Healthcare and Medicine Reference
success of intervention). Some, like risks, are important for primary preven-
tion. Others are important for different levels of health intervention such as
prevalence for secondary prevention or prognostic frequencies and fractions
beyond primary prevention.
220.127.116.11 Quantifying Our Uncertainties
An absolute certainty in medicine is rare. Our experience, understanding,
and decision making vary due to our previous experience and doings, differ-
ent patients, and settings in which health phenomena happen. Such uncer-
tainties are worth measuring, interpreting, and using for practical decisions.
We are well aware that all our observations, analyses, and studies are
subject to various random and systematic errors. Repeating studies of herd
immunity (what proportion of the population, “herd,” is immune after an
immunization program) or integrating results from clinical trials yield ranges
of results. Errors, samplings, detection, counting, health professionals' activi-
ties, and clinical practices may all contribute to the variety of findings from
one study and population to another. So, can we be certain that a finding of
70% of herd immunity is appropriate to guide us to implementing an immu-
nization program or not?
For all these and other reasons, we make decisions on the basis of point
and interval estimates.
Point estimates provide an idea of “exact” frequencies, proportions, rates, inci-
dence, and so on, such as 70% in the case of our herd immunity study (the term
“herd immunity” is used in infectious disease epidemiology for the proportion
of subjects in a given community who are resistant to the disease of interest).
Interval estimates , for example, 65%-75%, offer ranges within which a
“true” value lies. More precisely, they are intervals within which the true
estimates would fall given the variations of such estimates that might be
seen in multiple studies of the same problem in the same group of patients
or community. 95% or 99% CI specify our certainty regarding where our
observations might be found.
The range of observations also counts. If multiple studies indicate a herd
immunity range of 50%-90% (too wide in this example for the sake of
explicitness), a true value within such an interval may have practical implica-
tions. If we know in the case of a particular infectious disease that a herd
immunity of 80% is enough to make immunization unnecessary, in the case
of a true value of 60%, immunization would be necessary and in the case of
85%, it would not.