The challenge of assessing the impact of an intervention without a
control group is not to be underestimated and the authors are to be
congratulated on the attempt. However the choice of diabetes as the area
to be looked at has added additional complication to the analysis which
has not been completely acknowledged in the paper.
Firstly may I point out an error in the type two diabetes prevalence
chart in figure one where the scale is wrong by an order of magnitude.
Secondly the method of data extraction is not specified. A snapshot
of the medical record will give dates of diagnosis rather than dates of
entry. For instance a patient could be labeled today with a diagnosis of
type one diabetes from the first of January 2004. It cannot be assumed
from the snapshot that this code was present in, say, 2005.
This is very apparently when looking at the diagnostic codes. The
diagnostic codes changes from 2005 to 2006 to a smaller list of codes that
was largely homonymous with many of the codes that were used before. It
was predicted that, without code changes, there would be a 22% drop in
prevalence[1]. This did not occur, indeed it would appear that diabetes
continued to rise in prevalence at about the same rate as previously.
What actually happened was the most practices used their practice
systems to change their previous codes to the new ones with the same date.
The data presented here shows no sudden shift to the new codes - there is
not "step" on the graph. The conclusion that I would draw is that the data
in this study does not reflect the actual data used in 2003 and 2004 but
rather subsequently altered coding. This in turn makes drawing conclusions
about these two years rather more difficult.
1. Hippisley-Cox J, O’Hanlon S. Identifying patients with diabetes in
the QOF—two steps forward one step back. www.bmj.com/cgi/eletters/333/7570/672-a. Response to: Tanne J. Diabetes,
not obesity, increases risk of death in middle age. BMJ 2006; 333: 672
Competing interests:
General Practitioner working under QOF
I run the QOF Database website at http://www.gpcontract.co.uk
Competing interests:
No competing interests
12 June 2009
Gavin M Jamie
GP
Whalebridge Practice, Swindon Health Centre, Carfax Street, Swindon, SN1 1ED
Rapid Response:
Coding Changes
The challenge of assessing the impact of an intervention without a
control group is not to be underestimated and the authors are to be
congratulated on the attempt. However the choice of diabetes as the area
to be looked at has added additional complication to the analysis which
has not been completely acknowledged in the paper.
Firstly may I point out an error in the type two diabetes prevalence
chart in figure one where the scale is wrong by an order of magnitude.
Secondly the method of data extraction is not specified. A snapshot
of the medical record will give dates of diagnosis rather than dates of
entry. For instance a patient could be labeled today with a diagnosis of
type one diabetes from the first of January 2004. It cannot be assumed
from the snapshot that this code was present in, say, 2005.
This is very apparently when looking at the diagnostic codes. The
diagnostic codes changes from 2005 to 2006 to a smaller list of codes that
was largely homonymous with many of the codes that were used before. It
was predicted that, without code changes, there would be a 22% drop in
prevalence[1]. This did not occur, indeed it would appear that diabetes
continued to rise in prevalence at about the same rate as previously.
What actually happened was the most practices used their practice
systems to change their previous codes to the new ones with the same date.
The data presented here shows no sudden shift to the new codes - there is
not "step" on the graph. The conclusion that I would draw is that the data
in this study does not reflect the actual data used in 2003 and 2004 but
rather subsequently altered coding. This in turn makes drawing conclusions
about these two years rather more difficult.
1. Hippisley-Cox J, O’Hanlon S. Identifying patients with diabetes in
the QOF—two steps forward one step back.
www.bmj.com/cgi/eletters/333/7570/672-a. Response to: Tanne J. Diabetes,
not obesity, increases risk of death in middle age. BMJ 2006; 333: 672
Competing interests:
General Practitioner working under QOF
I run the QOF Database website at http://www.gpcontract.co.uk
Competing interests: No competing interests