OFFICE OF APPLIED STUDIES

Drug Abuse Warning Network, 2006:
National Estimates of Drug-Related Emergency Department Visits

U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES
Substance Abuse and Mental Health Services Administration
http://DAWNinfo.samhsa.gov

ACKNOWLEDGMENTS

This publication was prepared by the Substance Abuse and Mental Health Services Administration (SAMHSA), Office of Applied Studies (OAS) with RTI International under Contract No. 280-03-2602. Judy K. Ball, Ph.D., M.P.A. (DAWN Project Director, SAMHSA/OAS) and Victoria Albright, M.A. (Project Director, RTI) wrote the publication. Other significant contributors included CAPT Kathy M. Poneleit, M.P.H. (SAMHSA/OAS), Erin Mallonee, M.S. (Analyst, RTI), Karol Krotki, Ph.D. (Statistician, RTI), Leyla Stambaugh, Ph.D. (Analyst, RTI), and Peter Frechtel, M.A. (Statistician, RTI). The DAWN data collection was conducted by Westat under Contract No. 283-02-9025 under the direction of Josefina Moran, M.A.

PUBLIC DOMAIN NOTICE

All material appearing in this publication is in the public domain and may be reproduced or copied without permission from the Substance Abuse and Mental Health Services Administration (SAMHSA). However, this publication may not be reproduced or distributed for a fee without the specific, written authorization of the Office of Communications, SAMHSA. Citation of the source is appreciated. Suggested citation:

Substance Abuse and Mental Health Services Administration, Office of Applied Studies. Drug Abuse Warning Network, 2006: National Estimates of Drug-Related Emergency Department Visits. DAWN Series D-30, DHHS Publication No. (SMA) 08-4339, Rockville, MD, 2008.

OBTAINING ADDITIONAL COPIES OF PUBLICATION

This publication may be downloaded from http://DAWNinfo.samhsa.gov or from http://oas.samhsa.gov.

Hard copies may be obtained from http://oas.samhsa.gov/copies.cfm.

Or please call SAMHSA's Health Information Network at:

1-877-SAMHSA-7 (1-877-726-4727)
(English and Español).

ORIGINATING OFFICE

Substance Abuse and Mental Health Services Administration
Office of Applied Studies
1 Choke Cherry Road, Room 7-1044, Rockville, MD 20857
August 2008

CONTENTS

Acknowledgments

Highlights
ED visits involving drug misuse/abuse
Illicit drugs in ED visits
Alcohol and drug-related ED visits
Alcohol in combination with other drugs
Alcohol in patients under age 21
Nonmedical use of pharmaceuticals
Comparisons in drug misuse and abuse: 2004, 2005, and 2006
Special types of drug-related ED visits
Suicide attempts
Seeking detox

Introduction
Major features of DAWN
What is a DAWN case?
What drugs are included in DAWN?
What is covered in this publication?
Hospital participation in 2006
Estimates in this publication
Margin of error for estimates
Comparisons across years
Margin of error for comparisons across years
Estimates adjusted for population size

Drug Misuse and Abuse in ED Visits

Illicit Drugs in ED Visits

Alcohol in ED Visits
Alcohol in combination with other drugs (Tables 5-6, Figure 3)
Alcohol-related ED visits in patients under the age of 21 (Table 7)
ED visits for underage alcohol use (Tables 4, 8, Figure 4)

Nonmedical Use of Pharmaceuticals
Nonmedical use of pharmaceuticals (Tables 9-10, Figure 5)

Comparisons of ED Visits: 2004, 2005, and 2006
Drug misuse and abuse in ED visits (Table 11)
Illicit drugs in ED visits (Table 12)
Alcohol in ED visits (Tables 13-14)
Nonmedical use of pharmaceuticals (Table 15)

Special Types of Drug-Related ED Visits
Suicide attempts (Tables 16-17, Figure 6)
Suicide attempt ED visits: 2004, 2005, and 2006 (Table 18)
Seeking detox (Tables 19-20, Figure 7)
Seeking detox ED visits: 2004, 2005, and 2006 (Table 21)

List of Tables

Table 1. Drug misuse and abuse in ED visits in the U.S., by type of drug involvement: 2006
Table 2. Illicit drugs in ED visits: 2006
Table 3. Illicit drugs, by patient characteristics: 2006
Table 4. Alcohol in drug-related ED visits: 2006
Table 5. Alcohol in combination, by patient and visit characteristics: 2006
Table 6. Drugs most frequently reported with alcohol: 2006
Table 7. Alcohol in drug-related ED visits in patients under age 21: 2006
Table 8. Alcohol only (age < 21), by patient and visit characteristics: 2006
Table 9. Nonmedical use of pharmaceuticals: 2006
Table 10. Nonmedical use of pharmaceuticals, by patient and visit characteristics: 2006
Table 11. Drug misuse and abuse in ED visits in the U.S., by type of drug involvement: 2004, 2005, and 2006
Table 12. Illicit drugs in ED visits: 2004, 2005, and 2006
Table 13. Alcohol in drug-related ED visits: 2004, 2005, and 2006
Table 14. Alcohol in drug-related ED visits in patients under age 21: 2004, 2005, and 2006
Table 15. Nonmedical use of pharmaceuticals ED visits: 2004, 2005, and 2006
Table 16. Suicide attempts: 2006
Table 17. Suicide attempts, by patient and visit characteristics: 2006
Table 18. Suicide attempts: 2004, 2005, and 2006
Table 19. Seeking detox: 2006
Table 20. Seeking detox, by patient and visit characteristics: 2006
Table 21. Seeking detox: 2004, 2005, and 2006

List of Figures

Figure 1. Rates of ED visits involving selected illicit drugs: 2006
Figure 2. Illicit drugs, ED visit rates by age and gender: 2006
Figure 3. Alcohol with other drugs, ED visit rates by age and gender: 2006
Figure 4. Alcohol only (age < 21), ED visit rates by age and gender: 2006
Figure 5. Nonmedical use of pharmaceuticals, ED visit rates by age and gender: 2006
Figure 6. Suicide attempts, ED visit rates by age and gender: 2006
Figure 7. Seeking detox, ED visit rates by age and gender: 2006

List of Appendixes

Appendix A: Multum Lexicon End-User License Agreement
Appendix B: Glossary of Terms
Appendix C: DAWN Data Collection and Statistical Methods
Appendix D: Race and Ethnicity in DAWN

HIGHLIGHTS

This publication presents national estimates of drug-related visits to hospital emergency departments (EDs) for 2006, based on data from the Drug Abuse Warning Network (DAWN). Also presented are comparisons of 2006 estimates with those for 2004 and 2005. DAWN is a public health surveillance system that monitors drug-related ED visits for the Nation and for selected metropolitan areas. DAWN estimates pertain to the entire United States, including Alaska, Hawaii, and the District of Columbia. The Substance Abuse and Mental Health Services Administration (SAMHSA) is the agency responsible for DAWN. SAMHSA is required to collect data on drug-related ED visits under section 505 of the Public Health Service Act.

DAWN relies on a national sample of general, non-Federal hospitals operating 24-hour EDs. The sample is national in scope, with oversampling of hospitals in selected metropolitan areas. In each participating hospital, ED medical records are reviewed retrospectively to find the ED visits that were related to recent drug use. All types of drugs—illegal drugs, prescription and over-the-counter (OTC) pharmaceuticals, dietary supplements, and nonpharmaceutical inhalants—are included. Alcohol, when it is the only drug implicated in a visit, is included for patients younger than age 21; alcohol, when it is present in combination with another drug, is included for patients of all ages.

The 2006 estimates introduce several improvements made to the DAWN sampling and estimation methodology. In order to ensure comparability between estimates for 2004, 2005, and 2006, the improvements were applied retrospectively to 2004 and 2005 data, resulting in small changes as compared to early published estimates.

ED visits involving drug misuse/abuse

In 2006, hospitals in the United States delivered a total of 113 million ED visits, and DAWN estimates that 1,742,887 (confidence interval [CI]: 1,451,086 to 2,034,688)¹ ED visits were associated with drug misuse or abuse. Of those ED visits:

• 31% involved illicit drugs only,
  
• 28% involved pharmaceuticals only,
  
• 7% involved alcohol only in patients under the age of 21,
  
• 13% involved illicit drugs with alcohol,
  
• 10% involved alcohol with pharmaceuticals,
  
• 8% involved illicit drugs with pharmaceuticals, and
  
• 3% involved illicit drugs with pharmaceuticals and alcohol.
  

Illicit drugs in ED visits

For 2006, DAWN estimates that 958,164 (CI: 690,218 to 1,226,110) ED visits involved an illicit drug. Thus, over half (55%) of all the drug misuse/abuse ED visits during the year involved an illicit drug either alone or in combination with other types of drugs. DAWN estimates that:

• Cocaine was involved in 548,608 ED visits (CI: 374,579 to 722,636),
  
• Marijuana was involved in 290,563 ED visits (CI: 238,737 to 342,388),
  
• Heroin was involved in 189,780 ED visits (CI: 119,525 to 260,035),
  
• Stimulants, including amphetamines and methamphetamine, were involved in 107,575 ED visits (CI: 66,105 to 149,046), and
  
• Other illicit drugs, such as PCP, Ecstasy, and GHB, were much less frequent than any of the above.
  

Taking the margin of error into account, cocaine was more frequent than any of the other illicit drugs. The stimulants (amphetamines and methamphetamine) were less frequent than marijuana and as frequent as heroin.

After taking population size and the margin of error into account:

• The rates of ED visits involving cocaine, marijuana, and heroin were higher for males than females, but the rates for stimulants did not differ by gender,
  
• For cocaine, the rates for patients aged 18 to 54 were similar,² with lower rates for younger and older patients,
  
• For heroin, the rates were highest for patients aged 21 to 54,
  
• For marijuana, the rates were highest for patients aged 18 to 24, and
  
• For stimulants, the rates were highest for patients aged 18 to 44.
  

Alcohol and drug-related ED visits

For 2006, DAWN estimates that 577,521 (CI: 501,944 to 653,048) ED visits involved either alcohol in combination with another drug (all ages) or alcohol alone for patients under the age of 21. This is about one third (33%) of all drug misuse/abuse ED visits. Since DAWN does not account for ED visits involving alcohol alone in adults, the actual number of ED visits involving alcohol is higher. Alcohol is reported to DAWN when it is present in combination with other drugs, regardless of the patient's age.

Alcohol in combination with other drugs

In 2006, DAWN estimates 450,817 (CI: 383,818 to 517,816) ED visits related to use of alcohol in combination with other drugs. Alcohol was most frequently combined with:

• Cocaine alone (101,588 visits),
  
• Marijuana alone (41,653 visits),
  
• Cocaine and marijuana (21,241 visits), and
  
• Heroin alone (14,958 visits).
  

Alcohol in patients under age 21

DAWN estimates 76,760 (CI: 60,318 to 93,202) alcohol-related ED visits for patients aged 12 to 17, and 105,675 (CI: 82,757 to 128,593) alcohol-related ED visits for patients aged 18 to 20. Alcohol is an illegal drug for both of these age groups.

About two thirds (69%) of the alcohol-related ED visits for minors involved alcohol and no other drug. Taking population size and the margin of error into account:

• The rate of alcohol-only ED visits for patients aged 18 to 20 (581 visits per 100,000 population) was 2.8 times that for patients aged 12 to 17 (204 per 100,000), and
  
• Males and females had similar rates.
  

Nonmedical use of pharmaceuticals

For 2006, DAWN estimates that 741,425 (CI: 674,198 to 808,652) ED visits involved nonmedical use of prescription or OTC pharmaceuticals or dietary supplements. The majority of these visits (54%) involved multiple drugs.

Central nervous system (CNS) agents (present in 50% of nonmedical-use visits) and psychotherapeutic agents (44%) were the most frequent drugs reported in the nonmedical-use category of ED visits.

Among the CNS agents, the most frequent drugs were opiate/opioid analgesics (present in 33% of nonmedical-use visits), including single-ingredient formulations (e.g., oxycodone) and combination forms (e.g., hydrocodone with acetaminophen). Methadone, together with single-ingredient and combination forms of oxycodone and hydrocodone, were the most frequent opioids. Once the margin of error is taken into account, these three opioids appeared in similar numbers of visits:

• Hydrocodone/combinations in 57,550 ED visits (CI: 43,701 to 71,398),
  
• Oxycodone/combinations in 64,888 ED visits (CI: 49,746 to 80,030), and
  
• Methadone in 45,130 ED visits (CI: 35,870 to 54,389).
  

It is not possible to know, based on the documentation available in ED medical records, the extent to which the source of these drugs is a legitimate prescription, as opposed to other sources, nor is it possible to distinguish methadone used for treatment of opiate addiction from the methadone in pill form, which is prescribed for pain. In fact, methadone may be one of the most ambiguous drugs to categorize in DAWN. When a patient on opioid replacement therapy presents to an ED, methadone may be routinely documented in the medical record, but without sufficient detail to distinguish whether the methadone specifically was related to the ED visit.

Among the psychotherapeutic agents, the anxiolytics (anti-anxiety agents), sedatives, and hypnotics were the most frequent, occurring in almost a third (32%) of visits associated with nonmedical use of pharmaceuticals. ED visits involving benzodiazepines clearly outnumber those involving any of the other types of psychotherapeutic agents. DAWN estimates that 195,625 (CI: 167,789 to 223,461) ED visits associated with nonmedical use of pharmaceuticals involved benzodiazepines in 2006.

Taking population size and the margin of error into account:

• ED visit rates for nonmedical use of pharmaceuticals did not differ between females and males, and
  
• ED visit rates were highest for patients aged 18 to 44.
  

Comparisons in drug misuse and abuse: 2004, 2005, and 2006

In 2006, hospitals in the United States delivered a total of 113 million ED visits, an increase of 3.9% over 2004. The population of the United States increased 2.9%, from 294 million to 302 million, over the same period.

According to DAWN, the total number of ED visits attributable to drug misuse and abuse was stable across 2004, 2005, and 2006. That is, the apparent difference is within the margin of error. Across the different types of drug involvement, changes were detected for visits involving:

• pharmaceuticals alone (i.e., with no other type of drug), which increased 44% from 2004 to 2006;
• pharmaceuticals used in combination with illicit drug(s), which increased 36% from 2004 to 2006; and
• pharmaceuticals used in combination with alcohol, which increased 22% from 2005 to 2006.
  

Regarding the significant increases detected, it is worthwhile to consider that the number of pharmaceuticals dispensed for legitimate therapeutic uses may be increasing over time, and DAWN estimates are not adjusted to take such increases into account. Nor do DAWN estimates take into account the increases in the population or in ED use between 2004 and 2006.

No significant changes in ED visits from 2004 to 2006, or from 2005 to 2006, were detected for any of the major illicit drugs (cocaine, heroin, marijuana, and stimulants) or for alcohol.

ED visits related to nonmedical use of pharmaceuticals increased 38% in the period from 2004 to 2006. Among the drugs most frequently implicated in nonmedical use, notable changes from 2004 to 2006 occurred for psychotherapeutic agents (31%) and CNS agents (32%). Within these two categories, visits involving benzodiazepines increased 36%, and visits involving opiate/opioid analgesics increased 43%. Among the opiates/opioids, visits involving hydrocodone/combinations increased 44%, and visits involving oxycodone/combinations increased 56%.

DAWN is not able to assess whether increases or decreases in ED visits associated with specific pharmaceuticals are related to changes in the quantity of these pharmaceuticals being prescribed for therapeutic uses.

Special types of drug-related ED visits

Suicide attempts

DAWN estimates 182,805 (CI: 154,185 to 211,424) ED visits for drug-related suicide attempts in 2006.³ Nearly two thirds (65%) of ED visits for drug-related suicide attempts involved multiple drugs.

In these ED visits for drug-related suicide attempts in 2006:

• The large majority of drug-related suicide attempts (92%) involved pharmaceuticals,
  
• More than half (58%) included psychotherapeutic agents, such as benzodiazepines or antidepressants,
  
• Slightly less than half (45%) involved CNS agents, primarily analgesics (pain relievers), including both prescription and OTC formulations,
  
• Nearly one third (30%) involved alcohol (but DAWN data exclude visits for adults when alcohol is the only drug), and
  
• About one quarter (23%) involved an illicit drug.
  

Overall, there was no significant change in ED visits for drug-related suicide attempts during the time period from 2004 to 2006, but an increase was detected from 2005 to 2006. Increases were evident from 2004 to 2006, as well as 2005 to 2006, for some of the pharmaceuticals frequently involved in suicide attempts, such as benzodiazepines and opiate/opioid pain relievers.

Seeking detox

DAWN estimates 118,355 (CI: 90,171 to 146,540) drug-related ED visits for patients seeking detoxification or substance abuse treatment services during 2006. However, these visits tend to be concentrated in hospitals with administrative policies that require medical clearance in the ED for admission to these specialized units within the hospital. Therefore, these visits do not encompass the full extent of the demand for these services.

Nearly two thirds (65%) of the seeking detox ED visits involved multiple drugs. Illicit drugs and alcohol were common in these visits:

• Cocaine (49% of visits) and heroin (29%) were followed in frequency by marijuana (19%) and amphetamine or methamphetamine stimulants (7%), and
  
• Alcohol in combination with another drug was implicated in 40% of seeking detox ED visits.
  

Among the seeking detox ED visits, nearly 7 out of 10 (69%) received some type of follow-up care, either inpatient admission, referral elsewhere for detox or substance abuse treatment services, or transfer to another health care facility. However, more than one quarter (27%) of seeking detox cases might not have received the care they sought, because they were discharged to home.

No significant changes in ED visits from 2004 to 2006, or from 2005 to 2006, were detected for seeking detox ED visits overall, or for alcohol or the illicit drugs involved in these visits.

INTRODUCTION

This publication presents estimates of drug-related emergency department (ED) visits from the Drug Abuse Warning Network (DAWN) for 2006, with comparison of estimates for 2004 and 2005. DAWN is a public health surveillance system that monitors drug-related ED visits for the Nation and for selected metropolitan areas. The Office of Applied Studies (OAS) of the Substance Abuse and Mental Health Services Administration (SAMHSA), U.S. Department of Health and Human Services, has been responsible for DAWN operations since 1992.

Major features of DAWN

What is a DAWN case?

One of the most important features of DAWN is its expansive definition of a case:

A DAWN case is any ED visit related to recent drug use.

To be a DAWN case, the relationship between the ED visit and the drug use need not be causal; the drug needs only to be implicated in the visit. This approach accommodates cases where one or more drugs were involved but may or may not have directly caused the condition generating the ED visit, but at the same time avoids inclusion of current medications that are unrelated. Only recent drug use is included;⁴ the reason a patient used a drug need not be specified; and the criteria are broad enough to encompass all types of drug-related events, including explicit drug abuse. See Appendix C: DAWN Data Collection and Statistical Methods for a full description of DAWN cases and data collected on those cases.

What drugs are included in DAWN?

DAWN collects data on all types of drugs, including:⁵

• Illegal drugs, such as heroin, cocaine, marijuana, and Ecstasy,
  
• Prescription drugs, such as Prozac^®, Vicodin^®, OxyContin^®, alprazolam, and methylphenidate,
  
• Over-the-counter (OTC) medications, such as aspirin, acetaminophen, ibuprofen, and multi-ingredient cough and cold remedies,
  
• Dietary supplements, such as vitamins, herbal remedies, and nutritional products,
  
• Psychoactive, nonpharmaceutical inhalants,⁶
• Alcohol in combination with other drugs, and
  
• Alcohol alone, in patients aged less than 21 years.⁷

In some cases, the medical record includes only nonspecific drug categories (for instance, "benzodiazepines" or "opiates"). These situations are noted in DAWN publications with the abbreviation "NOS" or "Not Otherwise Specified." In addition, due to the multitude of drugs included in the DAWN drug vocabulary, it is not possible to show all drugs or even all types of drugs in this publication. In these situations, a residual group of drugs may be referred to in a table with the abbreviation "NTA" or "Not Tabulated Above."

What is covered in this publication?

While the full array of drug-related ED visits covered by DAWN is very broad, this publication focuses primarily on ED visits involving drug misuse and abuse. The national estimates of ED visits associated with drug misuse and abuse are presented in terms of the following three categories:

• Illicit drugs,
  
• Alcohol, and
  
• Nonmedical use of pharmaceuticals.
  

The illicit drugs category covers ED visits involving the use of substances that are generally illegal. The alcohol category includes alcohol used in combination with other drugs and alcohol used alone in patients under 21, but excludes alcohol used alone in patients aged 21 and over. Nonmedical use of pharmaceuticals includes ED visits related to the misuse or abuse of prescription or OTC medications or dietary supplements. Nonmedical use includes taking a higher than prescribed or recommended dose of a pharmaceutical (i.e., contrary to directions or labeling), taking a pharmaceutical prescribed for another individual, malicious poisoning of the patient by another individual, and substance abuse involving pharmaceuticals.⁸

In addition, this report includes a separate section on two special types of ED visits: drug-related suicide attempts and patients "seeking detox." The latter includes patients who present to the ED seeking detoxification services or entry into a substance abuse treatment program. These visits tend to be concentrated in hospitals with administrative practices requiring medical clearance in the ED for admission to detox or substance abuse treatment units within the hospital. Drug-related ED visits involving suicide attempts or seeking detox are excluded from the category of nonmedical use of pharmaceuticals.

Hospital participation in 2006

For 2006, 205 hospitals submitted data that were used for estimation. The overall weighted response rate was 26.1%. For the 12 oversampled metropolitan areas and divisions, the individual response rates ranged from 30.8% in the Houston metropolitan area to 71.2% in the Detroit metropolitan area. Additional detail on response rates is provided in Appendix C.

DAWN cases are found through a retrospective review of medical records in participating hospitals. Across all participating hospitals in 2006, 9.8 million charts were reviewed to find the drug-related ED visits that met the DAWN case criteria. Based on the review of charts, 346,946 drug-related visits were found and submitted. On average, a DAWN member hospital submitted 1,077 DAWN cases. However, the number of submitted cases varied widely across hospitals, from 7 cases to 6,782 cases (median 820) in a single hospital during 2006.

Estimates in this publication

Estimates in this publication were calculated by applying sampling weights, nonresponse adjustments, and poststratification adjustments to data from the probability sample of hospitals. Only national estimates pertaining to the entire United States—50 States and the District of Columbia—are provided in this publication.

Hospitals eligible for the DAWN sample are non-Federal, short-stay, general medical and surgical hospitals in the United States that operate 24-hour EDs. The American Hospital Association's (AHA) 2001 Annual Survey is the source of the sampling frame. Subsequent AHA surveys are used annually to refresh the sample. For a definition of sampling frame and other technical terms used in this publication, see Appendix B: Glossary of Terms.

The DAWN sample of hospitals includes an oversampling of hospitals in selected metropolitan areas (referred to as oversample areas), supplemented with a sample of hospitals from the remainder of the United States, which includes other metropolitan areas, as well as nonmetropolitan and rural areas. The metropolitan area boundaries correspond to the definitions issued by the Office of Management and Budget (OMB) in June 2003. National estimates are calculated as the sum of the estimates from oversample areas plus the estimate for the remainder of the United States after taking into account nonresponse, the volume of ED visits delivered by the universe of eligible hospitals in each area, and data quality factors. A more detailed discussion of the DAWN sample of hospitals and estimation procedures is provided in Appendix C.

Margin of error for estimates

Since DAWN relies on a sample of hospitals, each estimate produced from the DAWN ED data is subject to sampling variability, referred to as the "margin of error." This is the variation in the estimate that would be observed naturally if different samples were drawn from the same population using the same procedures. The sampling variability of an estimate in this publication is measured by its relative standard error (RSE). The precision of an estimate is inversely related to its sampling variability, as measured by the RSE: the greater the RSE, the lower the precision.

DAWN estimates with RSE values greater than 50% and quantities less than 30 are regarded as too imprecise for publication and are not shown. Three periods ("…") are displayed in the place of suppressed estimates. Ratios (percentages or rates per 100,000 population) based on suppressed estimates are likewise suppressed. Gray shading in a cell indicates that the cell is not applicable. For example, drugs other than alcohol cannot be present in an "alcohol-only" category.

In this publication, 95% confidence intervals (CIs) are included in many of the tables and are cited in the text along with the estimates. A 95% CI means that if repeated samples were drawn from the same population of hospitals using the same sampling and data collection procedures, the true population value would fall within the confidence interval 95% of the time. A CI, which is expressed as a range of values, is useful because the interval reflects both the estimate and its particular margin of error.

For readers unfamiliar with these statistical concepts, additional descriptions and examples are provided in Appendix C.

Comparisons across years

In this publication, between-year changes are assessed by comparing estimates for 2006 to those for 2004 and to those for 2005.

Major changes to DAWN were instituted at the beginning of 2003 as the result of a redesign that altered most of DAWN's core features. Changes included the design of the hospital sample, the drug-related cases eligible for DAWN, the data items submitted on these cases, and the protocol for case finding and quality assurance. These improvements created a permanent disruption in trends. As a result, comparisons cannot be made between old DAWN (2002 and prior years) and the redesigned DAWN (2004 and forward).⁹

Margin of error for comparisons across years

DAWN analysts tested each set of estimates between two years to ascertain if the difference exceeded its margin of error. To be reported in this publication as a change, the difference between estimates for two years must be statistically significant, that is, the difference must exceed the margin of error. Differences that are not statistically significant indicate that there was no real change between the two years. Appendix C provides additional detail on the method of testing for significant differences between estimates for different years.

Estimates adjusted for population size

Standardized measures are needed to make valid comparisons of ED visits across age and gender categories that differ in population size. For age in particular, the size of the underlying population differs considerably across age groups; for example, the number of individuals aged 18 to 20 in the United States is much lower than the number of individuals aged 35 to 44.

This publication reports rates of ED visits per 100,000 persons for unsuppressed estimates. Rates are generated using population data from the U.S. Bureau of the Census.¹⁰ An example of how these rates are generated and the 2006 population estimates used for this publication are found in Appendix C.

Standardized rates are not calculated for race and ethnicity subgroups, because the race and ethnicity categories available to DAWN are much less detailed and contain considerably more missing data than the race and ethnicity categories in the Census data. Appendix D: Race and Ethnicity in DAWN, describes the race and ethnicity data reported to DAWN.

DRUG MISUSE AND ABUSE IN ED VISITS

For 2006, DAWN estimates that over 1.7 million ED visits were associated with drug misuse or abuse (Table 1). This estimate includes:

• 958,164 ED visits (CI: 690,218 to 1,226,110) that involved illicit drugs alone or in combination with other drugs,
  
• 577,521 ED visits (CI: 501,994 to 653,048) that involved the use of alcohol alone or in combination with other drugs, and
  
• 860,108 ED visits (CI: 787,298 to 932,918) associated with nonmedical use of pharmaceuticals alone or in combination with other drugs.¹¹

Of the 1.7 million drug misuse/abuse visits, about two thirds (66%) were associated with a single drug type (illicit drugs, alcohol, or nonmedical use of pharmaceuticals). ED visits involving illicit drugs alone accounted for 31% of all visits related to drug misuse/abuse in 2006. ED visits involving nonmedical use of pharmaceuticals alone accounted for another 28%. About 7% of drug misuse/abuse visits were related to consumption of alcohol (and no other drug) by a minor.¹² The remaining visits (34%) involved some combination of illicit drugs, alcohol, and/or nonmedical use of pharmaceuticals.

This does not suggest that the majority of ED drug misuse/abuse visits involved a single drug. In fact, the typical drug-related ED visit involves multiple drugs, but these may be of a common type. For example, an ED visit involving illicit drugs alone often involves more than one illicit drug (e.g., cocaine and heroin).

ED visits in each of the three major categories—illicit drugs, alcohol, and nonmedical use of pharmaceuticals—are discussed in greater detail in separate sections of this publication.

ILLICIT DRUGS IN ED VISITS

To better understand the role of specific drugs and types of drugs in ED visits, this publication provides more detailed analysis of three drug categories: illicit drugs, alcohol, and nonmedical use of pharmaceuticals. This section focuses on ED visits involving illicit drugs.

For 2006, DAWN estimates that 958,164 (CI: 690,218 to 1,226,110) ED visits involved an illicit drug (Table 2). Thus, over half (55%) of all the drug misuse/abuse ED visits during the year involved illicit drugs, either alone or in combination with another drug type.

DAWN estimates that cocaine was involved in 548,608 (CI: 374,579 to 722,636) ED visits. In other words, close to one in three drug misuse/abuse ED visits (31%) involved cocaine.

Marijuana was involved in 290,563 (CI: 238,737 to 342,388) ED visits. Although it was associated with the second highest number of drug misuse/abuse ED visits for illicit drugs, marijuana was involved in approximately half as many ED visits as cocaine.

Heroin was involved in 189,780 (CI: 119,525 to 260,035) ED visits, or approximately 11% of drug misuse/abuse ED visits overall. This is likely an underestimate, though, because some portion of heroin use has been unavoidably classified as an "unspecified opiate." Heroin is an opiate, and some drug screens test for opiates only as a class. Nearly two thirds (64%) of reports of "opiates" submitted to DAWN for 2006 came from toxicology findings, so some unknown quantity of these may have been heroin. The number of drug misuse/abuse ED visits involving unspecified opiates is estimated at 55,674 (CI: 42,590 to 68,757) visits.

Stimulants, including amphetamines and methamphetamine, were involved in 107,575 (CI: 66,105 to 149,046) ED visits, about 6% of drug misuse/abuse ED visits. Amphetamines and methamphetamine are combined for this analysis because some drug screens test for amphetamines only as a class. Consequently, an amphetamine-positive result could indicate amphetamine or methamphetamine. Nearly all (99%) of the reports of amphetamines submitted to DAWN came simply as "amphetamine" and 65% of those were derived from toxicology findings.

Other illicit drugs appeared at much lower frequencies. For 2006, DAWN estimates:

• MDMA (Ecstasy) in 16,749 ED visits (CI: 11,470 to 22,028),
  
• PCP in 21,960 ED visits (CI: 5,518 to 38,403),
  
• Miscellaneous hallucinogens in 3,898 ED visits (CI: 2,591 to 5,205),
  
• LSD in 4,002 ED visits (CI: 1,945 to 6,059),
  
• GHB in 1,084 ED visits (CI: 517 to 1,652), and
  
• Ketamine in 270 ED visits (CI: 29 to 511).
  

When considered in relation to the population of the United States, ED visits associated with illicit drugs vary across major drugs of abuse (Figure 1):

• 182 visits per 100,000 population for cocaine,
  
• 96 visits per 100,000 population for marijuana,
  
• 63 visits per 100,000 population for heroin, and
  
• 36 visits per 100,000 population for stimulants.
  

Figure 1
Rates of ED visits involving selected illicit drugs: 2006

   D

SOURCE: Office of Applied Studies, SAMHSA, Drug Abuse Warning Network, 2006 (03/2008 update).

The rates of ED visits involving cocaine, marijuana, and heroin were higher for males than for females, after taking population size and the margin of error into account, but there was not a gender difference for stimulants (Figure 2). For cocaine the rates per 100,000 population were highest among patients aged 18 to 54, with lower rates for younger and older patients (Table 3, Figure 2). For heroin, the rates were highest for patients aged 21 to 54, while the rates for marijuana were highest for patients aged 18 to 24, and the rates for stimulants were highest for patients aged 18 to 44.

In terms of race/ethnicity, 44% of the visits related to any illicit drug use involved patients who were white. However, evaluating the relative frequencies across the race/ethnicity groups is impeded by missing data; race/ethnicity was unknown in 11% of illicit drug-related visits overall, and the percentage was higher for some drugs (e.g., 13% for heroin and 20% for stimulants).

Figure 2
Illicit drugs, ED visit rates by age and gender: 2006

   D

SOURCE: Office of Applied Studies, SAMHSA, Drug Abuse Warning Network, 2006 (03/2008 update).

ALCOHOL IN ED VISITS

Among all the drugs collected by DAWN, alcohol is unique. An ED visit related to alcohol use qualifies as a DAWN case under only two conditions: (1) the alcohol is found in combination with other drugs, regardless of patient age; or (2) the alcohol is found alone (i.e., not in combination with other drugs) in a patient under the age of 21. ED visits associated with alcohol use, particularly among underage patients, represent a significant public health and policy concern and are examined in detail in this section.

For 2006, DAWN estimates that 577,521 (CI: 501,994 to 653,048) ED visits involved either alcohol in combination with another drug (all ages) or alcohol alone for patients under the age of 21. This is about one third (33%) of all drug misuse/abuse ED visits (Table 4). Of all these ED visits involving alcohol, about one fifth (22%) involved patients under the age of 21 who used alcohol alone, that is, with no other drug.

Alcohol in combination with other drugs (Tables 5-6, Figure 3)

DAWN estimates 450,817 (CI: 383,818 to 517,816) ED visits related to use of alcohol in combination with another drug(s) in 2006. Alcohol in combination with other drugs is reported to DAWN regardless of the patient's age. These are the only alcohol reports received for patients aged 21 and older. It is these adult patients who account for nearly 9 out of 10 ED visits (87%) implicating alcohol with another drug (Table 5).

Males accounted for 63% of visits involving alcohol in combination with other drugs (Table 5). Taking population size and the margin of error into account, males had higher rates of such visits than females (Figure 3). There was little variation in rates across the age groups from ages 18 to 54. ED visit rates were lower for older and younger patients.

In terms of race/ethnicity, 57% of the visits with alcohol in combination involved patients who were white. Evaluating the relative frequencies across the race/ethnicity groups is impeded by missing data; in 10% of visits race/ethnicity was unknown.

Figure 3
Alcohol with other drugs, ED visit rates by age and gender: 2006

   D

SOURCE: Office of Applied Studies, SAMHSA, Drug Abuse Warning Network, 2006 (03/2008 update).

Alcohol was most frequently combined with (Table 6):

• Cocaine alone (101,588 visits),
  
• Marijuana alone (41,653 visits),
  
• Cocaine and marijuana (21,241 visits), and
  
• Heroin alone (14,958 visits)

Alcohol-related ED visits in patients under the age of 21 (Table 7)

For individuals under age 21, alcohol is an illegal drug, and ED visits related to both alcohol alone and alcohol in combination are reported to DAWN for this age group. Considering alcohol alone and alcohol in combination with other drugs, DAWN estimates:

• 76,760 (CI: 60,318 to 93,202) alcohol-related ED visits for patients aged 12 to 17, and
  
• 105,675 (CI: 82,757 to 128,593) alcohol-related ED visits for patients aged 18 to 20.
  

Nearly 7 in 10 (69%) of the alcohol-related ED visits for minors involved alcohol alone, a finding that is similar for patients aged 12 to 17 and patients aged 18 to 20 (Table 7).

ED visits for underage alcohol use (Tables 4, 8, Figure 4)

For 2006, DAWN estimates 126,704 (CI: 95,766 to 157,642) ED visits related to use of alcohol alone (i.e., not in combination with another drug) by patients who were younger than age 21 (Table 4). Nearly all (99%, or 125,888 visits) of those visits represent underage drinking that was not related to either a suicide attempt or a request for admission to detox or substance abuse treatment program (Table 8).

Taking population size into account, the rate of these alcohol-only ED visits for patients aged 18 to 20 (581 visits per 100,000 population) was 2.8 times that for patients aged 12 to 17 (204 per 100,000). Males and females had similar rates (Figure 4).

In terms of race/ethnicity, 58% of these alcohol-only visits involved patients who were white. Evaluating the relative frequencies of the race/ethnicity groups is impeded by missing data; in 16% of visits, race/ethnicity was unknown (Table 8).

Most (90%) of the alcohol-only ED visits resulted in patients' being treated and released, usually to home; another 6% were admitted to inpatient units (Table 8).

Figure 4
Alcohol only (age < 21), ED visit rates by age and gender: 2006

   D

SOURCE: Office of Applied Studies, SAMHSA, Drug Abuse Warning Network, 2006 (03/2008 update).

NONMEDICAL USE OF PHARMACEUTICALS

Use of illicit drugs is, by definition, substance abuse. For pharmaceuticals, however, distinguishing medical from nonmedical use is more complicated.¹³ In DAWN, "medical use" means taking a prescription or over-the-counter (OTC) pharmaceutical as prescribed or recommended, and "nonmedical use" is use that does not meet the definition of medical use.¹⁴ Thus, nonmedical use of pharmaceuticals includes taking more than the prescribed dose of a prescription pharmaceutical or more than the recommended dose of an OTC pharmaceutical or supplement; taking a pharmaceutical prescribed for another individual; deliberate poisoning with a pharmaceutical by another person; and documented misuse or abuse of a prescription or OTC pharmaceutical or dietary supplement. Nonmedical use of pharmaceuticals may involve pharmaceuticals alone or pharmaceuticals in combination with illicit drugs or alcohol.

A cautionary note: DAWN tries to capture only drugs that are related to the ED visit and actively discourages reporting of current medications that are unrelated to the visit. It is important to understand, however, that it is not possible, given the limitations of medical record documentation, to eliminate completely the reporting of current medications, and this should be considered when one interprets these findings. Also, it is not possible to know, based on the documentation available in ED medical records, the extent to which pharmaceuticals came from legitimate prescriptions versus other sources.

Nonmedical use of pharmaceuticals (Tables 9-10, Figure 5)

For 2006, DAWN estimates that 741,425 (CI: 674,198 to 808,652) ED visits involved nonmedical use of prescription or OTC pharmaceuticals or dietary supplements (Table 9). The majority of these visits (54%) involved multiple drugs (Table 10):

• Approximately one fifth (19%) of all nonmedical-use ED visits involved alcohol,
  
• One fifth (21%) involved pharmaceuticals in combination with an illicit drug, and
  
• About 6% involved pharmaceuticals in combination with both alcohol and an illicit drug.
  

Central nervous system (CNS) agents (50% of nonmedical-use visits) and psychotherapeutic agents (44%) were the most frequent drugs reported in the nonmedical-use category of ED visits (Table 9). Respiratory agents (4%), cardiovascular agents (5%), and all other types of pharmaceuticals were much less frequent.

Among the CNS agents, the most frequent drugs were opiate/opioid analgesics (33% of nonmedical-use visits), including single-ingredient formulations (e.g., oxycodone) and combination forms (e.g., hydrocodone with acetaminophen). Methadone, together with single-ingredient and combination forms of oxycodone and hydrocodone, were the most frequent opioids. Once the margin of error is taken into account, these three opioids appeared in similar numbers of visits:

• Hydrocodone/combinations in 57,550 ED visits (CI: 43,701 to 71,398),
  
• Oxycodone/combinations in 64,888 ED visits (CI: 49,746 to 80,030), and
  
• Methadone in 45,130 ED visits (CI: 35,870 to 54,389).
  

Note that ED records frequently do not distinguish methadone used for treatment of opiate addiction from the methadone in pill form that is prescribed for pain. In fact, methadone may be one of the most ambiguous drugs to categorize in DAWN. When a patient on opioid replacement therapy presents to an ED, methadone may be routinely documented in the medical record, but without sufficient information to distinguish whether the methadone was related to the ED visit.

The opioids were followed in frequency by the nonopioid analgesics containing acetaminophen (6% of nonmedical-use visits), muscle relaxants (5%), anticonvulsants (4%), and nonsteroidal anti-inflammatory agents (NSAIDs, 4%). DAWN estimates 44,314 (CI: 36,482 to 52,146) nonmedical-use visits involving nonopioid acetaminophen products. The most frequent muscle relaxant in nonmedical-use visits was carisoprodol, which was involved in 24,505 (CI: 13,352 to 35,658), or 3%, of nonmedical-use ED visits in 2006.

Among the psychotherapeutic agents, the anxiolytics (anti-anxiety agents), sedatives, and hypnotics were the most frequent, occurring in 233,875 (CI: 207,294 to 260,457) or about a third (32%) of visits associated with nonmedical use of pharmaceuticals. This category of pharmaceuticals includes barbiturates and benzodiazepines. ED visits involving benzodiazepines clearly outnumber those involving any of the other types of psychotherapeutic agents. DAWN estimates that 195,625 (CI: 167,789 to 223,461) ED visits associated with nonmedical use of pharmaceuticals involved benzodiazepines in 2006.

According to DAWN, the most frequently named benzodiazepines were alprazolam in 65,236 (CI: 50,911 to 79,561) ED visits and clonazepam in 33,557 (CI: 25,580 to 41,534) ED visits. Benzodiazepines without a specific ingredient named appeared in similar numbers: 58,348 (CI: 33,575 to 83,120) ED visits. Benzodiazepines occurring less frequently but still in substantial numbers included lorazepam in 23,720 (CI: 19,079 to 28,361) ED visits and diazepam in 19,936 (CI: 16,108 to 23,764) ED visits.

Among the other anxiolytics, sedatives, and hypnotics, the following drugs appeared in similar numbers of nonmedical-use ED visits:

• Zolpidem in 17,257 ED visits (CI: 11,633 to 22,882),
  
• Barbiturates, which are primarily unnamed, in 10,991 ED visits (CI: 7,068 to 14,913), and
  
• Diphenhydramine¹⁵ in 12,291 ED visits (CI: 9,643 to 14,939).
  

For the ED visits associated with nonmedical use of pharmaceuticals, other psychotherapeutic agents of interest include antidepressants and antipsychotics. DAWN estimates:

• Antidepressants in 79,682 ED visits (CI: 66,026 to 93,338), and
  
• Antipsychotics, such as quetiapine, in 44,733 ED visits (CI: 37,080 to 52,387).
  

Methylphenidate, a CNS stimulant that has captured much attention, occurs much less frequently. DAWN estimates that 2,192 (CI: 980 to 3,404) nonmedical-use ED visits involved methylphenidate.

Taking population size and the margin of error into account, visits for nonmedical use of pharmaceuticals did not differ between females (253 visits per 100,000 population) and males (237 visits per 100,000 population) (Figure 5). In terms of age, visit rates were highest for patients aged 18 to 44 and were lowest for patients aged 11 and younger.

In terms of race and ethnicity, 69% of visits related to nonmedical use of pharmaceuticals involved patients who were white (Table 10). Evaluating the relative frequencies of the race/ethnicity groups is impeded by missing data; in 12% of visits, race/ethnicity was unknown.

Patients were treated and released in about half (58%) of ED visits associated with nonmedical use of pharmaceuticals, with most discharged home (91%) and only 5% referred to detox or substance abuse treatment. In one third (30%) of all nonmedical-use visits, patients were admitted to inpatient hospital units (Table 10). Of those admitted to the hospital, about one third (29%) were sent to a critical care unit, about 22% to a psychiatric unit, and about half (47%) to other inpatient units. About 7% of ED visits for nonmedical use of pharmaceuticals resulted in transfers to another health care facility.

Figure 5
Nonmedical use of pharmaceuticals, ED visit rates by age and gender: 2006

   D

SOURCE: Office of Applied Studies, SAMHSA, Drug Abuse Warning Network, 2006 (03/2008 update).

COMPARISONS OF ED VISITS: 2004, 2005, AND 2006

This chapter presents the comparison of the estimates of ED visits for 2006 to prior years. Differences between years are presented in terms of the percent increase or decrease in visits in 2006 compared to the estimates for 2004 or 2005. Only statistically significant changes are discussed and are shown as changes in the tables. However, with only three years' estimates to compare, caution is urged in interpreting these significant differences as trends.

Drug misuse and abuse in ED visits (Table 11)

In 2006, hospitals in the United States delivered a total of 113 million ED visits, an increase of 3.9% over 2004. The population of the United States increased 2.9%, from 294 million to 302 million, over the same period.

According to DAWN, the number of ED visits attributable to drug misuse and abuse was stable from 2004 to 2006 (Table 11).¹⁶ That is, the apparent difference is within the margin of error. Across the different types of drug involvement, changes were detected for visits involving:

• pharmaceuticals alone (i.e., with no other type of drug), which increased 44% from 2004 to 2006;
  
• pharmaceuticals used in combination with illicit drug(s), which increased 36% from 2004 to 2006; and
  
• pharmaceuticals used in combination with alcohol, which increased 22% from 2005 to 2006.
  

