An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Substance Abuse and Mental Health Services Administration. Impact of the DSM-IV to DSM-5 Changes on the National Survey on Drug Use and Health [Internet]. Rockville (MD): Substance Abuse and Mental Health Services Administration (US); 2016 Jun.
Substance use disorders (SUDs), as described in DSM-IV, are part of a class of disorders (substance-related disorders) that are “related to the taking of a drug of abuse (including alcohol)” [1 p. 175]. Although the transition from DSM-IV to DSM-5 involves changes at multiple levels for SUDs, this basic definition remains unchanged [1,2]. However, changes have occurred at the class level (the specific disorders considered within the overall group of disorders), at the substance level (which substances are considered “drugs of abuse”), at the disorder level (the template of criteria that are applied, with some deviations, across all substances), and at the individual criteria level (the number and types of symptoms needed to meet criteria for a disorder). The following section delineates the specific changes from DSM-IV to DSM-5 and evaluates, to the extent possible given the available data, how these changes may affect the measurement of substance use disorders in the National Survey on Drug Use and Health (NSDUH) and the Mental Health Surveillance Study (MHSS). Table 2.1 provides a cross-cutting comparison of diagnostic criteria for both DSM-IV and DSM-5, which is important for framing the discussion of diagnostic changes across versions.
Comparison of DSM-IV, DSM-5, and NSDUH Substance Use Disorder Assessment.
NSDUH SUD assessments map closely to the DSM-IV criteria; however, NSDUH diverges occasionally from DSM-IV. Changes discussed below note the differences in NSDUH from DSM-IV criteria and the changes from DSM-IV to DSM-5. Currently available literature focuses on changes from DSM-IV to DSM-5; thus, in places where NSDUH diverges from the DSM-IV criteria, the impact of DSM-5 criteria changes on NSDUH may be difficult to quantify. These limitations are noted, as are the limitations in the current literature overall.
A disorder “class” is the term used to describe groups of similar disorders in the DSM (e.g., anxiety disorders and mood disorders). The DSM-5 contains numerous changes in the classification system and one of these changes has been to the classification of SUDs. In DSM-IV, SUDs belonged to the class substance-related disorders, which included only substance/drug-based disorders. In DSM-5 this classification has been broadened to include gambling disorder, and the section has been renamed Substance-Related and Addictive Disorders. Although this change will have no impact on prevalence estimates for SUDs, ensuring precise language in Substance Abuse and Mental Health Services Administration (SAMHSA) documentation and reports will be important to avoid equating prevalence estimates for any substance use disorder with prevalence estimates of the broader category of substance-related and addictive disorders.
NSDUH contains 13 modules assessing specific substances and substance types (tobacco, alcohol, marijuana, cocaine, crack, heroin, hallucinogens, inhalants, prescription pain relievers [opioids], tranquilizers, stimulants, and sedatives) and 1 module assessing special substance use, which assesses primarily methods of use.3 It also has the capacity to determine polysubstance dependence. The substances assessed vary somewhat from DSM-IV specified substances (see Table 2.1). Specific differences in NSDUH from DSM-IV include the following:
Changes from DSM-IV to DSM-5 in the types of substances assessed have been minor, but some reclassification has occurred. Primarily, cocaine (including crack) and amphetamines have been combined with other stimulants (excluding caffeine) into a single stimulant class based on evidence that they have similar mechanisms of action (increasing synaptic dopamine), symptom profiles, consequences, and prognoses. NSDUH-based reports could provide the estimates of the newly combined categories in addition to substance-specific estimates (e.g., providing an estimate for amphetamine use disorder and providing the estimate for all stimulants combined) in the future to enable comparisons with other datasets, track trends for the new diagnostic categories, and improve consistency with DSM-5 conventions.
NSDUH assesses substance abuse and substance dependence. DSM-IV and DSM-5 also assess substance intoxication, intoxication delirium, withdrawal syndrome, and withdrawal delirium for relevant substances. Because NSDUH does not assess those additional disorders, this report focuses only on changes to substance abuse and dependence criteria.
A major change from DSM-IV to DSM-5 is the combination of substance abuse disorder and substance dependence disorder into a single SUD. The DSM-IV substance abuse diagnosis required the endorsement of one or more symptoms (out of four, at any time) and no history of substance dependence for that category of substances (see Table 2.1 for the specific criteria). The substance dependence criteria required the endorsement of three or more symptoms (out of seven) in a 12-month period. DSM-IV diagnostic hierarchy rules also specified that people who met criteria for both substance abuse and substance dependence for a particular substance were diagnosed as having substance dependence only. The purpose of this was to reflect the increased severity of dependence over the abuse diagnosis [13]. The DSM-5 has eliminated the distinct abuse and dependence disorders for several reasons: (1) the distinction provided little guidance for treatment; (2) the distinction created “diagnostic orphans” (individuals who endorsed two dependence symptoms and no abuse symptoms and therefore did not meet any diagnostic criteria); (3) the hierarchical structure did not follow the anticipated relationship between abuse and dependence (that abuse was largely a less severe prodrome4 of dependence); and (4) the separation caused the abuse diagnosis to suffer from significant reliability problems [13,15–21]. The DSM-5 combines the abuse and dependence criteria under the new rubric substance use disorder, which requires 2 out of 11 criteria in a 12-month period for diagnosis. In addition, the DSM-5 has eliminated the abuse criterion related to recurrent substance-related legal problems (e.g., arrests for substance-related disorderly conduct [1]) and added a craving criterion. The legal problems criterion was dropped due to low endorsement, poor fit with other items, and the poor discrimination of this item (almost all people endorsing the legal criteria endorsed other criteria also) [13]. This was further verified by estimates using NSDUH data, discussed further in Section 2.5, Specific SUDs.
