To examine potential measurement effects for assessing the prevalence of NMPOU in Canada, we tested an alternative NMPOU item constructs in the Centre for Addiction and Mental Health (CAMH) Monitor surveys [14, 15].The 2008, 2009 and 2010 CAMH Monitor surveys employed a two-stage, probability sample based, computer-assisted telephone interviewing, random-digit dialing method, and sampled the Ontario general adult population (18+ years). Each month of the survey a sampling frame of all active area codes and exchanges in Ontario was obtained from ATI long lines tape. Telephone numbers from this source, as well as telephone numbers between or on either side of listed numbers were included in the sampling frame. This method allowed for any person with a home phone (listed or unlisted) or a cellphone to be included in the sampling frame; however, a proportion of people with an unlisted number and/or a cellphone were not included in the sampling frame. The sampling frame was then stratified into regions (based on counties) as follows: Toronto, Central West Ontario, Central East Ontario, West Ontario, East Ontario, and North Ontario (see [14–16] for the exact counties included within each region). Within each stratum, a random sample of telephone numbers was selected with equal probability. To increase the response rate in the stratum of Toronto, a letter was sent to the selected household informing them that they would be contacted and describing the history and objectives of the CAMH Monitor survey. Within selected households, the respondent who was 18 years of age or older, who had the most recent birthday of the eligible household members, and who could complete the interview in either English or French was selected to participate in the survey. A minimum of 12 call-backs were placed to unanswered numbers and all households who refused to participate on the first contact were re-contacted and asked again to participate. No incentive was offered for participation. The procedures and interviews of each cycle of the CAMH Monitor surveys were approved by the CAMH Research Ethics Board [14, 15].
NMPOU questions in the 2008 and 2009 waves of the CAMH Monitor surveys were asked from July to December, whereas the 2010 NMPOU data were collected from January to December of the respective year. Response rates for the CAMH Monitor surveys were calculated using the method proposed by the American Association for Public Opinion Research (response rate calculation #3) which takes into account in the response rate calculation the number of telephone numbers which may be eligible to be included in the survey [14
]. This formula for calculating the response rate is as follows:
where I represents the number of completions, P represents the number of partial completions, R represents the refusals/breakoffs, NC represents non-contacts, e represents the estimated proportion of cases of unknown eligibility that are eligible, and UH represents the number of telephone numbers which may be eligible to be included in the survey.
In the 2008, 2009 and 2010 waves of the CAMH Monitor surveys POAs were defined as “pain relievers that are obtained by a prescription from a doctor or dentist such as Percocet, Percodan, Demerol, OxyContin, Tylenol #3 or other products or pain relievers with codeine that are obtained in a pharmacy. Some people use these medications to treat pain resulting from an illness, injury.” NMPOU in the 2008 and 2009 waves of the CAMH Monitor surveys was measured similarly to the CADUMS questions described above as follows: 1) “Thinking about all the pain relievers you have used during the past 12 months did you get any of them” i) “from a prescription written for someone else such as a family member or a friend”, ii) “bought from someone else, without a prescription” and/or iii) “from any other source (defined as a source other than the previously mentioned sources and a prescription written for you),” and 2) “During the past 12 months, did you ever use pain relievers for the feelings it caused or to get high?” The alternative NMPOU item tested in the 2010 wave of the CAMH Monitor survey was based on the question: “In the past 12 months how many times, if at all, have you used any such pain relievers without a prescription or without a doctor telling you to take them?”
The items used in the 2008 and 2009 waves of the CAMH Monitor surveys were initially pretested with 25 respondents in November 2007 and included as a pilot study in December of the 2007 wave of the CAMH Monitor survey (N = 175). Applicable questions were asked at the end of the POA items in the 2009 cycle of the CAMH Monitor survey to evaluate the response process; 96.4% of respondents declared that the questions on POA use were not difficult to understand, and 96.1% of respondents did not need the questions to be repeated to them before answering . The items used in the 2010 wave of the CAMH Monitor survey were also initially asked in the 2007 and the 2009 cycles of the Ontario Student Drug Use and Health Survey. These questions were revised for the 2010 wave of the CAMH Monitor survey based on pretesting which confirmed item comprehension and feasibility.
For our analysis of the difference in the measurement effect of the alternative NMPOU survey items on the estimated prevalence of NMPOU, we controlled for the following demographic variables: age (grouped into three categories: 18–29, 30–54, 55+), region (living in Toronto, the rest of Ontario), and household income (<$30,000, $30,000–79,000, $80,000+, not stated). The substance use variables we controlled for in our analysis were tobacco use (defined as either daily or occasional (during the last 12 months) cigarette smoking), weekly binge drinking (defined as drinking five or more drinks on one occasion at least once a week in the previous 12 months), and cannabis use (defined as using cannabis at least once in the previous 12 months).
We also controlled for psychological distress, as measured by the 12-item General Health Questionnaire (GHQ-12). The GHQ-12 is a screening instrument that evaluates depression/anxiety and problems with social functioning  and we used a cut-off score of 3 or more on the GHQ-12 as an indication of elevated psychological distress.
All statistical analyses were performed using population expansion weights. Population expansion weights were constructed using census data from Statistics Canada on the number of people living in each stratum by age (using the age categories of 18-24, 25-44, 45-64, 65+) and by sex [14–16]. Significant differences between prevalence estimates were calculated using chi-square tests with a second-order Rao-Scott adjustment for survey data for the unadjusted analysis . For the unadjusted and adjusted regression analyses, we employed a general linear model adjusted for survey design. All statistics were performed using the statistical software package R version 2.15.1 , and the statistical software package Survey (a statistical package for R) .