Abstract
In a variety of decision settings when categorical information is the data in hand, the correspondence between an indicator test and a true condition (i.e., an accepted “gold standard”) is often assessed by the use of comparison or “truth” tables. The results from such a table can be considered an informational yield and allow a practitioner to assess the utility of the indicator test in various ways, such as specificity and sensitivity, false positive and negative rates, and the positive and negative predictive values. There are many reasons for the use of an indicator test. This article proposes a novel application of an informational measure in determining the utility of hair and urine specimens for clinical decision making. It does this for two commonly abused substances, cocaine and cannabinoids. It suggests that in clinical applications the ability to assess informational yield is an important factor in developing clinical strategies for treatment of addictive disorders.
Keywords :
Notes
1 By “true condition” we mean the physical presence of a drug or drug metabolite in the sample. In practical clinical terms we mean that an illicit drug has been consumed and its presence has been accurately identified by an analytic chemical test.
2 This results in a number of interesting effects in calculating the various measures. For example, the sensitivity measure in one table becomes the equivalent of the PPV in the subsequent table, and the specificity becomes equivalent to the NPV.
3 Since October, 2006 the criteria for CE and NOR were increased from >2% to >5%. Approximately 93.1% of the samples were analyzed under the updated metabolite criteria.
4 Comparing each cell of the cocaine and cannabinoid tables by a binominal single-sample test show that the differences in all cases are significant at p = .000.