Under DSM-5 criteria, craving is defined as a “strong desire or urge to use the substance.” DSM-5 text further adds that the phenomenon of craving “makes it difficult to think of anything else” and “often results in the onset” of use [2, p. 492]. Examination of general population studies indicated that craving, as an indicator of an SUD, did not add to the total information offered by other dependence criteria. That is, other dependence criteria (e.g., tolerance, withdrawal, and continuing use despite health problems) overlapped with craving so that the addition of craving identified very few people who did not already meet the threshold for a disorder through the other dependence criteria. However, the inclusion of craving with the abuse criteria added significantly to the diagnostic information and there is some indication that craving may become a target for biological treatments [22]. Notably, craving was already a component of the International Classification of Diseases, 10th revision (ICD-10), diagnostic system, which is used outside of the United States, and thus the DSM-5 craving addition improves consistency across classification systems.
NSDUH does not assess the craving criterion. If no changes were made to NSDUH questions and existing data were used to approximate the estimates by modifying the diagnostic algorithm, the impact would be seen on threshold-level cases (i.e., individuals who endorsed only one criterion). Data from other studies that assessed craving (such as the National Epidemiologic Study of Alcohol and Related Conditions [NESARC] and the National Longitudinal Alcohol Epidemiologic Survey [NLAES]) could be used to impute the frequency of people who would reach diagnostic threshold if the craving criterion was present. Imputation could be done with the population-level estimates or at the individual level. However, the current data that would be used to generate these estimates are between 10 (NESARC) and 20 (NLAES) years old and do not include adolescents. Imputation could increase the chances of misclassification, which may bias statistical results, particularly in analyses involving a large number of people close to the diagnostic threshold. In addition, the prevalence and characteristics of people endorsing the craving criterion may vary by substance. This is discussed in more detail later in this report in the context of the individual substances.
An additional criterion that has undergone some revisions in DSM-5 is the dependence criterion of withdrawal. Unlike other criteria, withdrawal symptoms are specific to the physiological effect of the substance (Table 2.2). In both DSM-IV and DSM-5, withdrawal is manifested by either (1) a person having the characteristic withdrawal symptoms for the substance, or (2) a person using the same or closely related substance to avoid the substance-specific withdrawal symptoms. DSM-IV and DSM-5 withdrawal criteria are unchanged for all substances except cannabis. Research conducted after the publication of the DSM-IV has identified a cluster of symptoms associated with cannabis withdrawal, and this new information has been included in the DSM-5 [2]. Cannabis withdrawal syndrome is defined by the presence of three or more symptoms developing within approximately 1 week of cessation of heavy and prolonged cannabis use. Symptoms can include (1) irritability, anger, or depression; (2) nervousness or anxiety; (3) sleep difficulties (e.g., insomnia or disturbing dreams); (4) decreased appetite or weight loss; (5) restlessness; (6) depressed mood; and (7) at least one physical symptom that causes significant discomfort (abdominal pain, shakiness/tremors, sweating, fever, chills, or headache).
DSM-IV to DSM-5 Withdrawal Symptom Comparison.
NSDUH assesses withdrawal (except cannabis withdrawal) as one part of the dependence criteria. However, NSDUH departs from DSM-IV criteria in several ways. With respect to specific substances, NSDUH diverges in three cases (see Table 2.2). First, NSDUH does not assess tobacco withdrawal. Second, there are some deviations from the withdrawal criteria for sedatives, hypnotics, or anxiolytics. DSM-IV specifies two or more symptoms to meet criteria for sedative, hypnotic, or anxiolytic withdrawal, although the NSDUH instrument specifies only that one symptom is necessary. This particular deviation is addressed in the future redesign, planned for 2015, and to conform with DSM-IV and DSM-5 withdrawal criteria. Moreover, DSM-IV criteria for sedative, hypnotic, or anxiolytic withdrawal specify only the symptom of insomnia, but NSDUH includes insomnia or hypersomnia. These differences may have led to an overestimate of the number of people who met criteria for sedative, hypnotic, or anxiolytic withdrawal and therefore substance dependence. Third, NSDUH has not historically assessed tranquilizer withdrawal leading to a potential underestimate of tranquilizer dependence.
Conceptually, NSDUH diverges from DSM (IV and 5) withdrawal criteria in several ways (for all substances with a withdrawal component), which may lead to estimates that do not completely reflect DSM withdrawal criteria. Withdrawal criteria in DSM-5 consist of two items:
Criteria A and B differ slightly across substances but follow a general template. Criterion A specifies that the person has to have ceased (or reduced) heavy and prolonged use of the substance. Criterion B specifies that a certain number of symptoms, from a list provided (see Table 2.2) developed within several hours of a few days after the cessation (or reduction) from Criterion A. Deviations from this general pattern occur for sedative, hypnotic, or anxiolytics and stimulants, wherein the Criterion A specifies only prolonged use (not heavy), and cannabis, which specifies that the Criterion B symptoms develop within approximately 1 week of ceasing or reducing use.
In NSDUH, withdrawal is assessed with a series of questions that follow a unified template for all substances with withdrawal symptoms (except the new cannabis withdrawal). First, respondents have to answer affirmative to a question about having tried to cut down or quit before they are routed to the main withdrawal questions. This is potentially problematic because the DSM does not specify that the person had to intentionally try to go without the substance to meet withdrawal criteria (e.g. if a person runs out of their supply of a substance they may experience withdrawal even though they did not try to cut down or quit). This routing pattern may lead to an underestimate of the prevalence of withdrawal in NSDUH.
