Prevalence Predictions

I am grateful for an email pointing me to some work done to try to predict the prevalence of QOF related diseases. The prediction is based on the age, sex, ethnicity, deprivation and smoking status of a practice's population. Readers with long memories may remember something similar being done by the North East Health Observatory a few years ago. This new work, however, covers more disease areas and is rather easier to use as the predictions are now featured on NHS Comparators (that link only works on NHS computers, sorry)

The methodology is interesting as the "expected" prevalence is based on various household surveys rather than GP data. The national prevalences are all rather lower than the "expected" prevalences except for cancer. The high relative prevalence of cancer in the QOF data is likely due to the predictions being based on one year and QOF registers being based on several years.

It is possible to argue with the methods of determining prevalence. It is not entirely clear that it is desirable that medically recorded prevalence should be the same as some of these surveys. Data about pratice populations is generally limited to age and sex and so various assumptions and approximations have been used in the model with data from other sources. Whilst there is some validity in these objections they do not justify writing off this work. For the first time at an accessible national level there is an attempt to produce corrected prevalence figures for practices. Comparison of one practice with another is still not simple but it is a little simpler. It is only thing that has made me log on to NHS Comparators recently.

No comments: