Comparing and Evaluating Substance Use Treatment Utilization Estimates from the National Survey on Drug Use and Health and Other Data Sources

1.1. Background

Substance use disorder (alcohol or drug dependence or abuse) is a serious public health threat in the United States and is associated with psychosocial and legal problems, such as mental and physical health conditions, lower educational attainment, underemployment or unemployment, involvement with the criminal justice system, victimization by and perpetration of violence, and homelessness (Chermack, Fuller, & Blow, 2000; Hser, Hoffman, Grella, & Anglin, 2001; Kubiak, Arfken, Swartz, & Koch, 2006; Mansell et al., 2006; Mertens, Lu, Parthasarathy, Moore, & Weisner, 2003; Stein, Dixon, & Nyamathi, 2008; CBHSQ, 2012). Substance use treatment has been shown to be effective in reducing substance use and can produce positive psychosocial and physical outcomes for persons who receive it (O'Brien & McLellan, 1996). In general, persons with substance dependence who receive treatment have been found to experience less disability related to substance use in their lifetime than untreated persons (Hasin, Stinson, Ogburn, & Grant, 2007). Given the public health implications of substance use treatment, obtaining accurate counts of the number of persons who need and receive substance use treatment is critical.

1.2. Overview of Data Sources

1.2.1 NSDUH

NSDUH, formerly called the National Household Survey on Drug Abuse (NHSDA), is the Federal Government's primary source of information on the nature and extent of substance use and abuse in the United States. Conducted since 1971, the survey collects data by administering questionnaires to a representative sample of persons aged 12 or older at their places of residence. The survey is sponsored by SAMHSA and is planned and managed by CBHSQ. Data collection is currently conducted under contract with RTI International, Research Triangle Park, North Carolina.¹ Approximately 67,500 persons are surveyed each year through face-to-face interviews. Data from the survey are used extensively by policymakers and researchers to measure the prevalence and correlates of licit and illicit substance use; to identify and monitor trends in substance use, treatment need, and treatment; and to analyze differences in substance use patterns by population subgroups. NSDUH also collects data on mental disorders and mental health treatment. This section briefly describes the survey methodology; Appendix A provides a more complete description. Reports on results from NSDUH data are available on the SAMHSA Web site (https://www.samhsa.gov/data/data-we-collect/nsduh-national-survey-drug-use-and-health).

NSDUH collects information from residents of households and noninstitutional group quarters (e.g., shelters, rooming houses, dormitories) and from civilians living on military bases. The survey excludes homeless persons who do not use shelters, military personnel on active duty, and residents of institutional group quarters, such as jails and hospitals.

From 1971 through 1998, the survey used paper-and-pencil data collection. Since 1999, the NSDUH interview has been carried out in English or Spanish using computer-assisted interviewing (CAI). Most of the questions are administered with audio computer-assisted self-interviewing (ACASI). ACASI is designed to provide the respondent with a highly private and confidential mode for responding to questions in order to increase the level of honest responses to questions about illicit drug use and other sensitive behaviors. Less sensitive items are administered by interviewers using computer-assisted personal interviewing (CAPI).

This review uses NSDUH data collected in the 6-year period from 2005 through 2010. Multiple years of data were used in order to provide a sufficient sample size to produce estimates for demographic subgroups. During those years, NSDUH employed a State-based design with an independent, multistage area probability sample within each State and the District of Columbia. The 8 States with the largest populations (California, Florida, Illinois, Michigan, New York, Ohio, Pennsylvania, and Texas, which together account for about half of the total U.S. population aged 12 or older) were designated as "large sample States" and had an annual sample size of about 3,600 respondents each. For the remaining 42 States and the District of Columbia, the annual sample size was about 900 respondents per State. The design oversampled adolescents aged 12 to 17 and young adults aged 18 to 25.

A 5-minute screening procedure using a handheld computer involves listing all household members along with their basic demographic data. Depending on the composition of the household, zero to two persons in each household are selected for an interview. The household screening response rates for the survey years 2005 to 2010 ranged from 88.8 percent in 2010 to 91.3 percent in 2005. Interview response rates for persons aged 12 to 17 ranged from 84.8 percent in 2007 to 87.1 percent in 2005; interview response rates for persons aged 18 to 25 ranged from 79.8 percent in 2007 to 83.1 percent in 2005; and interview response rates of persons aged 26 or older ranged from 71.4 percent in 2005 to 73.5 percent in 2007. Appendix B of the 2012 NSDUH report on national findings discusses how NSDUH response rates are calculated (see CBHSQ, 2013).

Because of the change in interviewing mode in 1999, the estimates from the pre-1999 surveys are not comparable with estimates from the current CAI-based surveys. Although the design of the 2002 through 2010 NSDUHs is similar to the design of the 1999 through 2001 surveys, there are also important methodological differences that affect the comparability of the 2002 to 2010 estimates with estimates from prior surveys. The most important change was the incentive payment that started in 2002 and continued in subsequent surveys. Each NSDUH respondent completing the interview is given $30. Also, the name of the survey was changed in 2002, from National Household Survey on Drug Abuse (NHSDA) to National Survey on Drug Use and Health (NSDUH). Improved data collection quality control procedures were introduced in the survey starting in 2001, and updated population data from the 2000 decennial census were incorporated into the sample weights beginning with the 2002 estimates. Analyses of the effects of these factors on NSDUH estimates have shown that 2002 and later data should not be compared with 2001 and earlier data from the survey series to assess changes over time. Appendix C of the 2004 NSDUH report on national findings discusses these changes to the survey in more detail (see OAS, 2005).

NSDUH questions about receipt of substance use treatment are asked of all respondents who report having used alcohol or an illicit substance at least once during their lives. In addition to alcohol, the illicit drugs include the following:

• marijuana or hashish,
• cocaine (including crack),
• heroin,
• hallucinogens,
• inhalants,
• prescription pain relievers used nonmedically (i.e., that were not prescribed for the respondent or were taken only for the experience or feeling they caused),
• prescription tranquilizers used nonmedically,
• prescription stimulants used nonmedically, and
• prescription sedatives used nonmedically.

Data analyzed in this report include those from interviews completed with persons aged 12 or older in the 2005, 2006, 2007, 2008, 2009, and 2010 NSDUHs. Approximately 406,900 completed interviews were obtained from persons aged 12 or older in the 2005 to 2010 NSDUHs.

1.2.2 TEDS

TEDS is a compilation of data detailing the demographic and substance use characteristics of admissions to and discharges from substance use treatment. TEDS is part of the Behavioral Health Services Information System (BHSIS), formerly known as the Drug and Alcohol Services Information System (DASIS), a cooperative program among SAMHSA and State substance abuse agencies to collect data on substance use treatment services. BHSIS is coordinated and managed by CBHSQ. TEDS data collection and analysis are currently conducted under contract with Synectics for Management Decisions, Inc. This data collection effort was developed in response to the 1988 Comprehensive Alcohol Abuse, Drug Abuse, and Mental Health Amendments (P.L. 100-690), which established a revised Substance Abuse Prevention and Treatment Block Grant (SAPTBG) and mandated Federal data collection on clients receiving treatment for either alcohol or drug abuse. This section briefly describes the data collection methodology; Appendix A provides a more complete description. Reports on results from TEDS data are available on the SAMHSA Web site (https://www.samhsa.gov/data/data-we-collect/teds-treatment-episode-data-set).

The TEDS system comprises two major components—the Admissions Data Set and the Discharge Data Set—which are linked at the record level for approximately 85 percent of admissions included in these analyses. Information on substance use treatment admissions and discharges that State administrative systems routinely collect is submitted to TEDS in a standard format on a monthly or quarterly schedule. Data are reported on approximately 2 million annual admissions to treatment by approximately 10,000 facilities, programs, or administrative units in the 50 States, the District of Columbia, and Puerto Rico.

The scope of facilities included in TEDS is affected by differences in State licensure, certification, and accreditation practices, as well as disbursement of public funds. Most State substance abuse agencies require facilities that receive any State/public funding (including Federal block grant funds) for the provision of alcohol and/or drug treatment services to report TEDS data to the State. States differ widely in the amount of public funding available for substance use treatment and in the constraints placed on the use of funds. Some State substance abuse agencies license or monitor facilities operating in their State that do not receive any public funding and also require them to report TEDS data. In States where not all facilities are required to report TEDS data, some facilities do so voluntarily. Facilities operated by Federal agencies (e.g., the Bureau of Prisons [BOP], the Department of Defense [DOD], and the Department of Veterans Affairs [VA]) generally do not report TEDS data to the State, although some facilities operated by the Indian Health Service (IHS) are included. Hospital-based substance use treatment facilities are frequently not licensed through the State substance abuse agency (Single State Authority [SSA]) and typically do not report TEDS data. To the extent that hospital-based facilities do report to TEDS, the records include those from specialty substance use treatment units within hospitals. Correctional facilities (State prisons and local jails) report TEDS data in some States but not in others. A table summarizing the State data system reporting characteristics for 2007, 2008, and 2009 is provided in Appendix A (see Table A.1).

The primary goal of TEDS is to monitor the characteristics of clients admitted to planned, continuing treatment regimens. Thus, TEDS excludes early intervention and crisis intervention programs that do not lead to enrollment in continued treatment.

This review focuses on TEDS data collected for the years 2007, 2008, and 2009. TEDS data used in this report are from the Admissions Minimum Data Set, a core set of data elements collected at admission by all States, and the Discharge Data Set, for which 44 States, the District of Columbia, and Puerto Rico reported clients discharged in 2007; 48 States, the District of Columbia, and Puerto Rico reported clients discharged in 2008; and 48 States, the District of Columbia, and Puerto Rico reported clients discharged in 2009. Ultimately, analyses were conducted on data from 47 States. Data from Alabama, Alaska, the District of Columbia, and Georgia were not included in these analyses because they submitted no or incomplete data for 1 or more of the 3 years analyzed. In order to establish as much comparability between TEDS and NSDUH as possible, data from Puerto Rico are not included in the estimates from TEDS shown in this report.

The TEDS Minimum Data Set includes the following information for each admission: date of admission; type of service; age; gender; race; ethnicity; number of prior treatment episodes; education; employment status; principal source of referral; primary, secondary, and tertiary substance problems and their usual route of administration, frequency of use, and age of first use; and planned use of medication-assisted opioid therapy. Information on use of primary, secondary, and tertiary problem substances is collected for the following substances:

• alcohol;
• cocaine/crack;
• marijuana/hashish;
• heroin;
• nonprescription methadone;
• other opiates and synthetics (including codeine, hydrocodone, hydromorphone, meperidine, morphine, opium, oxycodone, pentazocine, propoxyphene, tramadol, and any other drug with morphine-like effects);
• phencyclidine (PCP);
• other hallucinogens (including but not limited to LSD, DMT, STP, hallucinogens, mescaline, peyote, psilocybin);
• methamphetamine;
• other amphetamines (including amphetamines, MDMA, phenmetrazine, and other unspecified amines and related drugs);
• other stimulants (including methylphenidate and any other stimulants);
• benzodiazepines (including alprazolam, chlordiazepoxide, clonazepam, clorazepate, diazepam, flunitrazepam, flurazepam, halazepam, lorazepam, oxazepam, prazepam, temazepam, triazolam, and other unspecified benzodiazepines);
• other nonbenzodiazepine tranquilizers (including meprobamate and other nonbenzodiazepine tranquilizers);
• barbiturates (including but not limited to amobarbital, pentobarbital, secobarbital);
• other nonbarbiturate sedatives or hypnotics (including chloral hydrate, ethchlorvynol, glutethimide, methaqualone, and other nonbarbiturate sedatives or hypnotics);
• inhalants (including but not limited to chloroform, ether, gasoline, glue, nitrous oxide, paint thinner);
• over-the-counter medications (including aspirin, cough syrup, diphenhydramine and other antihistamines, sleep aids, any other legally obtained nonprescription medication); and
• other (including but not limited to diphenylhydantoin, GHB/GBL, ketamine).

TEDS observations represent admissions to or discharges from substance use treatment, not individuals. For example, one individual admitted to treatment twice within a calendar year would be counted as two separate admissions. For the purposes of comparability with the other data sets, we have used data from multiple years of the TEDS Admission Data Set, Discharge Data Set, and the unique person identifiers to produce estimates of the unduplicated number of persons who received treatment. Admission and discharge records linked by a unique identifier represent clients who had been both admitted and discharged. Admission records with no linked discharge should indicate that a client had been admitted but not yet discharged. However, because the discharge system was not fully operational in all States during the years examined, it was assumed that a number of these admissions had, in fact, been discharged. For each admission with no linked discharge, a probability of having been discharged was computed in order to provide estimates of the unduplicated number of persons who received treatment. More information on this is provided in Section 1.3 of this report.

1.2.3 N-SSATS

N-SSATS is an annual census of all known drug and alcohol abuse treatment facilities in the United States. N-SSATS, along with TEDS, is part of BHSIS, a cooperative program between SAMHSA and State substance use treatment agencies to collect data on substance abuse services. BHSIS is coordinated and managed by CBHSQ. N-SSATS data collection and analysis are currently conducted under contract with Synectics for Management Decisions, Inc., and Mathematica Policy Research, Inc.

N-SSATS is a paper/Web/telephone census designed to collect information from all facilities in the United States, both public and private, that provide substance use treatment. N-SSATS is designed to collect data on the location, characteristics, and utilization of services at public and private alcohol and drug use treatment facilities throughout the 50 States, the District of Columbia, and other U.S. jurisdictions. N-SSATS provides a mechanism for tracking the changing character and composition of the U.S. substance use treatment delivery system. The N-SSATS questionnaire covers the characteristics of the treatment facility, including client payment sources, services provided, hospital and residential capacity, as well as the number of treatment admissions in the past 12 months and a single-day client census. The objective of the census is to collect multipurpose data that can be used to

• assist SAMHSA and State/local governments in quantifying the nature and extent of services provided in State-supported and other substance use treatment facilities and in forecasting substance use treatment resource requirements;
• update SAMHSA's Inventory of Behavioral Health Services (I-BHS), formerly known as Inventory of Substance Abuse Treatment Services (I-SATS), which includes all known drug and alcohol use treatment facilities and mental health treatment facilities; I-BHS is used as a sampling frame for N-SSATS as well as other special surveys of treatment providers and their clients; and
• analyze trends in substance use treatment services and perform comparative analyses for the United States, regions of the country, States, counties, and populated areas (metropolitan statistical areas [MSAs] and core-based statistical areas [CBSAs]).

This section briefly describes the survey methodology; Appendix A provides a more complete description. Reports on N-SSATS data are available on the SAMHSA Web site (https://www.samhsa.gov/data/DASIS.aspx).

This review focuses on N-SSATS data collected in the 3-year period from 2007 through 2009. N-SSATS collects data on approximately 14,000 facilities, with some 1.2 million clients in treatment on the survey reference date (the last working day of March) and 3.6 million admissions to treatment in the previous 12 months. Information is collected at the site of delivery of services rather than according to administrative structure. A Web-based version of the paper questionnaire was introduced in 2002. Participation using the Web-based survey has increased over the years; in 2007, about 40 percent of survey responses were submitted via the Web and in 2009 about 58 percent of surveys were completed on the Web.

