Section A: Overview and Findings

A.1. Introduction

This report presents estimates of the prevalence of substance use or mental health problems in substate regions based on data from the combined 2004-2006 National Surveys on Drug Use and Health (NSDUHs). An annual survey of the civilian, noninstitutionalized population aged 12 or older, NSDUH is sponsored by the Substance Abuse and Mental Health Services Administration (SAMHSA). It collects information from persons residing in households, noninstitutionalized group quarters (e.g., shelters, rooming houses, dormitories), and civilians living on military bases. In 2004-2006, NSDUH collected data from 203,870 respondents aged 12 or older and was designed to obtain representative samples from all 50 States and the District of Columbia. The survey was planned and managed by SAMHSA's Office of Applied Studies (OAS), and data collection was conducted under contract with RTI International.¹

This report marks the third time that detailed estimates for substate regions (also referred to as planning regions, substate areas, or regions) in all 50 States and the District of Columbia have been presented by SAMHSA. The first report to provide such estimates used data from the 1999-2001 surveys (OAS, 2005b). The second report presented estimates for 22 measures or outcomes based on the 2002-2004 NSDUHs (OAS, 2006). This report presents estimates for 23 measures of substance use or mental health problems based on the 2004-2006 NSDUHs. In addition to the 22 measures reported in the 2002-2004 substate report, a past year major depressive episode (MDE) measure was added to this report. These reports provide a more detailed perspective on the variations in substance use rates both within and across States than was possible with prior State reports (e.g., Hughes, Sathe, & Spagnola, 2008; Wright, Sathe, & Spagnola, 2007).

Estimates were generated for 363 substate regions representing collectively the 50 States and the District of Columbia. These regions were defined by officials from each State and were typically based on the substance abuse treatment planning regions specified by the States in their applications for the Substance Abuse Prevention and Treatment (SAPT) Block Grant administered by SAMHSA.

A.2. Format of the Report

Section A provides a brief background on the survey, how substate regions were formed, the general methodological approach, and a brief discussion of selected findings. A complete list of the 23 substance use measures presented in this report is given in Section B, which also provides further information on the small area estimation methodology used to develop substate estimates. Section C includes tables with estimates for each of the 23 measures and the corresponding prediction intervals for all substate regions. It also contains a set of national maps that show the prevalence of each outcome measure for each substate region. The substate regions in the tables in Section C have been ordered alphabetically within each State. There are 10 separate tables, each having two or three related substance use or mental health measures. Estimates for aggregate regions (also specified by certain States) also are included in these tables. Section D contains definitions of the substate regions. Section E includes the population counts for persons aged 12 or older and the combined 2004, 2005, and 2006 NSDUH sample sizes and response rates for each substate region. Users may find the population counts helpful in calculating the weighted average prevalence estimate for any combination of substate regions or to determine the number of people using a particular substance in a substate region. For example, the number of persons aged 12 or older who used marijuana in the past month in Alabama's Region 1 (41,956 persons) could be obtained by multiplying the prevalence rate from Table C2 (4.0 percent—shown as 3.96 percent in the table) and the population estimate from Table E1 (1,059,503). Section F lists the references.

A.3. Overview: Substate Regions, Ranking Regions, and Small Area Estimation Methods

The substate regions for each State were developed in a series of communications between SAMHSA staff and State treatment representatives in late 2007. The goal of the project was to provide substate-level estimates showing the geographic distribution of substance use prevalence for regions that States would find useful for treatment planning purposes.²The final substate region boundaries were based on the State's recommendations assuming that the NSDUH sample sizes were large enough to provide estimates with adequate precision. Most States defined regions in terms of counties or groups of counties. A few States defined the regions in terms of census tracts. Several States also requested estimates for aggregate planning regions along with the estimates for their substate planning regions. An aggregate planning region is made up of two or more substate planning regions. A few of these States wanted the maps to be produced for the aggregate regions instead of their substate planning regions. For example, New York has 15 substate regions, and those 15 regions were combined to create 4 aggregate regions that are used in the maps. Hence, for each measure in this report, maps were produced for 345 planning regions and not for 363 regions. The discussion of findings in this chapter (Section A.5) also is restricted to these 345 planning regions.

These 345 substate regions used in the maps were ranked from lowest to highest for each measure and were divided into seven categories designed to represent distributions that are somewhat symmetric, like a normal distribution. Colors were assigned to all regions such that the third having the lowest prevalence are in blue, the middle third are in white, and the third with the highest prevalence are in red. The only exceptions were the three perception-of-risk outcomes, which have the highest estimates represented in blue and the lowest represented in red. To further discriminate among the regions that display relatively higher prevalence, the "highest" third has been further divided into three categories: dark red for the 15 substate regions with the highest estimates, medium red for the 31 substate regions with the next highest, and light red for the 69 substate regions in the third highest group. The "lowest" third is categorized in a similar way using three distinct shades of blue. Due to tied values of prevalence, the number of substate regions in each category may vary a little.

