STATISTICAL EVALUATION OF LAB TESTS

TP = True Positive

FP = False Positive

TN = True Negative

FN = False Negative

Sensitivity = __ TP __ x 100 = The frequency of positive test results among all the true positives. A test with high sensitivity will have a low number of false-negatives; but possibly a high number of false-positives.
TP + FN


Specificity = __ TN __ x 100 = The frequency of negative test results among all the true negatives. A test with high specificity will have a low number of false-positives; but possibly a high number of false-negatives.
FP + TN


Positive Predictive Value = __ TP__ x 100 = The probability that a positive test result will be correct.
TP + FP


Negative Predictive Value = __ TN __ x 100 = The probability that a negative test result will be correct.
TN + FN


Test Efficiency = ____ TP + TN ____ x 100 = Specifies a test's overall accuracy. The probability that a test result, either positive or negative, correctly predicts the presence or absence of disease.
total number of tests

Glossary of Statistical Terms

Accuracy = Describes the closeness of a test result to the actual value.

Analysis of Variance (ANOVA) = Estimates total standard deviation of control results analyzed over many days.

Bias = Difference between two quantities; "mean difference for all pairs of results".

Coefficient of Variation = The standard deviation divided by the mean.

Correlation Coefficient (r) = Assesses the reliability of regression estimates of slope and intercept; "describes the ellipse that encloses the results".

Imprecision = Coefficient of variation of the results in a replication experiment. (3)

Inaccuracy = Disagreement between an estimate and its true value. (3)

Line of Identity = Perfect relationship between a test method and reference method.

Mean = Arithmetic average.

Medical Decision Level (Xc) = Concentration where medical interpretation is critical for patient care.

Precision = Describes the reproducibility or consistency of a series of test results.

Reliability = Describes the degree of accuracy and precision of a test procedure.

Mean = The arithmetic "average". The sum of all values divided by the number of values.

Median = The middle value of a group of values i.e., halfway between the lowest and highest value.

Random Error (RE) = "difference between results due to random effects;" unpredictable.

Range = The difference between the highest and lowest value.

Regression Line (best fit) = Reduces the effect of RE and allows calculations of bias; estimates the meter value.

Sample Size (n) = total number of observations.

Slope (b) = Proportional relationship between two sets of results.

Standard Deviation = A measure of the dispersion of a set of data around the mean. An estimate of the degree of uniformity of the data.

Standard Error of the Estimate (Sy/x) = Measures the dispersion around the regression line.

Systematic Error (SE) = Disagreement between methods; error in one direction; consistently high or low.

Total Error = Effect seen when random and systematic errors are combined.

xi' of yi = Individual observations.

Y-intercept (a, y0) = The y value when x = 0.



REFERENCES

1. Campbell, J.M., and Campbell, J.B. 1990. Laboratory Mathematics , 4th ed. Mosby.

2. McClatchey, Kenneth D. 1994. Clinical Laboratory Medicine , Williams & Wilkins.

3. Buttner, J., et al. 1976. Clin. Chem. ; 22:532-540.


051109md