August 3, 2012

Accessing National and State Data from the National Survey on Drug Use and Health

The National Survey on Drug Use and Health (NSDUH), sponsored by the Substance Abuse and Mental Health Services Administration (SAMHSA) and managed by the Center for Behavioral Health Statistics and Quality (CBHSQ) within SAMHSA, is a national survey of the U.S. civilian, noninstitutionalized population aged 12 or older. NSDUH is one of the Federal Government's primary sources of data on the prevalence of substance use and mental illness in the United States.

NSDUH was first conducted in 1971 and has been fielded annually since 1990.¹ Approximately 67,500 in-person interviews are conducted each year. Topics examined in NSDUH include substance use, symptoms of substance use and mental disorders, treatment services for substance use and mental disorders, risk and protective factors, initiation of substance use, and demographic and socioeconomic characteristics of respondents.

Starting in 1999, the NSDUH sample was expanded to produce State-level estimates. The samples in each State were selected to represent proportionately the geography and demography of that State. This design allows for the production of both design-based and model-based State-level estimates. Design-based State estimates are computed using the same methods that create the estimates in the national findings report and detailed tables (see the following section on NSDUH publications). Model-based estimates are produced using a small area estimation (SAE) procedure in which State-level NSDUH data are combined with local-area county and census block group/tract-level data from the State. These model-based estimates provide a more precise representation of substance use and mental health outcomes at the State level than can be produced based solely on the NSDUH survey data, given the smaller State sample sizes.

Substate-level estimates are also produced using SAE methodology. Substate areas are defined by staff at SAMHSA and State substance abuse treatment representatives. State officials made the final decisions on the geographic boundaries. In most States, the substate regions are defined in terms of counties or groups of counties; in some States, the regions are defined in terms of census tracts (or parishes in Louisiana and boroughs/census areas in Alaska).

SAMHSA is committed to making NSDUH data as widely accessible as possible, disseminating the data through publications, data tables, public-use data files, and online data analysis systems. In addition, SAMHSA provides extensive methodological and technical documentation to guide the analysis and interpretation of NSDUH data. This approach to data dissemination is consistent with Federal guidelines and principles such as the Office of Management and Budget's policy on the release and dissemination of statistical products produced by Federal statistical agencies, as well as the President's memorandum on Transparency and Open Government.^2,3

SAMHSA produces national-, State-, and substate-level publications and tabular compendiums using NSDUH data, as well as methodological reports and technical documentation. SAMHSA makes these products available in print formats and electronically on its Web site. Reports can be accessed at http://www.samhsa.gov/data/. Main findings reports are produced on a fixed schedule (e.g., annually or biennially), whereas other reports are produced periodically with no fixed schedules.

Regular Reports of Main Findings
SAMHSA produces several reports that provide the basic NSDUH findings at the national, State, and local levels:

• annual national findings report,
• annual mental health national findings report,
• annual State report, and
• biennial substate report.

In addition, two extensive sets of tables (referred to as "detailed tables") associated with the national findings report and mental health findings report are produced each year. These tables contain prevalence estimates and estimated totals for a large variety of substance use and mental health characteristics by demographic and geographic categories. Tables showing trends from 2002 to the most recent year are included, and breakdowns are provided for demographic characteristics and other special groups such as college students, pregnant women, and persons on probation or parole. The detailed tables also include tables of standard errors (SEs) that correspond with each prevalence table, as well as p values associated with tests of statistical significance of year-to-year change.

SAMHSA produces an annual report with selected estimates for all 50 States and the District of Columbia using 2 years of pooled NSDUH data and a biennial report for between 300 and 400 substate areas using 3 years of NSDUH data.

Periodic Topical Reports
Reports focusing on specific topics related to substance use and mental health problems using NSDUH data are published periodically by CBHSQ. Some examples of these reports include:

• Data Spotlight: 1-page reports that highlight a single finding and are directed to a general audience;
• The NSDUH Report: 4- to 8-page short reports that highlight a single substance use and/or mental health issue and are directed to a general audience;
• CBHSQ Data Review: analytic reports of varying length that focus on one or more behavioral health issues; these reports sometimes explore specific topics in more depth than Spotlights or short reports and are often directed to professional audiences; and
• Analytic Series Reports: large-scale studies covering different aspects of a major behavioral health issue.

