In Wales

Authors: Rob Poole, Catherine A Robinson, John Bailey, Charles Shelton, Susan Ruben

Funded By: Welsh Assembly Government

Started: 2010

Finished: 2011

Contact: John Bailey

Document Download »

Summary

The study used records in six Coroners’ jurisdictions in Wales as the primary source of data.  The participating jurisdictions covered urban and rural environments within North, South and Mid-Wales. Our sample includes former residents of the catchment area of five of the current Welsh Health Boards. The study examined deaths that were reported in 2008.

An iterative and heuristic process was used to identify DIAL cases. The assignment of cases to three categories, DIAL case, DIAL borderline case and non-DIAL case, was made using probabilistic criteria, on the basis of judgements made by a research team with medical, sociological, psychological and risk assessment expertise.  All judgement processes, from primary screening to final categorical assignment, were subject to within-team validation. All judgements on clinical issues were subject to medical review.

Just under 6,000 files were subjected to primary screening. 858 deaths were selected for further scrutiny because of an initial indication that the deceased might have misused alcohol.  Data was collected for comparison on a further 502 deaths related to falls, fires, drivers in road traffic accidents and self-harm in the absence of evidence of alcohol misuse.

The secondary process of application of DIAL criteria was subject to multi-disciplinary assessment, involving a minimum of three team members. All complex or uncertain cases were subject to full team review. As in the primary screening process, it was possible to develop robust and transparent rules that could be applied.

Within the “achieved” sample (i.e. those cases where full data collection was possible), 384 deaths subject to application of DIAL criteria were categorised as definite DIAL cases. There were 26 DIAL borderline cases. A lack of information was the usual cause of uncertainty, though in a handful of cases, cause of death was complex or unknown. The role of alcohol in contributing to death was usually unambiguous in DIAL cases, and there were few, if any, disagreements when these cases were subject to joint team review.

 

Conclusions

  • Coroners’ data is rich and useful, but there are practical challenges for researchers working in Coroners’ offices. Variation in criteria for Inquests and in data collection and storage makes this an unsuitable source for routine epidemiological data on this subject. However, each office shows consistency of practice over time, and selection of particular jurisdictions allows the collection of important qualitative and trend information.
  • This study shows that the DIAL probabilistic approach to causation, when applied to Coroners’ data in Inquest cases, can be reliably and practically applied by non-medical researchers with medical support to ascertain cases where alcohol has played a significant causal role in death.
  • Our findings show that the DIAL method generates important information about lifestyle, patterns of drinking and premature deaths. It provides a mechanism to produce evidence regarding patterns of alcohol consumption, life circumstances and mortality that could not be ascertained using established methods.

< Return to project list