Although N-SSATS is a voluntary census, response rates are consistently about 95 percent. The incentive for participation is the opportunity to be included in SAMHSA's online Substance Abuse Treatment Facility Locator and annual National Directory of Drug and Alcohol Treatment Services. This searchable directory of drug and alcohol treatment programs shows the location of facilities around the country that treat alcohol use and drug use problems.

1.2.4 Key Differences between Data Sources

NSDUH, TEDS, and N-SSATS have important differences in data collection methodology and inclusion or exclusion criteria that can affect the interpretation of data from these sources. Table 1.1 summarizes some key differences between these three data sources.

Data Collection Methodology/Unit of Analysis. NSDUH provides self-reported data at the person level among noninstitutionalized civilians who live in households. TEDS provides information at the admission level about the characteristics of persons served by (mainly) publicly funded programs; for the increasing number of States that submit unique individual identifiers, TEDS data can also be analyzed at the person or episode level. As mentioned previously, analyses in this report mainly focus on the person-level TEDS data. N-SSATS provides information at the facility level about characteristics of treatment facilities, their operations, a single-day client census, and an estimate of admissions in the previous 12 months.

Inclusion and Exclusion Criteria. All three data sources have differing inclusion and exclusion criteria. NSDUH collects data on treatment received at both public and private facilities, as well as treatment received from solo practitioners, whereas it excludes active duty military members and persons who do not live in households (homeless persons not in shelters and persons in institutions such as jails or prisons, hospitals, or residential treatment facilities) during the majority of the quarter of the year in which they were interviewed. The facilities that report TEDS data are primarily those that receive State alcohol and/or drug agency funds for the provision of drug or alcohol services. These facilities often do not include facilities that do not receive funding through the SSA, IHS facilities, or hospital-based programs. TEDS does not include VA and military treatment facilities or Federal prisons. N-SSATS is a voluntary census with a very high response rate that includes IHS, DOD, and VA facilities but excludes most solo practitioners and facilities in jails, prisons, or other correctional facilities that exclusively treat incarcerated clients. These variations should be considered when interpreting differences in substance use treatment data between these data sources or when specific subgroups are being considered for analysis.

Seasonal Variation. Because of differences in reference dates used to compute the NSDUH estimates and TEDS/N-SSATS counts, it is possible that differences between TEDS/N-SSATS and NSDUH in the single-day counts may reflect seasonal variation in treatment utilization. Chapter 2 of this report includes analyses of the seasonality of admissions based on TEDS data and a discussion of this topic.

1.3. Measures of Receipt of and Need for Substance Use Treatment

This section provides descriptions of measures of substance use treatment receipt and need for treatment from NSDUH, TEDS, and N-SSATS that are included in this report. Three main measures of receiving substance use treatment were examined—total number of persons receiving substance use treatment in a given year, average number of persons receiving treatment on a single day in a given year, and total number of substance use treatment admissions in a given year. All three data sets allow for classification of persons who received alcohol treatment only, drug treatment only, or both alcohol and drug treatment. Among those who received alcohol and drug treatment, NSDUH and TEDS allow for the identification of which substances (alcohol or drugs) were the primary and secondary reasons for treatment.

In addition to the total number of persons who received treatment in a given year, another important measure when tracking changes in substance use treatment is the number of persons who received treatment for the first time in a given year. Both the NSDUH and TEDS provide information that allows for the estimation of the number of persons who received substance use treatment for the first time in a given year.

1.3.1 Receipt of Substance Use Treatment in the Past Year

NSDUH and TEDS data were used to derive estimates of the number of persons in specialty treatment in the past year.

NSDUH. The NSDUH measure of receipt of substance use treatment in the past year is derived from a series of questions about treatment received for alcohol use or illicit drug use (including the nonmedical use of prescription type drugs). NSDUH asks about treatment received in the past year at any location, such as a hospital (inpatient), rehabilitation facility (outpatient or inpatient), mental health center, emergency room, private doctor's office, prison or jail, or self-help group, such as Alcoholics Anonymous or Narcotics Anonymous. Persons could report having received treatment at more than one location. For comparison purposes, this report focuses only on "specialty treatment," or treatment for substance use at a specialty treatment facility. Specialty treatment is defined as treatment received at any of the following types of facilities: hospitals (inpatient only), drug or alcohol rehabilitation facilities (inpatient or outpatient), or mental health centers. The definition of specialty treatment does not include treatment at an emergency room, private doctor's office, self-help group, prison or jail, or hospital as an outpatient.

TEDS. TEDS data provide an overall count of the annual number of admissions for substance use treatment at facilities including outpatient, residential, and hospital inpatient facilities. The total number of persons admitted to treatment in a calendar year and the total number of persons who received treatment in a calendar year (including those who were admitted to treatment in a prior year but remained in treatment during the target year) were determined using the unique person identifier and linked admission and discharge data or, if there was no unique identifier, they were estimated using the probabilities of discharge prior to the start of the reference year. A unique identifier was defined as having a unique combination of State of residence, client ID, age, gender, and race/ethnicity. Multiple treatment episodes in a given year linked by a single unique identifier were counted as one person. For those admissions with no linked admission and discharge data, no attempt was made to account for multiple admissions by a single individual, which means that the annual count of persons who received treatment may be an overcount. Records for codependents (persons who do not have a substance use disorder but who receive treatment because a loved one has a substance use disorder) were excluded.

For the count of the number of persons receiving treatment in the past year, records were used for persons admitted on or before December 31 of that year and discharged on or after January 1 of that year. Each estimate is made up of two parts: (1) a direct count from linked admission/discharge records and (2) an estimate, based on the probability of being in treatment on January 1 of the given year if admitted in a different year, for those admissions not linked to discharges. For States with linked admissions and discharge data, this algorithm should identify any client who has been admitted and not yet discharged. However, because the discharge system was not fully operative in all States for the years we examined, it can be assumed that a number of these admissions without linked discharges had, in fact, been discharged. Therefore, a probability of having been discharged by each reference date (January 1) was computed for each unlinked admission record, based on the assumption that the unlinked admission records would have the same probability of being discharged on a given day following treatment as the linked admission records. The probability of being in treatment on the reference date was calculated as 1 minus the cumulative frequency. That is, if 0.34 (34 percent) of clients had been discharged by day 10, then a record with date of admission 10 days before the reference date had a 1.00 − 0.34 = 0.66 (66 percent) chance of being in treatment on the reference date. Probabilities of being in treatment were summed for each reference date. Thus, if 100 clients were admitted 10 days before the reference date, each would have a probability of 0.66, yielding a total of 66 clients likely to be in treatment on the reference date.

N-SSATS. N-SSATS data provide an overall count of the annual number of admissions for substance use treatment at all substance use treatment facilities that respond to the census and cannot be used to determine the number of persons who received treatment services in the prior 12 months.

1.3.2 Single-Day Counts of the Number of Persons in Drug or Alcohol Treatment

All three sources of data were used to derive single-data counts or estimates of the number of persons in drug or alcohol treatment.

NSDUH. For this report, two different estimates of the number of persons in specialty substance use treatment (i.e., substance use treatment received from a hospitals [inpatient only], drug or alcohol rehabilitation facility [inpatient or outpatient], or mental health center) on a single day were derived from NSDUH. One estimate was based on whether a respondent reported current treatment at the time of the survey.² Respondents who indicated that they had received specialty alcohol or drug treatment in the past year were asked if, at the time of the survey, they were currently receiving treatment for alcohol or drug use. The second measure of single-day treatment counts was derived from a question in which respondents who had received drug or alcohol treatment in their lifetime were asked if they had been enrolled in a treatment program at a hospital, drug rehabilitation facility, or mental health center on October 1 of the year prior to the survey. From this question, an estimate was derived of the number of persons in treatment on that specific day.

TEDS. For this report, estimates were computed for the number of clients in treatment on the last working day in March each year from 2007 to 2009 (March 30, 2007; March 31, 2008; and March 31, 2009) using linked admission and discharge records from TEDS data processed through October 10, 2011. Records for codependents (persons who do not have a substance use disorder but who receive treatment because a loved one has a substance use disorder) were excluded. The last working day of March was selected for these counts in order to be as consistent as possible with the counts provided by the N-SSATS data. Counts included total clients and the numbers of clients by treatment for alcohol only, drugs only, alcohol and drugs, primary alcohol/secondary drug, and primary drug/secondary alcohol. Admission records with date of admission on or after January 1, 2000, were included. Even though annual averages for 2007 to 2009 are presented in the report, as a data quality check, estimates for reference dates of the last working day in March were calculated for each year from 2000 to 2008 to check that this method yielded plausible data even when discharge records were sparse.

Estimates comprised the sum of two parts:

1. Linked records representing clients who had been both admitted and discharged. The number of clients who were admitted on or before the reference date and discharged on or after the reference date was calculated.
2. Admissions records with no linked discharge. This situation should indicate that a client has been admitted and not yet discharged. However, because the discharge system was not fully operative in all States for the years we examined, it can be assumed that a number of these admissions without linked discharges had, in fact, been discharged. Therefore, a probability of having been discharged by each reference date was computed for each unlinked admission record, based on the assumption that the unlinked admission records would have the same probability of being discharged on a given day following treatment as the linked admission records. The probability of being in treatment on the reference date was calculated as 1 minus the cumulative frequency. That is, if 0.34 (34 percent) of clients had been discharged by day 10, then a record with date of admission 10 days before the reference date had a 1.00 − 0.34 = 0.66 (66 percent) chance of being in treatment on the reference date. Probabilities of being in treatment were summed for each reference date. Thus, if 100 clients were admitted 10 days before the reference date, each would have a probability of 0.66, yielding a total of 66 clients likely to be in treatment on the reference date.

N-SSATS. The N-SSATS census provides a single-day count of the number of persons in hospital inpatient, residential (nonhospital), and outpatient substance use treatment on the last working day of March of each year. For outpatient clients, the count is the number of active clients; an active outpatient client is defined as someone who was seen at the facility for substance use treatment or detoxification at least once during the month of March and was still enrolled in treatment on the last working day of March.

1.3.3 Need for and Receipt of Substance Use Treatment

NSDUH. Using NSDUH data, an individual is defined as needing treatment for an alcohol or drug use problem if he or she met the diagnostic criteria for dependence on or abuse of alcohol or illicit drugs in the past 12 months specified in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) (American Psychiatric Association, 1994), or if he or she received specialty treatment for alcohol use or illicit drug use in the past 12 months. The questions related to dependence ask about having health and emotional problems associated with substance use, making unsuccessful attempts to cut down on use, developing tolerance, experiencing withdrawal, reducing other activities to use substances, spending a lot of time engaging in activities related to substance use, or using the substance in greater quantities or for a longer time than intended. The questions on abuse ask about problems at work, home, and school; problems with family or friends; physical danger; and trouble with the law because of substance use. Dependence is considered to be a more severe substance use disorder than abuse because it involves the psychological and physiological effects of tolerance and withdrawal.

For the purpose of this study, individuals needing treatment for an illicit drug use problem are defined as receiving treatment for their drug use problem only if they reported receiving specialty treatment for drug use in the past year. Thus, individuals who needed treatment for illicit drug use but only received specialty treatment for alcohol use in the past year or who received treatment for illicit drug use only at a facility not classified as a specialty facility are not counted as receiving treatment for drug use. Similarly, individuals who needed treatment for an alcohol use problem are only counted as receiving alcohol use treatment if the treatment was received for alcohol use at a specialty treatment facility. Individuals who reported receiving specialty substance use treatment but are missing information on whether the treatment was specifically for alcohol use or drug use are not counted in estimates of specialty drug use treatment or in estimates of specialty alcohol use treatment; however, they are counted in estimates for "drug or alcohol use" treatment.

1.3.4 Admissions in a Given Year

Substance use disorders are chronic and remitting conditions that often require multiple treatment episodes before they are effectively managed. Annual admissions are another important measure of substance use treatment receipt. N-SSATS and TEDS data provide counts of the number of admissions to specialty treatment in a given year.

1.4. Demographic Variables

This section describes key measures of demographic and geographic characteristics that are presented in this report to compare the distributions of numbers of persons in treatment.

Comparisons between NSDUH and TEDS of the number of persons who received substance use treatment in the past year are presented by racial/ethnic group. Data are presented for racial/ethnic groups based on current guidelines for collecting and reporting race and ethnicity data (Office of Management and Budget [OMB], 1997). Because respondents were allowed to choose more than one racial group, a "two or more races" category is presented that includes persons who reported more than one category among the basic groups listed in the survey question (white, black or African American, American Indian or Alaska Native, Native Hawaiian, other Pacific Islander, Asian, other). Respondents choosing both the Native Hawaiian and the other Pacific Islander categories but none of the other categories are classified in the combined "Native Hawaiian or other Pacific Islander" category instead of the "two or more races" category. Except for the "Hispanic or Latino" group, the racial/ethnic groups discussed in this report include only non-Hispanics; the category "Hispanic or Latino" includes Hispanics of any race. Data reported to TEDS³ allow for reporting of the same race/ethnicity categories included in NSDUH. Race and ethnicity data about clients are not collected as part of N-SSATS.

Comparisons of single-day treatment counts are presented for States and for four U.S. geographic regions. These regions and divisions, defined by the U.S. Census Bureau, consist of the following groups of States:

• Northeast Region—New England Division: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont; Middle Atlantic Division: New Jersey, New York, and Pennsylvania
• Midwest Region—East North Central Division: Illinois, Indiana, Michigan, Ohio, and Wisconsin; West North Central Division: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, and South Dakota
• South Region—South Atlantic Division: Delaware, District of Columbia, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia, and West Virginia; East South Central Division: Alabama, Kentucky, Mississippi, and Tennessee; West South Central Division: Arkansas, Louisiana, Oklahoma, and Texas
• West Region—Mountain Division: Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah, and Wyoming; Pacific Division: Alaska, California, Hawaii, Oregon, and Washington

It should be noted that NSDUH States/geographic regions reflect where the respondents lived at the time they completed the survey, which is not necessarily the State or region in which they received substance use treatment during the past year. N-SSATS data give the facility location, which is not necessarily where the person receiving treatment lived. The geographic data in TEDS correspond to the location of the facility to which the person was admitted, but may also reflect the location of, for example, the administrative unit of a multicounty health service region.

NSDUH estimates of the number of persons who needed and received treatment are presented for those classified as having "low income." Individuals are classified as low income if their family income is less than 100 percent of the Federal poverty threshold. Poverty level is based on income level, size of family, number of children in family, and additional family relationships such as foster children.

1.5. Organization of This Report

This report contains separate chapters that provide estimates of persons receiving substance use treatment and persons needing but not receiving substance use treatment. Chapter 2 presents national, regional, and State estimates from the NSDUH, TEDS, and N-SSATS data about the numbers of persons receiving past year and single-day substance use treatment as well as total admissions to treatment, including a discussion of the variations in seasonality of substance use treatment admissions and comparisons between NSDUH and TEDS in the number of persons who received treatment for the first time. Chapter 3 focuses on national and State estimates of the number of adults who needed but did not receive substance use treatment, with a focus on a new technique of estimating this statistic by combining NSDUH and TEDS data. A final chapter discusses the information provided by these three data sources and how this information can be used to provide a better understanding of substance use treatment in this country and the current unmet need for treatment services.

