|
Overview :
A clinical study may evaluate those with and without disease
using a test whose results are ranked into multiple categories or levels. The
distribution of the results for each group can be interpreted in a number of
ways. One method involves estimating the receiver operating characteristic
(ROC) curve, while another is to look at the rates shown by each group.
Assumptions: The rates shown at each level for each group
are constant.
Requirements:
(1) Study groups large enough to provide reliable estimates.
(2) Intervals defined to allow for optimum resolution.
percent of people in study group showing a value in a given
range =
= (number of people showing value within range) / (total
group size) * 100%
This can be transported to a general population based on the
prevalence of the disease.
number of people with disease =
= (prevalence in percent) * (population size)
number of people without disease =
= (population size) – (patients with disease) =
= (100% - (prevalence)) * population size
number of people with disease in general population expected
to show a result in a given range =
= (percent of people with disease in study population
showing a value in a given range) * (number of people with disease)
number of people without disease in general population
expected to show a result in a given range =
= (percent of people without disease in study population
showing a value in a given range) * (number of people without disease)
percent of people with a result in a given range who have
the disease =
= (number expected with disease) / ((number expected with
disease) + (number expected without disease)) * 100%
|
percent of people in given range
(category) with disease
|
Interpretation of Result Relative to
Diagnosis
|
|
100%
|
patient diseased
(diagnostic)
|
|
> 80%
|
patient probably
diseased (supports)
|
|
20 – 80%
|
patient may or may
not have disease (not informative)
|
|
< 20%
|
patient probably
not diseased (does not support)
|
|
0%
|
patient not
diseased (excludes the diagnosis)
|
|