Screening oncology patients for clinically significant emotional distress is a recommended standard of care in psycho-oncology. However, principles regarding the interpretation of screening and diagnostic tests developed in other areas of medicine have not been widely applied in psycho-oncology. This paper explores the application of the concepts of likelihood ratios and post-test probabilities to the interpretation of psychological screening instruments and demonstrates the development of an algorithm for screening for emotional distress and common psychopathology. Three hundred forty oncology/haematology outpatients at the Calvary Mater Newcastle, Australia completed the Distress Thermometer (DT), the PSYCH-6 subscale of the Somatic and Psychological Health Report and the Kessler-10 scale. The Hospital Anxiety and Depression Scale (HADS) (cutoff 15+) was used as the gold standard. Likelihood ratios showed that a score over threshold on the DT was 2.77 times more likely in patients who were cases on the HADS. These patients had a 53% post-test probability of being cases on the HADS compared with the pretest probability of 29%. Adding either the PSYCH-6 (3+) or the Kessler-10 (22+) to the DT (4+) significantly increased this post-test probability to 94% and 92%, respectively. The significance of these improvements was confirmed by logistic regression analysis. This study demonstrated the application of probability statistics to develop an algorithm for screening for distress in oncology patients. In our sample, a two-stage screening algorithm improved appreciably on the performance of the DT alone to identify distressed patients. Sequential administration of a very brief instrument followed by selective use of a longer inventory may save time and increase acceptability.
Supportive Care in Cancer Vol. 17, Issue 12, p. 1455-1462