Methodological Reports and Technical Documentation
SAMHSA makes available methodological reports and technical documentation on the NSDUH survey and estimation process to help ensure the accurate interpretation of NSDUH estimates and use of NSDUH data. Each year, SAMHSA produces a NSDUH Methodological Resource Book (MRB), which documents changes in the NSDUH questionnaire; how the NSDUH data were collected; response rates; procedures for sampling, editing, imputing, and weighting; and details on estimation. SAMHSA also produces methodological reports that cover a wide range of topics such as comparison of estimates with other studies, effect of questionnaire changes on estimates, bias associated with NSDUH estimates, and alternate sampling procedures.

In addition to publications, SAMHSA makes NSDUH public-use and restricted-use data available using different mechanisms in order to allow analysts to conduct their own research. The methods of dissemination vary depending on the confidentiality associated with the data and on the level of expertise and needs of the data user. SAMHSA has set up a Substance Abuse and Mental Health Data Archive (SAMHDA) to make NSDUH data available to data users.⁴

Information on accessing the public-use file (PUF), the forthcoming restricted-use data analysis system (R-DAS), and other future data access options are discussed in the following sections.

Access to Public-Use Data
The approximately 67,500 NSDUH respondents each year provide personal and sensitive information with the understanding that SAMHSA and its data collection contractor will protect their confidentiality. Survey interviewers and the explanatory materials they give to respondents pledge that the information the respondents provide will be used only for statistical purposes, and that information that could identify individual respondents and the data they reported will never be released to unauthorized persons. The Confidential Information Protection and Statistical Efficiency Act (CIPSEA) of 2002 provides a legal basis for offering this protection to all individually identifiable data collected for statistical purposes under a pledge of confidentiality.⁵

To make NSDUH data files available to analysts interested in conducting their own research, SAMHSA creates a PUF that includes most of the data NSDUH collects but protects respondents' personal information, as required by CIPSEA. In addition to removing specific identifying variables—such as date of birth, names, addresses, and geographic location (e.g., State and county)—from the PUF, several methods are used to prevent the disclosure of information about specific individuals. For example, the link between respondents in the same household is removed to reduce the likelihood that a survey participant could identify the data from another family member who also participated in the survey. In addition, a specific disclosure limitation process involving subsampling, substitution, and reweighting is implemented to protect the files from disclosure.⁶ It should be noted that a PUF and restricted files could give different results. However, these differences are typically small.

NSDUH public-use data files for 1979, 1982, 1985, 1988, and each year from 1990 to 2010 are currently available, free of charge, through SAMHDA. SAMHDA allows analysts to download the PUFs or to use the online data analysis system (DAS) to run statistical analyses (cross-tabulations, means, and regressions) directly from a PUF and obtain results online without downloading the data file. The PUF for each survey is generally available within 12 months of the completion of data collection. Instructions for accessing and analyzing NSDUH data online and finding specific variables within NSDUH are available at http://www.icpsr.umich.edu/icpsrweb/SAMHDA/. SAMHDA also provides access to the other CBHSQ data sources (i.e., from the Drug Abuse Warning Network [DAWN], Treatment Episode Data Set [TEDS], and National Survey on Substance Abuse Treatment Services [N-SSATS]).

Access to Restricted-Use Data
SAMHSA receives many requests for State estimates that are not available in SAMHSA publications or on the PUF. To meet this need while complying with the CIPSEA requirement to protect the confidentiality of NSDUH respondents, SAMHSA has developed an online DAS that will allow users to create their own custom tables of State-level estimates without directly accessing the restricted-use data files that contain State identifiers. The R-DAS has disclosure limitation methods embedded in it so that any output that does not meet those criteria is suppressed. This means that all produced output has no confidentiality-based restrictions and can be considered public use, and therefore there are no limitations on who can access the R-DAS.

Future Options for Access to Data
There are limitations to the analyses that the R-DAS is currently designed to support. For example, the first release will not support regressions or substate estimates. In order to further expand access to NSDUH restricted-use data, SAMHSA is also developing a data portal. Approved analysts will have access to restricted-use data from approved off-site locations. The data will remain in a central repository and cannot be locally downloaded, copied, or saved. All output will have to undergo disclosure analysis before the researcher can remove them from the system. The data portal is expected to be available in 2013. SAMHSA is currently pilot testing the portal and establishing criteria for accessing restricted-use data through the data portal.

This page was last updated on July 19, 2012.