Tables, text, and figures in this report present estimates of numbers of past year substance use treatment admissions, numbers of persons receiving treatment in the past year (including those admitted during the past year and those admitted prior to the past year), treatment outcomes, and numbers of persons needing but not receiving substance use treatment. In all tables and figures, estimates are presented based on data combined from 2 or more years to increase precision of the estimates; those estimates are annual averages based on multiple years of data.

Statistical tests have been conducted for all statements appearing in the text of the report that compare estimates between data sources (e.g., NSDUH vs. TEDS) or between subgroups within a given survey (e.g., by age group within NSDUH). Unless explicitly stated that a difference is not statistically significant, all statements that describe differences are significant at the .05 level. Differences are described using terms such as "higher," "lower," "more likely," and "less likely." Statements that use terms such as "similar," "no difference," or "same" to describe the relationship between estimates denote that a difference is not statistically significant. In addition, a set of estimates for a given data source or for population subgroups may be presented without a statement of comparison; in these instances, a statistically significant difference between these estimates is not implied, and testing may not have been conducted.

All estimates presented in the report have met the criteria for statistical reliability. Estimates that do not meet these criteria are suppressed and do not appear in tables, figures, or text. Subgroups with suppressed estimates are not included in statistical tests of comparisons. Appendix B of the 2012 NSDUH report on national findings discusses these suppression criteria in more detail (see CBHSQ, 2013).

All statistical tables produced for this report are provided in Appendix C.

2. Comparisons between Data Sources in Receipt of Substance Use Treatment

This chapter presents comparisons of estimates of the number of persons who received substance use treatment from NSDUH, TEDS, and N-SSATS and counts of the number of substance use treatment admissions reported in TEDS and N-SSATS. The first section of this chapter presents estimates of the number of persons who received substance use treatment in a given year based on annual averages from 2005 to 2010 NSDUH data and combined 2007 to 2009 TEDS data. The second section presents estimates of the number of persons who received substance use treatment for the first time in the past year based on NSDUH and TEDS data. The third section presents estimates of single-day counts of the number of persons in treatment for alcohol use, drug use, or both alcohol and drug use, based on annual averages for combined 2008 to 2010 NSDUH, combined 2007 to 2009 TEDS, and combined 2007 to 2009 N-SSATS data. The fourth section presents counts of annual treatment admissions for 2007 through 2009, based on annual averages for counts from TEDS and N-SSATS. Combining data from multiple years increases the sample size to support detailed estimates and is particularly useful for examining demographic and geographic correlates of receipt of alcohol use and drug use treatment.

2.1. Number of Persons Who Received Treatment in the Past 12 Months

This section presents comparisons between NSDUH and TEDS estimates of the numbers and characteristics of persons who received substance use treatment in the past year.

2.1.1 NSDUH Estimates

Description of NSDUH Estimates

In NSDUH, estimates of the number of persons who received substance use treatment in the past year are based on self-reports of treatment received. So that NSDUH estimates would be more comparable to TEDS data, we have used NSDUH estimates of the number of persons who received treatment in the past year in a specialty treatment facility, which includes treatment received at any of the following types of facilities: hospitals (inpatient only), drug or alcohol rehabilitation facilities (inpatient or outpatient), or mental health centers. Specialty treatment does not include treatment at an emergency room, private doctor's office, self-help group, prison or jail, or hospital as an outpatient. Persons could report receiving treatment at more than one location; persons who reported receiving treatment from at least one specialty treatment location were counted as having received treatment at a specialty treatment facility. Likewise, persons could report receiving treatment for more than one type of substance (e.g., alcohol, marijuana, and cocaine).

NSDUH data can also be used to estimate the number of persons who received treatment in the past year at any location. This includes specialty treatment facilities, as defined previously, or nonspecialty treatment facilities, such as an emergency room, a private doctor's office, a prison or jail, or a self-help group such as Alcoholics Anonymous or Narcotics Anonymous. For comparison, we have included a table in Appendix C that provides estimates and percentages of persons who received substance use treatment at any location based on NSDUH data (see Table C.2.1AL).

Limitations of the NSDUH Estimates

NSDUH is a household survey that relies on the sampled respondents to provide accurate and honest reports of their substance use and substance use treatment experiences. The NSDUH sample is drawn to represent the noninstitutionalized population of the United States aged 12 or older. It excludes persons who are in jail or prison, persons who are in a hospital or residential treatment during the majority of the quarter when the survey is conducted, and homeless persons who do not reside in shelters.

2.1.2 TEDS Counts

Description of TEDS Counts

The TEDS counts of the number of persons who received past year substance use treatment are derived from linked admission and discharge records from specialty substance use treatment facilities. More specific information about how these counts are derived is provided in Section 1.3 of this report. Persons could receive treatment at more than one type of treatment facility during a given year. Likewise, TEDS records include up to three types of substances (e.g., alcohol, marijuana, and cocaine) that a person could have received treatment for.

Limitations of the TEDS Counts

TEDS does not include all admissions to substance use treatment. It typically includes admissions to facilities that are licensed or certified by an SSA to provide substance abuse treatment, as well as admissions that are administratively tracked for other reasons. TEDS typically does not include clients treated in a jail/prison not run with State agency-administered funds, emergency room, private doctor's office, or self-help group. As Section 1.2 notes, many States require only publicly funded facilities to report data to TEDS, and some States only require facilities to report TEDS data for patients for whom treatment is publicly funded. Also noted in Section 1.2, hospital inpatient substance use treatment facilities are frequently not licensed through the SSA and may not be required to report TEDS data. A detailed list of institutions/clients included or excluded in each State is provided in Table A.1 in Appendix A.

2.1.3 National Estimates

This section compares the numbers and demographic/treatment characteristics of persons who received substance use treatment services in a specialty treatment facility in the past year. Alabama, Alaska, the District of Columbia, and Georgia were excluded from TEDS estimates because no or incomplete TEDS data were submitted for 1 or more of the 3 years examined (2007 to 2009) in these areas. In addition to NSDUH estimates based on all 50 States plus the District of Columbia, NSDUH estimates were also restricted to the subset of 47 States that were used to generate TEDS estimates; these are referred to as the "TEDS States." These estimates are presented in Appendix C (Table C.2.1). The NSDUH estimate for all States is approximately 4.5 percent higher than the NSDUH estimate for the TEDS States. In terms of the total count/estimate of the number of persons who received substance use treatment from specialty treatment facilities in the past year, the NSDUH estimate for the TEDS States was greater than the TEDS count (2,357,118 vs. 1,928,578).

Treatment Characteristics

A comparison of the treatment characteristics of persons who received substance use treatment from a specialty facility in the past year also shows several differences between the two data sources. NSDUH estimates were higher than TEDS counts across the three principal modality categories (hospital inpatient, residential rehabilitation—inpatient, and outpatient). Although the NSDUH estimate and the TEDS count reflected a similar proportion of persons who received treatment in an outpatient setting (81.3 and 78.2 percent, respectively), the NSDUH estimate included a notably greater percentage of persons who received treatment in an inpatient hospital setting than the TEDS count (31.3 vs. 3.2 percent). The NSDUH estimate also reflected a greater percentage of persons who received treatment in a residential rehabilitation facility than the TEDS count (40.6 vs. 28.2 percent).

Figure 2.1 shows the percentages of persons treated for specific substances or substance groups. The percentage of persons in the NSDUH estimate who were treated for marijuana use problems was lower than the percentage in the TEDS count who reported marijuana as a problem drug at the time of treatment admission (30.9 vs. 39.7 percent for NSDUH TEDS States only and TEDS, respectively). The NSDUH estimates for TEDS States only included greater percentages of persons treated for prescription drugs, inhalants, hallucinogens, and alcohol than did the TEDS counts (26.3 vs. 12.5 percent for prescription drugs, 6.0 vs. 0.2 percent for inhalants, 10.9 vs. 0.6 percent for hallucinogens, and 69.3 vs. 59.5 percent for alcohol). One possible explanation for the difference between the percentage of persons treated for prescription drugs reflected in the NSDUH estimate and the TEDS count is that TEDS queries treatment for methamphetamine use as a separate category, whereas NSDUH includes treatment for methamphetamine use as treatment for prescription drugs. The NSDUH estimate included a lower proportion of injection drug users than the TEDS count did (7.3 vs. 12.4 percent; data not shown). Note that restricting the NSDUH findings to the TEDS States does not have an appreciable effect on the percentage of persons receiving treatment for any of the substances.

Demographic Characteristics

There was some variation between the treatment counts from the two data sources on several demographic characteristics. Figure 2.2 shows the distribution across age groups of persons receiving specialty treatment. The NSDUH estimate included a lower percentage of youths and a greater percentage of older persons. Specifically, in the NSDUH estimate, 6.4 percent were aged 12 to 17 (150,083 persons), and in the TEDS count, 8.8 percent were aged 12 to 17 (170,289 persons). In the NSDUH estimate, 29.0 percent were aged 45 or older (683,673 persons), and in the TEDS count, 22.5 percent were aged 45 or older (433,924 persons). The restriction of the NSDUH sample to include only the TEDS States did not affect the percentage distribution across age groups; the percentages of the four age groups were nearly identical for the full NSDUH sample and for the NSDUH TEDS States only sample.

Differences were also seen between NSDUH and TEDS distributions across racial ethnic groups and educational attainment. As shown in Figure 2.3, the NSDUH estimate included a greater proportion of non-Hispanic whites and lower percentages of non-Hispanic blacks and Hispanics than the TEDS count. The NSDUH estimates were greater than the TEDS counts for both genders, although the proportions of males and females were similar between the two data sources. A comparison of the education status reflected in the two data sources shows the NSDUH estimate included a higher percentage of persons with more than a high school education and a lower percentage of persons with 8 or fewer years of education than the TEDS count. In the NSDUH estimate, 34.3 percent of persons had more than 12 years of education, and 5.7 percent had 0 to 8 years of education; in the TEDS count, 22.1 percent had more than 12 years of education, and 9.5 percent had 0 to 8 years of education.

Figure 2.4 shows the results of a comparison of the NSDUH 2005 to 2010 specialty treatment data for employment status,⁴ with the TEDS data reported for those variables (CBHSQ, 2011a). When compared with estimates from TEDS, estimates from NSDUH show a higher percentage of persons employed full time (37.3 vs. 16.3 percent for NSDUH and TEDS, respectively) and part time (12.9 vs. 7.6 percent for NSDUH and TEDS, respectively). TEDS estimates include a higher percentage than NSDUH of persons who were unemployed (39.0 vs. 12.8 percent for TEDS and NSDUH, respectively). One possible explanation for these findings is that many States require only publicly funded facilities to report data to TEDS, and some States only require facilities to report TEDS data for patients for whom treatment is publicly funded.

The findings for health insurance status were consistent with the pattern for employment, with a higher percentage of TEDS counts being uninsured (59.8 percent) compared with the NSDUH estimate (30.9 percent; data not shown).⁵

2.1.4 State Estimates

State-level NSDUH and TEDS estimates for the eight most populous States in the Nation (California, Florida, Illinois, Michigan, New York, Ohio, Pennsylvania, and Texas) are presented in Tables C.2.1.1-C.2.1.8. Estimates from the other States and the District of Columbia are not presented because of the lack of precision of NSDUH estimates from these States due to the relatively small sample sizes; there were approximately 3,600 interviews per year in the eight most populous States and approximately 900 interviews per year in the remaining States and the District of Columbia.

In six of the eight States examined, the TEDS count of the number of persons who received substance use treatment in the past year was smaller than the NSDUH estimate. Notable differences between the overall TEDS count and the NSDUH estimate were found in five States. For example, in Florida, the NSDUH estimate (138,035) was greater than the TEDS count (67,305). There were similar disparities between the NSDUH estimates and TEDS counts in Illinois (95,950 vs. 69,483), Michigan (102,203 vs. 57,956), Pennsylvania (108,068 vs. 46,990), and Texas (84,936 vs. 41,930). In two States, the TEDS count was larger than the NSDUH estimate, but the disparities were not statistically significant. In New York, the TEDS count was 245,898 persons, and the NSDUH estimate was 207,852. In Ohio, the TEDS count was 105,443 persons, and the NSDUH estimate was 96,729.

Treatment Characteristics

A comparison of the treatment characteristics showed several differences between the counts from the two data sources at the State level. Similar to the comparisons made at the national level, the State-level NSDUH estimates for all eight States reflected greater percentages of persons who received treatment in an inpatient hospital setting or residential rehabilitation setting than the TEDS counts. The State-level NSDUH estimates for five of the eight States reflected greater percentages of persons who received treatment in an outpatient setting than the TEDS counts, though only the differences in Pennsylvania were statistically significant; the exceptions were California (72.6 vs. 79.9 percent), Florida (70.2 vs. 81.7 percent), and Ohio (83.6 vs. 94.7 percent).

In some States, we found the two data sources produced similar percentages of persons treated for most substances, whereas in others, there was a large gap between many of the NSDUH estimates and the TEDS counts. The more notable differences include the following:

• In California, the NSDUH estimate showed that 32.5 percent of persons in specialty treatment were treated for prescription drug use problems, whereas the TEDS count was 6.5 percent. For inhalants, the NSDUH estimate was 8.8 percent, and the TEDS count was 0.1 percent. Similarly, for hallucinogens, the NSDUH estimate was 12.5 percent, and the TEDS count was 0.2 percent. The percentage of persons in specialty treatment in California for alcohol use problems based on the NSDUH data (72.8 percent) was greater than the percentage based on TEDS counts (39.7 percent).
• In Florida, the NSDUH estimate showed that 34.0 percent of persons in specialty treatment were treated for marijuana use problems, whereas the TEDS count was 47.7 percent. For heroin, the NSDUH estimate was 11.2 percent, and the TEDS count was 3.8 percent. For inhalants and hallucinogens, the NSDUH estimates were that 9.3 and 15.3 percent, respectively, were treated for problems with these drugs, and the TEDS counts were 0.2 and 0.5 percent, respectively. The percentage of persons in specialty treatment in Florida for alcohol use problems based on the NSDUH data (67.0 percent) was also greater than the percentage based on the TEDS counts (46.7 percent).
• In Illinois, the NSDUH estimate showed that 22.9 percent of persons in specialty treatment were treated for marijuana use problems, whereas the TEDS count was 40.5 percent. For heroin, the NSDUH estimate was 15.5 percent, and the TEDS count was 24.1 percent. For prescription drugs, the NSDUH estimate was 17.0 percent, and the TEDS count was 5.6 percent. The percentage of persons in specialty treatment in Illinois for hallucinogen use problems based on the NSDUH data (2.9 percent) was also greater than the percentage based on the TEDS counts (0.3 percent).
• In Michigan, the NSDUH estimate showed that 30.3 percent of persons in specialty treatment were treated for marijuana use problems, whereas the TEDS count was 41.9 percent. For inhalants and hallucinogens, the NSDUH estimates were that 4.2 and 8.7 percent, respectively, were treated for problems with these drugs, and the TEDS counts were 0.1 and 0.3 percent, respectively.
• In New York, the NSDUH estimate showed that 32.7 percent of persons in specialty treatment were treated for marijuana use problems, whereas the TEDS count was 42.8 percent. For inhalants and hallucinogens, the NSDUH estimates were that 3.4 and 12.1 percent, respectively, were treated for problems with these drugs, and the TEDS counts were 0.1 and 0.3 percent, respectively.
• In Ohio, the NSDUH estimate showed that 31.5 percent of persons in specialty treatment were treated for marijuana use problems, whereas the TEDS count was 46.2 percent. For prescription drugs, the NSDUH estimate was 23.5 percent, and the TEDS count was 11.3 percent. Similarly, for inhalants and hallucinogens, the NSDUH estimates were that 5.3 and 9.1 percent, respectively, were treated for problems with these drugs, and the TEDS counts were 0.2 and 0.5 percent, respectively.
• In Pennsylvania, the NSDUH estimate showed that 5.9 percent of persons in specialty treatment were treated for inhalant use problems, whereas the TEDS count was 0.2 percent. For hallucinogens, the NSDUH estimate was 9.6 percent, and the TEDS count was 0.7 percent.
• In Texas, the NSDUH estimate showed that 9.5 percent of persons in specialty treatment were treated for inhalant use problems, whereas the TEDS count was 0.2 percent. For hallucinogens, the NSDUH estimate was 18.4 percent, and the TEDS count was 0.4 percent. The percentage of persons in specialty treatment in Texas for alcohol use problems based on the NSDUH data (66.8 percent) was also greater than the percentage based on the TEDS counts (48.1 percent).