Regarding the significant increases detected, it is worthwhile to consider that the number of pharmaceuticals dispensed for legitimate therapeutic uses may be increasing over time, and DAWN estimates are not adjusted to take this into account. Nor do DAWN estimates take into account the increases in the population or in ED use between 2004 and 2006.

Illicit drugs in ED visits (Table 12)

No changes from 2004 to 2006, nor from 2005 to 2006, were detected for ED visits involving major illicit drugs: cocaine, marijuana, heroin, and stimulants (Table 12).

Alcohol in ED visits (Tables 13-14)

No significant changes in alcohol-related ED visits occurred during the period from 2004 to 2006, nor for the period 2005 to 2006 (Tables 13-14). These findings were consistent for all the alcohol-related ED visits: alcohol overall, alcohol in combination with other drugs, and alcohol alone in underage patients.

Nonmedical use of pharmaceuticals (Table 15)

ED visits related to nonmedical use of pharmaceuticals increased 38% in the period from 2004 to 2006 (Table 15). Among the drugs most frequently implicated in nonmedical use, the following changes from 2004 to 2006 are notable:

• Psychotherapeutic and CNS agents increased 31% and 32%, respectively.
  
• Anxiolytics, sedatives, and hypnotics increased 32% overall, with benzodiazepines increasing 36%.
  
• Opiate/opioid analgesics increased 43%, with hydrocodone/combinations increasing 44%, oxycodone/combinations increasing 56%, and unspecified opiates increasing 60%.
  

For the period 2005 to 2006, no change was detected in the overall number of ED visits related to nonmedical use of pharmaceuticals nor were changes noted for the substances most frequently implicated in nonmedical-use visits.

SPECIAL TYPES OF DRUG-RELATED ED VISITS

This chapter profiles two special types of drug-related ED visits captured by DAWN: drug-related suicide attempts and seeking detox cases. These are analyzed by DAWN as separate and distinct classes of drug misuse or abuse.

Suicide attempts (Tables 16-17, Figure 6)

DAWN estimates 182,805 (CI: 154,185 to 211,424) ED visits for drug-related suicide attempts in 2006 (Table 16). Although DAWN includes only suicide attempts that involve drugs, these attempts are not limited to drug overdoses. If there is drug involvement in a suicide attempt by other means (e.g., by gun), the case is included as drug related. However, suicide attempts not involving drugs at all (e.g., by gun alone) are excluded. Also excluded are suicide-related behaviors documented as something other than actual attempts (e.g., suicidal ideation, suicidal gesture, or suicidal thoughts).

Nearly two thirds (65%) of ED visits for drug-related suicide attempts involved multiple drugs (Table 17). Alcohol, either in combination with other drugs or alcohol alone in patients under age 21, was the most frequently implicated drug and was involved in one third (30%) of the ED visits for drug-related suicide attempts. Since DAWN excludes visits for adults when alcohol is the only drug, the role of alcohol in suicide attempts is probably larger. Illicit drugs were involved in just under one quarter (23%) of the ED visits for drug-related suicide attempts. The most frequently reported illicit drugs were cocaine (15% of visits) and marijuana (8% of visits), but the margins of error for the illicit drugs are quite large and the numbers are relatively small when compared with the pharmaceuticals.

Pharmaceuticals were involved in the majority (92%) of ED visits for drug-related suicide attempts, and it is not possible, based on ED medical record documentation, to measure the extent to which these pharmaceuticals may have been prescribed to the patient for a preexisting condition. More than half (58%) of ED visits for drug-related suicide attempts involved psychotherapeutic agents, and 45% involved central nervous system (CNS) agents. The most commonly used psychotherapeutic agents were benzodiazepines (48% of suicide-attempt visits involving psychotherapeutics) and antidepressants (35%). The CNS agents were primarily analgesics (pain relievers) and included both prescription and over-the-counter (OTC) formulations. DAWN estimates that the most commonly used pain relievers were opiates/opioids and acetaminophen/combinations, which were each present in approximately a third (33% and 31%, respectively) of suicide-attempt visits involving CNS agents, followed by nonsteroidal anti-inflammatory agents (NSAIDs, such as ibuprofen and naproxen, 19%) and salicylates/combinations (aspirins, 7%).

About half (55%) of the suicide attempts were admitted for inpatient hospital care. A fifth (21%) were admitted to an ICU/critical care unit; others were admitted to psychiatric units (18%) or other inpatient units (16%). Another 25% were transferred to another health care facility; only 14% were discharged home. Very few died in the ED. However, DAWN does not record deaths for patients who died before arriving at the ED or patients who died after admission to inpatient units of the hospital.

After accounting for population size and the margin of error, the rate of drug-related suicide visits for females (73 visits per 100,000 population) was higher than that for males (48 per 100,000) (Figure 6). In general, the rates for patients aged 18 to 54 exceeded the rates for younger and older age groups. The rate for patients aged 12 to 17 (81 visits per 100,000) exceeded the rates for patients aged 55 and over.

Figure 6
Suicide attempts, ED visit rates by age and gender: 2006

   D

SOURCE: Office of Applied Studies, SAMHSA, Drug Abuse Warning Network, 2006 (03/2008 update).

In terms of race/ethnicity, 63% of the suicide attempts involved patients who were white. Evaluating the relative frequencies of the race/ethnicity groups is impeded by missing data; in 10% of visits, race/ethnicity was unknown.

Suicide attempt ED visits: 2004, 2005, and 2006 (Table 18)

Overall, there was no significant change in ED visits for drug-related suicide attempts from 2004 to 2006, but an increase was detected from 2005 to 2006 (Table 18). Increases were evident from 2004 to 2006 for some of the drugs frequently involved in suicide attempts, such as benzodiazepines, which increased 36%, and opiate/opioid analgesics, which increased 44%.

Seeking detox (Tables 19-20, Figure 7)

DAWN estimates 118,355 (CI: 90,171 to 146,540) drug-related ED visits for patients seeking detoxification or substance abuse treatment services during 2006. These "seeking detox" visits tend to be concentrated in hospitals with administrative practices that require medical clearance in the ED for admission to detox or substance abuse treatment units within the hospital. Therefore, this estimate does not encompass the full extent of the demand for these services.

Nearly two thirds (65%) of the seeking detox ED visits involved multiple drugs, and more than one third (40%) of all seeking detox ED visits involved alcohol (Table 19). However, the role of alcohol may be underrepresented here, because for adults aged 21 and older this includes only alcohol in combination with other drugs. Among the illicit drugs, cocaine (49% of visits) and heroin (29% of visits) occurred most frequently, followed by marijuana (19% of visits) and amphetamine or methamphetamine stimulants (7% of visits). Among the pharmaceuticals, psychotherapeutic agents, which were primarily benzodiazepines (13%), and CNS agents, which were primarily opiate/opioid analgesics (26%), were notable. Among the opiates/opioids, hydrocodone/combinations and oxycodone/combinations were most frequent.

Among the seeking detox ED visits, nearly 7 out of 10 (69%) received some type of follow-up care, either inpatient admission, referral elsewhere for detox or substance abuse treatment services, or transfer to another health care facility (Table 20). However, about one quarter (27%) of seeking detox cases may not have received the care they sought because they were discharged to home.

Taking population size and the margin of error into account, the rates of seeking detox visits were similar across all age groups in the 18 to 44 range. The rate of seeking detox visits for males (51 per 100,000 population) was higher than that for females (27 per 100,000 population).

In terms of race/ethnicity, the majority (59%) of seeking detox visits involved patients who were white. Evaluating the relative frequencies of the race/ethnicity groups is impeded by missing data; in 9% of visits race/ethnicity was unknown.

Figure 7
Seeking detox, ED visit rates by age and gender: 2006

   D

SOURCE: Office of Applied Studies, SAMHSA, Drug Abuse Warning Network, 2006 (03/2008 update).

Seeking detox ED visits: 2004, 2005, and 2006 (Table 21)

No significant changes in ED visits from 2004 to 2006, or from 2005 to 2006, were detected for seeking detox ED visits overall, or for alcohol or the illicit drugs involved in these visits (Table 21).

APPENDIX A

MULTUM LEXICON
END-USER LICENSE AGREEMENT

1. Introduction

1. This License Agreement (the "License") applies to the Multum Lexicon database (the "Database"). This License does not apply to any other products or services of Cerner Multum, Inc. ("Multum"). A "work based on the Database" means either the Database or any derivative work under copyright law; i.e., a work containing the Database or a substantial portion of it, either verbatim or with modifications. A translation of the Database is included without limitation in the term "modification". Each end-user/licensee is addressed herein as "you".
   
   
2. Your use of the Database acknowledges acceptance of these restrictions, disclaimers, and limitations. you expressly acknowledge and agree that Multum is not responsible for the results of your decisions resulting from the use of the Database, including, but not limited to, your choosing to seek or not to seek professional medical care, or from choosing or not choosing specific treatment based on the Database.
   
   
3. Every effort has been made to ensure that the information provided in the Database is accurate, up-to-date, and complete, but no guarantee is made to that effect. In addition, the drug information contained herein may be time sensitive.
   
   
4. Multum does not assume any responsibility for any aspect of healthcare administered or not administered with the aid of information the Database provides.
   
   

2. Terms and Conditions for Copying, Distribution and Modification

1. You may copy and distribute verbatim copies of the Database as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Database a copy of this License (the readme.txt file) along with the Database and anything else that is part of the package, which should be identified.
   
   
2. You may modify your copy or copies of the Database or any portion of it to form a derivative work, and copy and distribute such modifications or work under the terms of Section 2.A. above, provided that you also meet all of these conditions:
   

1. You must cause the modified files to carry prominent notices stating that they are derived from the Multum Lexicon database from Cerner Multum, Inc. and that you changed the files and the date of any change(s).
   
   
2. If you incorporate modified files into a computer program, you must cause it, when started running for interactive use in the most ordinary way, to print or display an announcement including an appropriate copyright notice, a notice that you have modified the Multum Lexicon database from Cerner Multum, Inc., and a notice that there is no warranty (or that you provide the warranty) and telling the user how to view a copy of this License.
   

3. It is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Database.
   
   
4. You may copy and distribute the Database (or a work based on it, under Section 2.B.) in an encoded form under the terms of Sections 2.A. and 2.B. above provided that you also do one of the following:
   

1. Accompany it with the complete corresponding machine-readable plain text, which must be distributed under the terms of Sections 2.A and 2.B. above on a medium customarily used for software interchange; or,
   
   
2. Accompany it with a written offer to give any third party, for no charge, a complete machine-readable copy of the Database (and the entirety of your derivative work based on it, under Section 2.B.), to be distributed under the terms of Sections 2.A. and 2.B. above on a medium customarily used for software interchange.
   

5. You may not copy, modify, sublicense, or distribute the Database except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense or distribute the Database will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance.
   
   
6. You are not required to accept this License. However, nothing else grants you permission to copy, modify or distribute the Database or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by copying, modifying or distributing the Database (or any work based on the Database), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Database or works based on it.
   
   
7. Each time you redistribute the Database (or any work based on the Database), the recipient automatically receives a license from Multum to copy, distribute or modify the Database subject to these terms and conditions. you may not impose any further restrictions on the recipients' exercise of the rights granted herein. you are not responsible for enforcing compliance by third parties to this License.
   

3. Disclaimer of Warranties; Limitation of Damages

1. BECAUSE THE DATABASE IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE PROGRAM OR DATA, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING. MULTUM AND/OR OTHER PARTIES PROVIDE THE DATABASE "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED, STATUTORY OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE DATABASE IS WITH YOU. SHOULD THE DATABASE PROVE DEFECTIVE, INCOMPLETE, OR INACCURATE, YOU ASSUME THE RESPONSIBILITY AND COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
   
   
2. IN NO EVENT (UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING) WILL MULTUM, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE DATABASE AS PERMITTED ABOVE, BE LIABLE FOR ANY SPECIAL, INCIDENTAL, CONSEQUENTIAL, OR INDIRECT DAMAGES, INCLUDING DAMAGES FOR LOSS OF PROFITS, LOSS OF BUSINESS, OR DOWN TIME, EVEN IF MULTUM OR ANY OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
   
   
3. IN ADDITION, WITHOUT LIMITING THE FOREGOING, THE DATABASE HAS BEEN DESIGNED FOR USE IN THE UNITED STATES ONLY AND COVERS THE DRUG PRODUCTS USED IN PRACTICE IN THE UNITED STATES. MULTUM PROVIDES NO CLINICAL INFORMATION OR CHECKS FOR DRUGS NOT AVAILABLE FOR SALE IN THE UNITED STATES AND CLINICAL PRACTICE PATTERNS OUTSIDE THE UNITED STATES MAY DIFFER SUBSTANTIALLY FROM INFORMATION SUPPLIED BY THE DATABASE. MULTUM DOES NOT WARRANT THAT USES OUTSIDE THE UNITED STATES ARE APPROPRIATE.
   
   
4. You acknowledge that updates to the Database are at the sole discretion of Multum. Multum makes no representations or warranties whatsoever, express or implied, with respect to the compatibility of the Database, or future releases thereof, with any computer hardware or software, nor does Multum represent or warrant the continuity of the features or the facilities provided by or through the Database as between various releases thereof.
   
   
5. Any warranties expressly provided herein do not apply if: (i) the end-user alters, mishandles or improperly uses, stores or installs all, or any part, of the Database, (ii) the end-user uses, stores or installs the Database on a computer system which fails to meet the specifications provided by Multum, or (iii) the breach of warranty arises out of or in connection with acts or omissions of persons other than Multum.
   

4. Assumption of Risk, Disclaimer of Liability, Indemnity

1. THE END-USER ASSUMES ALL RISK FOR SELECTION AND USE OF THE DATABASE AND CONTENT PROVIDED THEREON. MULTUM SHALL NOT BE RESPONSIBLE FOR ANY ERRORS, MISSTATEMENTS, INACCURACIES OR OMISSIONS REGARDING CONTENT DELIVERED THROUGH THE DATABASE OR ANY DELAYS IN OR INTERRUPTIONS OF SUCH DELIVERY.
   
   
2. THE END-USER ACKNOWLEDGES THAT MULTUM: (A) HAS NO CONTROL OF OR RESPONSIBILITY FOR THE ENDUSER'S USE OF THE DATABASE OR CONTENT PROVIDED THEREON, (B) HAS NO KNOWLEDGE OF THE SPECIFIC OR UNIQUE CIRCUMSTANCES UNDER WHICH THE DATABASE OR CONTENT PROVIDED THEREON MAY BE USED BY THE END-USER, (C) UNDERTAKES NO OBLIGATION TO SUPPLEMENT OR UPDATE CONTENT OF THE DATABASE, AND (D) HAS NO LIABILITY TO ANY PERSON FOR ANY DATA OR INFORMATION INPUT ON THE DATABASE BY PERSONS OTHER THAN MULTUM.
   
