Screening for alcohol problems: what makes a test effective?

Alcohol Research & Health, Wntr, 2004 by Scott H. Stewart, Gerard J. Connors

Specific measures that help assess the usefulness of a screening test are its sensitivity, specificity, and overall accuracy.

Sensitivity

The term "sensitivity" refers to the ability of a test to correctly identify those people in a population who actually have the disorder. That is, sensitivity represents the probability that a test for a specific disorder will be positive when the disorder truly is present; it ranges in value from 0 to 1 (or equivalently, from 0 percent to 100 percent). The phrase "specific disorder" is important in this context because a screening test can perform differently depending on which disorder or group of disorders is being examined. The AUDIT, for example, will have a different sensitivity when screening for alcohol dependence than when screening for hazardous drinking, and yet another sensitivity when screening for both conditions (Fiellin et al. 2000).

Sensitivity is calculated as the proportion of people with a disease who have a positive screening test. In terms of the four groups of people defined when a screening test is compared with a gold standard, sensitivity is the ratio of true positives over all people who actually have the disorder (that is, true positives plus false negatives) (see figure 1). (1) A highly sensitive test is desirable when the cost of missing people who actually have a disorder (i.e., who have a false negative screening result) is high. For example, if a screening test is not sensitive enough to correctly identify a commercial airline pilot who exhibits "harmful drinking," the results (e.g., an intoxicated pilot flying a plane) can lead to potentially catastrophic consequences.

In screening for AUDs, sensitivity can be enhanced by lowering the cutoff scores used to define a positive screening result. For example, the AUDIT consists of 10 questions. Respondents can score between 0 and 4 points on each question, so the total score ranges between 0 and 40 points. (For more information on the AUDIT, see the sidebar "Screening Tests," on page 28.) Generally, a score of 8 points or higher is considered suggestive of a diagnosis of "hazardous alcohol use." However, if the cutoff score for hazardous use is lowered to 4 or more points, the sensitivity of the test increases significantly--that is, more people with a drinking problem would have a positive screening result. Such a lowered cutoff score rarely is used, however, because it also would increase the number of false positive results, thereby reducing the test's specificity, as described in the next section.

Specificity

Specificity is the test's ability to identify people in a group who do not have the disorder under investigation. That is, specificity is the probability that a test for a specific disorder will be negative when the disorder is truly absent. Like sensitivity, specificity values range from 0 to 1 (or 0 percent to 100 percent). Specificity is the ratio of people without the disease who screen negative (or true negatives) over all people who actually are without the disease (true negatives plus false positives) (see figure 1). The more specific the test is (i.e., the closer the specificity value is to 1), the fewer people will screen positive for the disease when they do not have it (i.e., the number of false positives approaches 0). A highly specific screening test is desirable when the cost of a false positive result is high. This is less of a problem when screening for drinking problems, because additional testing typically would be performed after a positive screen. Any additional diagnostic evaluations also require additional resources, however, and such resources often are limited.