There is intense interest in identifying and addressing substance use in general healthcare settings. Considerable federal resources are now devoted to promoting ‘screening and brief intervention (SBI)’ approaches for alcohol and other drugs, which involve screening and assessment, followed by brief counseling interventions that can be carried out by medical providers in the course of a regular office visit.1,2,3,4,5 Having a short and accurate substance use screening and assessment instrument is essential for the successful implementation of SBI in time-pressured general medical settings. In response, researchers have moved rapidly to develop and validate brief screening tools.6,7,8,9,10,11
Recent developments highlight the importance of identifying not only use of alcohol and illicit drugs, but also misuse of prescription drugs. Prescription drug misuse, and in particular the misuse of potent prescription opioids, has skyrocketed in the past decade.12 Overdose deaths attributed primarily to prescription opiates are now the leading cause of accidental deaths among adults,13,14 and emergency room visits due to prescription painkiller misuse have doubled in the past five years.15 Currently there is no standardized approach to accurately and efficiently screening adult patients for misuse of prescription drugs.
Many individuals who misuse prescription drugs are not engaged in use of any illicit drugs. 16 To capture this increasingly prevalent type of drug use, screening tools need to be able to detect prescription drug misuse independently from illicit drug use. Yet the screening and assessment instruments that are generally considered for implementation in medical settings (e.g. DAST-10, ASSIST) were developed at a time when prescription drug misuse was less prevalent, and little is known about their effectiveness for identifying it. Newer instruments, such as the NIDA-modified ASSIST and the single item drug screening question developed by Smith et al. (2010) have included prescription drugs.6,17 But the NIDA-modified ASSIST has not been validated, and the single item drug screening question, which does have good sensitivity and specificity for detecting any drug use, has not been evaluated for its accuracy in specifically detecting prescription drug misuse.
Concisely and clearly communicating the meaning of prescription drug misuse on a short screening instrument poses considerable challenges. Unlike illicit drugs, which are defined by their legal status, prescription drugs can be used appropriately or can be misused, depending on the conditions under which they are used and the intentions of the user. To draw a distinction between appropriate medical use and misuse of prescription medications, the language adopted by both the single item drug screening question and the NIDA-modified ASSIST is ‘nonmedical’ use. Yet little is known about how patients interpret this term.
As a first step in developing a screening instrument that could quickly and accurately identify both illicit and prescription drug misuse, we sought to gain a better understanding of how primary care patients interpret and answer screening questions about their use of these substances. We employed cognitive interviewing, an approach developed in the 1980s by survey methodologists and psychologists to evaluate sources of response error in questionnaires. Cognitive interviewing is supported by a large body of methodological research, and is one of the primary methods used by survey researchers to test the accuracy with which items are understood and answered by respondents. 18,19,20,21,22 Given the similarity between surveys and structured screening instruments, we applied cognitive interviewing techniques to elucidate how primary care patients understand items in a substance use screening questionnaire.
Cognitive interviewing recognizes that the question answering process can be complex, involving the cognitive steps of comprehending the question, retrieving relevant information from memory, making a decision about how to answer, and then mapping the response onto the options given in the survey question. We postulated that answering questions about illicit and prescription drug use primarily poses difficulty in the areas of comprehension and deciding on an answer. Comprehending the question requires an understanding of the language used to describe the drug or behavior, such as ‘nonmedical’ or ‘illegal’ use. Making a decision about how to answer is influenced by the participant’s comfort with the way in which the question is asked, and is important to consider when the item includes stigmatized behavior such as drug use.