Rapid Responses to:

PAPERS: Mike Damiani, Carol Propper, and Jennifer Dixon Mapping choice in the NHS: cross sectional study of routinely collected data BMJ 2005; 330: 284 [Abstract] [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

Analysis only as good as the data

Matthew Walmsley   (7 February 2005)

Trust addresses not enough

John R. Davies   (10 February 2005)

Analysis only as good as the data 7 February 2005
Matthew Walmsley, GP Marsden Road Health Centre, South Shields, NE34 6RE Send response to journal: Re: Analysis only as good as the data	Living in the North of England, I was drawn to this article by the large red patches on the maps that covered what I think of as my local area. There appear to be large areas of the North of England that are an unacceptably long travelling time from the nearest hospital. However, looking a little closer I realised that large hospitals at Ashington, South Shields, Carlisle, Hexham, Whitehaven, Barrow, Lancaster and Preston are missing from the map. This lack of accuracy on hospital locations in the part of the country that I know does make me worry about the accuracy of hospital locations plotted in the rest of the country. The authors state that they "obtained the postcodes of all NHS trusts dealing with acute conditions". I suspect that they may have obtained a list of postcodes of trust headquaters. Unfortunately, this is an extremely inaccurate method of locating hospitals, particularly when some trusts are responsible for a number of acute hospitals many miles apart. I remember being taught that any analysis is only as good as the data on which it is based. In this article, a potentially interesting analysis appears to be based on fundamentally flawed data on hospital locations. I only hope (for the sake of all who work in the NHS north of Watford) that policy-makers do not base decisions about our long-term future on similar data. Competing interests: None declared
Trust addresses not enough 10 February 2005
John R. Davies, Consultant Anaesthetist Royal Lancaster Infirmary (LA1 4RP) and Westmorland General Hospital, Kendal (LA9 7RG) Send response to journal: Re: Trust addresses not enough	A serious flaw in this analysis may be the use of Trusts', rather than hospitals', postcodes. The location of the Morecambe Bay Hospitals NHS Trust is correctly shown at Kendal, but the Trust includes two other hospitals at Barrow-in-Furness and Lancaster, both of which are larger and offer wider facilities than the Kendal site. These two sites are both over 20 miles from Kendal. This means that access to a hospital in this area is much better than that indicated on the paper's maps as 100 minutes travel or more, and one hospital or less within 60 minutes. A more serious error is the complete omission of the North Cumbria Acute Hospital Trust, whose sites at Carlisle and Whitehaven serve the area of Cumbria coloured red, for poor access. If such errors are repeated throughout England, the conclusions, though intuitively correct, cannot be substantiated. John Davies john.r.davies@btinternet.com Competing interests: I am employed by the Morecambe Bay Hospitals NHS Trust.