   
3. MULTUM SHALL NOT BE LIABLE TO ANY PERSON (INCLUDING BUT NOT LIMITED TO THE END-USER AND PERSONS TREATED BY OR ON BEHALF OF THE END-USER) FOR, AND THE END-USER AGREES TO INDEMNIFY AND HOLD MULTUM HARMLESS FROM ANY CLAIMS, LAWSUITS, PROCEEDINGS, COSTS, ATTORNEYS' FEES, DAMAGES OR OTHER LOSSES (COLLECTIVELY, "LOSSES") ARISING OUT OF OR RELATING TO (A) THE END-USER'S USE OF THE DATABASE OR CONTENT PROVIDED THEREON OR ANY EQUIPMENT FURNISHED IN CONNECTION THEREWITH AND (B) ANY DATA OR INFORMATION INPUT ON THE DATABASE BY END-USER, IN ALL CASES INCLUDING BUT NOT LIMITED TO LOSSES FOR TORT, PERSONAL INJURY, MEDICAL MALPRACTICE OR PRODUCT LIABILITY.
   

5. Miscellaneous

1. You warrant that you have authority within the organization you identified during registration for the Database to enter into license agreements with other organizations including Multum.
   
   
2. You agree that Multum may identify you and/or your organization by name as a "licensee", "licensed user", or "licensing organization" of the Database or a "client" of Multum in Multum's external market communications. you also agree that Multum may issue, if it desires, a press release stating that you and/or your organization have licensed the Database.
   
   
3. If conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other obligations, then as a consequence you may not distribute the Database at all.
   
   
4. If any portion of this License is held invalid or unenforceable under any particular circumstance, the balance of this License is intended to apply and the License as a whole is intended to apply in other circumstances.
   
   
5. If the distribution and/or use of the Database is or becomes restricted in certain countries either by patents or by copyrighted interfaces, Multum may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License.
   

Multum Lexicon © 2007

Cerner Multum, Inc.
Colorado Center Tower One
2000 South Colorado Boulevard
Suite 11000
Denver, Colorado 80222
Document revised November 30, 2000

License downloaded 9/1/2006 from: http://www.multum.com/license.htm
Lexicon Registration

APPENDIX B

GLOSSARY OF TERMS

This glossary defines terms used in data collection activities, analyses, and publications associated with the emergency department (ED) component of the Drug Abuse Warning Network (DAWN).

Accidental ingestion: This category of drug-related ED visits includes those involving the accidental use of a drug, for example, childhood drug poisonings and individuals who take the wrong medication by mistake.

Adverse reaction: This category of drug-related ED visits represents the consequences of using a prescription or over-the-counter (OTC) pharmaceutical for therapeutic purposes and includes visits related to adverse drug reactions, side effects, drug-drug interactions, and drug-alcohol interactions. Adverse reactions that involve a pharmaceutical with an illicit drug are exceptions that are excluded from this category.

Alcohol only (age less than 21): This category of drug-related ED visits includes those in which alcohol was the only drug involved and the patient was aged less than 21. Although alcohol is an illegal drug for minors, combining these cases with other cases involving illicit drugs tends to mask rather than highlight their importance for prevention and treatment efforts. Most instances of alcohol as the only drug in patients under age 21 are classified in the alcohol only (age < 21) case type. However, some are classified as suicide-attempt or seeking detox, case types that precede alcohol only (age < 21) in sequence.

Case description: A description of how the drug(s) was related to the patient's ED visit. The case description, in conjunction with other documentation in the ED medical record, is used to determine if the ED visit is reportable to DAWN. It is copied verbatim from the patient's chart when possible.

Case type: See Type of case.

Case type other: See Drug misuse and abuse.

Confidence interval (CI): An interval estimate, that is, a range of values around a point estimate that takes sampling error into account. Ninety-five percent is an accepted standard of confidence. Technically, a 95% CI means that, if repeated samples were drawn from the same population of hospitals using the same sampling and data collection procedures, the true population value would fall within the confidence interval 95% of the time. Practically, a 95% CI summarizes both the estimate and its margin of error in a straightforward way with a reasonable degree of confidence. Calculation of 95% CIs is discussed in Appendix C.

Diagnosis: The condition(s) for which the patient was treated as determined by the clinician after study.

Disposition: The location or facility to which an ED patient was referred, transferred, or released.

Treated and released includes three categories:

• Discharged home—"Home" is used as a broad category to mean discharged to the patient's residence. Home is generally used for people who live locally; however, for students at nearby universities, home means their university; for travelers who get sick on the road, it may mean their hotel or wherever they are staying, and so forth.
• Released to police/jail
• Referred to detox/treatment—The chart indicates that the patient was referred to a substance abuse treatment or detox program, facility, or provider.
  

Admitted to this hospital includes five categories of inpatient units:

• ICU/critical care
• Surgery
• Chemical dependency/detox
• Psychiatric unit
• Other inpatient unit—The inpatient unit was not specified or does not match one of the preceding units.
  

Other disposition includes five categories:

• Transferred—The patient was transferred to another health care facility.
• Left against medical advice—The patient left the treatment setting without a physician's approval.
• Died—The patient died after arriving in the ED but before being discharged, admitted, or transferred.
• Other—The discharge status is documented in the chart but does not fit into any of the preceding categories.
• Not documented—The discharge status was not documented in the medical chart.
  

Drug: A substance that was recorded in a DAWN case report. Substances accepted by DAWN include alcohol, illicit drugs, prescription and over-the-counter pharmaceuticals, dietary supplements, and nonpharmaceutical inhalants. Multiple substances ("drugs") can be reported for each DAWN case. Therefore, the total number of drugs exceeds the total number of DAWN cases reported. (See also single-drug case.)

Drug category: A generic grouping of related pharmaceuticals or other substances reported to DAWN, based on the classification of Multum Information Services. Multum Information Services is a subsidiary of the Cerner Corporation and a developer of clinical drug information systems and a drug knowledge base. More information is available at http://www.multum.com. In general, the Multum categories follow the therapeutic uses for prescription and over-the-counter pharmaceuticals.

Additional clarification is provided for the following drug categories:

• Alcohol alone—DAWN collects data on alcohol when used alone only if the patient is under age 21.
• Alcohol in combination—The category for alcohol present with another reportable substance. DAWN does not gather data on alcohol used alone if the patient is aged 21 years or older. For patients 21 and older, alcohol must be used with another substance to be reported to DAWN. Alcohol in combination is reportable for all ages.
• Amphetamines—This class of substances has been extracted from the category of central nervous system (CNS) stimulants because of its importance as a major substance of abuse. For purposes of classification, "amphetamines" (plural) includes a class of compounds derived from or related to the drug amphetamine. Although some "designer" drugs fall into the class of amphetamines, we choose to report some of them individually as major substances of abuse (e.g., methamphetamine). This category does not include other CNS stimulants, such as caffeine or methylphenidate.
• Combinations not tabulated above (NTA)—This category includes combinations composed of two or more major substances of abuse that are mixed and taken together. For example, "speedball," which usually refers to the combination of heroin and cocaine taken at once, would be classified as a combination NTA, whereas heroin and cocaine used separately would be classified separately in the categories heroin and cocaine. Combinations consisting of a major substance of abuse and another substance are classified in the category of the major substance (e.g., heroin with scopolamine is classified as heroin).
• Inhalants—This category includes anesthetic gases and psychoactive nonpharmaceutical substances for which the documented route of administration was inhaling, sniffing, or snorting. Psychoactive nonpharmaceuticals fall into one of the following three categories: (1) volatile solvents—adhesives (model airplane glue, rubber cement, household glue), aerosols (spray paint, hairspray, air freshener, deodorant, fabric protector), solvents and gases (nail polish remover, paint thinner, correction fluid and thinner, toxic markers, pure toluene, cigar lighter fluid, gasoline, carburetor cleaner, octane booster), cleaning agents (dry cleaning fluid, spot remover, degreaser), food products (vegetable cooking spray, dessert topping spray such as whipped cream, whippets), and gases (butane, propane, helium); (2) nitrites—amyl nitrites ("poppers," "snappers") and butyl nitrites ("rush," "locker room," "bolt," "climax," "video head cleaner"); or (3) chlorofluorohydrocarbons (freons). Anesthetic gases (e.g., nitrous oxide, ether, chloroform) are presumed to have been inhaled.
• Stimulants—This category includes amphetamines and methamphetamine. Since some drug screens test for amphetamines only as a class, an amphetamine-positive result could indicate amphetamine or methamphetamine. For this reason, amphetamines and methamphetamine are combined for analysis into the category "stimulants." This category does not include other CNS stimulants, such as caffeine or methylphenidate.
  

Drug misuse and abuse: A group of ED visits defined broadly to include all visits associated with illicit drugs, alcohol use in combination with illicit drugs or alcohol alone among those younger than 21 years, and nonmedical use of pharmaceuticals. Nonmedical use of pharmaceuticals includes prescription and OTC pharmaceuticals in ED visits that are of the following case types:

• Overmedication—This category was designed to capture nonmedical use, overuse, and misuse of prescription and OTC medications that are not documented as drug abuse in the medical chart.
• Malicious poisoning—This category was designed to capture cases of drug use in which the patient was administered a drug by another person for a malicious purpose. Drug-facilitated sexual assault is one type of malicious poisoning, but other types of malicious poisonings, such as product tampering, would be classified in this category as well.
• Case type Other—This category includes all drug-related ED visits that could not be assigned to any of the other seven types. By design, most cases of documented drug abuse will fall into this category.
  

Drug-related ED visit: Any ED visit related to recent drug use. This is the definition of a DAWN case effective January 1, 2003. To be a DAWN case, a drug needs only to be implicated in the visit; the drug does not have to have caused the visit. One patient may make repeated visits to an ED or to several EDs, thus producing a number of visits. The number of unique patients involved in the reported drug-related ED visits cannot be estimated, because no direct patient identifiers are collected by DAWN.

Estimate: A statistical estimate is the value of a parameter (such as the number of drug-related ED visits) for the universe that is derived by applying sampling weights to data from a sample.

Hospital emergency department (ED): Only hospitals that meet eligibility criteria for DAWN are recruited to participate. To be eligible, hospitals must be non-Federal, short-stay, general medical and surgical facilities that operate one or more EDs 24 hours a day, 7 days a week, and be located in the United States. Specialty hospitals, hospital units of institutions, long-term care facilities, pediatric hospitals, hospitals operating part-time EDs, and hospitals operated by the Veterans Health Administration and the Indian Health Service are excluded. The universe of EDs is identified from the American Hospital Association's Annual Survey Database. (See also Universe.)

Malicious poisoning: See Drug misuse and abuse.

Metropolitan area: An area comprising a relatively large core city or cities and the adjacent geographic areas. Conceptually, these areas are integrated economic and social units with a large population nucleus. This DAWN publication utilizes areas defined by the Office of Management and Budget (OMB) in June 30, 2003, based on population data from the 2000 decennial Census.

Not otherwise specified (NOS): The catch-all category for substances that are not specifically named. Terms are classified into an NOS category only when assignment to a more specific category is not possible based on information in the source documentation (ED patient charts).

Not tabulated above (NTA): The designation used when categories are not presented in complete detail; smaller units are combined in the NTA category.

Overmedication: See Drug misuse and abuse.

p-value: A measure of the probability (p) that the difference between two estimates could have occurred by chance, if the estimates being compared were really the same. The larger the p-value, the more likely the difference could have occurred by chance. For example, if the difference between two DAWN estimates has a p-value of 0.01, it means that there is a 1% probability that the difference observed could be due to chance alone.

Population: See Universe.

Precision: The extent to which an estimate agrees with its mean value in repeated sampling. The precision of an estimate is measured inversely by its standard error (SE) or relative standard error (RSE). In DAWN publications, estimates with RSEs greater than 50% are regarded as too imprecise to be published. ED table cells where such estimates would have appeared contain the symbol "…" (3 dots). (See also Relative standard error.)

Race/ethnicity: According to the standard protocol issued by OMB in 1997, the race/ethnicity categories on the DAWN data collection forms are as follows:

• White—A person having origins in any of the original peoples of Europe, the Middle East, or North Africa.
• Black or African American—A person having origins in any of the black racial groups of Africa.
• Hispanic or Latino—A person of Cuban, Mexican, Puerto Rican, South or Central American, or other Spanish culture or origin, regardless of race.
• Asian—A person having origins in any of the original peoples of the Far East, Southeast Asia, or the Indian subcontinent including, for example, Cambodia, China, India, Japan, Korea, Malaysia, Pakistan, the Philippine Islands, Thailand, and Vietnam.
• American Indian or Alaska Native—A person having origins in any of the original peoples of North and South America (including Central America), and who maintains tribal affiliation or community attachment.
• Native Hawaiian or Other Pacific Islander—A person having origins in any of the original peoples of Hawaii, Guam, Samoa, or other Pacific Islands.
• Not documented—Used when documentation of race is not available from source records.
  

Despite the detail allowed by these categories, race and ethnicity are often not documented with this level of specificity in patient/decedent records. As a result, categories used to tabulate race and ethnicity data in the publications are:

• White—Anyone meeting the definition of white (above). Those who are identified as white and Hispanic are classified as Hispanic.
• Black—Anyone meeting the definition of black or African American (above). Those who are identified as black or African American and Hispanic are classified as Hispanic.
• Hispanic—Anyone whose ethnicity is Hispanic or Latino (above) is placed in the category Hispanic, regardless of race.
• Race/ethnicity NTA—This includes those categories that are too small to report independently, including the following: two or more races, American Indian or Alaska Native, Asian, and Native Hawaiian or Other Pacific Islander.
• Unknown—Race/ethnicity is unknown. Those who are identified only as Hispanic are classified as Hispanic.
  

Rate: A measure of the incidence of drug-related ED visits per 100,000 population. A rate can be calculated for the total population or for any subset defined by characteristics such as age and gender.

Relative standard error (RSE): A measure of an estimate's relative precision. The RSE of an estimate is equal to the estimate's standard error (SE) divided by the estimate itself. For example, an estimate of 2,000 cocaine visits with an SE of 200 visits has an RSE of 10%. The larger the RSE, the less precise the estimate. Estimates with an RSE of 50% or greater are not published by DAWN. (See also Precision and Standard error.)

Sampling: Sampling is the process of selecting a proper subset of elements from the full population so that the subset can be used to make inference to the population as a whole. A probability sample is one in which each element has a known and positive chance (probability) of selection. A simple random sample is one in which each member has the same chance of selection. In DAWN, a sample of hospitals is selected in order to make inference to all hospitals; DAWN uses simple random sampling within strata.

