> The issue is how to apply the "evidence" that we find in our searches
> to a single patient...
I suggest the role of evidence-based practice is to filter for the best available evidence, then present it in a meaningful manner. In terms of presenting statistical evidence that relates to an individual rather than a population, a so-called "five number summary" common in the business statistics world (viz. MBA or MPA program teachings) might be worth considering. Measures of central tendency do describe the average; confidence intervals do describe the precision with which we can estimate average tendencies. However, individuals exhibit a wider range of responses. Box plots and five number summaries can be used to show the extreme low value, 25th percentile value, median value, 75th percentile value, and extreme high value of the data. As an epidemiologist faced with a serious personal health issue, I found solace in knowing that while median and lower half survival statistics were grim, the upper end was skewed by individuals with 100% survival over long periods of time. That not only gave hope, but was a clue that most case series data sets were inadequately stratified to be predictive of all cases for that particular relatively rare disease. Subsequent search for a better risk index confirmed that belief, and outcome for the patient concerned was very positive.
David.
--
David Birnbaum, PhD, MPH
Adjunct Professor
School of Nursing
University of British Columbia
Principal, Applied Epidemiology
British Columbia, Canada
|