In two of the eight States (California and Texas), the NSDUH estimate included a lower proportion of injection drug users than the TEDS counts, reflecting the pattern seen in the national comparisons. In Ohio, Michigan, and Illinois, the NSDUH estimate indicated a lower proportion of injection drug users than the TEDS count for the State, but the difference between the two data sources was not statistically significant. In Florida and Pennsylvania, the two data sources had similar proportions of injection drug users (5.2 percent in both surveys for Florida, and 11.8 percent in TEDS and 12.0 percent in NSDUH for Pennsylvania). In New York, no NSDUH estimate of injection drug use was reported due to low precision.

Demographic Characteristics

Across the eight States, we observed some patterns of demographic characteristics of persons in treatment that were consistent with national comparisons, as well as some patterns that differed from national comparisons. For example, in five of the eight States (California, Florida, Illinois, Ohio, and Pennsylvania), the NSDUH estimates showed lower percentages of youths than the TEDS counts, as we found in the national estimates/counts. In the remaining three States (Michigan, New York, and Texas), the percentages of youths identified in the two sources were more similar to one another.

In the national comparisons, we found that the NSDUH estimate included a greater proportion of non-Hispanic whites and lower percentages of non-Hispanic blacks (not statistically significant) than the TEDS count. In Illinois, we found the same pattern. In Pennsylvania, the NSDUH estimate included a greater proportion of whites and similar proportion of blacks than the TEDS count. However, in the six remaining States, we found that the NSDUH estimates and TEDS counts of non-Hispanic whites were similar to each other. When comparing the two sources for each State, we found that in six States (California, Illinois, Michigan, New York, Ohio, and Pennsylvania), the NSDUH estimate included a lower percentage of non-Hispanic blacks than the TEDS count, although this difference was only statistically significant in Illinois and Ohio. In Florida and Texas, no NSDUH estimate of non-Hispanic blacks was reported due to low precision. In looking at State-level estimates of Hispanics, we found that in Pennsylvania, the NSDUH estimate had a lower percentage of Hispanics than the TEDS count, although this difference was not statistically significant. We found the converse to be true in Ohio, and the difference was statistically significant. In New York and Texas, no NSDUH estimates of Hispanics were reported due to low precision; in California, Florida, Illinois, and Michigan, the NSDUH estimates and TEDS counts were similar.

In the national comparisons, the NSDUH estimate included a higher percentage of persons with more than a high school education and a lower percentage than the TEDS count of persons with 8 or fewer years of education. The discrepancy between the two sources with regard to the percentage that had 8 or fewer years of education was present in all States. The most drastic, and only statistically significant difference between sources, was found in Pennsylvania, where the NSDUH estimate was that 6.7 percent of persons treated in specialty treatment had 8 or fewer years of education and the TEDS count reflected that 57.1 percent had that level of education. One possible explanation for this difference is that there were technical problems with the education variable in Pennsylvania during 1 or more of the years examined. It may also be possible that persons with little formal education are the recipients of publicly funded substance use treatment in that State or that persons with little formal education in Pennsylvania do not participate in the NSDUH. The pattern of NSDUH estimates having higher percentages of persons with more than a high school education was also found in seven of the eight States. In Texas, no NSDUH estimate of more than 12 years of education was reported due to low precision.

2.1.5 Conclusions and Limitations

At the national level, there were some differences between the treatment characteristics of the TEDS counts and those of the NSDUH estimates. As would be expected, the national TEDS estimate for persons receiving treatment in a hospital inpatient setting was much lower than the NSDUH estimate (61,382 vs. 738,366).

Treatment Characteristics

Because privately funded treatment facilities often are not included in TEDS but NSDUH collects data from persons who received treatment from both publicly and privately funded facilities, the TEDS count of persons who received substance use treatment in the past year was lower than the estimate from NSDUH for the 47 TEDS States (approximately 1.9 million vs. 2.4 million). A comparison of treatment modality also produces several expected differences between the two data sources. Although the TEDS count and the NSDUH estimate reflected a similar proportion of persons who received treatment in an outpatient setting (78.2 and 81.3 percent, respectively), the TEDS count included a notably lower proportion of persons who received treatment in an inpatient hospital setting than the NSDUH estimate (3.2 vs. 31.3 percent). This may be because hospital-based substance use treatment facilities are frequently not licensed through the State and therefore are often not required to report TEDS data. The TEDS count of persons receiving treatment in a residential rehabilitation setting also reflected a lower percentage than did the NSDUH estimates (28.2 vs. 40.6 percent).

To provide a comparison of the two data sources that is more equivalent with regard to inclusion/exclusion of hospital-based inpatient facilities, two additional tables are included in Appendix C (see Table C.2.1NH) that show demographic and other characteristics of persons who received substance use treatment from specialty treatment facilities excluding persons receiving treatment solely from hospital-based facilities. Despite the exclusion, the NSDUH estimate for the TEDS States (2,227,200) was greater than the TEDS count presented in Table C.2.1 (1,928,728).

Methodological differences between NSDUH and TEDS in the collection of data on specific substances of abuse may partially explain the differences found between data sources in the percentages of persons treated for specific substances. In the NSDUH, respondents who report past year treatment are asked whether they received that treatment for each of the substances (including alcohol) that they had used in their lifetime. If a respondent reports that he or she was treated for more than one substance, he or she is asked to identify the one substance that was the main reason for their treatment. The nature of substance use treatment might encourage the inclusion of most or all substances that have been abused. Treatment in many cases is generic and in instances of broad spectrum drug use is not linked to a specific substance. In TEDS, the client's primary, secondary, and tertiary problem substances are reported, which we infer to be the substances for which they are receiving treatment. The specific problem substances are provided by the client, but if more than one substance is reported, there are no specific guidelines concerning how to determine which substance is primary or which substances to include if a client has been misusing more than three substances. There may also be administrative reasons that influence which drugs are classified as primary and which are classified as secondary or tertiary.

The NSDUH percentages are significantly higher than the TEDS percentages for prescription drugs, inhalants, hallucinogens, and alcohol. One possible explanation is that these substances are less likely to be identified as problem substances at the time of treatment admission when other drugs (e.g., heroin and cocaine) have been identified as primary or secondary problem drugs. Being specifically asked whether or not a respondent was treated for use of each substance used in his or her lifetime may lead to a more inclusive list of substances than making a report of the primary, secondary, and tertiary problem substances at admission to treatment. In nearly 15 percent of cases, NSDUH respondents reported four or more substances for which treatment was received over the past year (which could have included multiple admissions).

Additionally, differences exist between the ways that NSDUH and TEDS define inhalants. In NSDUH, respondents are given a general description of inhalants (substances persons sniff or inhale for kicks or to get high) and then are provided with a list of the more common inhalants when they are asked about use of these types of drugs. The NSDUH list contains: amyl nitrite ("poppers," locker room odorizers, "rush"); correction fluid, degreaser, or cleaning fluid; gasoline or lighter fluid; glue, shoe polish, or toluene; halothane, ether, or other anesthetics; lacquer thinner or other paint solvents; lighter gases, such as butane or propane; nitrous oxide or "whippits"; spray paints; or other aerosol sprays. In TEDS, problem substance information is collected during the admission process. The TEDS guidelines for which drugs should be categorized as inhalants include the following: chloroform, ether, gasoline, nitrous oxide, and paint thinner. It is possible that the greater percentage of persons treated for inhalant use in NSDUH when compared with TEDS could be due, at least in part, to the more inclusive list of inhalants provided in NSDUH. However, it could also not be a reason because the definition in the manual provides examples rather than an exhaustive list, but States would still cross-walk the type of substances in the NSDUH list to the inhalants category in the TEDS. The person being admitted to treatment would not see or be limited to the examples in the TEDS manual.

The proportion of injection drug users in TEDS was higher than the proportion of injection drug users in the NSDUH. The TEDS estimates included only those admissions for which injection was the usual route of administration of their primary, secondary, or tertiary problem drug, whereas the NSDUH estimates included anyone who had injected one of the three queried substances (heroin, methamphetamine or Desoxyn^®, and Methedrine^®) one or more times in the past year. Although asking about injection as the usual route of administration in TEDS versus asking about any injection use in NSDUH would suggest that NSDUH estimates would be higher than TEDS counts; however, the opposite pattern was found. One possible explanation is that NSDUH respondents were queried about injection administration of only three drugs, whereas injection of any drug was included in TEDS counts. For example, cocaine was frequently named as an injection drug in TEDS, but it was not included in the NSDUH list of queried drugs for injection administration. Another possible explanation is that NSDUH estimates included a higher percentage of persons in treatment for alcohol use than did the TEDS counts, and injection is not a common route of administration of alcohol. The NSDUH injection questions were also included in a different module than the questions about the substances used and may be subject to reporting bias (respondents not wanting to reveal behaviors that are stigmatized or socially undesirable). Injection drug use might also be underreported in NSDUH due to conditioning (decreasing tendency of respondents to answer "yes" to questions that came later in an interview as a way of avoiding being asked additional follow-up questions) and social desirability concerns of respondents. Finally, in TEDS, accurate reporting of the typical route of drug administration at admission was relevant to the client's treatment plan and services received (i.e., injection drug use may increase the risk of contracting certain illnesses such as hepatitis C and HIV), whereas this was not the case when NSDUH respondents were retroactively reporting on their injection of drugs. It is likely that there are multiple such variations in methodology between NSDUH and TEDS, as well as variations in methodology in the collection of substance use data between States, facilities, and admissions in TEDS, which could impact these comparisons. It is not possible to determine the magnitude or the direction of the influence of each of these types of potentially biasing factors on these comparisons.

Demographic Characteristics

Differences between TEDS and NSDUH estimates of specific racial or ethnic groups might be related to the broader definition of treatment in NSDUH but might also be associated with differences in the way that data concerning race and ethnicity were collected for the two data sources. In NSDUH, the interviewer asked two questions about racial and ethnic identity, whereas in TEDS, the questions, definitions, and process used to identify a person's race/ethnicity varied somewhat between individual treatment facilities.

State-Level Estimates

Differences between TEDS and NSDUH State estimates of the number of persons who received past year treatment were likely to have been largely a consequence of the broader inclusiveness of treatment used in NSDUH. For example, in TEDS, hospital inpatient cases include only those treated in the chemical dependence unit of the hospital, whereas NSDUH does not make that distinction. State-level differences might also reflect requirements within the State for reporting TEDS data. (For a more detailed explanation of differences in State reporting requirements, see Table A.1 in Appendix A.) One limiting factor might include the types of facilities that were required to report TEDS data within an individual State. For example, in Texas, only facilities that received State/public funding were required to report, and those facilities were only required to report data on clients whose individual treatment was funded by State/public funds. For Texas, the NSDUH estimate of persons who received treatment in the past year was twice as great as the estimate from TEDS (84,936 vs. 41,930). In New York, all facilities that received State/public funding as well as those that were licensed/certified by the SSA were required to report TEDS data on all clients in their facilities. For New York, the NSDUH estimate of persons treated in the past year was less than the estimate from TEDS (207,852 vs. 245,898), although this difference was not statistically significant. Notably, the NSDUH estimates at the State level are less precise than the national NSDUH estimates due to the smaller sample size.

2.2. Numbers of Persons Who Received Treatment for the First Time in a Given Year

2.2.1 NSDUH Estimates

Description of NSDUH Estimates

NSDUH respondents who indicated that they had ever received treatment or counseling for their use of alcohol or any drug were asked to report the age at which they first received treatment or counseling for alcohol or other drug use. By subtracting the current age of the respondent from the age of first treatment, it is possible to identify respondents who received treatment for the first time in a given calendar year, and therefore estimate the number of persons who received treatment for the first time in any calendar year. Note that data from different survey years could be used to provide different estimates of the number of persons who received treatment for the first time in any one calendar year. For example, the number of persons who received treatment for the first time in 2006 could be estimated by determining the number of persons whose current age in 2007 was 1 year older than their age at first treatment or by determining the number of persons whose current age in 2008 was 2 years older than their age at first treatment. In this report, NSDUH estimates of the number of persons who received treatment for the first time in different years are based on combined 2008 and 2009 NSDUH data and represent the number of persons who had received treatment in their lifetime who first received treatment in each year from 1995 to 2007. The questions about first substance use treatment included treatment received at any location. This included treatment received at

• a hospital (inpatient);
• a rehabilitation facility (outpatient or inpatient);
• a mental health center;
• an emergency room;
• a private doctor's office;
• a prison or jail; or
• a self-help group, such as Alcoholics Anonymous or Narcotics Anonymous.

Limitations of the NSDUH Estimates

The limitations of these estimates are similar to the limitations discussed previously for the number of persons who received treatment in the past year. The NSDUH is a household survey that relies on sample members to provide accurate and honest reports of their substance use and substance use treatment experiences; this may be especially problematic if asking persons who have been in treatment multiple times over many years to recall their age when they first received treatment. As a household survey, NSDUH also excludes persons who are in jail or prison, persons who are in a hospital or residential treatment, and homeless persons who do not reside in shelters for the majority of the quarter that their dwelling unit has been sampled. Finally, it should be noted that the NSDUH sample has been weighted to produce accurate estimates of substance use among the noninstitutionalized population of the United States aged 12 or older.

2.2.2 TEDS Counts

Description of TEDS Counts

Each TEDS admission record included the number of previous treatment episodes the client had received in any drug or alcohol treatment facility. Each admission for which there were no reported prior admissions for substance use treatment was considered a first-time admission. In this report, TEDS counts include the number of admissions in each year from 1995 to 2007 that reported no previous substance use treatment admissions. The TEDS counts of the number of persons who first received substance use treatment in a given year are derived from records for admissions.