Sampling frame: A list of units from which the ED sample is drawn. All members of the sampling frame have a probability of being selected. A sampling frame is constructed such that there is no duplication and each unit is identifiable. Ideally, the sampling frame and the universe are the same. The sampling frame for the DAWN hospital ED sample is derived from the American Hospital Association (AHA) Annual Survey Database.

Sampling unit: A member of a sample selected from a sampling frame. For the DAWN sample, the units are hospitals, and data are collected for all drug-related ED visits at the responding hospitals selected for the sample.

Sampling weights: Numeric coefficients used to derive population estimates from a sample by adjusting for deviations from the original sample design due to unequal probability sampling, variable nonresponse, and other potential sources of bias.

Seeking detox: This category of drug-related ED visits captures patients seeking substance abuse treatment, drug rehabilitation, or medical clearance for admission to a drug treatment or detoxification unit. They are classified separately because they often reflect administrative practices that vary across hospitals and may vary over time within the same hospital. Seeking detox visits tend to be concentrated in those facilities that operate specialized inpatient units providing substance abuse treatment or detoxification services, and the largest numbers are found in facilities that require medical clearance for entry into such treatment to be granted in their EDs.

Single-drug case: A single-drug case is one in which only one drug was involved. Because multiple substances may be recorded for each DAWN case (see Drug), readers should be cautious in interpreting the relationship between a given drug and the number of associated visits or deaths. For example, if the source record for a patient/decedent documented marijuana use, this does not mean that marijuana was the only drug involved in the visit/death or that the marijuana caused the visit/death. One should always consider whether and how many other drugs were used in combination. Even then, attributing a causal relationship between the visit/death and a particular drug may not be possible. DAWN captures single-drug visits/deaths involving alcohol only if the patient /decedent was younger than age 21.

Standard error (SE): A measure of the sampling variability or precision of an estimate. The SE of an estimate is expressed in the same units as the estimate itself. For example, an estimate of 10,000 visits with an SE of 500 indicates that the SE is 500 visits.

Statistically significant: A difference between two estimates is said to be statistically significant if the value of the statistic used to test the difference is larger or smaller than would be expected by chance alone. For DAWN ED estimates, a difference is considered statistically significant if the p-value is less than 0.05. (See also p-value.)

Strata (plural), stratum (singular): Subgroups of a universe within which separate ED samples are drawn. Stratification is used to increase the precision of estimates for a given sample size, or, conversely, to reduce the sample size required to achieve the desired level of precision. The DAWN ED sample is stratified into metropolitan area cells plus an additional cell for the remainder of the United States. To ensure thorough coverage within metropolitan areas, the universe of hospitals in each is allocated into substrata identified by (a) two types of hospital ownership (public, private) and (b) up to four size categories (measured in terms of annual ED visits), creating up to eight substrata in each metropolitan area stratum. Hospitals in the stratum that covers the rest of the United States are stratified first by Census region, and then by state, type of ownership, and size (also measured in terms of ED visits). A systematic sample is selected from each of the geographic strata.

Suicide attempt: This category of drug-related ED visits captures suicide attempts (e.g., "attempted suicide," "tried to kill self") documented in the medical record in which a drug was involved. Suicidal gestures, thoughts, or ideation, including attempts to "harm" self, are assigned to another case type.

Type of case: A classification used to group similar DAWN cases from the diverse set of all drug-related ED visits. Each case is coded into one and only one category, the first that applies from the following hierarchy: suicide attempt, seeking detox, alcohol only (age < 21), adverse reaction, overmedication, malicious poisoning, accidental ingestion, and other. The rules for assignment of DAWN cases to types of cases are defined in the DAWN ED decision tree in Appendix C.

Universe: The entire set of units for which generalizations are drawn. The universe for the DAWN ED sample is all non-Federal, short-stay, general medical and surgical hospitals in the United States that operate one or more EDs 24 hours a day, 7 days a week. Specialty hospitals, hospital units of institutions, long-term care facilities, pediatric hospitals, hospitals operating part-time EDs, and hospitals operated by the Veterans Health Administration and the Indian Health Services are excluded. The universe of EDs is identified from the American Hospital Association's Annual Survey Database.

APPENDIX C

DAWN DATA COLLECTION AND STATISTICAL METHODS

Introduction

The Drug Abuse Warning Network (DAWN) is a public health surveillance system that has monitored drug-related emergency department (ED) visits to hospitals since the early 1970s. DAWN was initially established by the Drug Enforcement Administration. Then DAWN was transferred to the U.S. Department of Health and Human Services (USDHHS), where the National Institute on Drug Abuse (NIDA) conducted DAWN from 1980 to 1992. Since 1992, the Office of Applied Studies (OAS) of the Substance Abuse and Mental Health Services Administration (SAMHSA), USDHHS, has been responsible for DAWN operations and reporting.

Since its inception, DAWN has relied on data collected from a sample of hospitals. However, over the years, the exact survey methodology has been adjusted to improve the quality, reliability, and generalizability of the information produced by DAWN. When NIDA assumed responsibility for DAWN in 1980, implementation of a sample of hospitals to produce representative estimates for the Nation and for selected metropolitan areas became a priority. This sample, refreshed with annual maintenance, continued to support DAWN estimates for the coterminous United States and 21 metropolitan areas until 2002. By that time, major population shifts and changes in the hospital industry over the preceding two decades made apparent the need for a redesign of the sample of hospitals, which was undertaken as part of a wholesale redesign of most major features of DAWN.

Currently, the DAWN survey relies on a longitudinal probability sample of hospitals located throughout the United States, including Alaska and Hawaii. Hospitals eligible for selection into the DAWN sample must be non-Federal, short-stay, general surgical and medical hospitals located in the United States, with at least one 24-hour ED. This current approach was first implemented in the 2004 data collection year.

DAWN uses the data from the visits classified as DAWN cases in the selected hospitals to calculate various estimates of drug-related visits for the Nation as a whole, as well as for specific metropolitan areas. To calculate these estimates and measure their precision, the DAWN survey requires the application of sampling and weighting methodologies.

This appendix documents the data collection methods and the sampling, weighting, and variance estimation methodologies used to develop estimates for the DAWN data collected for 2004, 2005, and 2006.

Target population

The target population is drug-related ED visits in non-Federal, short-stay, general surgical and medical hospitals in the United States with at least one 24-hour ED.

Hospital sample frame

DAWN uses the American Hospital Association (AHA) Annual Survey Database as the basis for its sampling frame. The AHA maintains an updated national registry of U.S. hospitals that is estimated to have a coverage rate of 99%.¹⁷ A health care organization must meet several criteria to be classified as a hospital by the AHA. These include the provision of patient services, diagnostic or therapeutic, for general or specific medical conditions; licensed medical staff; and accreditation by organizations such as the Joint Commission on Accreditation of Health Care Organizations. A hospital is considered to be eligible for inclusion in the DAWN sampling frame if it is a non-Federal, short-stay, general surgical and medical hospital in the United States, with at least one 24-hour ED. Many DAWN hospitals operate multiple EDs.

Determination of DAWN eligibility

A hospital is considered ineligible if any one of the key criteria that define eligibility (non-Federal, short-stay, general surgical and medical hospitals located in the United States, with at least one 24-hour ED) is not met. Only those hospitals that meet all the criteria are considered eligible. For hospitals where critical eligibility data are missing from the AHA database, any one of the nonmissing criteria can render it ineligible. Otherwise, the hospital is considered to have unknown eligibility. For any hospital with unknown eligibility, other variables in the AHA Annual Survey Database are used to determine eligibility. If the hospital's eligibility remains unknown, additional data sources are consulted to determine eligibility.

DAWN data collection

DAWN ED data are collected through a retrospective review of ED medical records for patients treated in the ED. Patients or families are never interviewed. The review of source records is performed by a trained DAWN Reporter in each member facility. Depending on the needs of the facility, the DAWN Reporter may be an employee of the hospital or an employee of the DAWN operations contractor.

For each facility that participates in DAWN, the designated DAWN Reporter reviews all medical records to find ED visits related to drug use. The DAWN Reporter submits an electronic case report to the DAWN system for each ED visit that meets the specific case selection criteria. DAWN Reporters also track, on a copy of the ED registration log, their progress in reviewing the universe of ED visits.

Data items collected by DAWN

The case report form showing all the collected DAWN data items is provided in Figure C1.

Figure C1
DAWN ED case form

   D

DAWN features that enhance data quality and reliability

Several methods are used to improve the quality and reliability of DAWN data, including the following:

• Retrospective review of ED medical records for every patient treated in participating EDs,
  
• Electronic reporting with automated prompts and data validation,
  
• Inclusion of data items on the health effects of drug use and additional detail on patient disposition,
  
• Elimination of incidental drug reporting,
  
• Accurate, specific, and nonredundant drug reporting,
  
• Inclusion of data items to identify drugs confirmed by laboratory testing,
  
• Rigorous training and certification of DAWN Reporters, and
  
• In-house review and cleaning of DAWN case reports (see later section, Reduction of Bias, for more details).
  

ED visits eligible for DAWN

A DAWN case is any ED visit related to recent drug use. DAWN includes ED visits associated with substance abuse and misuse, both intentional and accidental. DAWN also includes ED visits related to the use of drugs for legitimate therapeutic purposes. To be a DAWN case, the relation between the ED visit and the drug need not be causal; the drug needs only to be implicated in the visit.

The case criteria are intended to be broad and inclusive, and to have few exceptions. Broad criteria take into account the fact that documentation in medical records varies in clarity and comprehensiveness across hospitals and among clinicians within hospitals. Broad criteria minimize the potential for judgments that could cause data to vary systematically and unexpectedly across reporters and hospitals. In addition, broad criteria are designed to capture a very diverse set of drug-related visits that can be aggregated or disaggregated to serve a variety of analytical purposes and the interests of multiple audiences. In DAWN, only recent drug use is included,¹⁸ the reason a patient used a drug is irrelevant, and the criteria are broad enough to encompass all types of drug-related events, including, but not limited to, explicit drug abuse.

There are a few clearly delineated exceptions to the DAWN eligibility criteria. An ED visit is not a DAWN visit if:

• There is no evidence of recent drug use.
• The patient left the ED without being treated.
• The patient consumed a nonpharmaceutical substance but did not inhale it.
• The patient has a history of drug use but no recent use.
• Alcohol is the only substance involved and the patient is an adult (aged 21 or over).
• The only documentation of a drug is in toxicology test results.
• The only drugs listed (e.g., current medications) are not related to the visit.
• The patient is being treated as a consequence of undermedication (i.e., taking too little of a drug).
  

Types of cases in DAWN

By design, DAWN's broad case criteria yield a diverse set of visits. To bring order to this heterogeneous mix of ED visits, each visit is assigned to one of eight types, which may be analyzed separately or in purposeful combinations. The eight types of visits are:

• Suicide attempt,
  
• Seeking detoxification,
  
• Alcohol only in patients under age 21,
  
• Adverse reaction,
  
• Overmedication,
  
• Malicious poisoning (including drug-facilitated sexual assault or product tampering),
  
• Accidental ingestion, and
  
• Other.
  

DAWN Reporters assign each DAWN case to one, and only one, of the eight case types, based on a series of questions and decision rules. The questions and rules are organized as a decision tree (Figure C2). Starting at the top, each case is assigned to the first case type that applies, even if the case might also meet the rules for a subsequent category. The eight case types were ordered with this in mind.

The final category in the decision tree is called Other and it is reserved for DAWN visits that do not meet any of the rules for classification into one of the first seven types. By design, most cases of drug abuse are classified as case type Other. This approach, which never directly identifies drug abuse, comes from the recognition that medical records frequently lack explicit documentation of substance abuse. This lack of documentation may occur for several reasons. First, the distinctions among use, misuse, and abuse are often subjective. Second, if there is a low index of suspicion for drug abuse in some types of patients, ED physicians may be unlikely to label those types of patients as drug abusers. Third, in many States, insurers may legally deny payment for ED visits related to substance abuse. Thus, financial incentives may be a powerful factor to influence documentation practices.

Drugs included in DAWN

DAWN includes all types of drugs:¹⁹

• Illegal drugs, such as heroin, cocaine, marijuana, and Ecstasy,
  
• Prescription drugs, such as Prozac^®, Vicodin^®, OxyContin^®, alprazolam, and methylphenidate,
  
• Over-the-counter (OTC) medications, including aspirin, acetaminophen, ibuprofen, and multi-ingredient cough and cold remedies,
  
• Dietary supplements, including vitamins, herbal remedies, and nutritional products,
  
• Psychoactive, nonpharmaceutical inhalants,
  
• Alcohol in combination with other drugs, and
  
• Alcohol alone, in patients aged less than 21 years.
  

Figure C2
Type of case decision tree

   D

To be reportable, a nonpharmaceutical substance must be consumed by inhalation, sniffing, or snorting, and it must have a psychoactive effect when inhaled. An ED visit involving inhalation of a nonpharmaceutical, psychoactive substance and no other drug qualifies as a DAWN case. Carbon monoxide is excluded from the inhalants. Beginning in 2004, cases involving accidental exposures (e.g., exposure to paint fumes while one is painting a closet) are excluded as well.

Hospital participation

For 2006, 205 hospitals submitted data on 269,339 drug-related ED visits that were used for estimation (Tables C1 and C2). The overall weighted response rate was 26.1%. For the 12 oversampled metropolitan areas and divisions, individual response rates ranged from 30.8% in the Houston-Baytown-Sugar Land, TX Metropolitan Statistical Area to 71.2% in the Detroit-Warren-Livonia, MI Metropolitan Statistical Area.

For 2005, 224 hospitals submitted data on 268,128 drug-related ED visits that were used for estimation. The overall weighted response rate in 2005 was 28.9%. There were 13 oversampled areas in 2005 with response rates ranging from 31.1% to 77.3%.

For 2004, 220 hospitals submitted data on 168,841 drug-related ED visits that were used for estimation. The overall weighted response rate in 2004 was 23.9%. There were 13 oversampled areas in 2004 with response rates ranging from 35.3% to 70.8%.

Charts reviewed for drug-related ED visits

DAWN cases are found through a retrospective review of medical records in participating hospitals. Across all participating hospitals in 2006, 9,837,481 charts were reviewed to find the drug-related ED visits that met the DAWN case criteria. Based on the review of charts, 346,946 drug-related visits²⁰ were found and submitted to the DAWN database, a case rate of 3.5%. On average, a DAWN member hospital submitted 1,077 DAWN cases. However, the number of submitted cases varied widely across hospitals, from 7 cases to 6,782 cases (median 820) in a single hospital during 2006.

Across all participating hospitals in 2005, 11,472,887 charts were reviewed to find the drug-related ED visits that met the DAWN case criteria. Based on the review of charts, 374,276 drug-related visits were found and submitted, a case rate of 3.3%. On average, a DAWN member hospital submitted 843 DAWN cases. The number of DAWN cases varied widely, from 1 case to 9,021 (median 525) for a single hospital.

