The Washington Post/Kaiser Family Foundation Survey of Long-Term Prescription Painkiller Users and Their Household Members
The Washington Post/Kaiser Family Foundation Survey Project is a partnership combining survey research and reporting to better inform the public. The Post-Kaiser Survey of Long-Term Prescription Painkiller Users and Their Household Members, the 30th in this series was conducted by telephone October 3 – November 9, 2016, among a representative random national sample of 809 adults age 18 and over who they themselves, or a household member, have taken strong prescription painkillers for a period of two months or more1 at some time in the past two years, other than to treat pain from cancer or terminal illness. Interviews were administered in English and Spanish, combining random samples of both landline (n=266) and cellular telephones (n=543).
Sampling, data collection, weighting and tabulation were managed by SSRS in close collaboration with The Washington Post and Kaiser Family Foundation researchers.
The SSRS Omnibus survey (detailed below) estimates that about seven percent of adults in the U.S. have either themselves used strong prescription painkillers in the past two years for two months or more, other than to treat pain from cancer or terminal illness (five percent), or have a household member that has (two percent). Due to the low-incidence of this study population, the sampling was designed to increase efficiency in reaching this group by using the following sample sources:
- Cell and Landline Phone Random Digit Dialing (RDD) (n=354): The dual frame landline and cellular phone sample was generated by Marketing Systems Group (MSG) using RDD procedures. Interviewers calling landline phone numbers asked to speak with an adult currently at home on a random rotation. Interviews calling cellular phones interviewed the person answering the phone after verifying eligibility.
- Respondents Previously Completing Interviews on the SSRS Omnibus Survey (n=455). Weekly, RDD landline and cellular phone surveys of the general public were used to identify eligible respondents. Individuals who had previously indicated on the SSRS omnibus survey that they fit the eligibility criteria for this study were re-contacted.
Regardless of the sample source, all respondents were screened to verify that they have taken a strong prescription painkiller for a period of two months or more in the past two years or have a household member who has, other than to treat pain from cancer or terminal illness. If a respondent indicated that both themselves and a household member qualified, they completed the questionnaire about their own personal use. The screening questions are identified at the beginning of this document as S3 through S5.
A multi-stage weighting design was applied to ensure an accurate representation of the population of long-term strong prescription drug users and their household members. The first stage of weighting involved corrections for sample design, including accounting for non-response for the re-contact sample. In the second weighting stage, demographic adjustments were applied to account for systematic non-response along known population parameters. No reliable administrative data were available for creating demographic weighting parameters for this group. Therefore, demographic benchmarks were derived by compiling a sample of all respondents interviewed on the SSRS Omnibus survey between August 18, 2016 and November 9, 2016 (N=15,944) and weighting this sample to match the national adult population based on the 2016 U.S. Census Current Population Survey March Supplement parameters for age, gender, education, race/ethnicity, region, phone status, and population density. This sample was then filtered to include respondents qualifying for the current survey (N=1,122), and the weighted demographics of this group were used as post-stratification weighting parameters for the total sample (including age by gender, education, race/ethnicity, region, population density, self or household member prescription painkiller user, and phone status).
All sampling error margins and tests of statistical significance have been adjusted to account for the survey’s design effect, which is 1.6 for this survey. The design effect is a factor representing the survey’s deviation from a simple random sample, and takes into account decreases in precision due to sample design and weighting procedures. Sample sizes and margin of sampling errors for key groups are shown below; other subgroups are available by request. Note that sampling error is only one of many potential sources of error in this or any other public opinion poll.
|Table 6: Sample Size and Margin of Sampling Error
|Margin of sampling error (percentage points)
|Among personal users
|Report being addicted or dependent
|Do not report being addicted or dependent
This questionnaire was administered with the exact questions in the exact order as appears in this document. If a question was asked of a reduced base of the sample, a parenthetical preceding the question identifies the group asked. Current users were asked the questions in the present tense, indicated where applicable in the first part of the parenthetical text within the question. Past users were asked the questions in the past tense and read the language in the second half of the parentheses. In addition, those personally using were asked the questions directly about their use, whereas household members were asked about the user in most cases, as indicated in the questions.
Since some of the questions could be sensitive and due to the nature of the survey content, interviewers were given specific instructions on how to cope with respondents who seemed agitated or distressed by the questions, including offering resources to which respondents could turn for support.
The Washington Post and the Kaiser Family Foundation each contributed financing for the survey, and representatives of each organization worked together to develop the survey questionnaire and analyze the results. Each organization bears the sole responsibility for the work that appears under its name. The project team from the Kaiser Family Foundation included: Mollyann Brodie, Ph.D., Bianca DiJulio, and Bryan Wu. The project team from The Washington Post included: Scott Clement and Emily Guskin. Both The Washington Post and the Kaiser Family Foundation public opinion and survey research are charter members of the Transparency Initiative of the American Association for Public Opinion Research.