Limitations of the TEDS Counts

As is true for the other counts discussed in this chapter, the generalizability of the TEDS counts is limited by the inclusion/exclusion criteria for the TEDS database and also by the State-level differences in reporting requirements that affect which data are reported to TEDS from the individual States. TEDS counts do not represent a census of all admissions to substance use treatment in a given year because TEDS includes primarily facilities that receive State-administered public funds. Privately funded facilities are generally not required to report to TEDS and are unlikely to do so; those that are operated by the Federal Government (e.g., those in VA hospitals or Federal prisons) are also unlikely to report to TEDS. TEDS explicitly excludes clients treated in an emergency room, a private doctor's office, a self-help group, or a jail/prison that is not run with funds administered by an SSA. Second, there are instances in which data submission is suspended temporarily for 1 or more years in some States or jurisdictions because of changes to their data collection systems (e.g., in 2006, Alaska, the District of Columbia, Georgia, and Vermont did not submit data to TEDS). Third, not every State that participates in TEDS submits all of the data about each client for all of the variables in the data set. For example, in 2006, there were incomplete or missing data from five States/districts for the previous treatment variable. In that year, there were 699,008 clients admitted to treatment who had not had a prior treatment episode and there were 339,995 cases with missing data for the prior treatment episode variable. The information collected on the TEDS is also mostly self-reported by the persons admitted for treatment; however, the recall of certain data (e.g., prior treatment or substances used) may be different depending on the saliency of the topic and the differences in the context within which the data are being collected.

2.2.3 Annual Estimates from 1995 to 2007

Figure 2.5 compares the annual estimates for the number of persons receiving substance use treatment for the first time in the years 1995 through 2007. Because the estimates from NSDUH included treatment or counseling received at any type of facility, whereas the TEDS counts included treatment received only at specialty treatment facilities, the number of persons who received treatment for the first time in any given year would be expected to be higher in NSDUH estimates than in TEDS counts. This was true for the estimates from 2006 and 2007 but was not true for estimates from previous years. In each year from 1997 to 2005, either there was no difference between the NSDUH and TEDS estimates or the TEDS estimates were higher than the NSDUH estimates. The NSDUH estimates for each year that were based on having to recall treatment from more than 3 or 4 years ago generally became smaller as the time between the interview and the year of first treatment increased (smaller numbers for earlier years), whereas the TEDS estimates were more consistent over time.

2.2.4 Conclusions and Limitations

There are multiple reasons to expect the NSDUH estimates to be higher than the TEDS estimates for the recent years (i.e., where the recall period is 1 or 2 years). First, the NSDUH estimates include any type of substance use treatment or counseling in each State, whereas the TEDS estimates are limited to treatment in a specialty substance use treatment facility and often include only facilities that receive State-administered public funds. This suggests that, as expected, due to its design, the TEDS counts of admissions with no prior treatment are much lower than the estimate of total number of persons who received any treatment for the first time in any given year.

There are also multiple possible explanations for the lower numbers in the NSDUH estimates in nearly all but the most recent years. Because NSDUH is a self-report survey, it relies on respondents' memory for when certain events, such as going to substance use treatment, occurred. Consequently, these estimates are susceptible to memory biases such as "forward telescoping," in which respondents tend to report that an important event from their past occurred more recently than it actually did (Johnson & Schultz, 2005) or simply forgetting treatment that occurred further back in the past. Another possible explanation is that in the early years shown in Figure 2.5 (e.g., 1995), a subset of NSDUH respondents in 2008 and 2009 who had received treatment in their lifetime were not at an age at which substance use treatment would have been a likely issue. For example, a respondent who was 18 years old in 2008 when he participated in the NSDUH was born in 1990. That respondent would have been 5 years old in 1995 and thus would have been very unlikely to have received substance use treatment for the first time in that year. A third possible explanation is that because of the increased risk of morbidity and mortality among persons with substance use problems, some of the persons who were in substance use treatment in 1995 might not have been eligible to have participated in the 2008 or 2009 survey due to incarceration, hospitalization, or death. These results indicate that estimates for a given year using NSDUH should use data from no more than 1 or 2 survey years after the year of interest (e.g., estimates for 2005 should use data from 2006 or 2007, not data from 2008 or later) because of declining data quality.

2.3. Single-Day Treatment Counts

Single-day treatment counts are estimates of the number of persons who were receiving substance use treatment on a given day of the year. The single-day treatment measures included in this section are

• estimates from the 2008 to 2010 NSDUH data of the number of persons in substance use treatment at a hospital, drug rehabilitation facility, or mental health center on October 1 of the year prior to the survey;
• estimates from the 2007 to 2009 NSDUH data of the number of persons in specialty substance use treatment on an average day of the year (estimates are limited to treatment received at specialty facilities in order to increase their comparability with the NSDUH October 1 estimates and the TEDS counts);
• counts from the 2007 to 2009 TEDS data of the number of persons in specialty substance use treatment on the last working day of March of a given year to match N-SSATS;
• counts from the 2007 to 2009 N-SSATS data of the number of persons in substance use treatment in all treatment facilities on the last working day of March of a given year; and
• counts from the 2007 to 2009 N-SSATS data of the number of persons in substance use treatment facilities, except for hospitals, on the last day of March of a given year (estimates are limited to outpatient facilities in order to increase their comparability with TEDS).

In the sections that follow, national and regional single-day treatment counts from these three data sources are presented, followed by counts from the eight most populous States. Tables containing single-day treatment counts for the Nation, the four census regions, and the eight most populous States are provided in Appendix C, Tables C.2.2-C.2.6. For comparison purposes, tables containing single-day treatment counts for the 50 States plus the District of Columbia are also included in Appendix C (see Tables C.2.7-C.2.11). The State estimates other than for the eight most populous States are not discussed in the text for this section.

Individual single-day treatment counts were generated for alcohol or drugs, alcohol only, drugs only, both alcohol and drugs, as well as with alcohol as the primary substance, and with a drug as the primary substance. N-SSATS data do not include sufficient detail to determine whether treatment received for both alcohol and drugs was primarily for alcohol or primarily for drugs; therefore, the alcohol primary and drug primary counts were not generated using N-SSATS data.

2.3.1 NSDUH Estimates

Description of NSDUH Estimates

Two different single-day treatment estimates can be generated using NSDUH data. One estimate uses data from a question about whether or not the respondent was receiving substance use treatment at the time of the interview,⁶ as well as the outcome of their last treatment episode. Respondents who indicated that they were currently receiving treatment were asked whether or not they were receiving this treatment for each of the substances they had used during their lifetime. This estimate will be referred to as the "average-day single-day treatment estimate."

In NSDUH, respondents who were receiving treatment at the time of the interview were asked whether they were receiving treatment for each of the substances they had used during their lifetime.⁷ If the respondent indicated this treatment was for more than one substance, they were asked to identify the main substance for which this treatment was received. These responses were used to determine which substances were the focus of the treatment they received during their current treatment episode.

The second estimate is based on responses to a question that asks respondents whether or not they were in substance use treatment on October 1 of the year prior to the survey. This estimate will be referred to as the "October 1 single-day treatment estimate." Due to a significant percentage (36.2 percent) of cases with unknown information about the substance(s) for which treatment was received, we present only estimates of treatment for either alcohol or drug use.

Both the average-day single-day treatment estimates and the October 1 single-day treatment estimates include only reports of specialty substance use treatment, which includes treatment received at a hospital, a drug rehabilitation facility, or a mental health center.

The overall average-day single-day treatment count was 1,196,460 persons. Among those persons, 36.7 percent were treated for alcohol use only, 26.3 percent were treated for drug use only, 33.2 percent were treated for both alcohol and drug use, and for 3.8 percent, this information was unknown. The October 1 single-day treatment count was 1,434,851 persons, who were treated for either alcohol or drug use.

Limitations of the NSDUH Estimates

As with the previous two sets of estimates discussed, the NSDUH sample is drawn to be representative of the noninstitutionalized population of the United States. In addition, NSDUH relies on respondents to remember accurately and report honestly their substance use and substance use treatment experiences.

The NSDUH questions do not specifically ask whether a person was in specialty treatment on the day of the interview. Respondents are asked what service types (modalities) they received in the past year and what the main type of treatment they are currently receiving is (for those who are still in treatment). Those who received specialty treatment in the past year and were in treatment on the day of the interview have been included in the average-day single-day treatment counts. Constructing the average-day single-day treatment counts in this way likely includes persons who were in specialty treatment in the past year but were not receiving specialty treatment at the time of the interview (e.g., had been receiving outpatient treatment at a mental health facility 6 months ago and is only participating in a self-help group at the time of the interview). Likewise, constructing the average-day single-day counts in this manner likely excludes persons who were receiving in-patient or residential treatment on the day of the interview (persons staying in a hospital or residential treatment center would not be available for an interview).

2.3.2 TEDS Counts

Description of TEDS Counts

The TEDS single-day treatment counts were computed for the number of persons in treatment on the last working day in March from 2007 to 2009 (March 30, 2007; March 31, 2008; and March 31, 2009). TEDS person-level treatment counts were obtained for 47 States, either through linking the admissions and discharge data processed through October 10, 2011, or through an alternate estimation method. A more detailed description of the methods used to link admission and discharge records is provided in Section 1.2. The TEDS single-day treatment counts are derived from counts of the number of admissions to a specialty substance abuse facility.

TEDS admission data include the primary, secondary, and tertiary problem substances. These data are used to determine which substances were the focus of the treatment received. The average single-day treatment count for 2007 to 2009 was 532,109 persons. Among those persons, 19.3 percent were treated for alcohol use only, 40.9 percent were treated for drug use only, 37.4 percent were treated for both alcohol and drug use, and 2.4 percent had no reported substance use at the time of admission.

Limitations of the TEDS Counts

As mentioned previously, the generalizability of the TEDS counts is somewhat limited by the inclusion/exclusion criteria for the TEDS database and also by State-level differences in reporting requirements that affect which data are reported to TEDS from the individual States. TEDS explicitly excludes clients treated in an emergency room, a private doctor's office, a self-help group, or a jail/prison not run with State agency-administered funds. As Section 1.2 notes, many States require only publicly funded facilities to report data to TEDS, and some States only require facilities to report TEDS data for patients for whom treatment is publicly funded.

2.3.3 N-SSATS Counts

Description of N-SSATS Counts

N-SSATS includes a single-day census of the number of persons in hospital inpatient substance use treatment units, residential (nonhospital) treatment, or outpatient substance use treatment on the last working day of March of each year (March 30, 2007; March 31, 2008; and March 31, 2009). For outpatient facilities, the count includes only active clients (those who were seen at the facility for substance use treatment or detoxification at least once during the month of March and who were still enrolled in treatment on the last working day of March).

N-SSATS data include the percentages of clients treated for only alcohol use, only drug use, or both alcohol and drug use, but it does not include information on whether alcohol or drugs were primary for those in treatment for both alcohol and drug use. The average single-day treatment count for 2007 to 2009 among all facilities was 1,153,617 persons. Among those persons, 18.8 percent were treated for alcohol use only, 36.0 percent were treated for drug use only, and 45.2 percent were treated for both alcohol and drug use.

2.3.4 National and Regional Estimates

Single-Day Treatment Counts for Alcohol or Drug Treatment

The single-day treatment counts in Figure 2.6 include persons who received treatment for either alcohol or drug use (also see Table C.2.2). As would be expected, the TEDS count (532,109) was significantly lower than either of the NSDUH or the N-SSATS counts. The NSDUH October 1 count (1,434,851) was the highest count, followed by the NSDUH average-day count (1,196,460). The N-SSATS all-facilities count (1,153,617) was somewhat lower than the NSDUH counts, followed by the N-SSATS count that excluded persons receiving treatment in hospitals (1,139,670). Regional counts for alcohol or drug treatment are consistent with the national counts for alcohol or drug treatment in that the NSDUH October 1 counts for alcohol or drug treatment were highest in all regions except the Northeast, followed by the NSDUH average-day counts, and the N-SSATS all-facilities counts. The TEDS counts were lowest in all regions.

Single-Day Treatment Counts for Alcohol-Only Treatment

Figure 2.7 and Table C.2.3 show single-day treatment counts (NSDUH average day, TEDS, and N-SSATS) of persons who were receiving treatment for alcohol use problems but not for other drugs. There was considerable variation across the four estimates. The NSDUH average-day count (438,665) was the highest, followed by the N-SSATS count that included all facilities (216,832). The TEDS count was less than one third of the NSDUH average-day count and less than half of the N-SSATS all-facilities count. Because the facilities that report to TEDS are a subset of the facilities that report to N-SSATS, it is not surprising that the TEDS counts were less than the N-SSATS counts; as with all tables included in Appendix C, excluding hospitals from N-SSATS had a minimal impact on the single-day treatment counts. TEDS also represents mainly admissions to publicly funded substance use treatment facilities, perhaps explaining why this count was lower for TEDS than for NSDUH, which reflected treatment received at both public and private facilities.

Table C.2.3 also shows regional single-day counts for alcohol-only treatment. The regional counts followed the same pattern as the national counts. The NSDUH average-day counts were significantly higher than the N-SSATS all-facilities counts except the Northeast, and the TEDS counts were appreciably lower than the N-SSATS counts.

Single-Day Treatment Counts for Drug-Only Treatment

The pattern of estimates across data sources for single-day treatment counts for persons receiving treatment for drug use but not alcohol use problems was different than the pattern for alcohol or drug treatment and alcohol-only treatment (Figure 2.8; Table C.2.4). For example, the N-SSATS counts including all facilities (414,845) and excluding hospitals (411,654) were higher than the NSDUH average-day count (314,806). As was true of the alcohol-only and alcohol or drug counts, the TEDS count (217,502) was appreciably lower than the other counts. One possible explanation of this shift between the counts for alcohol-only treatment and drug-only treatment in NSDUH and N-SSATS is that the increased stigma of receiving treatment for a drug use problem relative to that for receiving treatment for an alcohol use problem suppressed the self-reports of drug treatment receipt in NSDUH more than self-reports of alcohol treatment receipt. There was considerable variation at the regional level in the pattern of single-day counts for drug-only treatment. In the West, for example, the N-SSATS counts were the highest (96,774 and 96,368), followed by the TEDS count (68,868), with the NSDUH average-day count (62,355) as the lowest. In contrast, in the South, the N-SSATS counts were the highest (128,179 and 127,076), but the TEDS count (31,364) was less than one third of the NSDUH average-day count (Table C.2.4).

Single-Day Treatment Counts for Alcohol and Drug Treatment

Figure 2.9 presents single-day treatment counts for persons receiving treatment for both alcohol and drug use problems (also see Table C.2.5). These counts show a pattern similar to those seen in the drug-only counts in that the N-SSATS counts (521,940 and 514,128) were significantly higher than both the NSDUH average-day count (397,732) and the TEDS count (198,971), and the NSDUH average-day count was significantly higher than the TEDS count. Several of the regional counts for alcohol and drug treatment show a different pattern than do the national counts. For example, in the Midwest, the TEDS count (61,125) was not significantly lower than the NSDUH average-day count (63,211), although both were appreciably lower than the N-SSATS counts (109,918 and 108,793). In the South, the NSDUH average-day count (120,097) was not significantly lower than the N-SSATS counts (134,368 and 131,470), and both were more than 4 times higher than the TEDS count (26,589). These regional differences are difficult to interpret, which may be due to regional variations in collecting N-SSATS and/or TEDS data.

