Application: To judge the usefulness of a screening or diagnostic test for a particular disease
To assess the quality of a diagnostic test:
validity and reliability sensitivity and specificity positive and negative predictive values
When using quantitative test resultssuch as measurements
of fasting glucose, serum cholesterol, or hematocrit levels the physician will also need to know the accuracy and precision of the measurement as well as the normal reference values Validity Validity: the extent to which it actually tests what it claims to testin other words, how closely its results correspond to the real state of affairs Validity is synonymous with accuracy Accuracy is the ability of a test to produce a true value of the measurements and true classifications for the samples under study Reliability Reliability: Reliability is synonymous with repeatability and reproducibility; it is the level of agreement between repeated measurements of the same variable. Hence, it is also called testretest reliability Reliability corresponds to precision Precision is the extent to which random variability is absent from a test result. 2-by-2 table Sensitivity and specificity Sensitivity: Ability to detect people who do have the disease A sensitive test is obviously required in situations in which a false negative result has serious consequences. High sensitivity is required of tests used to screen donated blood for HIV, for cytologic screening tests (Pap smears) for cervical cancer, and for mammograms Very sensitive tests are therefore used for screening or for ruling out disease; if the result of a highly sensitive test is negative, it allows the disease to be ruled out with confidence Sensitivity and specificity Specificity: Ability to detect people who do not have the disease High specificity is required in situations in which the consequences of a false positive diagnosis are serious If the result of a highly specific test is positive, the disease is almost certainly present Very specific tests are therefore appropriate for confirming or ruling in the existence of a disease If the result of a highly specific test is positive, the disease is almost certainly present Sensitivity and specificity
In clinical practice, sensitivity and specificity
are inversely related: an increase in one causes a decrease in the other
Sensitivity and specificity are both measures
of a tests validity i.e. its ability to correctly detect people with or without the disease in question False-positive and False-negative rates False-positive rate: False-positive rate= B/ (B+D) False-negative rate: False-negative rate = C/ (A+C) Receiver Operating Characteristic curve Positive-predictive value The positive predictive value (PPV) of a test is the proportion of positive results that are true positives, that is, the likelihood that a person with a positive test result truly has the disease Negative-predictive value The negative predictive value (NPV) of a test is the proportion of negative results that are true negatives, that is, the likelihood that a person with a negative result truly does not have the disease Cut-off points TYPES OF VARIABLES Nominal variables Naming or categorical variables For example, blood groups, occupations Variables are assigned a numerical value Dichotomous variables Binary variables Variables have only two values For example, normal/abnormal; female/male; living/dead Ordinal variables Ranked variables Characterized in terms of more than two values and have a clearly implied direction from better to worse For example, respiratory distress (mild/moderate/severe); abnormal murmurs (1+ to 6+) Continuous variables Dimensional variables Medically important data For example, heights, weights, serum glucose levels, systolic and diastolic blood pressures THANK YOU FOR YOUR ATTENTION