Substance Use & Misuse, 45:1509–1524
Copyright © 2010 Informa Healthcare USA, Inc.
ISSN: 1082-6084 (print); 1532-2491 (online)
DOI: 10.3109/10826081003682107
Medicinal Misuse
Validity of Self-Reported Misuse of Prescription
Opioid Analgesics
MEREDITH SMITH,1 ANDREW ROSENBLUM,2 MARK
PARRINO,3 CHUNKI FONG2 AND SALVATORE COLUCCI1
1Purdue Pharma L.P., One Stamford Forum, Stamford, USA
2National Development and Research Institutes, Inc., New York, USA
3American Association for the Treatment of Opioid Dependence, New York,
USA
Aim: To determine concurrent validity of self-reported misuse of prescription opi-
oids. Design and Setting: Cross-sectional study in five U.S. methadone maintenance
programs. Participants: 92 addicts. Measurements: Self-reported questionnaire assess-
ing past-month misuse of 14 opioid analgesics, and color photographs of five opioid
analgesics with instructions to mark those used in the past month “to get high.” Con-
cordance between self-report and photograph endorsement was assessed via Kappa
statistic. Findings: 29 respondents completed both questionnaire and photograph en-
dorsements. Kappas were 0.62 (OxyContin R©), 0.59 (methadone), 0.49 (Dilaudid R©),
and 0.46 (generic extended-release oxycodone). Conclusions: Good-to-fair concurrent
validity of self-reported abuse was seen for OxyContin R©, methadone, Dilaudid R©, and
generic extended-release oxycodone.
Keywords validity; self-reported drug misuse; opioid analgesic abuse; OxyContin R©;
Dilaudid R©
Introduction
One of the most significant public health developments in the United States over the past
decade has been the marked rise in the nonmedical use and abuse1 of prescription drugs
(Compton and Volkow, 2006). The majority of research documenting this phenomenon has
relied on individual self-reports collected via questionnaire (Johnston et al., 2007; McCabe,
2004; McCabe et al., 2007; Substance Abuse and Mental Health Administration, 2008).
Confidence in these estimates depends on an understanding of the degree and source(s)
of measurement error associated with such self-report. While there is substantial literature
Address correspondence to Salvatore Colucci, Purdue Pharma L.P., One Stamford Forum, Stam-
ford, 06901-3431 USA; E-mail: salvatore.colucci@pharma.com.
1The journal’s style utilizes the category substance abuse as a diagnostic category. Substances
are used or misused; living organisms are and can be abused. Editor’s note.
1509

1510 Smith et al.
examining the validity of self-reported substance use, the majority of such work to date has
focused on illicit drugs (Del Boca and Noll, 2000; Maisto et al., 1990). Comparatively less
is known concerning the validity of self-reported abuse of prescription psychotherapeutic
agents, including opioid analgesics, the most widely misused of all such medications
(Compton and Volkow, 2006; Substance Abuse and Mental Health Administration, 2008;
Zacny et al., 2006).
Self-report of behaviors related to substance use is a complex and demanding mental
activity that is known to be influenced by an array of factors. Such factors include respondent
attributes (e.g., physical and psychological state, demographics, personality type), task-
related characteristics (e.g., question form, wording, and sequencing; length of recall period;
format, mode, and setting of survey administration), motivation (e.g., fear of negative social
or legal sanctions; desire to present oneself in a positive light), and cognitive processes (e.g.,
attention, comprehension, memory retrieval, integration) (Del Boca and Noll, 2000; Magura
et al., 1987; Maisto et al., 1990; Page et al., 1977).
In terms of self-reported misuse of prescription psychotherapeutic medications, there
are a number of unique considerations that may further exacerbate the complexity of this
assessment task. First, prescription medications go by several different identifiers: the
chemical substance, the brand or generic name, and one or more “street” monikers. Study
respondents, particularly those who are relatively unsophisticated or inexperienced users,
may not be equally familiar with these various descriptors, and response inaccuracies may
result if a questionnaire item refers to only one such term. Misattribution may also occur
when the chemical name for one particular drug is the same as, or similar to, that for
another drug (e.g., “hydromorphone” and “hydrocodone”), or when a highly publicized,
branded product is mistaken for a generic counterpart. Lastly, responses may be particularly
susceptible to social desirability bias due to the perceived social cachet associated with using
certain prescription drugs (Corliss, 2001; Podsakoff et al., 2003).
Overall, the empirical literature suggests that the validity of self-reported misuse of
prescription psychotherapeutic drugs may be lower than that for heroin and other illicit
drugs (Aiken and Lo Sciuto, 1985; Calhoun et al., 2006; Cisin and Parry, 1979; Cordingley
et al., 1990; Hser, 1997; Johnston and O’Malley, 1997; Lennox et al., 2006; Magura
et al., 1987; Page et al., 1977; Solbergsdottir et al., 2004; Wish et al., 1997). However,
interpretation of findings is complicated by the fact that research efforts to date have varied
widely in terms of type of population studied, assessment context (e.g., drug treatment,
school, corrective facility, general-population-based household survey), and measurement
tools and procedures, all factors known to influence accuracy of self-reported drug use
(Aiken and Lo Sciuto, 1985; Calhoun et al., 2006; Cisin and Parry, 1979; Co´lon et al.,
2001; Cordingley et al., 1990; Fendrich et al., 2005; Harrison et al., 2007; Hser, 1997;
Hubbard et al., 1977; Johnston and O’Malley, 1997; Lennox et al., 2006; Magura et al.,
1987; Page et al., 1977; Solbergsdottir et al., 2004). Moreover, studies typically have not
been designed to collect product-specific data. Rather, measurement has generally been at
the level of the entire psychotherapeutic drug class (e.g., benzodiazepines). In particular,
prescription opioids have frequently been grouped together with heroin for assessment
purposes under the rubric of “opiates” (Aiken and Lo Sciuto, 1985; Cordingley et al., 1990;
Hser, 1997; Magura and Kang, 1996; Magura et al., 1987).
To date, four different types of validity criteria have been employed: (1) biometric mea-
sures (i.e., urinalysis, saliva tests, and hair assays), (2) collateral reports, (3) administrative
records, and, (4) recanting rates. Each of these accuracy criteria has well-documented lim-
itations (Maisto et al., 1990; Page et al., 1977; Wish et al., 1997). Biometric tests, due to
an array of technological and procedural reasons, can vary markedly in their sensitivity to