Multiple imputation for completion of a national clinical audit dataset

The Myocardial Ischaemia National Audit Project (MINAP) is a register of heart attacks covering 234 acute admitting hospitals in England and Wales. It is used to assess the extent to which hospitals are attaining the government targets for patients with heart attacks (myocardial infarction). MINAP is therefore of national importance in coronary care and of potential international importance for research. As with most observational databases, there is missing data in MINAP, which has the potential to bias statistical analyses. In this paper, we use multiple imputation to reduce the impact of missing data and we give details of how our imputation scheme was implemented. The key contribution of this paper is the provision of multiply completed datasets, suited to a range of analyses, that can be used to make efficient inferences without the distractions of missing data. Our work will assist MINAP in achieving its priority goal of providing useful data with which to analyse patient care.