Across all participating hospitals in 2004, 13,299,739 charts were reviewed to find the drug-related ED visits that met the DAWN case criteria. Based on the review of charts, 307,094 drug-related visits were found and submitted, a case rate of 2.3%. On average, a DAWN member hospital submitted 609 DAWN cases. The number of DAWN cases varied widely, from 1 case to 7,485 (median 377) for a single hospital.

DAWN data in this publication

For analysis, three categories of ED visits related to drug misuse and abuse were defined. These categories, designed to parallel the approach of the National Survey on Drug Use and Health, are based on:

• Use of illicit drugs,
  
• Use of alcohol, in combination with other drugs, and alcohol alone in patients under the age of 21, and
  
• Nonmedical use of pharmaceuticals (e.g., prescription or OTC drugs).
  

These three categories are defined by drug and case type as shown in Table C3. Because multiple drugs may be involved in a single visit, these categories are not mutually exclusive. A drug-related ED visit involved 1.6 drugs, on average, in 2004, 2005, and 2006.

Sampling and estimation

DAWN sample design

The redesign of the DAWN system introduced in 2003 altered most of the major features of the DAWN data collection and included a new sample of hospitals that constituted the DAWN. The new sampling plan, fully implemented for the first time for the 2004 estimates, formed a nationally representative panel of hospitals to be followed longitudinally for the indefinite future. Briefly, this new design is a probability-based, stratified, one-stage sample. A complete and accurate list of all hospitals in the United States was drawn and, from that, all hospitals meeting the criteria for the target sample frame were identified. Samples were drawn to provide the capability to make estimates for the Nation as well as select Metropolitan Statistical Areas (MSAs) and divisions (Table C4). Each year the sample frame is updated to account for new hospitals.

The stratified design called for drawing oversamples of hospitals in 48 MSAs; in four of those 48 MSAs, additional oversamples were drawn for a total of nine divisions.²¹ In effect, there are 53 nonoverlapping geographic areas (44 whole MSAs and nine divisions). (See Table C4 for list of MSAs and divisions²² where oversamples were drawn.) These areas are collectively referred to as oversample areas, or OS areas.

Metropolitan Statistical Areas and subdivisions

In order to accommodate a planned expansion of the metropolitan areas covered by DAWN, a maximum set of metropolitan areas, based on the definitions issued by the Office of Management and Budget (OMB) in June 2003, was selected. Which metropolitan areas to include was a topic of the DAWN redesign.²³ Retention of the existing 21 metropolitan areas was important because there was significant demand for estimates for those areas, and addition of the five most populous metropolitan areas in each of the nine Census divisions was deemed important to improve DAWN's geographic and population coverage. This yielded a total of 48 metropolitan areas. For many of the 48 metropolitan areas, the June 2003 definitions resulted in larger metropolitan areas. In some cases, these larger areas represented a merger of previously separate metropolitan areas. However, there continued to be strong interest among users of DAWN statistics in the areas covered by the original 21 metropolitan areas. In order to address the needs of these users, four of the merged areas were subdivided.²⁴ For each of these areas, there was a sample for the metropolitan area, as well as a sample for each subdivision. This would enable DAWN to produce estimates for the metropolitan areas and for the subdivisions. As a result of this process, the final metropolitan-area sample included a total of 53 geographic units: 48 metropolitan areas, 2 subdivisions each for 3 of these metropolitan areas, and 3 subdivisions for one of these metropolitan areas.

Sampled hospitals in each of the OS areas were stratified by hospital size (up to four categories based on volume of ED visits) and ownership type (public and private). (Size categories were determined independently for each OS area.) The stratification plan included an additional geographic construct for the remainder of the United States outside the OS areas. Hospitals in the remainder area were stratified into 24 remainder strata based on four regions (Northeast, South, Midwest, West), hospital size (three size categories based on volume of ED visits), and ownership type (public and private).

To begin, a cross classification was created by categories of ownership type and geographic unit. Within each combination of geographic area and ownership type, the number of hospitals determined the number of unique size categories. If there were three or fewer hospitals, only one size category was defined. If there were four, five, six, or seven hospitals, two size categories were defined. If there were eight or more hospitals, four size categories were defined. In the remainder sample, within each combination of Census region and ownership, there were three size categories. This produced 24 unique strata from which to draw the hospitals for the remainder sample.²⁵

The DAWN national estimates are the sum of the estimates for OS areas and the remainder area. Using a formula, the national estimate is depicted as:

   D

where a_i is the estimate for OS area i, 53 is the number of OS areas, and b is the remainder area estimate.

It was never expected that DAWN would be able to expand data collection into all 53 OS areas. Instead, the expectation was that DAWN would build up gradually to the number of OS areas its budget could support. The DAWN sample design was conceived to provide the flexibility to change gradually over time in terms of the number of OS areas where data were collected, while providing the statistical infrastructure to enable the production of reliable and representative estimates for the Nation and select OS areas, regardless of their number.

To accomplish this, the DAWN design incorporates an approach whereby a subset of the hospitals within the OS areas was identified a priori as having a dual-purpose in estimation. Referred to as "dual-purpose hospitals," these designated hospitals can contribute to an estimate for the OS area in which they are located or they can contribute to the estimate for the remainder area. Dual-purpose hospitals carry two probabilities of selection (POS) and two stratum identifiers. One POS/stratum is associated with membership in an OS-area sample and the other is associated with membership in the remainder-area sample.²⁶

Figure C3 depicts the initial sample as it was drawn to provide:

• individual samples from a series of OS areas,
  
• dual-purpose hospitals within those areas, and
  
• a remainder sample to represent the rest of the country.
  

For estimation for each data year, the first step is to determine which role each sampled hospital will play in that year's estimates. To do this, the response rates and nonresponse patterns for each OS area are reviewed to determine data quality. Those OS areas with acceptable data quality are allowed to stand on their own as the basis for separate estimates. These are referred to as "stand-alone OS areas." All hospitals in stand-alone OS areas, including those originally designated as being in the dual-purpose subsample, are considered to be "oversample hospitals" in the OS areas, and they contribute to the OS-area estimate using their OS-area POS/stratum.

If it is determined based on response rates and bias analyses that an OS area cannot stand alone, the design provides that the OS area is eliminated as a separate area but becomes part of the remainder area. In this instance:

• only those dual-purpose hospitals that are designated a priori to contribute to the remainder-area estimate are retained in the remainder-area subset,
  
• these hospitals contribute to the remainder-area estimate using their remainder-area POS/stratum, and
  
• data from any other hospitals in the OS area are excluded from the national estimates.
  

Figure C4 depicts the assignment of dual-purpose hospitals to either an OS area or the remainder area and the exclusion of OS hospitals outside of stand-alone OS areas that are not designated as dual purpose.

After it is determined which OS areas will be stand alone, the DAWN national estimates as reported in this publication are the sum of the estimates for stand-alone OS areas plus the remainder area. Using a formula, the national estimate is depicted as:

   D

where a_i is the estimate for stand-alone OS area i, 53 is the number of stand-alone OS areas, and b is the remainder area estimate inclusive of dual-purpose hospitals in OS areas that are not stand alone.

It is important to note that the definition of the remainder area and the remainder sample of hospitals is designed to be fluid; hospital membership in the remainder sample changes from year to year depending on the response rates and data quality within the OS areas.

Sample maintenance

Because DAWN is a longitudinal survey that will be used to analyze trends in drug-related ED visits over time, annual updates to the sample are performed to ensure that the sample remains representative of the target population. The initial sample was selected in 2003 from a sampling frame created from the 2001 AHA Annual Survey Database. In every subsequent year, the sampling frame is updated to reflect new, closed, merged, and demerged hospitals, based on updates to the AHA files. These updates include newly eligible hospitals, which are those new hospitals or previously ineligible hospitals that are now eligible. Each year, the newly eligible hospitals are provided the opportunity to be selected into the sample, based on the sampling fraction of the stratum in which the newly eligible hospital is located.

Reduction of bias

Survey error is the extent to which findings from the survey sample differ from those of the population of interest. The statistical methodologies described above are designed to minimize error. There are additional sources of error, often referred to as "bias," that also contribute to overall error. Measuring bias is difficult because it requires accurate knowledge about corresponding population values. The DAWN survey methodology includes proven techniques, practices, and protocols that reduce the potential for introducing bias. For example, clearly defined criteria are used to construct the initial hospital sampling frame. Coverage bias is minimized, because the sampling frame has virtually 100% coverage of the target population. To minimize measurement bias, the individuals who collect data for DAWN are provided with specialized and intensive training; automated methods for data entry are used; and the data are subject to quality reviews at several points in the data collection process. Additional detail on the survey methodologies used to enhance DAWN data quality and reduce bias is provided in an earlier DAWN publication.²⁷

Sample size and sample allocation

DAWN defines precision in terms of the relative standard error (RSE) of an estimate. The RSE is the standard error of the estimate divided by the actual point estimate. DAWN is designed to have RSEs less than or equal to 10% for metropolitan-area estimates, and RSEs less than or equal to 15% for national estimates pertaining to total drug-related visits, cocaine visits, heroin visits, and marijuana visits. As discussed below, these desired precision levels are important drivers for setting sample size targets.

Sample sizes for each geographic area were determined by the area's targeted precision level in combination with the theory of optimal allocation for stratified samples. According to this approach, the variance of the sample estimates will be minimized when the sample size, n_h, in each sampling stratum is made proportional to the quantity W_hS_h /C_h, where W_h is the proportion of sampling units, S_h is the population standard deviation for the parameter being measured, and C_h represents the square root of the cost of sampling in stratum h.

Using these optimum allocation conditions, the minimum required sample sizes necessary to achieve the targeted levels of precision in each DAWN area were calculated using the following general considerations:

• Geographic units for which estimates are desired (national and metropolitan areas described under Stratification),
  
• Precision level desired (see Target levels of precision),
  
• Specific types of estimates for which minimum precision is desired (e.g., estimates of total, cocaine, heroin, and marijuana ED visits), and
  
• Cost.
  

In addition to these considerations, sampling rates (i.e., the number of sampled hospitals divided by the number of eligible hospitals) were also subject to the following constraints:

• First, if fewer than four hospitals existed in the stratum population, then all hospitals in the stratum were selected into the sample.
• Second, if the sampling rate for a particular stratum was greater than 90%, then all units in the stratum were selected into the sample.
• Finally, if any calculations produced a sample size smaller than two hospitals, then the sample size was set to two hospitals.
  

Response rate calculations

In 2006, the initial DAWN sample included 1,286 hospitals divided among 53 OS areas (48 MSAs and nine divisions) and one remainder area. Response rates and nonresponse bias analyses were assessed to determine which of these 53 OS areas could stand alone (Figure C3). Once this determination was made, hospitals that were neither dual-purpose nor located in a stand-alone OS area were treated as if they were not sampled. For 2006, this has the effect of reducing the sample from 1,286 hospitals to 544 hospitals, which is the number used for purposes of computing the unweighted response rates (Table C1).

Of the 53 original oversample areas, a total of 12 areas (nine metropolitan areas and three submetropolitan areas) were determined to be able to stand alone in 2006. A total of 13 areas were determined to be able to stand alone in 2004 and 2005. The 13th OS area was New Orleans-Metairie-Kenner, LA. These determinations yielded a sample size of 556 hospitals in 2004 and 562 hospitals in 2005.

Sampling weights

The DAWN hospitals are selected using stratified simple random sampling with oversampling in selected metropolitan areas. The stratum sample sizes were determined through an optimum allocation process. Sampling weights are first calculated as the inverse of the probability of selection and then adjusted for variable nonresponse and by a procedure known as "poststratification," or benchmark adjustment.

Within-hospital weighting adjustment

Within-hospital nonresponse occurs when a hospital provides incomplete data. To minimize the impact of within-hospital nonresponse, the DAWN weighting plan includes nonresponse adjustment factors that were developed and applied for each month of data collection within each facility. The within-hospital nonresponse adjustment factor is calculated as the total number of ED visits within a month within a facility divided by the total number of reviewed charts for that same facility-month.

The within-hospital weights are applied to the case data by month and by facility. That is, the visit counts for a given facility-month are first summed for each drug and then multiplied by the corresponding within-hospital adjustment factor for that facility-month. The weighted totals are then summed over all facilities and months to give a total weighted visit count for each drug for each hospital.

Weighting adjustment for hospital nonresponse

Hospital-level nonresponse occurs when hospitals fail to provide any data. To minimize the impact of hospital nonresponse, the DAWN weighting plan includes nonresponse adjustment factors that were developed and applied within each weighting class. Weighting classes were formed based on the aforementioned sampling stratification schemes. Within each weighting class, the nonresponse adjustment factor is calculated as the sum of the sampled hospital weights divided by the sum of the weights of the responding hospitals. The hospital nonresponse adjustment factors are checked to make sure the adjustments are within reasonable bounds. If a nonresponse adjustment factor is out of bounds (either too small or too large), adjacent weighting classes are collapsed and new nonresponse adjustment factors are calculated.

When the hospital-level nonresponse adjustment factors are considered final, a nonresponse-adjusted sampling weight was then calculated as the product of the nonresponse adjustment factor and the sampling weight. For each weighting class, a verification check was conducted to ensure that the sum of the nonresponse-adjusted sampling weights was equal to the sum of the sampled hospital weights.

Weighting adjustment for population benchmarks (poststratification)

The DAWN weighting plan also includes a poststratification adjustment factor that reconciles the weighted number of total visits for responding hospitals with the number of total visits from the most recent AHA Annual Survey Database. DAWN used a ratio adjustment within strata to implement this adjustment.

Poststratification strata were formed based on the aforementioned sampling stratification schemes. Within each stratum, the adjustment factor was calculated as the ratio of the AHA count of total visits to the weighted sum of total visits for responding hospitals. The factors were verified to ensure they were within reasonable bounds. If they were out of bounds (either too small or too large), adjacent poststratification strata were collapsed and new poststratification adjustment factors were calculated.

When the poststratification adjustment factors were considered final, a poststratified weight was then calculated. The final weight was calculated as the product of the poststratification adjustment factor and the nonresponse-adjusted sampling weight. For each poststratification stratum, a validity check was conducted to ensure that the sum of the poststratified weighted total visits was equal to the corresponding AHA count of total visits from each stratum.

Special consideration was given to New Orleans because Hurricane Katrina (on August 29, 2005) and its aftermath caused serious disruptions to the operations of hospitals and, consequently, the DAWN data collection process for the remaining four months of the year. Weight adjustments were implemented by referring to the AHA total counts in combination with a case-by-case study of each hospital in the New Orleans metropolitan area.²⁸ Based on these studies, estimates were made of the proportion of the year that each hospital was open and serving the public. These were used to adjust the AHA totals, which in turn serve as input for population benchmark counts for New Orleans.