The second question for assessing withdrawal asks: “Please look at the symptoms listed below. During the past 12 months, did you have” # (varies by substance) “or more of these symptoms after you cut back or stopped using substance?” This is followed by a list of the specific withdrawal symptoms for that substance. If respondents answer affirmatively, then they receive another question about withdrawal that varies slightly from the previous: “Please look at the symptoms listed below. During the past 12 months, did you have # or more of these symptoms at the same time that lasted for longer than a day after you cut back or stopped” using the substance? Respondents are only considered to have met withdrawal criteria if they endorsed the second question. These questions have several points of potential deviation from DSM criteria.
The final deviation from DSM (IV and 5) criteria is that NSDUH does not assess the second part of the withdrawal item, which is taking the substance or a closely related substance to avoid withdrawal. This would likely lead to an underestimate of withdrawal symptoms because of missing individuals who preemptively avoided withdrawal symptoms by using the substance or a closely related substance.
With the potential for adding an assessment of cannabis withdrawal, these deviations should be considered since new items will need to be developed for cannabis and there is the opportunity to better map NSDUH to DSM criteria while other changes are being implemented. Because these deviations are specific to NSDUH, it is difficult to quantify the impact of their revision on prevalence estimates. Overall, it is probable that estimates would increase, but the magnitude of increase in unknown.
The DSM-IV did not specifically assess the severity of SUDs, although in general, dependence was considered more severe than abuse and people receiving the dependence diagnosis did not receive an abuse diagnosis even if the criteria for abuse were met [13]. DSM-5 has added a symptom count-based severity indicator, with two to three symptoms being classified as mild, four to five symptoms classified as moderate, and six or more symptoms being classified as severe. The severity index addition was driven by research, which suggested a simple symptom count was as effective at measuring severity as more complicated algorithms [23]. Calculation of severity using NSDUH could be accomplished with little effort if the craving criterion were added to the survey.
Thus far, changes in diagnostic criteria have been evaluated at the overall diagnostic level (the template applied to each substance, Table 2.1). However, criteria changes may not affect estimates for every substance equally and some substances deviate slightly from the general SUD template. This section discusses the impact of the DSM-5 revisions to each substance-specific SUD in terms of prevalence and measurement considerations.
Assessment of alcohol use disorder (AUD) aligns with the DSM-IV and DSM-5 SUD template discussed previously (i.e., DSM-IV assessed alcohol abuse [1 or more of 4 criteria] and alcohol dependence [3 or more of 7 criteria], and DSM-5 assesses AUD [2 or more of 11 criteria], Table 2.1). The changes from DSM-IV to DSM-5 included the addition of the craving criterion, the removal of the legal problems criterion, and the change in threshold for diagnosis. Evidence for how the DSM-5 revision will affect this estimate is mixed. In an examination of responses in the second wave of the NESARC, a nationally representative household-based survey of over 34,000 U.S. adults, Agrawal, Heath, and Lynskey reported a DSM-IV past year AUD prevalence of 9.7 percent (weighted) and a prevalence of DSM-5 AUD of 10.8 percent, a relative increase of 11 percent [24]. Similarly, in a recent analysis of an aggregated dataset including 7,543 individuals from family-based and case-control genetic studies of substance dependence (86.6 percent of whom had an SUD), Peer et al. found that discordant subjects (those who had AUD under either DSM-IV or DSM-5 criteria, but not both), were more likely to have gone from no diagnosis under DSM-IV criteria to having a diagnosis under DSM-5 criteria rather than the reverse [25]. In the Peer et al. sample, the prevalence of AUD increased from 63.0 percent under DSM-IV to 63.5 percent under DSM-5 criteria. However, this study was primarily composed of individuals with an SUD and the impact on NSDUH estimates may not be the same. Results of a nationally representative study conducted in Australia, suggest that the increase may be more substantial in a more generalizable sample. The study, which included 10,641 adults sampled from representative households across Australia, found that there was a 61.7 percent increase (from 6.0 to 9.7 percent) in the prevalence of AUD among past year alcohol users when the criteria were changed from DSM-IV to DSM-5 [26].
Not all studies have found an increase in prevalence under DSM-5 criteria. In a population-based study of 5,443 current drinkers aged 18 to 64 in the Netherlands, 5.4 percent (weighted) of current drinkers had a past year AUD using DSM-IV criteria, whereas 4.4 percent of current drinkers met DSM-5 criteria (including craving) for past year AUD [27]. Comparisons across these studies are difficult because of the differences in the study population and also because of the denominator used to calculate estimates (e.g., all participants, participants who used alcohol in the past year, current drinkers).
Direct changes in AUD prevalence estimates resulting from each specific criteria change have also been examined in a handful of studies. The removal of the legal problems’ criterion is expected to have little impact on the prevalence estimates because studies have found that this item has a low rate of endorsement and contributes little information beyond the other AUD criteria [24,28]. This is consistent with data from NSDUH, which indicates that, although 0.8 percent (Table 2.3, weighted) of respondents endorsed the legal criterion item in NSDUH, only 0.1 percent of respondents endorsed only the legal criterion (additional tables providing weighted Ns can be found in Appendix A, and corresponding prevalence estimates among past year substance using respondents as opposed to all respondents that are presented below are in Appendix B).