In addition to this report, the following substate region tables will be available on SAMHSA's website:

• Age Group Tables, by Substate Region: Tables of prevalence estimates for youths aged 12 to 17, young adults aged 18 to 25, adults aged 26 or older, and persons aged 18 or older for each substance use measure for substate regions having sufficient precision.
• Substate Region Age Group Comparison Table: A table showing ranges of prevalence estimates for each outcome categorized into seven groups from lowest to highest for age groups 12 to 17, 18 to 25, 26 or older, and 18 or older.
• State-Specific Substate Region Tables and Maps: State-specific tables and maps showing substate region estimates for each State separately for all persons aged 12 or older.
• Highest and Lowest Substate Region Estimates within State: Tables showing substate regions with the highest and lowest estimates for each measure within a State, along with an indication of statistical significance of the difference.
• Comparison of 2002-2004 and 2004-2006 Substate Region Estimates: Tables showing the 2002-2004 and 2004-2006 substate estimates, along with an indication of statistical significance of the difference among regions, with common geographic definitions between the two time periods.

These additional tables will be posted at http://oas.samhsa.gov/substate.cfm as they become available.

Estimates in this report are based on hierarchical Bayes estimation methods that combine survey data with a national model. Applying this methodology to the State substance use measures has been shown to result in more precise estimates than using the sample-based results alone (Wright, 2002). The methodology used to produce estimates in this report is the same as that used to produce State estimates from the NSDUH data since 1999 and has been used for prior substate reports (see OAS, 2005b, 2006). Sample data have been combined across 3 years (2004-2006) in this report to improve the precision of substate region estimates. The estimate for each region is accompanied by a 95 percent prediction interval (for more details, see Section B, Substate Region Estimation Methodology).

In addition to the substate region estimates, comparable estimates are provided for the 50 States and the District of Columbia using the same methodology. Because these estimates are based on 3 consecutive years of data, they are not directly comparable with the State estimates in earlier reports that are based on only 2 consecutive years. Estimates for the Nation and the four census regions also are presented. These regions, defined by the U.S. Census Bureau, are defined as follows:

Northeast Region - Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, and Vermont.

Midwest Region - Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin.

South Region - Alabama, Arkansas, Delaware, District of Columbia, Florida, Georgia, Kentucky, Louisiana, Maryland, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia, and West Virginia.

West Region - Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, and Wyoming.

Because the small area estimation methods used here tend to borrow strength from both the national model and the State-level random effects, estimates for substate regions with sample sizes that were closer to the minimum (200) tend to be shrunk more toward the corresponding State prevalence estimate than substate regions with large sample sizes. This methodology tends to cluster the small sample substate estimates around their State means. Thus, relatively high estimates for small substate regions tend to shrink toward the State mean, while relatively low estimates tend to increase toward the State mean. On the other hand, for regions with large sample sizes, the methodology produces estimates that are close to the weighted average of the sample data. In addition, these estimates are design consistent so that as the sample size for a substate region increases, the estimate approaches the true population value.

A.4. Comparability with Past Estimates

For the 2002 NSDUH, a number of methodological changes were introduced, including a $30 incentive for participating in the survey, additional training for interviewers to encourage adherence to survey protocols, a change in the survey name, and a shift to the 2000 decennial census (from the 1990 census) as a basis for population counts used in estimation. An unanticipated result of these changes was that the prevalence rates for 2002 were in general substantially higher than those for 2001. These rates were substantially higher than could be attributable to the usual year-to-year trend. Additional information on these methodological changes is available in OAS (2005a).

Because of the changes in the survey that took place in 2002, estimates for 2004-2006 are not comparable with estimates for 1999-2001, and it is not possible to separate the effect of the methodological changes from the true trends in substance use. Therefore, one should not conclude that any differences between estimates from 1999-2001 and 2004-2006 represent true changes. However, estimates from 2002-2004 and 2004-2006 are comparable for outcomes that were defined in a similar manner and for substate regions defined consistently across these time periods. Such estimates of change will be posted at http://oas.samhsa.gov/substate.cfm as they become available.

There were 10 States that made changes to their 2002-2004 substate regions for producing the 2004-2006 estimates: Arkansas, Florida, Hawaii, Maine, Maryland, Michigan, New York (made changes to aggregate regions only), North Carolina, South Dakota, and Wisconsin. Except for South Dakota, all of these States had at least one substate region definition that was the same as the 2002-2004 definition. The remaining 40 States and the District of Columbia did not change their definitions of substate regions or aggregate regions. Pennsylvania, however, renamed its substate regions, but the definitions (based on counties) did not change from those shown in the 2002-2004 report.