Healthcare and Medicine Reference
In-Depth Information
Attributable risk (risk difference, absolute risk reduction) is the dif-
ference between the rate of events in a group of interest (exposed to a
noxious or beneficial factor) and the rate of events in a nonexposed group.
The rationale behind an attributable risk is that in the case of a multifacto-
rial cause (web of causes) of the disease or health event shared both by the
exposed or nonexposed group, the difference is due to the factor in ques-
tion, the rest of the web of causes, and their consequences being the same
for both groups compared. In our cohort study: 80/100,000 20/100,000 =
60/100,000.
For example, in searching for cause(s) of a foodborne infectious disease,
we want to detect the most probable contaminated dish from among many,
eaten in various combinations, by a group of people. By establishing an
attributable risk (risk difference) for each meal as consumed or not (attack
rate of disease in this dish's eaters compared to such an attack rate in non-
eaters), we may consider as the most probable vehicle of infection the dish
showing the greatest attributable risk among them all. Specificity of causal
association from one meal to another is shown by the biggest risk difference;
relative risks show its strength and etiological fractions (see below) show
the role and predominance of each dish in a given web of causes (all dishes
under consideration as potential sources/causes of infection in the entire
study). 15,39 Nosocomial (hospital) infections or other outbreaks of events with
underlying sets or webs of causes may also be analyzed this way.
Etiological fraction (attributable benefit fraction) is a proportion of
events in exposed subjects, which is due to the factor of interest from the
web of causes of the health problem under study. It is estimated either
from the proportion of the attributable risk from the total of the risk in the
exposed group (in the cohort or longitudinal studies) yielding an attributable
risk percent or as an odds ratio minus one (in case-control studies), attrib-
utable odds . These computations provide some insight into the specificity of
one causal factor and its prevalent role in the web of causes. We presume
that other causes form the web of causes are similarly present and mani-
fested in the exposed and unexposed groups of interest. The closer the etio-
logical fraction is to 100%, the more specific (exclusive, prevalent, dominant)
the role of this factor is among all other causal factors under consideration.
In our cohort study,
80 /100,000 20 /100,000
80 /100,000
×=
100
75%
Search Pocayo ::




Custom Search