Alcohol Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Preliminary evidence suggests that the addition of the craving item will increase the population prevalence of DSM-5 AUD. Table 2.4 contains a summary of the craving symptom endorsement prevalence reported in several studies. Estimates for the alcohol-craving symptom endorsement ranged from 1.3 to 7.0 percent for past year endorsement, but there was a much larger range for lifetime craving, which depending on the sample was from 3.5 percent in a nationally representative sample and 26 percent in a high-risk sample [13,25,29–31]. In the previously discussed study of primarily substance users, Peer et al., examined the causes of DSM-IV and DSM-5 criteria status changes on AUD estimates and found that only 1 percent of status changes were caused by the addition of the craving criterion [25]. However, the results of this nonnationally representative study must be interpreted with caution based on the oversampling of substance-dependent participants, which may not generalize to a population-based sample. Agrawal and colleagues’ analysis of the wave 2 NESARC data (a more generalized sample) suggested a slightly larger impact resulting from the addition of the craving criterion [24]. Of the 3,026 people who endorsed only 1 of the 10 DSM-5 criteria (excluding craving), 4.1 percent (unweighted, less than 1 percent of the total population) met DSM-5 criteria resulting from endorsement of the craving item.
Prevalence of DSM-5 Alcohol Use Disorder Craving Criterion from Available Studies.
Preliminary evidence suggests that the addition of the craving item will increase the population prevalence of DSM-5 AUD. Table 2.4 contains a summary of the craving symptom endorsement prevalence reported in several studies. Estimates for the alcohol-craving symptom endorsement ranged from 1.3 to 7.0 percent for past year endorsement, but there was a much larger range for lifetime craving, which depending on the sample was from 3.5 percent in a nationally representative sample and 26 percent in a high-risk sample [13,25,29–31]. In the previously discussed study of primarily substance users, Peer et al., examined the causes of DSM-IV and DSM-5 criteria status changes on AUD estimates and found that only 1 percent of status changes were caused by the addition of the craving criterion [25]. However, the results of this nonnationally representative study must be interpreted with caution based on the oversampling of substance-dependent participants, which may not generalize to a population-based sample. Agrawal and colleagues’ analysis of the wave 2 NESARC data (a more generalized sample) suggested a slightly larger impact resulting from the addition of the craving criterion [24]. Of the 3,026 people who endorsed only 1 of the 10 DSM-5 criteria (excluding craving), 4.1 percent (unweighted, less than 1 percent of the total population) met DSM-5 criteria resulting from endorsement of the craving item.
Evaluating response patterns from NSDUH provides some information regarding the impact of AUD criteria changes on prevalence estimates. Current estimates in NSDUH indicate that from 2002 to 2012 the past year prevalence of DSM-IV AUD was 7.4 percent weighted (Table 2.5). If the NSDUH diagnostic algorithm was altered to exclude the legal criterion and change the threshold to 2 or more symptoms (out of 10), without including the new craving criterion then the prevalence estimate for past year AUD would increase to 8.5 percent (a 1.1 percent increase). This increase was seen across all subgroups, including by sex, race, and age. This increase in prevalence is being driven by diagnostic orphans who endorsed two dependence criteria but no abuse criteria, and therefore meet DSM-5 but not DSM-IV criteria. NSDUH estimates suggest that 2.4 percent of the population were diagnostic orphans, which has implications both for the prevalence estimates of AUD, as well as “who” is meeting diagnostic criteria and their associated characteristics.
Alcohol Use Disorder among People Aged 12 or Older under DSM-IV and DSM-5 Criteria, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
The above change in diagnostic algorithm would introduce the possibility of misclassification as a result of not including the craving criterion, which would lead to an underestimate of disorder prevalence. In evaluating the prevalence of single criterion endorsement and who may have endorsed the missing craving criterion thereby being at risk for misclassification, NSDUH data indicate that up to 8.7 percent of the population is at risk for misclassification. Moreover, the prevalence of endorsing one AUD criterion was not equal across groups, suggesting a higher probability for misclassification among males and people aged 18 to 35, in particular. Analyses showed that tolerance (3.3 percent) and spending a great deal of time in activities necessary to obtain, use, or recover from alcohol (2.8 percent) were the two criteria most likely to be endorsed alone (Table 2.6), and based on data from Hasin and colleagues [22], tolerance is highly correlated with craving.
Respondents Who Endorsed Only One Alcohol Use Disorder Criterion among People Aged 12 or Older, by Criterion and Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
NSDUH and MHSS do not assess caffeine use; therefore, changes in diagnostic criteria are not discussed in depth. However, some changes between DSM-IV and DSM-5 are noteworthy. Caffeine use disorder is a new addition to DSM-5. DSM-IV diagnoses included only caffeine intoxication, caffeine-induced disorders (anxiety and sleep), and caffeine-related disorder not otherwise specified. First, caffeine withdrawal syndrome has been added as a substance-related and addictive disorder in DSM-5. Second, caffeine use disorder has been added to DSM-5 Section 3, Conditions for Further Study. The addition of caffeine use disorder to Conditions for Further Study was based on preliminary evidence indicating the clinical significance of withdrawal and dependence and concern over a rise in case reports of fatalities due to caffeine intoxication from energy drinks and diet pills, as well as concerns over the safety of caffeine–alcohol combination beverages that have drawn U.S. Food and Drug Administration (FDA) attention [13,29,32,33].
The DSM-IV assessed cannabis abuse and cannabis dependence, but no withdrawal syndrome was specified for dependence diagnosis [1]. This was based upon a lack of data identifying withdrawal symptoms. Since the DSM-IV’s initial publication, however, research has identified symptoms of withdrawal that produce clinically significant impairment, and the revised DSM-5 recognizes these symptoms [34]. Criteria for DSM-5 cannabis withdrawal symptoms include three or more symptoms occurring within approximately 1 week of cessation of heavy and prolonged cannabis use, including irritability/anger/aggression; nervousness/anxiety; sleep difficulty, such as insomnia or disturbing dreams; decreased appetite or weight loss; restlessness; depressed mood; and at least one physical symptom that causes significant discomfort: abdominal pain, shakiness/tremors, sweating, fever, chills, or headache (Table 2.2).
