A common problem for a clinician is to decide if a change in a laboratory test value is significant or not.
Sources of variability in a laboratory test result:
(1) change in patient due to disease
(2) change due to an intervention or manipulation (massaging the prostate, etc.)
(3) changes due to physiologic variability (circadian rhythm, etc.)
(4) changes due to diet, drugs or hydration
(5) change due to specimen collection, handling and transportation
(6) change due to variability in the testing itself
To decide if a change in a test value may be significant vs may just reflect analytical variability:
change in test value =
= (current test value) – (previous test value)
decision level =
= 1.96 * SQRT(2) * (standard deviation for test) =
= 2.77 * (standard deviation)
To simplify the calculation, 2.77 is usually rounded to 2.8 or 3.0.
If the absolute value for the change in test value is < decision level, then the change could be explained by variability in the analytical method.
If the absolute value for the change in test value is >= the decision level, then the change cannot be explained solely by variability in the analytical method. IF ALL OF THE OTHER SOURCES OF VARIABILITY ARE MINIMAL, then the change can be ascribed to a change in the patient.
NOTE: The calculations are similar to the reliable change index (above).
If the standard deviation for the test is variable, then the usual standard deviation (USD) can be used (Copeland). This is the mean of the interval standard deviations.
If the laboratory does not provide the standard deviation for the test, it can be estimated if it based on the mean value +/- 2 standard deviations:
estimated standard deviation =
= ((high end of reference range) – (low end of reference range)) / 4
(1) Do not make a crucial decision on the basis of a single laboratory value alone.
(2) Monitor trends (slope) in a series of values.
(3) Know and record what is happening to the patient.
(4) Standardize as much as possible.
(5) Use a laboratory that is consistent in its precision and accuracy.
To read more or access our algorithms and calculators, please log in or register.