Total drug-related ED visits

Estimates for the entire universe of DAWN-eligible hospitals in the United States are produced by applying poststratified weights to the data received from the sampled hospitals. Thus, for 2006, 269,339 submitted cases were extrapolated to an estimate of 3,441,855 drug-related ED visits. Considering the margin of error, this estimate may range from 3,033,110 to 3,850,600 drug-related ED visits out of more than 113 million total ED visits estimated for the United States (Table C2).

Calculation of estimates

All estimates produced for this publication were calculated using data that had been weighted according to the plan described above. Estimates for any variable of interest were determined by first summing the case totals within facility-month, applying the within-hospital weight, summing to the hospital level, applying the final hospital weight, and summing over all hospitals.

Variance estimation

Each hospital in the DAWN sample was selected through a random process, which theoretically could have been repeated many times, resulting in many hypothetical samples. "Sampling variance," or the margin of error, refers to the extent to which these samples vary. Two measures of this variability are the standard error (SE) and the relative standard error (RSE), which is defined as the SE of the estimate divided by the estimate itself. The precision of an estimate is inversely related to the sampling variance, as measured by the RSE. The greater the RSE value, the lower the precision.

For example, if there are 10,000 estimated visits involving a given drug, and this estimate has an SE of 500 visits, then the RSE value is 5%:

RSE = SE/Estimate
RSE = 500/10,000
RSE = 0.05, or 5%.

In this publication, "confidence intervals" (CIs) are included in many of the tables and are often cited in the text along with the estimates. The 95% CI is calculated as:

CI = Estimate ± (1.96 × RSE × Estimate)

where 1.96 comes from the table of normal distribution z-values. Ninety-five percent of the normal distribution lies within 1.96 standard deviations of the mean.

Applying the formula to the example above, the 95% CI would be:

10,000 ± (1.96 × 0.05 × 10,000) = 10,000 ± 980.0
Lower limit: 10,000 - 980 = 9,020
Upper limit: 10,000 + 980 = 10,980
95% CI: 9,020 to 10,980.

If repeated samples were drawn from the same population of hospitals, using the same sampling and data collection procedures, the true population values would fall within the confidence interval 95% of the time.

Both between- and within-hospital variance components were accounted for. Within-hospital variance was estimated using a replication strategy by which two random replicates were created within each hospital and the variance between the two replicates represented the within-hospital contribution. Typically, this component was considerably smaller than the between-hospital variance that was calculated as the variance between weighted hospital totals within each stratum.

Variance estimates reported in this publication were determined using Taylor Series Linearization. Variance estimates were calculated using SUDAAN^® software.

Standardized rates

Standardized measures are needed to make valid comparisons of estimates across age and gender categories. For age in particular, the size of the underlying population differs considerably across age groups; for example, the number of individuals aged 18 to 20 in the United States is much lower than the number of individuals aged 35 to 44. All other factors being the same, a higher estimate of ED visits would be expected to occur naturally for the group that is larger in the population.

To take the size of the underlying population into account, rates of ED visits per 100,000 population were calculated using population data from the U.S. Bureau of the Census.²⁹

For each age and gender category, the estimate for a category was divided by the population for that category, which was then divided by 100,000. For example, consider an estimate of 1,000 visits for an age group of 1,000,000 persons, and an estimate of 1,000 visits for an age group of 500,000 persons. The rates would be calculated as:

1,000 / (1,000,000/100,000) = 1,000/10
= 100 visits per 100,000 population

1,000 / (500,000/100,000) = 1,000/5
= 200 visits per 100,000 population.

Population estimates used to generate rates for 2006 are provided in Table C5.

Standardized rates were not calculated for race and ethnicity subgroups, because the race/ethnicity categories available to DAWN are much less detailed and contain considerably more missing data than the race and ethnicity categories in the Census data. Appendix D describes the race and ethnicity data reported for DAWN.

Determination of significant differences between years

Comparisons in the estimates of ED visits between years are presented in the form of percentage differences. These are calculated as the 2006 estimate minus the 2004 estimate divided by the 2004 estimate. For shorter term comparisons, these are calculated as the 2006 estimate minus the 2005 estimate divided by the 2005 estimate. The result is presented as a percentage, which is shown only if the difference between the two years is statistically significant. Tests for the significance of differences between two years consider the variance of each year's estimate and the covariance between the two. Thus, hospitals that appear in both samples and provide data in both years contribute to the covariance and, thus, decrease the overall sampling variance beyond the combined contribution of the two samples. The variance estimation process used to establish significance takes into account this overlap between the two annual samples.

Publication criteria

DAWN can produce estimates for thousands of drugs, patient characteristics, and visit characteristics, but some of these estimates are too imprecise, too small, or based on too little data to be reliable. In these situations, the estimate was replaced by three dots (…) in the published table. Estimates were suppressed (i.e., not published) according to the following rules:

• The RSE of the estimate was greater than 50%.
  

When the RSE is greater than 50%, the lower bound of the 95% CI approaches or includes the value zero. A CI that includes zero means that the estimate is not statistically different from zero at this precision level.

• The weighted or unweighted quantity of ED visits was less than 30.
  

Estimates this small constitute rare events, which are based on a small number of cases and have precision levels that are difficult to quantify. In many instances, such rare events have variances so large that the estimate would be suppressed because of its RSE alone. Rare events that meet RSE criteria for publication are nonetheless based on very little data and are deemed too unreliable for publication.

There are some estimates with an RSE equal to zero. This occurs when the number of ED visits being estimated is small and all the hospitals contributing to that estimate were selected with certainty, that is, their sampling probability is unity. Strictly speaking, there is no sampling error in such situations and the RSE is equal to zero. These results occur almost exclusively in situations with small numbers of ED visits, where the absence of any sampled hospital data is due to nonresponse and the small number of hospitals contributing to the estimates. In these situations, the necessary data are not available to approximate sampling errors.

APPENDIX D

RACE AND ETHNICITY IN DAWN

In October 1997, the Office of Management and Budget (OMB) issued a revised standard protocol for race and ethnicity categories used in Federal data collection systems.³⁰ The new protocol permitted separate reporting of race and Hispanic ethnicity, and it incorporated the ability to capture more than one race for an individual, a few modifications in nomenclature (e.g., "black" was changed to "black or African American"), division of certain categories ("Asian or Pacific Islander" was split into two categories, "Asian" and "Native Hawaiian or Other Pacific Islander"), and elimination of the "other" category. For data collections such as DAWN, where self-identification of the individual is not feasible (no patient is interviewed for DAWN), the OMB protocol also permitted a combined format, whereby race and Hispanic ethnicity would be recorded in a single data item, which could still record multiple entries for race and/or Hispanic ethnicity. The single data item for race and ethnicity is shown in the DAWN ED case form that has been used since 2003 (Appendix C, Figure C1).

Despite the increased detail allowed by the new categories and the provision for multiple entries, the actual race/ethnicity data extracted from source records and submitted to DAWN is quite limited. This is because the source documents (i.e., the ED medical records from which DAWN data are abstracted) rarely contain such detailed information on race/ethnicity of patients.

For reference, estimates of drug-related ED visits by race/ethnicity are presented in Table D1. This analysis, which is based on the most detailed coding of race/ethnicity in DAWN case reports, reveals that estimates for the following categories are too small to be meaningful:

• Multiple (i.e., two or more) races/ethnicities (i.e., two or more races/ethnicities were documented in the source record for the same individual),
  
• Hispanic or Latino ethnicity with any specific race indicated,
  
• American Indian or Alaska Native,
  
• Asian, and
  
• Native Hawaiian or Other Pacific Islander.
  

Therefore, in the tables of estimates in this and other DAWN publications we have retained a more limited set of categories: white, black, and Hispanic. A fourth category, called "Race/ethnicity not tabulated above (NTA)," is used to tabulate those categories that are too small to report independently.³¹ All cases reported to DAWN as Hispanic or Latino ethnicity are tabulated as Hispanic race/ethnicity, regardless of race.

This lack of detailed race and ethnicity data in DAWN case reports also prevents us from generating rates per 100,000 population for race and ethnicity categories. Data from the 2000 decennial Census were collected and are being tabulated according to the revised race and ethnicity protocol and are therefore incompatible with DAWN estimates.

End Notes

¹ The 95% confidence interval (CI) accounts for the margin of error of the estimate. It indicates, with a high degree of confidence, that the true population value was between 1,451,086 and 2,034,688 drug-related ED visits.
² That is, the rates for the age categories 18 to 20, 21 to 24, 25 to 29, 30 to 34, 35 to 44, and 45 to 54 were not significantly different.
³ Though a drug was implicated in each visit, these attempts are not limited to drug overdoses.
⁴ That is, patients with a history of drug use (and no recent use) are excluded.
⁵ The classification of drugs used in DAWN is derived from the Multum Lexicon, © 2007, Multum Information Services, Inc. The classification was modified to meet DAWN's unique requirements (2007). The Multum Licensing Agreement governing use of the Lexicon is provided in Appendix A.
⁶ To be reportable, a nonpharmaceutical substance must be consumed by inhalation, sniffing, or snorting and must have a psychoactive effect when inhaled. An ED visit involving inhalation of a nonpharmaceutical, psychoactive substance qualifies as a DAWN case even if no other drug was present. Carbon monoxide is excluded from the inhalants. Since 2004, cases involving accidental exposures (e.g., exposure to paint fumes while painting a closet) have been excluded as well.
⁷ ED visits related to alcohol use alone are excluded for patients aged 21 and over.
⁸ Excluded are suicide attempts involving pharmaceuticals, accidental ingestions, visits for patients seeking detoxification services or entry into a substance abuse treatment program, and visits associated with the therapeutic use of pharmaceuticals.
⁹ For DAWN, 2003 was a transition year: 2003 data reflected some of the new features (e.g., the expanded case criteria) but also some of the old (e.g., the old sample of hospitals). Full-year estimates were not published for 2003, and the estimates that were published are not comparable to those from prior or subsequent years.
¹⁰ Population estimates for 2006 are, as of 6/9/2008, from the U.S. Bureau of Census Postcensal Resident Population National Population Dataset, National estimates by demographic characteristics – single year of age, sex and race, and Hispanic Origin, Monthly Population Estimates. Link: http://www.census.gov/popest/datasets.html. File: NC-EST2007-ALLDATA-R-File14.csv.
¹¹ These three categories of ED visits are not mutually exclusive, and the sum of the estimates is greater than the total number of drug misuse/abuse visits. See Appendix C for additional detail on the type of ED visits included in each category.
¹² ED patients over the age of 21 for whom alcohol was the only drug associated with their ED visits are not considered DAWN cases.
¹³ DAWN cases are identified through a retrospective review of medical charts. Given the limitations of medical record documentation, we have concluded that distinguishing misuse from abuse reliably is not feasible.
¹⁴ Taking less than the prescribed or recommended dose is not considered "nonmedical use."
¹⁵ This includes only single-ingredient formulations. Many multi-ingredient pharmaceuticals containing diphenhydramine are classified elsewhere, e.g., as respiratory agents.
¹⁶ In this publication, drugs are classified using the Drug Reference Vocabulary that was current as of May 2007.
¹⁷ AHA Annual Survey Database, Fiscal year 2001 Health Forum LLC, Copyright 2003, One North Franklin Street, Chicago, IL 60606.
¹⁸ That is, patients with a history of drug use (and no recent use) are excluded.
¹⁹ The classification of drugs used in DAWN is derived from the Multum Lexicon, © 2007, Multum Information Services, Inc. The classification was modified to meet DAWN's unique requirements (2007). The Multum Licensing Agreement governing use of the Lexicon is provided in Appendix A and can be found on the Internet at http://www.multum.com.
²⁰ For 2004-2006, more hospitals participated in DAWN than were used in estimation. Therefore, the number of drug-related ED visits from all participating hospitals exceeded the number used for estimation.
²¹ The four MSAs where samples were drawn for divisions are Los Angeles, Miami, New York, and San Francisco. The division definitions used by DAWN follow Census Bureau definitions of Metropolitan Divisions, except in New York where the three submetropolitan areas were defined uniquely based on local input.
²² MSAs and Metropolitan Divisions follow the standard definitions issued by the Office of Management and Budget in June 2003 (available at http://www.whitehouse.gov/omb/bulletins/b03-04.html and http://www.census.gov/population/www/estimates/metrodef.html).
²³ Substance Abuse and Mental Health Services Administration, Office of Applied Studies. Drug Abuse Warning Network: Development of a New Design (Methodology Report). DAWN Series M-4, DHHS Publication No. (SMA) 02-3754, Rockville, MD, 2002.
²⁴ When metropolitan areas were redefined in June 2003 based on data from the 2000 decennial Census, several legacy MSAs were merged with other MSAs to form new, much larger MSAs. However, a strong constituency of DAWN data users still needed estimates for the pre-merger areas. Because of this, four of the 48 metropolitan areas-Los Angeles, Miami, New York, and San Francisco-were subdivided into a total of nine divisions, corresponding to the constituents' areas of interest.
²⁵ Four Census regions times two ownership categories times three size categories equals 24 strata.
²⁶ In addition, a portion of hospitals in the nine oversampled divisions were identified a priori to serve in their MSA-level oversample and were assigned an OS area level POS/stratum for that third purpose. Therefore, hospitals in the four MSAs with division-level oversampling can have up to three nonzero POS/strata: (1) a POS/stratum for membership in the MSA; (2) a POS/stratum for membership in the division; and (3) a POS/stratum for membership in the remainder area.
²⁷ See Appendix B, Technical Notes: Changes to Improve the Quality of DAWN Data in: Substance Abuse and Mental Health Services Administration, Office of Applied Studies. Drug Abuse Warning Network 2003: Interim National Estimates of Drug-Related Emergency Department Visits. DAWN Series D-26, DHHS Publication No. (SMA) 04-3972, Rockville, MD, 2004.
²⁸ Information provided by the Louisiana Hospital Association, the Louisiana Public Health Institute, the City of New Orleans Health Department, and available at numerous websites maintained by individual hospitals, news organizations, and State and Federal agencies was invaluable in conducting this assessment. We are grateful for their assistance.
²⁹ Population estimates for 2006 are, as of 6/9/2008, from the U.S. Bureau of Census Postcensal Resident Population National Population Dataset, National estimates by demographic characteristics – single year of age, sex and race, and Hispanic Origin, Monthly Population Estimates. Link: http://www.census.gov/popest/datasets.html. File: NC-EST2007-ALLDATA-R-File14.csv.
³⁰ See Office of Management and Budget, Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity, 62 Fed. Reg. 58,782 (October 30, 1997).
³¹ One exception is that, if two races are reported and the second is reported as unknown, the episode is coded for the known race.