Data from wave 1 of NESARC indicates that approximately 34.4 percent (weighted) of frequent cannabis using adults (≥ 3 times per week, n=2,613) reported three or more symptoms of cannabis withdrawal in their lifetime [34]. However, in the previously mentioned study of a sample of primarily substance users, Peer et al., reported a lower prevalence of lifetime cannabis withdrawal (18 percent) [25]. Only one study assessed the prevalence of past year cannabis use. Data from the NLAES, a household-based study of 42,862 adults in the United States, found that 7.4 percent of past year cannabis users met criteria for cannabis withdrawal [30].
Along with the addition of cannabis withdrawal in DSM-5, cannabis use disorder also underwent changes in diagnostic threshold, merging of cannabis abuse and dependence, removal of the legal criterion, and the addition of the craving criterion. Published comparisons suggest little overall difference in the prevalence of DSM-IV and DSM-5 cannabis use disorder despite criteria changes. In a sample of a primarily substance-dependent population, the prevalence of DSM-IV lifetime cannabis use disorder was 39.4 percent and for DSM-5 criteria was 41.0 percent [25]. In a more generalized sample in Australia, the prevalence of past year cannabis use disorder was lower using DSM-5 criteria (5.4 percent) compared with DSM-IV criteria (6.2 percent) [35].
The removal of the legal criterion is expected to have little effect on prevalence estimates of cannabis use disorder. The removal of this criterion was based on low endorsement rates and because repeated studies found that the criterion provided little information beyond what was captured by other criteria [17,35,36]. Examination of NSDUH data indicated an estimated 0.2 percent of respondents endorsed the legal criterion and less than 0.1 percent (Table 2.7, weighted) only endorsed the legal criterion.
Marijuana Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
The impact of craving on DSM-5 cannabis use disorder prevalence is poorly understood because of a paucity of research. No studies were identified that quantified the number of people meeting criteria only because they endorsed craving. The reduction in prevalence discussed above suggests that the impact of adding the craving criterion on prevalence estimates will be minor; however, the missing craving criteria may affect NSDUH prevalence estimates because of the fewer number of criteria available to meet requirements. Estimates from Compton et al. put the prevalence of past year craving of cannabis at approximately 13.9 percent [30]; however, studies of lifetime endorsement range from 2.5 to 18 percent, demonstrating an imprecision among existing studies (Table 2.8) [12,37]. Calculating the impact on NSDUH estimates of cannabis use disorder, specifically, is complicated by the fact that NSDUH does not assess cravings or the new withdrawal syndrome, which introduces two points of uncertainty leading to an inability to draw inferences with any degree of certainty.
Prevalence of DSM-5 Cannabis Use Disorder Craving Criterion from Available Studies.
Disorder criteria for phencyclidine (PCP) use disorder and other hallucinogen use disorder vary slightly from the overall SUD template in that a withdrawal syndrome has not been identified for these substances. This is consistent in DSM-IV and DSM-5 and with NSDUH assessment. DSM-IV had separate diagnostic subcategories for phencyclidine use disorder, hallucinogen use disorder, and their respective substance-related diagnoses; however, the DSM-5 includes phencyclidine use disorder, other hallucinogen use disorders, and their respective substance-related disorders under one diagnostic subcategory, hallucinogen-related disorders. NSDUH combines phencyclidine and other hallucinogenic substance use into one assessment and diagnosis, which is more closely aligned with DSM-5’s subcategory classification (hallucinogen-related disorders) than the DSM-IV; however, phencyclidine and other hallucinogens have separate SUD diagnoses. In order to be aligned with DSM-5, the NSDUH would need to assess phencyclidine use disorder separately from other hallucinogen use disorder. Alternatively, a more precise use of terminology in NSDUH reports and related documentation could be used to match the DSM-5 usage of the phrases “phencyclidine use disorder or other hallucinogen use disorder” or “any hallucinogen-related substance use disorders” to clarify which substances are included in the category.
The other diagnostic changes (i.e., combining abuse and dependence, changing the diagnostic threshold, and dropping and adding criteria) will still affect the estimates for the hallucinogen-related disorders; however, little data are available to help quantify the impact on current estimates. Consistent with other substances, removal of the legal criterion is expected to have little impact on the estimates of this disorder [38,39]. This was confirmed using data from NSDUH, which indicated that less than 0.1 percent (Table 2.9, weighted) of respondents endorsed the legal criterion item, and so few respondents endorsed only the legal criterion that the data are suppressed.
Phencyclidine or Other Hallucinogen Use Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Other information on the prevalence rates of the DSM-5 phencyclidine use disorder and other hallucinogen use disorder and the frequencies of craving endorsement for these two diagnoses have not been published. There is an indication that the DSM-IV hallucinogen use disorder diagnostic criteria produce more diagnostic orphans (i.e., individuals who endorse one or two dependence criteria but no abuse criteria) than do most of the other substances. Lynskey and Agrawal examined the prevalence of diagnostic orphans among the NESARC sample and found that 1.28 percent of people who had used phencyclidine or other hallucinogens in their lifetime would be considered diagnostic orphans (those meeting two dependence criteria and no abuse criteria), suggesting that the new criteria would increase the prevalence estimates for these two diagnoses [39].