Single-Day Treatment Counts for Primary Alcohol with Secondary Drug Treatment

Figure 2.10 presents single-day treatment counts (NSDUH average day and TEDS) for persons receiving treatment for both alcohol and drug use problems and for whom alcohol was the primary substance of abuse (information on the primary substance of abuse for persons in treatment for both alcohol and drug problems is not available from N-SSATS) (also see Table C.2.6). These counts show a pattern similar to those seen in all of the previously reported counts in that the NSDUH average-day count (165,571) was significantly higher than the TEDS count (86,898). In two of the regional counts, however, the TEDS counts were not significantly lower than the NSDUH counts. In the Northeast, the NSDUH count was 29,702, and the TEDS count was 25,999. In the Midwest, the NSDUH count was 22,992, and the TEDS was 28,262, although the difference was not statistically significant.

Single-Day Treatment Counts for Primary Drug with Secondary Alcohol Treatment

Figure 2.11 presents single-day treatment counts from NSDUH (average day) and TEDS for persons receiving treatment for both alcohol and drug use problems and for whom a substance other than alcohol was their primary substance of abuse (information on the primary substance of abuse for persons in treatment for both alcohol and drug problems is not available from N-SSATS) (also see Table C.2.7). Similar to what was found with other counts, the TEDS count (111,973) was significantly lower than the NSDUH average-day count (228,813). Regional counts for primary drug with secondary alcohol treatment are consistent with the national counts in that the TEDS counts were lower than the NSDUH average-day counts. These differences were statistically significant in all regions but the Midwest.

2.3.5 State Estimates

The single-day treatment counts presented previously at the national and regional levels were also estimated for each of the 50 States and the District of Columbia. This section of the report focuses on the single-day treatment counts from the eight States with the largest populations (California, Florida, Illinois, Michigan, New York, Ohio, Pennsylvania, and Texas). These States were selected for inclusion because the NSDUH sample size for these eight States was roughly 4 times the sample size for the other States, resulting in more precise estimates for these States. The estimates for the eight largest States are given at the bottom of Tables C.2.2-C.2.7, and single-day treatment counts for all 50 States plus the District of Columbia are presented in Tables C.2.8-C.2.12.

Single-Day Treatment Counts for Alcohol or Drug Treatment

The single-day treatment counts in Figure 2.12 include persons who received treatment for either alcohol or drug use (also see Table C.2.2). The estimates for these counts from seven of the eight States show a similar pattern to those of the national and several of the regional counts, in that the NSDUH October 1 count is the highest among the five data source estimates. The exception is New York, for which the N-SSATS count (117,075) is higher than both the NSDUH October 1 count (110,008) and the NSDUH average-day count (100,975), although neither difference is statistically significant. In six of the eight States, the TEDS estimates are the lowest, though not always statistically significantly so.

Single-Day Treatment Counts for Alcohol-Only Treatment

Figure 2.13 shows single-day treatment counts for persons receiving treatment for alcohol use problems but not for other drugs, among the eight largest States (also see Table C.2.3). As with the counts for the Nation as a whole, all States had a higher NSDUH average-day count than any other count, although the differences between NSDUH and N-SSATS all-facilities counts in Illinois and New York were not statistically significant. Note that there is less variability between the counts in New York than between those in the other States, with a range from 20,683 (the NSDUH average-day count for New York) to 13,131 (the TEDS count for New York). In other States, such as California, there is a much larger gap between estimates. These differences in variability between estimates are likely caused by variability in individual State reporting requirements for TEDS data. In New York, all facilities that receive State/public funding and all facilities that are licensed/certified by the SSA are required to report data on all clients in the facility, whereas in Florida, facilities that receive State/public funding are required to report data on all clients, and in Illinois, facilities that receive funding through the SSA are required to report data on State-funded/publicly funded clients only.⁸

Single-Day Treatment Counts for Drug-Only Treatment

Single-day treatment counts for persons receiving treatment for drug use but not alcohol use problems are shown in Figure 2.14 (and Table C.2.4). As was true for the national estimates, the single-day drug-only treatment counts from the eight States reflect a different pattern from the alcohol-only treatment counts. The N-SSATS count including all facilities is the highest count among the five estimates except those for Illinois, where the NSDUH average-day count is larger than the N-SSATS all-facilities count (20,963 vs. 16,527), and Pennsylvania, where the NSDUH average-day count is larger than the N-SSATS all-facilities count (29,081 vs. 19,470). Note that the differences between the N-SSATS count and NSDUH estimates in Florida, New York, Pennsylvania, and Texas are not statistically significant. There is considerable variation between the estimates for each of the eight States.

Single-Day Treatment Counts for Alcohol and Drug Treatment

Figure 2.15 shows single-day treatment counts (NSDUH average day, TEDS, and N-SSATS) for persons receiving treatment for both alcohol and drug use problems (also see Table C.2.5). There is a great amount of variation between estimates within individual States. The TEDS estimate is the lowest of the counts in all States, although not significantly so in Illinois, New York, Ohio, and Texas.

Single-Day Treatment Counts for Primary Alcohol with Secondary Drug Treatment

Figure 2.16 shows single-day treatment counts (NSDUH average day and TEDS) for persons receiving treatment for both alcohol and drug use problems and for whom alcohol was the primary substance of abuse (also see Table C.2.6). With the exceptions of Florida and New York, the NSDUH average-day estimates are higher than the TEDS counts. This difference was statistically significant only in California and Michigan. No NSDUH average-day estimate is reported for Illinois due to low precision.

Single-Day Treatment Counts for Primary Drug with Secondary Alcohol Treatment

Figure 2.17 shows single-day treatment counts for persons receiving treatment for both alcohol and drug use problems and for whom a substance other than alcohol was the primary substance of abuse (also see Table C.2.7). With the exception of Ohio, the NSDUH average-day estimates are higher than the TEDS counts. This difference was statistically significant in California, Florida, and Pennsylvania.

Seasonality of Admissions

One issue that could impact comparisons between the single-day treatment counts based on the NSDUH October 1 measure and the TEDS and N-SSATS measures based on the last working day in March is seasonality of treatment admissions. If there are differences in the overall number of admissions around the times of these single-day treatment counts—for example, more admissions in the fall than in the spring—this could help explain the variation found in the counts between these data sources. In order to investigate whether there is seasonal variation in treatment admissions, we analyzed the TEDS admissions data from 2006 and 2007. These data show variation in admissions across the year, and the admissions patterns in both years were similar (see Figure 2.18). Average daily admissions for 2006 and 2007 combined ranged from a low of 4,262 in December to a high of 5,606 in January.

Because substance dependence is a chronic disorder that often requires long-term treatment, we also looked at TEDS linked admission and discharge data from 2006 and 2007 to investigate whether there is seasonal variation in the number of clients receiving treatment. Among these clients, there were between 325,000 and 425,000 clients in treatment every day during 2006 and 2007. As Figure 2.19 shows, the patterns in both years were similar, with distinct dips in the number of persons receiving treatment at the end of June and the end of December. Note that one possible reason for the dip at the end of June is that the fiscal year ends at this time for some States, and resources for substance use treatment may be reduced during this period relative to other times of the year. Also note that the number of persons in treatment was somewhat higher at the end of March (an average of 410,906 persons in treatment on the last working day of March) than at the beginning of October (an average of 390,277 persons in treatment on October 1). These data indicate that although there are seasonal differences in the number of persons receiving substance use treatment, the differences found between the end of March and the beginning of October do not seem to be marked enough to explain the differences found between NSDUH single-day treatment estimates and the single-day treatment counts from TEDS and N-SSATS.

2.3.6 Conclusions and Limitations

Table 2.1 shows the single-day treatment counts for all alcohol or drug treatment, as well as those for alcohol only, drugs only, and alcohol and drugs, including, when available, treatment for alcohol and drugs in which alcohol was the primary substance and treatment for alcohol and drugs in which a drug was the primary substance. Figure 2.20 illustrates the proportion of the overall single-day treatment counts from each source that received treatment for alcohol only, drugs only, and alcohol and drugs. The N-SSATS and NSDUH single-day treatment counts for alcohol or drug treatment are nearly twice the TEDS count. It is unclear if this is solely because the facilities that report to TEDS and the clients that are reported about in TEDS are associated with the receipt of public funds, whereas NSDUH estimates and N-SSATS counts reflect treatment received at both publicly and privately funded facilities. Excluding clients receiving care at hospitals from N-SSATS had a minimal impact on the N-SSATS single-day treatment counts. The NSDUH single-day estimates range from more than 100 percent higher than the N-SSATS estimate (based on the average-day count for alcohol-only treatment) to 25 percent lower (based on the average-day count for drugs only).

The NSDUH single-day treatment count for alcohol-only treatment is more than 4 times the TEDS count, but the TEDS single-day treatment count for drug-only treatment is about two thirds the NSDUH count. One possible explanation for this shift in the difference between NSDUH and TEDS counts for alcohol-only and drug-only treatment is that the increased stigma of receiving treatment for a drug use problem relative to receiving treatment for an alcohol use problem suppressed the self-reports of drug treatment receipt in NSDUH more so than self-reports of alcohol treatment receipt. Among other possible explanations, TEDS may not include a representative proportion of facilities that treat alcohol in the absence of other drug problems, problems related to alcohol use are minimized in the context of other drug use at the time of treatment admission, or a disproportionate number of persons who seek alcohol treatment receive the treatment at private facilities that are less likely to be included in TEDS.

There are two issues that limit the utility of national TEDS single-day treatment counts. First, a great deal of variability exists between States in data reporting requirements and, consequently, in the data reported to TEDS. In some States (such as New York), nearly all substance use treatment facilities are required to report TEDS data, whereas in other States, the reporting requirements are not as broadly applied. In many States, the only treatment facilities that report TEDS data are those that receive public funding. The N-SSATS data for these estimates are based on counts of persons in treatment on a single day of the year, the last working day of March. About 80 percent of the N-SSATS single-day count is comprised of clients receiving outpatient treatment. For outpatient clients, all clients who received any outpatient service in the month of March who had not been discharged as of the last working day of March are included in the single-day count. Seasonal variation of admissions may affect the representativeness of these estimates.

Each of the national single-day treatment estimates (NSDUH, TEDS, and N-SSATS) provides unique information about the number of persons receiving treatment on a given day of the year. N-SSATS is designed to provide a single-day treatment count. The N-SSATS count across all facilities is likely the most inclusive and stable estimate. The counts from TEDS (with some exceptions, such as New York State) provide information that more likely reflects the number of persons receiving treatment in publicly funded substance use treatment facilities that provide rehabilitation treatment services. The NSDUH estimates provide an estimation of the number of persons from the noninstitutionalized household population who were in substance use treatment (or specialty substance use treatment) on a single day of the year (October 1 of the year prior to the survey) or an average day of the year. The single-day count from TEDS was calculated using the last working day of March to match the date used in N-SSATS. The relative lack of seasonal differences in TEDS admissions indicates that comparisons of single-day counts from N-SSATS or TEDS with single-day counts from NSDUH (which uses an October 1 date) are reasonable. At the national level, the October 1 single-day treatment estimate from NSDUH is higher than the average-day estimates, although this pattern does not hold for all regions and States. These differences may reflect seasonal differences particular to a given region of the country, or they may be due to differences in the way the two estimates are derived (i.e., one is based on a question asking about having been in treatment on a particular day in the past and the other is based on the respondent currently being in treatment).

2.4. Past Year Admissions Counts

This section presents comparisons between TEDS and N-SSATS counts of the total number of substance use treatment admissions for a given calendar year.

2.4.1 TEDS Counts

Description of TEDS Counts

The TEDS counts of the number of persons who were admitted to substance use treatment are derived from admission records. As previously noted, TEDS data are less likely to include admissions for privately funded treatment than admissions for publicly funded treatment. The limitations of these counts are detailed subsequently.

Limitations of the TEDS Counts

TEDS does not include all admissions to substance abuse treatment. In some States, it includes admissions to facilities that are licensed or certified by an SSA to provide substance abuse treatment (or that are administratively tracked for other reasons). With few exceptions, TEDS does not include clients treated in a jail/prison not run with State agency-administered funds, emergency rooms, private doctor's offices, or self-help groups. As Section 1.2 notes, many States require only publicly funded facilities to report data to TEDS, and some States only require facilities to report TEDS data for patients for whom treatment is publicly funded. Also noted in Section 1.2, hospital inpatient substance use treatment facilities are frequently not licensed through the SSA and may not be required to report TEDS data.

2.4.2 N-SSATS Counts

Description of N-SSATS Counts

The N-SSATS counts of the number of treatment admissions to substance use treatment are derived from a question in the N-SSATS survey that asks facilities to report how many admissions for substance abuse treatment the facility had in the 12-month period beginning April 1 of the prior year and ending the last working day of March in the current year. Most facilities report this information. The item response rate for this question was 97.7 percent in 2007, 98.1 percent in 2008, and 100.0 percent in 2009.

2.4.3 National Estimates

This section compares the counts of substance use treatment admissions for N-SSATS and TEDS. Alabama, Alaska, the District of Columbia, and Georgia were excluded from the TEDS counts estimates because no or incomplete TEDS data were submitted for one or more of the 3 years examined (2007-2009) in these areas. For the purpose of these comparisons, N-SSATS estimates were restricted to the subset of 47 States that were used to generate the TEDS counts. These counts, for 2007, 2008, 2009, and an average of the 3 years are presented in Appendix C (Table C.2.13). As Figure 2.21 shows, the total TEDS admissions count is less than 60 percent of the N-SSATS admissions count (1,953,448 vs. 3,483,632, respectively). This difference is seen in all geographic regions except the Northeast. The discrepancy between the TEDS count of national admissions and that of N-SSATS is even greater when the TEDS national count is compared with the N-SSATS count for all 50 States and the District of Columbia (1,953,448 vs. 3,606,668).

2.4.4 Conclusions and Limitations

The N-SSATS admission counts for alcohol or drug treatment are almost twice the TEDS counts. This is likely because the facilities that report to TEDS are a subset of the facilities that report to N-SSATS. There are less pronounced differences between the two counts in the Northeast. This is likely because the Northeast includes New York and Massachusetts, which require all substance use treatment facilities licensed or certified by their SSAs to report TEDS data on all of their clients.

3.1. Current Methods of Assessing Needing but Not Receiving Treatment

Currently in NSDUH, persons are classified as needing treatment for an alcohol or drug use problem if they met the diagnostic criteria of the DSM-IV (American Psychiatric Association, 1994) for dependence on or abuse of alcohol or an illicit drug in the past 12 months, or if he or she received specialty treatment for alcohol use or illicit drug use in the past 12 months. A person is classified as having received treatment for an illicit drug or alcohol use problem if he or she has reported receiving specialty treatment for his or her drug or alcohol use problem in the past year. Thus, an individual who needed treatment for illicit drug use but only received specialty treatment for alcohol use in the past year or who received treatment for illicit drug use only at a facility not classified as a specialty facility was not counted as having received treatment for drug use. Similarly, an individual who needed treatment for an alcohol use problem was only counted as having received alcohol use treatment if the treatment was received for alcohol use at a specialty treatment facility.

