This report also compares estimates of AMI based on combined 2012–2014 NSDUH data to estimates based on combined 2010–2012 NSDUH data. All changes across time in this report are statistically significant at the .05 level. Estimates are annual averages based on combined 2012–2014 NSDUH data from about 142,000 respondents aged 18 or older. Estimates were derived from a complex statistical model (i.e., small area estimation) in which NSDUH substate data were combined with other local area data to enhance statistical power and analytic capability.⁷

State Estimates of Any Mental Illness

As described previously, the 50 states and the District of Columbia were divided into quintiles based on the percentage of the population with past year AMI. As shown in Table 1, the 10 states in the lowest quintile of estimates of past year AMI included New Jersey (15.83 percent), Illinois (16.36 percent), North Dakota (16.45 percent), Florida (16.46 percent), South Dakota (16.46 percent), Maryland (16.80 percent), Connecticut (16.84 percent), Texas (17.33 percent), Hawaii (17.35 percent), and Iowa (17.40 percent).

The 10 states in the highest quintile of estimates of past year AMI included Oregon (22.72 percent), Utah (21.72 percent), West Virginia (21.70 percent), Maine (21.30 percent), Rhode Island (21.11 percent), Idaho (20.77 percent), New Hampshire (20.54 percent), Oklahoma (20.47 percent), Tennessee (20.29 percent), and Arkansas (20.13 percent).

This report also compares the combined 2012–2014 state estimates of past year SMI with 2010–2012 estimates of SMI to examine changes over time. The 2010–2012 data are based on information obtained from 138,300 adults aged 18 or older. The inclusion of a common year (i.e., 2012) in these comparisons increases the precision of the estimates and the ability to detect statistically significant differences between the two periods. Statistically significant differences between 2010–2012 and 2012–2014 indicate average annual change between 2010–2011 and 2013–2014. It is not possible to examine changes over time at the substate level because of changes to substate boundaries by the states between 2010–2012 and 2012–2014.

When the 2010–2012 state estimates of past year AMI for adults aged 18 or older were compared with the 2012–2014 state estimates, 4 states (California, Maine, North Carolina, and Rhode Island) experienced a statistically significant increase in the percentage of adults with past year AMI. The remaining 46 states and the District of Columbia experienced no change in the percentage of adults with AMI in the past year (Table 2). No states experienced declines.

SAMHSA works with state substance abuse/mental health agency representatives to define substate areas that meet state needs and reporting requirements while ensuring that the NSDUH sample sizes are large enough to provide estimates with adequate precision.¹¹ Combined 2012–2014 NSDUH data can be used to estimate past year AMI in 362 substate regions. The 2012–2014 estimates in this report are based on substate boundaries that reflect current state needs and reporting requirements and may not be comparable with estimates from substate regions from prior years. For substate region definitions, see the "2012–2014 National Survey on Drug Use and Health Substate Region Definitions" at https://www.samhsa.gov/data/. In most states, the substate regions are defined in terms of single counties or groups of counties; in some states, the regions are defined entirely in terms of census tracts (in Connecticut, the District of Columbia, and Massachusetts), parishes (in Louisiana), boroughs/census areas (in Alaska), a combination of counties and census tracts (in California and Delaware), and a combination of counties and independent cities (in Maryland, Missouri, Nevada, and Virginia).

Substate region estimates for past year AMI among people aged 18 or older are displayed on a U.S. map (Figure 2). In Table S1, substate region estimates are shown to two decimal places and are listed alphabetically by state. To produce the substate map in Figure 2, the substate estimates of past year nonmedical use of prescription pain relievers were ordered from highest to lowest percentage and were then divided into three approximately equal groups based on their percentage. There are 121 substate regions in the lowest third (i.e., with the lowest percentages) and there are 121 substate regions in the highest third (i.e., with the highest percentages). There are 120 substate regions in the middle third. The highest and lowest thirds were subdivided into thirds to further distinguish among the substate regions. Overall, the seven groups in each map were constructed to represent a somewhat symmetrical distribution.¹² In some cases, a category could have more or fewer substate regions because two (or more) substate regions have the same estimate (to two decimal places). When such ties occurred at the "boundary" between two groups, all substate regions with the same estimate were assigned to the lower group. Individual state maps at https://www.samhsa.gov/data/ provide more granularity in areas too small to display clearly on the U.S. maps. Table 2 provides estimates associated with each map. Ninety-five percent confidence intervals are included as a measure of precision for each estimate.¹³