Analyses of NSDUH data suggest a slight increase in the prevalence of phencyclidine and other hallucinogen use disorder. Between 2002 and 2012 an annual average of 0.2 percent of NSDUH respondents (weighted) met DSM-IV criteria (Table 2.10). Excluding the legal problems criterion from the diagnostic algorithm and changing the threshold to two or more symptoms indicates a similar percentage in the prevalence among NSDUH respondents (0.2 percent). However, this does not include the possibility of misclassification caused by the missing craving criterion. An estimated 0.2 percent of NSDUH respondents endorse one of the nine criteria and are therefore at risk for misclassification. Adolescents and adults aged 16 to 25, American Indian or Alaska Native adults, and adults of two or more races had the highest prevalence of single criterion endorsement. Tolerance was the most frequently endorsed single criteria (0.1 percent; Table 2.11).
Phencyclidine Use Disorder or Other Hallucinogen Use Disorder among People Aged 12 or Older under DSM-IV and DSM-5 Criteria, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Respondents Who Endorsed Only One Phencyclidine Use Disorder or Other Hallucinogen Use Disorder Criterion among People Aged 12 or Older, by Criterion and Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs. (more...)
Inhalant use disorder criteria are similar to DSM-IV and DSM-5 SUD templates with slight variation. DSM-IV and DSM-5 do not include a withdrawal syndrome as a dependence/SUD criterion. Changes from DSM-IV to DSM-5 include dropping the legal problems criterion, adding the craving criterion, and combining diagnoses for inhalant abuse and inhalant dependence into one diagnosis (inhalant use disorder) with the new threshold (e.g., 2 or more of 10).
To date, very little has been published on the impact of criteria changes on prevalence estimates of inhalant use disorder. One study using wave 1 data from NESARC found that similar to studies of other substances, a single dimensional construct of inhalant use disorder was a better fit to the underlying construct of inhalant addiction than the separate diagnostic categories of abuse and dependence [38]. Moreover, studies using both clinical treatment and community samples of adolescents and adults have repeatedly found that the legal criterion had low endorsement and added little additional information to the measurement of this disorder [39–41]. Data from NSDUH indicate that less than 0.1 percent of the population endorsed the legal criterion (Table 2.12).
Inhalant Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Speculating on what little data are available, combining the inhalant abuse and inhalant dependence criteria under one diagnosis will likely increase the prevalence of inhalant use disorder slightly over prevalence estimates based on the DSM-IV since more of the diagnostic orphans (those meeting two dependence criteria and no abuse criteria) are classified as cases under DSM-5. Past year prevalence estimates in NSDUH suggest an annual average of 0.1 percent of respondents (weighted) met DSM-IV criteria and a similar percent endorsed two or more of the nine criteria assessed in NSDUH (excluding legal criterion and craving). It is difficult to quantify the impact of adding the craving criterion because frequency estimates are not available at this time. However, data from the existing NSDUH instrument suggests that 0.1 percent of NSDUH respondents endorse only one of the symptom criteria and therefore up to 0.1 percent might be reclassified by endorsement of the additional craving criterion (Table 2.13). The potential for misclassification was not equal across all subgroups, with youth aged 12 to 17 having the highest rates of single criterion endorsement. Sample size estimates limit the evaluation of individual criterion endorsement, but tolerance appears to be the criterion most likely to be endorsed alone (Table 2.14).
Inhalant Use Disorder among People Aged 12 or Older under DSM-IV and DSM-5 Criteria, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Respondents Who Endorsed Only One Inhalant Use Disorder Criterion among People Aged 12 or Older, by Criterion and Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Opioid use disorder is an addiction to a class of chemicals that activate the body’s opioid receptors. This includes drugs like heroin and opioid-based prescription pain relievers (e.g., hydrocodone, morphine, and codeine). Changes in the DSM criteria for opioid use disorder follow the overall criteria for SUDs. Summarizing the impact of these changes on population prevalence at the population level are difficult because many studies present statistics for a particular substance (e.g., heroin or oxycodone) and these changes may not be applicable to the overall DSM category of opioid use disorder. In addition, NSDUH varies from DSM-IV and DSM-5 by assessing abuse and dependence separately for heroin and prescription pain relievers (which are primarily, but not completely opioid based due to an open ended response option). This makes generalizing findings from the heroin literature on NSDUH easier, but complicates the evaluation of DSM changes on prescription pain relievers.
Consistent with studies on other substances, studies of opioid abuse and dependence in both community and clinical samples consistently find that a single construct of addiction that combines abuse and dependence criteria is a better fit in measuring the underlying construct of opioid addiction, rather than a separate and hierarchical construct of abuse and dependence [22,36,42–44]. This supports the unification of abuse and dependence under DSM-5. However, the impact on opioid use disorder estimates of combining abuse and dependence into a single disorder, changing the diagnostic threshold, and replacing the legal criterion with craving is not entirely clear. One small study using a sample of 705 chronic pain patients suggested a 2 percent decrease in the prevalence of opioid use disorder under DSM-5 criteria compared with that found under DSM-IV criteria [45]. However, Peer et al. in their study consisting of primarily participants with a substance use disorder suggested almost no difference in the prevalence of DSM-IV and DSM-5 opioid use disorder (34.7 and 34.8 percent, respectively) [25]. No studies examining heroin use disorder and prescription opioid use disorder separately were identified, and studies using samples closer to the NSDUH sampling frame are not available.