SAMHSA publishes annual national estimates of the numbers and percentages of persons who needed substance use treatment in the past year, as well as the numbers and percentages of persons who received specialty treatment among those who needed treatment (CBHSQ, 2011b). SAMHSA has published only one report providing State-level estimates of the percentage of persons who received specialty treatment among those who needed treatment (Han, Clinton-Sherrod, Gfroerer, Pemberton, & Calvin, 2011). Due to the relatively small sample size at the State level of the persons who needed treatment, multiple years of data had to be used for this report. SAMHSA also publishes State estimates of the percentage of the full population who needed treatment but did not receive specialty treatment using 2-year rolling averages based on small area estimation models (OAS, 2010). This measure has the advantage of higher precision for State estimates because the entire population is used in the denominator rather than just those who needed substance use treatment. A disadvantage of this measure is that it is not possible to determine whether changes over time in this measure are due to changes in the number of persons who need treatment or the proportion of persons who need treatment and actually receive it.

This section focuses on persons who needed treatment but did not receive specialty treatment; this estimate was produced by dividing the number of persons who needed treatment by the number who received specialty treatment, and subtracting that figure from 100 percent.

3.2. An Alternative Method of Assessing Needing but Not Receiving Treatment

One possible criticism of the current measure of needing but not receiving treatment using NSDUH data is that it relies on estimates of the number of persons who have received specialty treatment in any year based on relatively small sample sizes. Available data sources, such as TEDS, that provide actual counts of the number of persons who have received substance use treatment present an alternative to using NSDUH data for estimates of both those who need treatment and those who receive treatment. It may be that using actual counts from TEDS of the number of persons who received treatment may result in a more accurate measure than estimating that number using NSDUH data. This may be of interest especially when producing measures at the State level, given the relatively low number of sampled persons who receive specialty treatment in each State.

In order to evaluate this alternate method, the percentages of persons needing but not receiving specialty substance use treatment were computed in two ways. The first method used the NSDUH estimate of treatment need, based on annual averages from the 2008 to 2010 surveys as the denominator and the NSDUH estimate of the number of persons receiving specialty treatment as the numerator. The second method used the NSDUH estimate of treatment need as the denominator and the TEDS count, based on annual averages from the 2007 to 2009 data, of the number of persons receiving treatment as the numerator.

3.3. National and State Estimates of Needing but Not Receiving Treatment

3.3.1 National and State Estimates

Table C.3.1 in Appendix C presents the estimates from NSDUH of the number of persons who needed treatment and the number of persons who received specialty treatment and counts from TEDS of the number of persons who received treatment. Table C.3.1 also includes two estimates of the percentage of persons who needed and received treatment, one using the NSDUH estimate of the number of persons who received specialty treatment as the numerator, and the second using the TEDS count of the number of persons who received treatment as the numerator. Because these counts reflect the number of persons who received treatment (as opposed to the number of treatment admissions), these estimates and percentages are available only for the 47 States for which we have TEDS data (Alabama, Alaska, the District of Columbia, and Georgia are excluded) for all 3 years examined.

Figure 3.1 compares the percentages of persons who did not receive treatment among those who needed treatment in NSDUH (NSDUH estimate of the number of persons who received specialty treatment in the numerator) and TEDS (TEDS counts of the number of persons who received treatment in the numerator). Section 2.1.2 of this report provides details of how counts of persons who received treatment were computed using the TEDS data. The figure presents combined estimates or the set of 47 States for which sufficient data were available in TEDS (the "Total" bars) and presents separate estimates for the 8 most populous States. The total estimates of the number of persons who needed but did not receive treatment from NSDUH and TEDS are quite similar (89.3 and 91.7 percent, respectively). With the exceptions of Ohio and New York, the percentages of persons who needed but did not receive treatment are higher in State estimates using the TEDS numerator than in those based on the NSDUH numerator, although the differences in California and Illinois are not statistically significant. In California, the estimates generated using the NSDUH data and those generated using the TEDS data are quite similar. The two States with the largest gaps between estimates are Michigan (92.9 vs. 85.4 percent unmet treatment need for NSDUH and TEDS estimates, respectively) and Pennsylvania (87.6 vs. 94.5 percent unmet treatment need for the NSDUH and TEDS estimates, respectively).

3.3.2 NSDUH Estimates for Persons with Low Income

The TEDS data in many States are representative of only the persons in that State who received substance use treatment in publicly funded treatment programs (e.g., Florida, Hawaii, Nevada) or only those clients in publicly funded programs whose individual treatment was publicly funded (e.g., Indiana, Louisiana, Nebraska). In an effort to make the NSDUH estimates of persons needing treatment more comparable with the TEDS counts of persons receiving treatment, an additional set of analyses were run using NSDUH data that were subset down to persons with a household income that is less than the Federal poverty threshold, and thus would be more likely to have received publicly funded treatment.

As a means of avoiding suppression of estimates due to lower sample sizes in this restricted group, we limited this analysis to the eight most populous States in the NSDUH data set (California, Florida, Illinois, Michigan, New York, Ohio, Pennsylvania, and Texas). Nonetheless, estimates for Florida and New York were suppressed due to low precision. We also conducted these analyses on the remaining States (the States other than the eight most populous States); however, due to lower sample sizes of respondents who received treatment, estimates were suppressed for all but one of these States.

Figure 3.2 compares the estimates of the percentage of persons who did not receive treatment among those who needed treatment generated using the NSDUH estimates among persons with low income and TEDS counts of persons who received treatment in six of the eight most populous States for which we had sufficient sample size. As noted previously, in these analyses, we used NSDUH data from 2008 through 2010 and TEDS data from 2007 through 2009. Unlike the results of the analyses presented earlier, which were not restricted to persons with low incomes, the TEDS-based estimates are lower than those generated with NSDUH-based treatment estimates; however, for the Michigan estimates, this difference is not statistically significant. In some States, such as California, Illinois, and Ohio, there is a large difference between the percentage calculated using the TEDS numerator and the one in which the NSDUH numerator was used. For other States, such as Pennsylvania and Texas, the difference is less pronounced.

3.4. Conclusions and Limitations

These findings indicate that the percentage of persons who needed but did not receive treatment was slightly higher overall and in most States for which we could conduct analyses when using the TEDS counts of the number of persons who received treatment as the numerator rather than NSDUH estimates of the number of persons who received specialty treatment. Importantly, there was little difference between the two estimates at the State level, suggesting that using the TEDS counts in the numerator does not provide an advantage in generating State estimates over using the NSDUH estimates. Given the considerable differences between NSDUH and TEDS described throughout this report, as well as the relatively similar findings when using the NSDUH and TEDS estimates as the numerator of this statistic, these analyses do not suggest that using TEDS counts in the numerator of this statistic provides an advantage over using NSDUH estimates.

Of the 46 State-level comparisons made, the TEDS and NSDUH estimates are within 2 percent of one another in 13 States. The NSDUH estimate of needing but not receiving treatment is smaller than the TEDS estimate in 23 States, and the TEDS estimate is smaller than the NSDUH estimate in the remaining 10 (Colorado, Iowa, Louisiana, Maine, Maryland, New Jersey, New York, South Dakota, Vermont, and Wyoming). Notably, all but 2 of these States (Vermont and Wyoming) require all facilities that are licensed or certified by the SSA to report data to TEDS on all clients, which generally indicates complete TEDS reporting from all facilities in the State. Among the other States that require more complete TEDS reporting, two had NSDUH estimates that were similar to the TEDS estimates (Arkansas and Connecticut), two had a NSDUH estimate of needing but not receiving treatment that was lower than the TEDS estimate (Montana and Rhode Island), and one had a TEDS estimate that was lower than the NSDUH estimate (New Jersey).

Regardless of which estimate of treatment receipt is used, the percentage of persons who did not receive treatment among those who needed it indicates that a large majority of persons who needed substance use treatment did not receive it. It may be that the broad criteria for classifying treatment need (i.e., having either substance abuse or dependence or having received specialty treatment in the past year) has an impact on this estimate. Specifically, it may be that some persons who are classified with substance abuse but not substance dependence may not meet the admissions criteria of chronic severity of substance use required by some types of specialty treatment facilities. The two measures of treatment receipt presented in this chapter focus on self-report of treatment at specialty facilities (i.e., a drug or alcohol rehabilitation facility [inpatient or outpatient], hospital [inpatient], or mental health center) or admission to one of the substance use treatment facilities that report to TEDS. Admission criteria for these types of programs often include some level of chronic severity of substance use disorder symptoms, most typically focusing on symptoms of dependence rather than abuse. In fact, for Opioid Treatment Programs, opioid dependence (not abuse) is one of the primary admission criteria. According to the 2007 NSDUH, 9.5 percent of the population was classified as needing substance use treatment, with 4.1 percent of the population meeting the criteria for abuse but not for dependence. Further research is needed to determine whether the sizable portion of those classified with treatment need who did not meet the criteria for substance dependence meet the admission criteria of different types of specialty treatment facilities and, if not, how this impacts the percentage of persons who needed but did not receive treatment.

4.1. Summary of Findings

4.1.1 Single-Day Treatment Counts

Single-day treatment counts, which provide a count or estimate of the number of persons enrolled in substance use treatment programs on a single day, can be computed using data from each of the three data sources. In general, NSDUH and N-SSATS estimates of total persons in treatment are similar, while estimates based on TEDS are lower by more than 50 percent (Table C.2.2). This was the case at the national level, and for most of the eight largest States, with some exceptions. A comparison of the NSDUH October 1 estimate and the NSDUH average-day estimate (i.e., the two methods that can be used to compute single-day treatment estimates from the NSDUH) shows that the average-day estimate is lower by 17 percent. Despite the similarity between the NSDUH and N-SSATS numbers, a concern with the NSDUH estimates is the inconsistencies in reporting across some treatment items. In particular, among the respondents whose interview occurred less than 12 months after the prior October 1 (i.e., quarters 1-3 of the following year), 25 percent of those respondents reporting that they had been in treatment on October 1 reported on separate questions that they had not received any treatment in the past 12 months.

While the overall single-day treatment counts between the N-SSATS and NSDUH are similar, there still is some variability among the single-day treatment counts for alcohol-only treatment (Table C.2.3) and drug-only treatment (Table C.2.4). For both drug-only and alcohol-only counts, the TEDS estimates are about 50 percent of the N-SSATS estimates, similar to the overall comparisons of persons in treatment. However, the NSDUH average-day alcohol-only estimate is higher than the corresponding N-SSATS estimate, while the NSDUH estimate for drug-only treatment is lower than the N-SSATS estimate. In terms of proportions, 37 percent of the estimated persons in treatment based on NSDUH were in treatment for alcohol only. Corresponding percentages for N-SSATS and TEDS were both 19 percent.

Figure 4.1 shows that the N-SSATS and TEDS single-day treatment counts were fairly stable across years, but the NSDUH estimates exhibited some variability from year to year. Overall, it appears that using a single-day treatment count to represent substance use treatment received on a typical day in a given year is acceptable because there is no significant seasonal variation in treatment receipt over the course of a year (Figure 2.18).

4.1.2 Number of Persons Who Received Treatment in the Past Year

For this measure, comparisons were made between NSDUH and TEDS (Table C.2.1). The NSDUH national estimate is about 28 percent higher than the TEDS estimate. Limiting the NSDUH data to just the 47 States that reported to TEDS, the NSDUH estimate is still about 22 percent higher than the TEDS count.

The comparisons by treatment modality (i.e., hospital inpatient, residential rehabilitation, or outpatient) provide some insight on the sources of the underestimation in TEDS, and possibly NSDUH as well. The NSDUH estimates of the number of persons receiving inpatient treatment were substantially higher than the TEDS counts. In particular, NSDUH estimated more than 10 times the number of persons treated as an inpatient in a hospital (738,366 vs. 61,382 in the 47 TEDS States). Again comparing with N-SSATS single-day counts, the TEDS number is an undercount. N-SSATS indicates that during 2007 to 2009 there were about 14,000 persons receiving substance use treatment in a hospital inpatient setting on the single day surveyed. Given the typical length of stays for this type of treatment is short, the N-SSATS count would project to several hundred thousand persons treated in a year. The NSDUH estimate for persons receiving outpatient treatment in the past year is higher than the TEDS estimate (1,916,780 and 1,507,988, respectively, in the 47 TEDS States), but both appear to be low when considering the annual N-SSATS single-day counts of clients in outpatient treatment (approximately 1 million) and the median length of stay across all outpatient clients. Likewise, a comparison of N-SSATS and TEDS counts of the number of past year treatment admissions (Table C.2.13) shows that TEDS admissions counts are more than 40 percent lower than those from N-SSATS. It is likely that the lower TEDS counts are due to (1) TEDS counts including mainly substance use treatment received at publicly funded facilities, whereas NSDUH and N-SSATS estimates include all facilities; and (2) the limited inclusion of hospital inpatient treatment patients in the TEDS sample.

The NSDUH estimates also were higher than the TEDS counts for receiving treatment for several substances: prescription drugs (620,335 vs. 240,288), inhalants (141,049 vs. 3,574), and hallucinogens (257,338 vs. 12,139). This is likely due to the methodological differences in the way these data are collected in the data sources. For example, in TEDS, the primary, secondary, and tertiary problem substances are recorded from the admission record, whereas in NSDUH, respondents are asked whether or not the treatment received in the past year was received for each substance they have used in their lifetimes.

Regarding client characteristics, the NSDUH estimates included a slightly lower percentage of youths aged 12 to 17 (6.4 vs. 8.8 percent) and a higher percentage of older adults (aged 45 or older; 29.0 vs. 22.5 percent) than the TEDS counts. The NSDUH estimates also included a higher percentage of persons with more than a high school education than the TEDS count (34.3 vs. 22.1 percent) and a lower percentage of persons with 8 or fewer years of education than the TEDS count (5.7 vs. 9.5 percent). Differences between TEDS and NSDUH estimates of specific racial or ethnic groups might be related to the broader inclusion of privately funded treatment in NSDUH but might also be associated with differences in the way that data concerning race and ethnicity were collected for the two data sources. Specifically, NSDUH respondents are asked to self-report the racial and ethnic groups with which they identify, whereas in TEDS, the processes for gathering this information vary between individual treatment facilities. NSDUH estimates show a higher proportion of clients who were white (66.6 vs. 60.4 percent) and lower proportions who were black (16.4 vs. 19.6 percent) or Hispanic (13.1 vs. 14.7 percent).

Comparisons of the client characteristics of NSDUH estimates of persons who received substance use treatment in the past year with TEDS data reported for those variables (CBHSQ, 2011a) reveal higher proportions of persons employed full time and part time in the NSDUH sample than in the TEDS counts (37.3 vs. 16.3 percent of persons employed full time for NSDUH and TEDS, respectively, and 12.9 vs. 7.6 percent of persons employed part time for NSDUH and TEDS, respectively). The findings for health insurance status were consistent with the pattern for employment, with a higher percentage of TEDS counts being uninsured (59.8 percent) compared with the NSDUH estimate (30.9 percent).

4.1.3 Receiving Treatment for the First Time

Though TEDS data are not typically presented in this manner, both NSDUH and TEDS can be used to compare estimates of the number of clients receiving substance use treatment for the first time in a given year. The NSDUH numbers are higher than the TEDS number for the year prior to interview, but the NSDUH numbers clearly decrease monotonically for estimates in previous years, indicating recall bias. A similar phenomenon has been shown for drug use initiation. Therefore, the NSDUH should not be used to look at first-time treatment retrospectively beyond 2 years.