Among the substate regions, estimates of past year AMI ranged from 23.95 percent in Oregon’s Region 3 located in the northwestern part of the state to 14.53 percent in Florida’s Southern region (Circuits 11 and 16) consisting of Miami-Dade and Monroe Counties. Of the 17 substate regions with the lowest estimates of past year AMI, 6 were in the Midwest (2 in North Dakota, 2 in South Dakota, and 2 in Illinois), 6 were in the South (3 in Florida, 2 in Maryland, and 1 in Virginia), and 5 were in the Northeast (3 in New Jersey, 1 in Connecticut, and 1 in Pennsylvania).

Of the 15 substate regions with the highest estimates of past year AMI, 8 were in the West (4 in Oregon, 3 in Utah, and 1 in Idaho), 4 were in the Northeast (2 in Maine, 1 in Rhode Island, and 1 in Vermont), 2 in the South (both in West Virginia), and 1 was in the Midwest (Ohio).

Among the substate regions, estimates of past year AMI ranged from 23.95 percent in Oregon’s Region 3 located in the northwestern part of the state to 14.53 percent in Florida’s Southern region (Circuits 11 and 16) consisting of Miami-Dade and Monroe Counties. Of the 17 substate regions with the lowest estimates of past year AMI, 6 were in the Midwest (2 in North Dakota, 2 in South Dakota, and 2 in Illinois), 6 were in the South (3 in Florida, 2 in Maryland, and 1 in Virginia), and 5 were in the Northeast (3 in New Jersey, 1 in Connecticut, and 1 in Pennsylvania).

Of the 15 substate regions with the highest estimates of past year AMI, 8 were in the West (4 in Oregon, 3 in Utah, and 1 in Idaho), 4 were in the Northeast (2 in Maine, 1 in Rhode Island, and 1 in Vermont), 2 in the South (both in West Virginia), and 1 was in the Midwest (Ohio).

The previous sections examined 2012–2014 NSDUH state and substate past year AMI estimates individually. Some substate areas are too small to display clearly on the U.S. national maps (Figure 2); therefore, individual state maps are particularly useful for seeing these small substate areas. SAMHSA produces individual NSDUH state maps that display the substate estimates of past year AMI and other behavioral health measures. In this section, one of the individual state maps is presented to illustrate the variability within states. For more state-specific NSDUH maps, see https://www.samhsa.gov/data/sites/default/files/NSDUHsubstateStateTabs2014/NSDUHsubstateSpecificStatesTOC2014.htm.

As previously noted, the assignments of the substate areas within states were created by dividing 362 substate regions, nationally, into 7 groups based on their past year AMI percentages. Figure 2 shows that states that are in the highest and lowest quintiles tend to be fairly uniform across the substate estimates. That is, states with the highest percentages of past year AMI tend to have substate areas with high percentages of past year AMI. For example, 6 of the 10 states in the highest quintile of AMI estimates had substate estimates that were all in the highest third. When all of the substate areas are in the same third, this is a probable indicator of low variability within those states. Likewise, 7 of the 10 states in the lowest quintile of AMI estimates had substate-level estimates that were all in the lowest third, indicating low variability within those states.

Across the states and the District of Columbia, the most variability in substate estimates occurred within states in the middle quintile. Stated another way, the states in the middle third in Figure 1 had the most variation at the substate level in Figure 2. Of the 11 states in the middle quintile, 8 states had substate-level estimates of past year AMI that were in the highest, middle, and lowest third, which may indicate more variability. An example of this variability can be seen in California (Figure 3).

In California, past year AMI for adults aged 18 or older ranged from 21.19 percent in Region 1R, consisting of the 15 counties in the northern section of the state to 16.26 percent in San Mateo (Region 9R). In California, 4 substate regions were in the highest third (Regions 1R, 20R, 15R, and LA SPA 2), indicating higher estimates of AMI. Nine substate regions were the lowest third (Regions 7R, 13 and 19R, 18R, 6, 14, and 9R, and LA SPAs 3, 8, and 7), indicating lower estimates of AMI. The remaining 13 substate regions (Regions 3R, 8R, 16R, 2R, 12R, 5R, 21R, 10, 17R, and 4R, and LA SPAs 4, 6, and 1 and 5) fell in the middle third. 

