Greetings ALLSTATERS from the Royal Infirmary of Edinburgh.
I'm involved with the Scottish National Stroke Audit at the Royal. NHS Quality Improvement Scotland require us to report monthly data on aspects of the process of stroke care e.g. proportion of acute stroke patients brain scanned within 48 hours of admission to hospital, proportion of ischaemic stroke patients administered aspirin within 48 hours of admission, proportion of acute stroke patients receiving care in an acute stroke unit etc. At the Royal we have been following the late John Tukey's advice that "Display is an obligation!" and complementing the tabular monthly summaries with Shewhart Individuals (X) charts of the monthly proportions.
The charts have been welcomed by the stroke clinicians and have demonstrated clearly the impact on the process of stroke care of changes to the structure of the stroke service e.g. the provision of new faster scanners has led to an increase in the proportion of patients brain scanned within 48 hours of admission and a consequent increase in the proportion of ischaemic stroke patients administered aspirin within 48 hours of admission.
We have TWO queries.
QUERY 1
For the two-year period July 2002 to June 2004: -
At Hospital A an Individuals chart of the monthly proportion of ischaemic stroke patients discharged on a statin yielded no signals of special cause variation. Thus we concluded that the proportion was stable and predictable at 70%.
At Hospital B there were many signals on an Individuals chart of the monthly proportion of ischaemic stroke patients discharged on a statin but the appearance of the chart suggested that the proportion had increased and then levelled off at around 70%.
If we create the chart for Hospital B with the centre line set at 70 is it sensible to say that when we reach the point in time where the chart ceases to yield signals we can say that at that time clinical practice at Hospital B "came into line" with that at hospital A?
Alternatively can we arbitrarily split the data for Hospital B into Phase 1 and Phase 2 and chart each phase separately and explore to find a split that yields no signals in Phase 2 and take that as an indication of the point in time when clinical practice in this respect stabilised at Hospital B?
QUERY 2
We are familiar with a paper by Benneyan JC, Lloyd RC and Plsek PE entitled Statistical process control as a tool for research and health care improvement. Quality and Safety in Healthcare. Dec 2003;12(6):458-464. They state "An advantage of SPC is that classical statistical methods typically are based on "time static "statistical tests with all data aggregated into large samples that ignore their time order-for example, the mean waiting time at intervention sites might be compared with that at non-intervention sites. Tests of significance are usually the statistical tool of preference used to see if one group is "significantly different" from the other. These are useful methods and have good statistical power when based on sufficiently large data sets. The delay in accumulating a sufficient amount of data, however, often limits the application of these methods in practice in health care and practitioners may resort to simple bar charts, line graphs, or tables to present the data. In this case the practitioner can only make a qualitative statement about whether or not there "seems" to be an improvement.
In contrast, SPC methods combine the rigour of classical statistical methods with the time sensitivity of pragmatic improvement. By integrating the power of statistical significance tests with chronological analysis of graphs of summary data as they are produced, SPC is able to detect process changes and trends earlier. While this may be a less familiar branch of statistics to many researchers, it is no less valid. SPC also distils statistical theory into relatively simple formulae and graphical displays that can easily be used by non-statisticians.".
We would be particularly interested in receiving any comments on the claim that SPC is able to detect process changes and trends earlier and/or details of any relevant published work.
Best Wishes
Robin Henderson
Stroke Audit Coordinator
Royal Infirmary of Edinburgh
**********************************************************************
The information contained in this message may be confidential or legally privileged and is intended for the addressee only, If you have received this message in error or there are any problems please notify the originator immediately. The unauthorised use, disclosure, copying or alteration of this message is strictly forbidden.
**********************************************************************
|