State Data Tables and Reports from the 2021-2022 NSDUH

SAMHSA decided to continue with multimode data collection moving forward, meaning that the 2021 NSDUH is based on both in-person and web interviews. In-person data collection remained limited to areas with low COVID-19 infection rates. As COVID-19 infection rates dropped in most areas of the country over the course of the year, the proportion of web interviews decreased. More than three quarters of interviews in Quarter 1 were completed via the web (76.6 percent). By Quarter 4, fewer than half of the interviews (41.5 percent) were completed that way. Altogether, 54.6 percent of the 2021 interviews were completed via the web.

• In order to assess comparability, we first tested, “Did the mode of data collection (i.e. whether a respondent completed the survey on the web or in-person) influence the estimates?” In other words, “Were there differences between responses from the two modes of collection, and were those differences significant enough to affect whether or not the data from each mode was comparable?”
  • We found that estimates based on web interviews are different from estimates based on in-person interviews. First, we found that the demographics of each mode differed significantly. That is to say, the composition of the people answering by web differed from the composition of the people answering in-person (i.e. by gender, race, education, etc.). Secondly, in addition to demographic differences, we also found that in-person respondents were more likely to be users of certain substances and were more likely to have experienced mental health issues. These differences are often called a mode effect because the differences are observed between the two modes of data collection.
• We then asked, “Given that we know there are differences between modes of data collection which make them incomparable, can we use demographics to weight the data in such a way that they ARE comparable?” In other words, “Can we make both groups of respondents similar enough in demographic composition to control for the differences in mode that we observe, and therefore allow us to compare estimates between the two modes?”
  • We know that while weighting (adjusting) the data to account for the differing demographics between respondents who responded via the web and in person helps to lessen the mode effect impact, it doesn’t completely eliminate the effect enough so that we can fairly compare the two modes statistically. Because of this, estimates based on both web and in-person interviews are not comparable to estimates based on either mode alone.
  • Therefore, because 2021 estimates are based on multimode data collection and estimates from 2019 or earlier are based on in-person data collection alone, 2021 estimates are not comparable with estimates from 2019 or earlier. Estimates of change from 2019 or earlier to 2021 would be too greatly influenced by the mode effect.

• Differences in substance use between web respondents and in-person respondents could be due to variability in COVID-19 infection rates in different areas. For example, many respondents could complete the interview only on the web (i.e., those living in areas with high COVID-19 infection rates where in-person data collection was not permissible). If individuals living in these areas are less likely to use substances, overall estimates of substance use in 2021 would be affected given the large proportion of interviews completed on the web (54.6 percent).
• There could be differences between web respondents and in-person respondents for reasons other than underreporting of substance use among web respondents. For example, some people prefer to complete an interview via web, while others prefer to complete an interview in person. Adult web respondents in NSDUH have completed higher levels of education, on average, than their in-person counterparts; people with higher educational levels are less likely to smoke cigarettes and are more likely to have access to the Internet. Weighting the data to match the demographic characteristics of the population controls helped to account for educational differences between web and in-person respondents.

• Offering a web mode enables collecting data from individuals who may not have otherwise completed an interview. For example, some individuals will complete an interview only on the web because they will not allow strangers in their homes. Some individuals’ only option is to complete the interview on the web because they live in areas inaccessible to field interviewers, such as gated communities, or they live in areas where in-person data collection is not permissible due to high COVID-19 infection rates.
• Likewise, offering an in-person mode enables collecting data from individuals who would not otherwise complete an interview. For example, some individuals’ only option is to complete an interview in person because they do not have a computer, a Smart phone, or a tablet or do not have Internet or a cellular data plan. Some individuals choose to complete an interview only in person due to a lack of computer proficiency or discomfort with entering personal information online.

• SUDs are characterized by impairment caused by the recurrent use of alcohol or other drugs (or both), including health problems, disability, and failure to meet major responsibilities at work, school, or home. SUD questions in the 2021 NSDUH assessed the presence of an SUD in the past 12 months based on criteria specified in the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM 5).* NSDUH respondents were classified as having an SUD if they met 2 or more of up to 11 applicable criteria in a 12-month period.
• DSM-5 SUD criteria apply to any use of prescription drugs, not just misuse. People who use but do not misuse prescription drugs could nevertheless meet SUD criteria. It is important for individuals who are using prescription drugs to receive ongoing medical monitoring, even if individuals are not misusing the prescription drugs. For these reasons, respondents in the 2021 NSDUH who reported any use of prescription pain relievers, tranquilizers, stimulants, or sedatives in the past 12 months were asked corresponding SUD questions.

*American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.).  https://doi.org/10.1176/appi.books.9780890425596

The number of applicable DSM-5 criteria for classifying respondents as having a prescription drug use disorder depends on whether respondents misused prescription drugs or whether they used prescription drugs in the past year but did not misuse them. Respondents who misused prescription drugs in the past year were classified as having a prescription drug use disorder if they met 2 or more of 11 DSM-5 SUD criteria. Respondents who used prescription drugs in the past year but did not misuse them were classified as having a prescription drug use disorder if they met two or more of nine criteria. Two criteria (tolerance and withdrawal) do not apply to respondents who used but did not misuse these prescription drugs in the past year; tolerance and withdrawal can occur as normal physiological adaptations when people use these prescription drugs appropriately under medical supervision.˚

˚Hasin, D. S., O’Brien, C. P., Auriacombe, M., Borges, G., Bucholz, K., Budney, A., Compton, W. M., Crowley, T., Ling, W., Petry, N. M., Schuckit, M., & Grant, B. F. (2013). DSM-5 criteria for substance use disorders: Recommendations and rationale. American Journal of Psychiatry, 170(8), 834-851. https://doi.org/10.1176/appi.ajp.2013.12060782

• Asking SUD questions in 2021 of all past year users of prescription drugs more than doubled the estimated numbers of people who had prescription drug use disorders, compared with the estimated numbers based only on people who misused prescription drugs in the past year.
• Asking SUD questions of all past year users of prescription drugs also tended to capture data from people with mild SUDs, defined as having two or three SUD criteria for a given substance.
• In contrast, asking SUD questions of all past year users of prescription drugs had less effect on SUD estimates for aggregate SUD measures that included substances other than prescription drugs. For example, 20.0 million people aged 12 or older in 2021 were estimated to have an illicit drug use disorder in the past year (including use of marijuana, cocaine, heroin, hallucinogens, inhalants, or methamphetamine and misuse of prescription drugs). The estimate increased for any drug use disorder to 24.0 million people when data included past year users of prescription drugs, or an increase of about 20 percent.