Although the prevalence of adults with mental illness among the states ranges widely, it is important to note that there are many people with AMI in every state, which is expected because mental illness is common.¹⁴ Substate regions with highest estimates of past year AMI among adults aged 18 or older are located in all U.S. regions. The presence of AMI in every state reinforces that mental illness is a major public health concern in the United States. Factors that potentially contribute to the variation may need further study; however, policymakers can use state- and substate-level information to help inform their assessments of mental health needs in their communities. Maps and tables presented in this report can help state policymakers quickly see where efforts are needed to address mental health in their state. For example, substate regions within states can vary in the prevalence of past year AMI (e.g., California). As data from several years of NSDUHs are accumulated, in-depth analysis of these state and substate data will continue to provide insight into the patterns of mental illness, such as variations over time and by age and gender within each state. SAMHSA provides information about where to find mental health treatment at https://findtreatment.samhsa.gov/. 

Other NSDUH Substate Measures

The combined 2012–2014 NSDUH estimates for past year AMI for adults aged 18 or older are available, along with 25 additional behavioral health measures for 384 substate areas, 50 states and the District of Columbia, 4 census regions, and the United States. Information on the methodology that generated these estimates is available online at https://www.samhsa.gov/data/. This report discusses one of the measures for the 362 substate areas displayed on the maps. The 25 additional measures include substance use and mental health issues, including use of illicit drugs, alcohol, and tobacco; substance use disorders; needing but not receiving treatment for a substance use problem; serious mental illness; depression; and suicidal thoughts. Also provided are national maps for all measures and detailed tables including percentages for each substate region, state, census region, and the nation for people aged 12 or older; tables by age group; and state-specific tables and maps. The state maps are particularly useful in areas too small to display clearly on the U.S. maps.

1. Center for Behavioral Health Statistics and Quality. (2013). Behavioral Health, United States, 2012 (HHS Publication No. SMA 13-4797). Rockville, MD: Substance Abuse and Mental Health Services Administration. Retrieved from https://www.samhsa.gov/data/  

2. Center for Behavioral Health Statistics and Quality. (2015). Behavioral health trends in the United States: Results from the 2014 National Survey on Drug Use and Health (HHS Publication No. SMA 15-4927, NSDUH Series H-50). Retrieved from https://www.samhsa.gov/data/   

3. Public Law No. 102-321, the Alcohol, Drug Abuse, and Mental Health Administration Reorganization Act of 1992, established a block grant for states within the United States to fund community mental health services for adults with serious mental illness. The law required states to include prevalence estimates in their annual applications for block grant funds. This legislation also required SAMHSA to develop an operational definition of serious mental illness. Information about SAMHSA's block grant programs can be found at https://www.samhsa.gov/. It should be noted that SAMHSA updated the definition of serious mental illness for use in mental health block grants to include mental disorders as specified in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders.

4. A discussion of the methodology used to generate AMI estimates can be found in Appendix B of the 2014 NSDUH mental health findings report. For information on mental illness and mental health service utilization, see Chapter 2 in the 2012 NSDUH mental health findings report.

5. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (DSM-IV) (4th ed.). Washington, DC: Author.

6. SAMHSA has been publishing estimates of the prevalence of past year serious mental illness and AMI among adults aged 18 or older since the release of the 2008 NSDUH national findings report. Estimates were based on a model developed in 2008. In 2013, SAMHSA developed a more accurate model for the 2012 data and subsequently revised the serious mental illness and AMI estimates for 2008, 2009, 2010, and 2011 based on the 2012 model. The combined 2010–2012 and 2012-2014 state and substate estimates in this report are based on the 2012 model. For additional information, see the NSDUH short report titled Revised Estimates of Mental Illness from the National Survey on Drug Use and Health at https://www.samhsa.gov/data/2k13/NSDUH148/sr148-mental-illness-estimates.pdf.

7. Estimates presented in this report are derived from a hierarchical Bayes model-based small area estimation (SAE) procedure in which NSDUH data at the substate level are combined with local area county and census block group/tract-level data from the area to provide more precise estimates of substance use and mental health outcomes. The precision of the SAE estimates can be improved significantly by combining data across 3 years (i.e., 2012 to 2014). With 3 years of combined NSDUH data, the sample sizes in the 362 substate regions ranged from 100 people to approximately 3,500 people.