4.1.4 Needing but Not Receiving Treatment

Percentages of persons who needed but did not receive treatment were computed in two ways: one using the NSDUH estimate of persons receiving treatment in the numerator (over the NSDUH estimate of persons needing treatment in the denominator), and the other using the TEDS count of persons receiving treatment (again using the NSDUH estimate of persons needing treatment in the denominator). Of the 46 State-level comparisons made, the TEDS and NSDUH estimates were similar in 12 States. In the remaining States, the NSDUH estimate of needing but not receiving treatment is smaller than the TEDS estimate in 25 States, and the TEDS estimate is smaller than the NSDUH estimate in 9 States. Using the TEDS numerator has the advantage that it does not have the sampling error that NSDUH estimates have, but given the coverage and response issues associated with TEDS, it is not clear that using TEDS data results in a substantial improvement of the estimates.

4.2. Implications for Using Substance Use Treatment Data from These Sources

4.2.1 Coverage

Across most measures, TEDS shows undercoverage relative to N-SSATS and NSDUH. Overall, N-SSATS appears to be a good source of data with regard to coverage for client counts nationally and at the State level for specialty treatment. If overall client counts are of interest, N-SSATS seems to be the best source of data. Although NSDUH national estimates are similar to the N-SSATS counts, internal inconsistencies in NSDUH reporting of treatment suggest caution and further exploration.

TEDS includes persons who are homeless and those who are institutionalized in treatment facilities long term, whereas homeless or institutionalized persons have little chance of inclusion in the NSDUH. NSDUH includes persons who are treated in privately funded facilities and persons whose treatment is privately funded as well as those whose treatment was publicly funded, whereas TEDS is mainly limited to those whose treatment was publicly funded. NSDUH also includes a great deal of information about individuals who received treatment that can be tied to other covariates collected in the survey, allowing for more specific analyses regarding coverage for different demographic groups. N-SSATS includes a census of all facilities regardless of funding, although in some cases the counts reported are estimates rather than actual counts.

4.2.2 Comparisons across States

All three data systems produce counts or estimates by State. State-level analyses not only are useful to local and State officials but also can provide policy assessments based on differing laws, policies, and conditions in States. These analyses require consistent data collection methods, definitions, and coverage across States to ensure the data are comparable. While comparability is achieved in NSDUH and N-SSATS, small sample sizes by State is an important limitation of NSDUH for State-level comparisons. Reporting anomalies in TEDS require extra caution in drawing conclusions about differences in treatment by State:

• The number and characteristics of TEDS client records depends, to some extent, on external factors, including the availability of public funds. In States with higher funding levels, a larger percentage of the substance-abusing population may be admitted to treatment, including the less severely impaired and the less economically disadvantaged.
• The way an admission is defined may vary from State to State such that the absolute number of admissions is not a valid measure for comparing States.
• States continually review the quality of their data processing. As systematic errors are identified, revisions may be enacted in historical TEDS data files. While this system improves the data set over time, reported historical statistics may change slightly from year to year.
• States vary in the extent to which coercion plays a role in referral to treatment. This variation derives from criminal justice practices and differing concentrations of abuser subpopulations.
• Public funding constraints may direct States to selectively target special populations (e.g., pregnant women or adolescents).

4.2.3 Implications Concerning Trends

Some of the basic questions about substance use treatment concern trends over time in service need or utilization. In TEDS, there are two principal sources of variations in estimates of variation over time: reporting anomalies and true changes in admission and discharge patterns. N-SSATS and NSDUH have relatively fewer methodological shifts or anomalies that may cause variations in the data over time. In the absence of those, differences in N-SSATS or NSDUH counts or estimates, when statistically significant and meaningful, may be expected to reflect actual changes. One limitation of NSDUH is that because it is a sample survey, trends in rare events and among subpopulations may be difficult to assess due to potentially large sampling error.

4.2.4 Unique Contributions of Each Data Source

Each of the three data sources makes a unique contribution toward our understanding of substance use treatment. TEDS is the only data source that provides information about both treatment admissions and discharges. NSDUH is the only source that provides the ability to link treatment or client characteristics with an extensive set of other relevant covariates; it is also the only source that has a measure of all substance use treatment received in the past year, not just specialty treatment. NSDUH is also the only source that provides data concerning treatment need and percentages of those in need who received treatment. N-SSATS provides the best overall single-day treatment counts nationally, for individual States, and across time.

4.3. Next Steps

The analyses presented in this report are the first step in refining analyses to identify the best ways to utilize CBHSQ data systems to answer some basic questions about substance use treatment, such as how many persons receive treatment in a year, how large is the gap between treatment received and treatment needed, and how have the numbers of persons receiving and needing treatment changed over time. The discussion and conclusions in this report are also a step toward understanding how to answer questions about the data sources, such as why do these studies give conflicting results, and which estimate should be used to answer questions about substance use treatment.

Some specific questions raised by this analysis need to be explored further to achieve maximum benefit from studies utilizing these three data sets. Avenues of investigation should include the following:

• What number/proportion of the TEDS count of persons in treatment are homeless persons or persons who have been institutionalized long term, and what are the characteristics of these persons?
• Can N-SSATS and TEDS be linked at some level to quantify the extent and the nature of the undercoverage in TEDS?
• Can the N-SSATS past year admissions numbers be used to generate person-level past year treatment counts?
• Are there changes in definitions, questionnaires, and eligibility rules in these three studies that will facilitate joint analyses?
• Can admissions be estimated from NSDUH, without adding new questions?
• Are there ways of utilizing TEDS and N-SSATS to enhance the precision of NSDUH small area estimates?
• How can NSDUH and TEDS be used to enhance N-SSATS so that more informative projections of admissions and persons receiving treatment can be made?

Finally, it is important to recognize that the nature of these data sets and how they may be used in the future will not remain constant. The movement toward more integrated care, mental health parity, and health care reform will undoubtedly lead to changing data needs and data systems. The results presented here will be valuable to policymakers and designers of data systems as the substance use treatment system evolves within the broader health care context.

A.1 Purpose of the Data Sets

While NSDUH is focused on alcohol and drug use in the general U.S. civilian, noninstitutionalized population, TEDS compiles information on client admissions to substance use treatment, and N-SSATS collects information from public and private facilities that provide substance use treatment. This section describes in more detail the objectives and scope of these three data sets.

NSDUH

NSDUH is the primary source of statistical information on the prevalence, patterns, and consequences of alcohol, tobacco, and illegal drug use and abuse in the general U.S. civilian, noninstitutionalized population aged 12 or older. In particular, NSDUH provides information about trends in the use of alcohol, tobacco, and illegal drugs in the population and trends within specific population subgroups. NSDUH yields estimates of substance use for all 50 States, the District of Columbia, and substate regions within each State. NSDUH also collects data on sociodemographic variables such as age, gender, pregnancy status, race/ethnicity, education, employment, and geographic area.

NSDUH also presents findings on mental health issues in the civilian, noninstitutionalized population. Specifically, NSDUH allows estimates to be made of the prevalence of mental illness and mental health service utilization, including the prevalence of mental disorders that co-occurred with substance use or with substance use disorders, and treatment for co-occurring mental disorders and substance use or substance use disorders.

TEDS

TEDS is a compilation of data on the demographic and substance abuse characteristics of admissions to and discharges from substance abuse treatment. TEDS is coordinated and managed by CBHSQ and is part of BHSIS, formerly known as DASIS, a cooperative program among SAMHSA and State substance abuse agencies to collect data on substance abuse services. The TEDS data collection effort was developed in response to the 1988 Comprehensive Alcohol Abuse, Drug Abuse, and Mental Health Amendments (P.L. 100-690), which established a revised SAPTBG and mandated Federal data collection on clients receiving treatment for either alcohol or drug abuse.

The primary goal of TEDS is to monitor the characteristics of clients admitted to planned, continuing treatment regimens. Thus, TEDS excludes early intervention and crisis intervention programs that do not lead to enrollment in continued treatment. A majority of States report data on all admissions to all eligible facilities, although some report only admissions financed by public funds.

N-SSATS

N-SSATS was designed to collect data on the location, characteristics, and utilization of alcohol and drug treatment facilities and services throughout the 50 States, the District of Columbia, and other U.S. jurisdictions.⁹ CBHSQ plans and directs N-SSATS. N-SSATS, along with TEDS, is a component of DASIS, a cooperative program among SAMHSA and State substance abuse agencies to collect data on substance abuse services. The 2009 survey was the 32nd in this series of national surveys begun in the 1970s.

N-SSATS provides the mechanism for quantifying the dynamic character and composition of the U.S. substance abuse treatment delivery system. The objectives of N-SSATS are to collect multipurpose data that can be used to

• assist SAMHSA and State and local governments in assessing the nature and extent of services provided in State-supported and other substance abuse treatment facilities, and in forecasting substance abuse treatment resource requirements;
• update SAMHSA's I-SATS, which includes all known drug and alcohol abuse treatment facilities;
• analyze substance abuse treatment services trends and conduct comparative analyses for the Nation, regions, and States;
• generate the National Directory of Drug and Alcohol Abuse Treatment Programs, a compendium of facilities approved by State substance abuse agencies for the provision of substance abuse treatment; and
• update the information in SAMHSA's Substance Abuse Treatment Facility Locator (http://findtreatment.samhsa.gov/), a searchable database of facilities approved by State substance abuse agencies for the provision of substance abuse treatment.

TEDS and N-SSATS in the Context of BHSIS

TEDS and N-SSATS are two of the three components of SAMHSA's BHSIS. Together, the components provide national- and State-level information on the numbers and characteristics of individuals admitted to alcohol and drug treatment programs and describe the facilities that deliver care to those individuals. BHSIS is the primary source of national data on substance use treatment. The core component of BHSIS is the I-SATS, a continuously updated comprehensive listing of all known public and private substance abuse treatment facilities. I-SATS is the list frame for N-SSATS. Facilities in I-SATS fall into two general categories and are distinguished by the relationship of the facility to its State substance abuse agency. These categories are described subsequently.

• Treatment facilities approved by State substance abuse agencies

The largest group of facilities (about 80 percent of all active I-SATS facilities) includes those that are licensed, certified, or otherwise approved by the SSA to provide substance abuse treatment. State DASIS representatives maintain this segment of I-SATS by reporting new facilities, closures, and address changes to SAMHSA. Some facilities are not licensed, certified, or otherwise approved by the State agency. Some private for-profit facilities fall into this category. This group also includes programs operated by Federal agencies, the VA, the DOD, and the IHS. I‑SATS records for federally operated facilities are updated annually through lists provided by these agencies.

• Treatment facilities not approved by State substance abuse agencies

This group of facilities (about 20 percent of all active I-SATS facilities) represents the SAMHSA effort since the mid-1990s to make I-SATS as comprehensive as possible by including treatment facilities that State substance abuse agencies, for a variety of reasons, do not fund, license, or certify. Many of these facilities are private for-profit, small group practices, or hospital-based programs. Most of them are identified through periodic screening of alternative source databases. State substance abuse agencies are given the opportunity to review these facilities and to add them to the State agency-approved list, if appropriate.

TEDS includes facilities that are licensed or certified by the SSA to provide substance abuse treatment (or are administratively tracked for other reasons) and that are required by the States to provide TEDS client-level data. N-SSATS is an annual survey of the location, characteristics, services offered, and utilization of alcohol and drug abuse treatment facilities in I-SATS.

A.2 Design

Fundamental design differences characterize the NSDUH, TEDS, and N-SSATS data sets due to their differences in purpose and scope. The following descriptions provide more detail on the design characteristics of each data set.

NSDUH

The 2005 through 2010 NSDUHs are part of a coordinated 5-year sample design (extended through 2010) providing estimates for all 50 States plus the District of Columbia for the years 2005 through 2010. The respondent universe is the civilian, noninstitutionalized population aged 12 years old or older residing within the United States. The survey includes persons living in noninstitutionalized group quarters (e.g., shelters, rooming/boarding houses, college dormitories, migratory workers' camps, halfway houses) and civilians living on military bases.

Although this population includes almost 98 percent of the total U.S. population aged 12 or older, it excludes some important and unique subpopulations that may have very different substance use and treatment patterns. For example, the survey excludes active military personnel who have been shown to have significantly lower rates of illicit drug use. Also, persons living in institutional group quarters, such as prisons and residential substance use treatment centers, are not included in NSDUH, and they have been shown in other surveys to have higher rates of illicit drug use. Also excluded are other types of institutional group quarters such as nursing homes, mental institutions, and long-term hospitals. Also excluded from the survey are persons with no fixed household address, such as homeless persons not living in a shelter for the majority of the quarter in which the shelter is sampled; homeless persons are another population shown to have higher than average rates of illicit drug use.

For the 50-State design, 8 States were designated as large sample States (California, Florida, Illinois, Michigan, New York, Ohio, Pennsylvania, and Texas) with target sample sizes of 3,600. For the remaining 42 States and the District of Columbia, the target sample size was 900. This approach ensures there is sufficient sample in every State to support small area estimation (SAE)¹⁰ while at the same time maintaining efficiency for national estimates.

More information on the sample design can be found in the 2008 NSDUH sample design report by Morton, Chromy, Hirsch, and Martin (2009) on the SAMHSA Web site (available as a PDF at https://www.samhsa.gov/data/Methodological_Reports.aspx).

TEDS

TEDS is designed to monitor the demographic and substance abuse characteristics of clients admitted to substance use treatment. TEDS admissions do not represent individuals but rather admission-level data. For example, one individual admitted to treatment twice within a calendar year would be counted as two admissions. However, data from some States include person-level identifiers that allow linkage of treatment episode records for an individual. This report uses these data to provide person-level rather than treatment episode-level estimates.

The national TEDS data are analyzed and published annually on a calendar year basis, although States submit TEDS data to SAMHSA on an ongoing basis as it is received from the provider facilities. The cutoff for receipt of data for each TEDS report is approximately August 31 of the year following the data year. Data received after that time does not appear in the initial annual report. For example, the annual report for calendar year 2009 admissions is prepared after the cutoff of August 31, 2010. For inclusion in the initial annual report, States must submit complete 2009 data before August 31, 2010.

TEDS data is not submitted by all States and jurisdictions each year. In 2007, Alabama, Alaska, Georgia, Mississippi, and West Virginia did not submit TEDS data; in 2008, Alabama and Georgia did not submit TEDS data; and in 2009, Georgia and the District of Columbia did not submit TEDS data.

Although TEDS does not exclude facilities, the facilities that report TEDS data are primarily those that receive State alcohol and/or drug agency funds (including Federal Block Grant funds) for the provision of drug and/or alcohol treatment services. The scope of facilities included is affected by differences in State licensure, certification, and accreditation practices, and disbursement of public funds. Although there is variation by State, the majority include only facilities receiving public funding. Therefore TEDS generally does not include data on private facilities, VA-operated facilities, DOD-operated military treatment facilities, hospital-based programs, and Federal prisons; some facilities operated by the IHS are included. A list of State reporting requirements, which is provided in the annual TEDS admissions reports, is presented in Table A.1.