Zeiss HUMPHREY 720i, User Manual

Page: 135 / 328

STATPAC ANALYSIS & PRINTING RESULTS 7-19
REV. B 02/03 PN 51680-1 HUMPHREY FIELD ANALYZER
II -
i
series
Modified linear The STATPAC linear regression of mean deviation, accompanied by the message “MD Slope
regression analysis Significant” or “Not Significant” appears on the Glaucoma Change Probability Analysis
printout as well as the Change Analysis.
STATPAC modifies this regression analysis whenever marked learning effects are present. When
at least five tests are chosen for analysis, the modified STATPAC regression analysis discards the
first test result in a series if its mean deviation is significantly out of line with and worse than
the trend shown in later tests (p < 5%). If the modified method is used, the value shown is
labeled “Modified MD Slope”.
When STATPAC determines that a first test result should be omitted from its calculations to
correct for learning effects, it also automatically discards this test result from the calculation of
the probability plots.
On statistics and When considering the probability statements in this statistical package, it is important to be
probability conscious of what they do and do not mean. They are an aid to interpretation, not a diagnosis.
The doctor’s judgement is still the most important element in determining the clinical
significance of perimetric findings.
The probability statements are based on the distribution seen in the normal population. Saying
that less than 5% of the normal population deviates from the norm by a certain amount means
just that and nothing more. It does not mean that there is only a 5% chance that the result is
normal.
The positive predictive rate depends, of course, on the prevalence of defective fields in the
population studied. The probability that a given result is abnormal depends on the relative
prevalence in the population of defects caused by disease versus the prevalence of the same
field “defect” in normals. If a certain field result is seen 5% of the time in normals, and similar
glaucomatous field defects are seen in 0.5% of the population, then the result is ten times as
likely to be associated with normality as with disease.
Certainly one should also be aware that some patients commonly seen in a clinical practice
may not meet the criteria of normality (for example, visual acuity) which had to be applied in
creating a normals data base. These patients may fall outside normal limits established in this
statistical package for reasons other than field loss, such as cataracts.
A note of caution Rules of common sense must be applied when using STATPAC. This statistical package
represents an attempt to aid the practitioner in making medical decisions. There will be
situations where it will not give the proper analysis either because of its own limitations or
because it was applied to inappropriate data. Obviously, the practitioner must bear the ultimate
responsibility for all decisions and must use STATPAC with its limitations in mind. In cases of
uncertainty, consultation with sub-specialists is often the prudent course.
7.htp 3/14/03, 3:17 PM 19