Variables affecting laboratory test results
Preanalytical variables include physiological factors, patient preparation, specimen collection and transport.
Analytical variables include the precision and accuracy of the test method and factors which may interfere with a particular assay eg, lipaemia, in vitro haemolysis, and medications.
Post-analytical variables include data entry and calculations by laboratory staff, result validation, interpretation of the result, data transfer and the method used to report the results (electronic, paper or telephone).
Pathology tests guide clinical decision making and the clinician should have some understanding of the factors which influence the reliability of a diagnostic test for such decisions to be valid.
The clinician has an important part to play in the avoidance, or control, of many of the preanalytical variables.
The clinician also needs to have an understanding of the sensitivity and specificity of tests and of their predictive value.
Profile testing, or 'screening', is an expensive process which, even with tests of high sensitivity and specificity, has a limited positive predictive value (PPV), if the prevalence is low. Highly accurate test results may be entirely meaningless or misleading when used in this fashion.
False positive results lead to unnecessary and expensive follow-up testing and patient anxiety.
False negative results place the patient at risk.
Diagnostic tests must supplement rather than be used as a substitute for clinical skills.
Careful clinical assessment followed by discretionary testing is cost-effective, efficient and leads to improved patient outcomes.
Diagnostic accuracy studies
||Positive test result
||Negative test result
||TP + FN
||FP + TN
||TP + FP
||FN + TN
||TP + FP + TN + FN
TP, true positive; FP, false positive; FN, false negative; TN, true negative.
Diagnostic sensitivity (‘positivity in disease’) of a test refers to the probability of obtaining a positive result for a subject with a given disease.
Diagnostic sensitivity = TP / (TP + FN)
High sensitivity corresponds to a high negative predictive value (NPV) and is the ideal property of a ‘rule out’ test.
Diagnostic specificity (‘negativity in health’) of a test is the probability of obtaining a negative result in a subject without a given disease.
Diagnostic specificity = TN / (TN + FP)
High specificity corresponds to a high PPV and is the ideal property of a ‘rule in’ test.
Receiver-operating characteristic curve
A receiver-operating characteristic (ROC) curve compares sensitivity versus specificity across a range of values to enable the best cut-off for clinical purpose to be assigned.
A major factor in improving the PPV of a test is to limit the use of the test to those patients who are likely to have the disease in question.
PPV is the proportion of positive results that are true positives, which represents the diagnostic value of a positive result for the test.
PPV = TP / (TP + FP)
NPV is the proportion of negative results that are true negatives.
NPV = TN / (TN + FN)
As the prevalence increases, PPV increases and NPV decreases.
Diagnostic efficiency of the test refers to the probability of all tested subjects being correctly classified with or without a given disease.
Diagnostic efficiency = (TP + TN) / (TP + FP + TN + FN)
The diagnostic efficiency of a test can be markedly increased by using it in a discretionary fashion in high risk groups or in patients with clinical features suggesting the disease in question.
This improved diagnostic efficiency does not significantly reduce its NPV.
Predictive values and diagnostic efficiency are significantly influenced by FP and FN rates and by the prevalence of the disease in the population being tested.
Prevalence = (TP + FN) / (TP + FP + TN + FN)
Thus, for a test with a diagnostic sensitivity of 95% and a diagnostic specificity of 95%, with a population prevalence of the disease of 1%, the PPV is only 16.1%.
Ultimately, the value of a diagnostic test will depend upon its ability to alter the pre-test probability of a disease into a post-test probability that will influence a clinical management decision, which can be achieved by the application of likelihood ratios (LR), but note that probability needs to be converted to odds.
LR+ = sensitivity / (1 – specificity)
LR– = (1 – sensitivity) / specificity)
A LR+ >10 and a LR– <0.1 are considered to exert highly significant changes in probability, such as to alter clinical management.
Florkowski CM. Sensitivity, specificity, receiver-operating-characteristic (ROC) curves and likelihood ratios: communicating the performance of diagnostic tests. Clin Biochem Rev 2008; 29: S83-7.