Article Text

Download PDFPDF
  1. MS Ahmad1,
  2. J Anius1,
  3. A Idowu1,
  4. S Ahmad2,
  5. T Kashem3
  1. 1Emergency Department, BHR Hospitals NHS Trust, Romford, Essex, UK
  2. 2Holy Family and Red Crescent Medical College and Hospital, Dhaka, BANGLADESH
  3. 3Bart's Health NHS Trust, London, UK


    Objectives & Background Although the Summary Hospital-level Mortality Indicator (SHMI) has been in use since 2011 as a performance indicator at a trust level, there seems to be little evidence of standardising emergency department (ED) mortality. To achieve this, a greater understanding of the patterns of ED mortality, and the factors that influence it, must be undertaken. The objectives of this retrospective observational study was to establish whether a correlation existed between the patterns of deaths in the ED, the Trust it belonged to, and the wider community it served. Ours is one of the largest and busiest acute hospital trusts in the country, serving a population of 750,000 that is socially and ethnically diverse. Two null hypotheses were proposed; firstly that there was no correlation between ED mortality and Trust mortality and similarly, there was no correlation between ED mortalilty and mortality in the community.

    Methods Data for the ED was compiled from Symphony, the department's electronic patients records. Trust figures were obtained from the Information Governance. Mortality figures from the three surrounding boroughs were sourced from the Office for National Statistics. Data was retrospectively compiled from the start of the financial year in April 2009 to March 2015. Trust figures were exclusive of ED deaths. A total of 1868 deaths were observed in the ED, 14246 in the Trust and 31358 in the community. These were then compared. ED deaths were multiplied by 10 for visual comparison.

    Results Despite a rising trend in the community, ED and Trust deaths are on the decline. Peaks are noted in both groups during the winter months, highlighted by the blue bars. Deaths across the community and the Trust show a unimodal distribution with a strikingly similar positive skew. The mortality distribution of the ED shows a bimodal distribution suggesting two distinct sub-populations. Nonetheless, the Pearson correlation coeffecient was found to be reasonable (ED and Trust r=0.563, P<0.001; ED and Borough r=0.454, P<0.001).

    Conclusion Despite the difference in distribution, there is a significant positive correlation between all three datasets, thus disproving both null hypotheses. It is clear that surges in winter are part of the normal mortality distribution. A further analysis will therefore need to be made in regards to the possible second sub-group affecting ED mortality.

    Figure 1

    Frequencies of deaths over a 6 year period with trends shown in a linear fashion. To match visually, ED deaths have been multiplied by a factor of 10. The winter months are highlighted in blue, and all three datasets reveal peaks during these periods.

    Figure 2

    A unimodal distribution of death in the community showing a positive skew.

    Figure 3

    Surprisingly, the distribution of deaths in the trust reflect that of the wider community with the same pattern of shift to the right.

    Figure 4

    The first histogram reveals the standard distribution of ED attendances despite winter surges. The second histogram displays a bimodal distribution in ED deaths. This would suggest that there are two distinct sub-populations being assessed. It would need to be determined as to what change in processes are taking place to contribute to this.

    • emergency departments

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.