More information about text formats
We write to highlight some questions which arise from the recently published paper entitled “How do time trends in in-hospital mortality compare? A retrospective study of England and Scotland over 17 years using administrative data” by María José Aragón and Martin Chalkley in BMJ Open 2018;8:e017195.
Seeking to understand variations in hospital associated mortality is a worthy endeavour and we welcome the authors’ contribution. They are clearly aware of the importance of case mix adjustment when studying comparative mortality, but may not have been able to take account of differences in the way hospital activity data are collected in the two health care systems.
The recording of comorbidity in English HES data may be more complete than in our SMR01s. The reason for this is the financial incentive to fully cover background risk in England. Whilst Scottish data are getting better, the practical consequence of this for the York study is that they will have underestimated the risk of death for our patients, thereby increasing the relative mortality parameters used to calculate the risk adjusted trends used in the plots shown.
The paper suggests “If for example alternative settings to which terminally ill patients can be discharged have expanded faster in England than in Scotland, we would observe the kind of differential trend of in-hospital mortality established by our analysis. The second, more worrying possibility is that there remains some element o...
The paper suggests “If for example alternative settings to which terminally ill patients can be discharged have expanded faster in England than in Scotland, we would observe the kind of differential trend of in-hospital mortality established by our analysis. The second, more worrying possibility is that there remains some element of the difference in trend that relates to the efficacy of hospital treatments in the two countries.”
In Scotland, we have focussed on health and social care integration. The aim of this is to ensure care is delivered to patients in their homes or a homely setting wherever possible, with hospital care reserved for only the most unwell. This would explain the observed higher rate of mortality for hospitals, but not reflect the improved care being delivered early in the patients journey or in the community settings.
The York study has raised important questions about comparative hospital mortality in England and Scotland, but does not allow us to draw strong conclusions about the variation we see. We would be very keen to collaborate to advance our understanding of these trends.