Data on the removal of the legal criterion are conflicting. Studies consistently report that it is the least frequently endorsed criterion; however, the rate of endorsement varies by study population and substance. Hasin et al. examined a substance-using sample recruited from inpatient hospitals, outpatient clinics, and methadone clinics and found that 7.97 percent of the 364 patients (a mixture of inpatient and outpatient participants) who had used heroin in the past 12 months endorsed the legal problem criterion [22]. A similar lifetime prevalence (8 percent) was found among the sample of substance-dependent participants examined by Peer et al. [25]. The next lowest rate of endorsement was for hazardous use, which was endorsed by 30.22 percent of past year users. However, Shand et al. examined a sample of opioid-dependent adults (any type) participating in opioid pharmacotherapy clinics and found that, although still the least frequently endorsed criterion, it was endorsed by 73 percent of the 1,511 participants [42]. In a more generalizable study using NESARC data, 2.26 percent of opioid-ever users endorsed the legal problems’ criterion [36]. Unfortunately, none of these studies reported on the number of participants who met criteria only through their endorsement of the legal criterion. Data from NSDUH indicates that less than 0.1 percent of respondents endorsed the legal criterion for heroin use (Table 2.15, weighted), only 0.1 percent endorsed the legal criterion for pain relievers (Table 2.16, weighted), and only 0.1 percent (Table 2.17, weighted) endorsed the legal criterion for either heroin or pain relievers when the data were combined.
Heroin Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Pain Reliever Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Heroin/Pain Reliever Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
The impact of adding the craving criterion is as yet undetermined. Two studies were identified that examined the prevalence of endorsement of the craving criterion (Table 2.18). In the Peer et al. study, 29.0 percent of adults who had used opioids in their lifetime endorsed the craving criterion. In a more generalizable sample using data from the nationally representative NLAES, past year craving was endorsed by 7.8 percent of past year opioid users [30]. No studies were identified that examined craving separately for heroin and prescription pain relievers, making the comparison more difficult.
Prevalence of DSM-5 Opioid Use Disorder Craving Criterion from Available Studies.
Existing NSDUH data suggest that the prevalence of heroin use disorder will change little between DSM-IV and DSM-5 criteria (Table 2.19). DSM-IV diagnosed heroin use disorder among NSDUH respondents was estimated at 0.1 percent annually from 2002 to 2012 (weighted). The same prevalence estimate was found when the legal criterion was excluded and the threshold was set at two or more symptoms. Furthermore, less than 0.1 percent of participants (actual estimate suppressed) only endorsed one criterion (primarily withdrawal only; Table 2.20) suggesting that the possibility of misclassification from the missing craving criterion is very low.
Heroin Use Disorder among People Aged 12 or Older under DSM-IV and DSM-5 Criteria, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Respondents Who Endorsed Only One Heroin Use Disorder Criterion among People Aged 12 or Older, by Criterion and Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Data from NSDUH examining prescription pain relievers show a different pattern of responses. The prevalence of DSM-IV defined prescription pain reliever use disorder was 0.7 percent in the population (Table 2.21, weighted). The prevalence under DSM-5 criteria, excluding the craving item, was slightly higher at 0.8 percent. One percent of respondents endorsed only one criterion and therefore were at risk of being misclassified without the craving criterion. This risk was not equivalent across groups. American Indian or Alaska Native respondents and those between the ages of 16 to 25 had the highest rates of single criterion endorsement. The most commonly endorsed single criterion was withdrawal (0.5 percent) followed by tolerance (0.3 percent; Table 2.22).
Pain Reliever Use Disorder among People Aged 12 or Older under DSM-IV and DSM-5 Criteria, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Respondents Who Endorsed Only One Pain Reliever Use Disorder Criterion among People Aged 12 or Older, by Criterion and Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
To mimic the DSM-5 criteria of opioid use disorder, diagnostic criteria were combined to generate a combined diagnosis. Data were combined at the criterion level. For example, if an individual endorsed hazardous use for heroin but not for pain relievers, the combined hazardous use criterion was considered met. The diagnostic algorithm was then run on the combined criteria; one or more abuse criteria (including the legal criterion) or three or more dependence criteria were considered DSM-IV heroin/pain reliever use disorder (this is not referred to as opioid use disorder because of the small possibility for endorsement of nonopioid pain relievers). Two or more of any of the combined criteria (excluding the legal criterion) was considered DSM-5 heroin/pain reliever use disorder. Weighted prevalence estimates for the combined diagnoses among NSDUH respondents indicated an increase from DSM-IV to DSM-5 criteria (0.8 to 0.9 percent; Table 2.23). Approximately 0.6 percent of the population had one of the combined criteria. The most frequently reported single criterion was withdrawal (0.5 percent) followed by tolerance (0.3 percent; Table 2.24).
Heroin/Pain Reliever Use Disorder among People Aged 12 or Older under DSM-IV and DSM-5 Criteria, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Respondents Who Endorsed Only One Heroin/Pain Reliever Use Disorder Criterion among People Aged 12 or Older, by Criterion and Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
DSM-IV groups sedatives, hypnotics, and anxiolytics into the same group for the purposes of assessing this SUD. Sedative is a general term for any substance that has a calming or sleep-inducing effect, including anxiolytics and hypnotics. The difference in terms between anxiolytics and hypnotics is based on purpose: hypnotics are sleep aids and anxiolytics are antianxiety medications. However, there is considerable overlap in function across substances (i.e., a substance can be both an anxiolytic and hypnotic) because they all work to activate GABA (gamma-aminobutyric acid) receptors in the brain. Currently, NSDUH assesses sedatives and tranquilizers. However, “tranquilizer” is not a term used in the DSM-IV or DSM-5. Tranquilizers are central nervous system depressant drugs classified as sedative-hypnotics, and are therefore captured under the broader category of sedatives/anxiolytic/hypnotics in the DSM.
