Dear James, dear Chris
Using percentage or patient years in reporting results is a
conseqence of approbriate study design.
To use the approbriate study design for your study is the most
crucial thing before you start your experiments or observations.
1) In general, percentages can be used if the observation times,
starting with the experiment, are the same for all observed
individuals (subjects). Then you can use the so called "Cumulative
Incidence" (CI) in both groups (experimental and controls). As this
is a dimensionles number (percentage) the resulting ratio,e.g. the
Relative Risk (RR) RR = CIexperimental / CIcontrols, or ARRs can
be presented also in percentages.
2) If observation times are not the same for all indivuduals, as it
is often in real life experiments (clinical settings in contrast to
laboratory) then you must know how long you could observe every
single individual. This obervation time must be entered in the
analysis. Then you use "person-time" in den denominator, you use
Incidence Rates (IR) (Events / persons x person-observation time) in
experiment and control group.
Then you CAN NOT use percentages, as the summ of person - times stay
in the denominator, and Incidence (IR) is not a dimensionless figure.
Using percentages in this case (reporting IRs) would lead to wrong
results.
However, when you have calculated the IRs for both groups
correctly, a resulting measure of association, the Relative Risk
(RR) (RR = IRexperimental / IRcontrol) is dimensionless again as the
person-times in the 2 denominators can be cancelled out. So the
Relative Risk, and ARR can be presented again as percentages even
in this "open cohort study" design.
I hope this helps to clear your not so easy "clinical epdimology"
question. If my short explanation confused you, there is
an enlightning little book on this by a leading epidemiologist of the
Harvard School of Public Health,
A.M.Walker: Obervation and Inference; ISBN 0-917227-07-7)
or the introductionary standard text on this:
Epidemiology in Medicine by C.H.Hennekens and J.E. Buring, Chapter 4,
Measures of Disease Frequency and Association; ISBN0-316-35636-0)
Yours Franz Piribauer
and
In general if tThis design
kind
of intervention
*****************************************************
how to reach Franz Piribauer:
----------------------------------------------------
any time, any place:
Mobile Tel.: +43 (664) 243 0886
Email: [log in to unmask]
Internet: http://members.eunet.at/pico
Email to SMS (Mobile Tel.): [log in to unmask]
-----------------------------------------------------------------
at PiCo office, Bernardgasse 39, A-1070, Vienna, Austria
Tel.&Fax: +43 (1) 524 6020
Internet: http://members.eunet.at/pico
-----------------------------------------------------------------
at Government Office: Paulustorgasse 4/II, A-8010 Graz, Austria
Tel: +43 (316) 877 3536
Fax: +43 (316) 877 3555
Email: [log in to unmask]
Internet: http://www.stmk.gv.at/gesundheit
================================================
Home Graz: Liebiggasse 12, A-8010, Graz, Austria
Tel.&Fax: +43 (316) 347 259
----------------------------------------------------
Home Vienna: Bernardgasse 39/30, A-1070, Vienna, Austria
Tel.: +43 (1) 526 7750
*****************************************************
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|