2020 National Survey on Drug Use and Health (NSDUH) Releases

• The coronavirus disease 2019 (COVID-19) pandemic necessitated changes in data collection, which may have had an impact on the measurement of behaviors. Changes in estimates of behavior may be due to this impact or may be due to true changes in prevalence. The primary methodological changes were (1) limited data collection from mid-March through September 2020, (2) introduction of web data collection in October 2020 with limited in-person data collection, and (3) questionnaire changes beginning in October 2020. These changes resulted in data collection mode effects and potentially other effects on the estimates. Because these changes in data collection coincided with the COVID-19 pandemic, the effects of methodological changes cannot be completely separated from actual changes in the outcomes.
• The comparability between 2020 estimates and previous years’ estimates was unknown when the 2020 estimates were first released in October 2021. For that reason, data users were cautioned against making comparisons between estimates from 2020 and prior years. Further research using the larger sample of multimode data (i.e., in-person or web) from the 2021 NSDUH data has confirmed that 2020 estimates should not be compared with estimates from prior years. Key contributors to the break in comparability were the loss of 2 quarters of data collection in 2020 and the introduction of web data collection in October 2020.

The criteria used to categorize SUD among NSDUH respondents changed from the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV)¹ to the fifth edition (DSM-5),²resulting in some differences in who is classified as having an SUD.

¹American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.).

²American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). https://doi.org/10.1176/appi.books.9780890425596

• Both are “right” relative to the diagnostic criteria in their respective DSM editions. Between DSM-IV and DSM-5, the SUD criteria were changed, resulting in differences in who is classified as having an SUD.
• Under DSM-IV, NSDUH respondents are classified as having an SUD if they meet at least one of four criteria of substance abuse (these criteria focus on behaviors that reflect consequences of misuse of the substance) and/or at least three of seven criteria of substance dependence (these criteria focus on behaviors that reflect indications that one has become dependent on the substance). The DSM-5 criteria do not distinguish between symptoms of substance abuse and those of substance dependence. Instead, a DSM-5 SUD classification requires that at least two of eleven criteria be met.
• There are two significant differences between the DSM-IV SUD criteria and the DSM-5 SUD criteria. First, one criterion of DSM-IV substance abuse (recurrent substance-related legal problems) is not part of the DSM-5 SUD criteria. Second, a new criterion (craving or a strong desire to use the substance) has been added to DSM-5 that was not present in the DSM-IV SUD criteria. Craving is a commonly experienced symptom for several classes of substances, such as opioids and sedatives. The other ten DSM-5 criteria correspond to DSM-IV criteria.

• Quarter 1 (January to March) data were collected like 2019 data were collected, with all household screenings and individual interviews conducted in person.
• Quarter 4 (October to December) data collection methodology, data processing, and data collection outcomes changed as follows:
  • Web-based data collection was introduced to avoid in-person contact during the COVID-19 pandemic.
  • Screening and overall response rates were considerably lower for web-based data collection.
  • Youth interview response rates declined because of the required procedures for obtaining parental permission for youth to complete the web-based interview.
  • Some respondents did not complete their web interview, resulting in more missing responses to questions later in the questionnaire.
  • Web respondents tended to have completed higher levels of education, on average, than in-person respondents and the general population. Weighting procedures incorporated education in an effort to correct the imbalance.
  • New questions were introduced in Quarter 4 for telehealth services, youth suicidality, and issues related to the COVID-19 pandemic.

• Data was not collected from mid-March through September 2020 and thus did not capture changes in people’s behavior and mental health during an unprecedented time in history (i.e., changing COVID-19 infection rates, lockdowns, and other societal changes). The sample is representative of January to March and October to December but is not representative of the entire year in which many people experienced changes in almost every aspect of life.
• The sample for each calendar quarter follows the same national/state design as each full-year sample, and the weights for sample records are adjusted so that the sums are consistent with known national totals. For these reasons, the combined sample from Quarters 1 and 4 is representative in terms of geography (total U.S. and states) and demographic characteristics, even though it is smaller than prior years’ samples.
• In general, larger sample sizes yield more precise estimates. Having a smaller national sample in 2020 means that many national estimates are not as precise as estimates from prior years, although the precision for 2020 national estimates is considered to be more than adequate.
• Compared with larger samples, smaller sample sizes are less likely to produce sufficiently precise estimates for small subpopulations and for rare behaviors. Estimates that were deemed insufficiently precise due to the small sample size were suppressed in tables and reports. More estimates were suppressed in 2020 than in prior years.