A.J.O'rourke posted:
> As part of some model answers I am writing for a masters course, the
>following question has come up:
>
> Where would the results of an audit sit in the "hierarchy of evidence"?
>
> My own feeling is that most audits fall into the category of quasi-
>scientific studies, with prospective audits, where the staff set out to
>collect specific focussed data having more validity than retrospective
>projects, which hope they can extract adequate data from records collected
>for quite different purposes. Maybe a prospective audit would have the
>standing of say a cohort study; a retrospective audit that of a
>case-control study.
Approach I take with my students is to create a 2-way table. One axis deals
with when observations are taken relative to an exposure or intervention
(just after, both before & after, or at multiple points in time). The other
axis deals with comparison groups (either none, or two or more with
non-random allocation, or two or more with random allocation). Since an
audit is really just a survey of certain types of individuals to see if
they've been treated in an expected manner, it is an ex-post facto
description (case series) type study in the top left cell (observation just
after, no comparison groups) or perhaps a simple before-and-after
description (observations taken before and after some intervention but no
external comparison group). This approach neatly divides quasi-experimental
or observational designs from experimental study designs (based on whether
comparison groups are allocated by randomization or not), and provides a
continuum: Designs toward top left are easiest to conduct but most
susceptible to bias and distortion; designs toward bottom right are most
difficult to conduct but least susceptible to bias and distortion. Cohort
and case-referent designs sit in the middle cell (before & after
observations, two or more non-randomly allocated comparison groups), but
since audits typically don't involve comparison groups I wouldn't agree with
your last premise.
David Birnbaum, PhD, MPH
Adjunct Professor
Dept. of Health Care & Epidemiology
University of British Columbia, Canada
|