8. In some cases, a "quintile" could have more or fewer states than desired because two (or more) states have the same estimate (to two decimal places). When such ties occurred at the "boundary" between two quintiles, all states with the same estimate were assigned to the lower quintile.

9. In this report, state estimates are discussed in terms of their observed rankings because they provide useful context. However, a state having a highest or lowest rate does not imply that the state's rate is significantly higher or lower than the rate of the next highest or lowest state. Similarly, the quintiles were not selected to represent statistical differences across quintiles or to correspond to proximity to a target public health threshold for a particular measure. For example, the division of states into quintiles does not indicate that states in the same quintile are statistically similar to each other. While a nearly equal number of states are contained in each quintile, the size of the intervals (i.e., the difference between the upper and lower limits of each quintile) that define the map boundaries is not necessarily uniform across each quintile.” When comparing two state prevalence rates, the method of overlapping confidence intervals is more conservative (i.e., it rejects the null hypothesis of no difference less often) than the standard method based on Z statistics when the null hypothesis is true. Even if confidence intervals for two states overlap, the two estimates may be declared significantly different by the test based on Z statistics. Hence, the method of overlapping confidence intervals is not recommended to test the difference of two state estimates. A detailed description of the method of overlapping confidence intervals and its comparison with the standard methods for testing of a hypothesis is given in the following articles: (a) Schenker, N., & Gentleman, J. F. (2001). On judging the significance of differences by examining the overlap between confidence intervals. American Statistician, 55(3), 182–186. (b) Payton, M. E., Greenstone, M. H., & Schenker, N. (2003). Overlapping confidence intervals or standard error intervals: What do they mean in terms of statistical significance? Journal of Insect Science, 3, 34. For details on a more accurate test to compare state prevalence estimates, please see Section B.12 in Appendix B of 2011-2012 National Survey on Drug Use and Health: Guide to state tables and summary of small area estimation methodology, located at https://www.samhsa.gov/data/report/2011-2012-nsduh-state-estimates-substance-use-and-mental-disorders.

10. The West has 13 states: AK, AZ, CA, CO, HI, ID, MT, NM, NV, OR, UT, WA, and WY. The South has 16 states plus the District of Columbia: AL, AR, DE, FL, GA, KY, LA, MD, MS, NC, OK, SC, TN, TX, VA, and WV. The Northeast has 9 states: CT, MA, ME, NH, NJ, NY, PA, RI, and VT. The Midwest has 12 states: IA, IL, IN, KS, MI, MN, MO, ND, NE, OH, SD, and WI.

11. Substance use and mental health officials from each of the 50 states and the District of Columbia typically define these substate areas to correspond to areas reported in their applications for the Substance Abuse Prevention and Treatment Block Grant (SABG) administered by SAMHSA. The SABG program provides financial and technical assistance to the 50 states, the District of Columbia, and other jurisdictions to support substance abuse prevention and treatment programs and to promote public health. States use NSDUH substate estimates for a variety of purposes, including strategic planning and program development, production of epidemiological profiles for briefing state legislatures and informing the public, allocation of funds to areas based on the need for services, and other uses.

12. The seven categories were not selected to represent statistical differences across categories or to correspond to proximity to a target public health threshold for a particular measure. For example, the division of substate regions into seven categories does not indicate that substate regions in the same category are statistically similar to each other. Furthermore, the size of the intervals (i.e., the difference between the upper and lower limits of each category) that define the map boundaries is not necessarily uniform across each category. The substate areas are uniquely defined based on the needs of each state and may not be demographically or geographically comparable to substate areas in other states.

13. When comparing two substate region percentages, the method of overlapping confidence intervals is more conservative (i.e., it rejects the null hypothesis of no difference less often) than the standard method based on Z statistics when the null hypothesis is true. Even if confidence intervals for two substate regions overlap, the two estimates may be declared significantly different by the test based on Z statistics. Hence, the method of overlapping confidence intervals is not recommended to test the difference of two substate region estimates. As percentages are standardized, they do not inform a reader when two states or substates have the same percentage but different population sizes.

14. World Health Organization. (2001). Strengthening mental health promotion (Fact sheet no. 220). Geneva, Switzerland: Author.

Lipari, R.N., Van Horn, S., Hughes, A. and Williams, M. State and substate estimates of any mental illness from the 2012–2014 National Surveys on Drug Use and Health. The CBHSQ Report: July 20, 2017. Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration, Rockville, MD.