Sedative, hypnotic, and anxiolytic use disorder underwent all of the criteria changes from DSM-IV to DSM-5 that have been previously discussed, including adding craving, removing the legal problems criterion, merging the abuse and dependence categories, and changing the threshold. However, like inhalant use disorder, phencyclidine use disorder, and other hallucinogen use disorder, there has been little review of the impact of criteria changes on generating prevalence estimates based on the DSM-5 compared with the DSM-IV [13]. No published estimates were identified for the prevalence of sedative, hypnotic, or anxiolytic use disorder based on the DSM-5. No studies were identified that assessed the frequency of the craving criterion for the sedative, hypnotic, or anxiolytic use disorder. The removal of the legal criteria is expected to have limited impact on DSM-IV to DSM-5 changes in estimates [36,39,46]. Data from NSDUH indicate that less than 0.1 percent of respondents endorsed the legal criterion for sedatives (Table 2.25, weighted), and less than 0.1 percent of the population met the legal criterion for tranquilizers (Table 2.26). Even when combined (Table 2.27), the endorsement of the legal criterion for sedative or tranquilizer abuse is less than 0.1 percent.
Sedative Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Tranquilizer Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Sedative/Tranquilizer Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
NSDUH’s tranquilizer abuse and dependence module does not currently assess tranquilizer withdrawal symptoms. This may lead to an underestimate of DSM-IV sedative/hypnotic/anxiolytic dependence and DSM-5 sedative/hypnotic/anxiolytic use disorder. Withdrawal symptoms for tranquilizers are the same as those assessed for other sedatives assessed in NSDUH. Therefore, the same questions used to assess withdrawal in the sedative module could be added to the tranquilizer module to align the assessment or tranquilizer dependence and tranquilizer use disorder with DSM criteria. This would marginally add to the assessment length and may disrupt trend estimates, but would improve face validity with the DSM.
NSDUH currently assesses sedatives and tranquilizers separately, whereas the DSM makes no diagnostic distinction between the two, which means that NSDUH’s reporting of sedative, hypnotic, or anxiolytic use disorder does not directly map onto the DSM categories. Although this would disrupt current trend estimates, the sedatives and tranquilizer modules could be combined to increase concordance with DSM and also reduce participant burden. Alternatively, future NSDUH-based publications could report the combined prevalence to facilitate comparison with other research if tranquilizer withdrawal were added to the assessment.
The impact of adding the craving criterion to the DSM-5 criteria for sedatives, hypnotics, or anxiolytics is unclear. There is a paucity of estimates on craving endorsement for sedatives, hypnotics, or anxiolytics. However, DSM-IV–defined sedative use disorder criteria are met by an estimated 0.1 percent of the population (Table 2.28) and a similar percentage of respondents met DSM-5 criteria, excluding the craving criterion. Approximately 0.1 percent of respondents endorsed only one criterion, suggesting the risk of misclassification because of the missing craving criterion is only 0.1 percent at most. Moreover, the risk of misclassification was largely similar across all subgroups. The withdrawal criterion was the most frequently endorsed solitary criterion (0.1 percent; Table 2.29). Data are unavailable for tranquilizers specifically since the lack of tranquilizer withdrawal makes estimates of the craving impact unreliable (as there are multiple points of uncertainty).
Sedative Use Disorder among People Aged 12 or Older under DSM-IV and DSM-5 Criteria, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Respondents Who Endorsed Only One Sedative Use Disorder Criterion among People Aged 12 or Older, by Criterion and Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
The DSM-IV to DSM-5 criteria changes for stimulant use disorder followed the overall SUD template changes with one additional change—the amphetamine use disorder and the cocaine use disorder (including crack use) were combined into a single stimulant use disorder diagnosis in DSM-5. This change was made based on the similarity of effect for amphetamines and cocaine [2]. NSDUH does not follow the DSM-IV criteria precisely in its assessment of stimulants. NSDUH collects data on cocaine, crack, and other stimulants in three separate modules (methamphetamine, although not a prescription drug, is assessed in the prescription drug stimulant module). In order to produce DSM-5 consistent estimates, NSDUH would need to either combine the cocaine and stimulant (which includes amphetamines) modules or provide separate and combined estimates for these substances in future publications. This would provide additional comparison options for researchers and provide data for service providers who may need estimates based on the DSM-5 for planning purposes.
The impacts of specific diagnostic changes have been examined in a few published studies, and the changes in population prevalence estimates have been examined in a few studies of cocaine addiction only. In a study conducted among a sample of recently incarcerated correctional inmates, past year cocaine use disorder prevalence based on the DSM-IV was 12.7 percent, whereas the prevalence using DSM-5 criteria was 11.0 percent [47]. Notably, about 50 percent of inmates who met the DSM-IV criteria for cocaine abuse did not meet the DSM-5 criteria for cocaine use disorder [48]. The prevalence of this disorder was largely stable as a result of capturing more diagnostic orphans (individuals who met one or two dependence criteria and no abuse criteria). However, it is unclear how consistent these results will be for other stimulant types.
As was found with other substances, the combination of abuse and dependence into a single diagnosis was demonstrated to be a more accurate measurement of the underlying construct of stimulant addiction [22,36,39]. The legal criterion was found to add little diagnostic information and had a low rate of endorsement; thus, its removal is expected to have little impact on prevalence estimates [22,36]. Data from NSDUH suggests that less than 0.1 percent of respondents endorsed the legal criterion for stimulants (Table 2.30, weighted), 0.1 percent endorsed the legal criterion for cocaine (Table 2.31, weighted), and 0.1 percent of the population endorsed the legal criterion for either stimulants or cocaine when combined (Table 2.32).
Stimulant Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Cocaine Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Stimulant/Cocaine Legal Criterion Endorsement among People Aged 12 or Older, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Data examining the impact of the craving criterion addition are only available for cocaine. Peer et al. reported a lifetime prevalence of cocaine craving of 45 percent among adults in a high-risk, largely substance-using sample [25]. Compton et al. reported a past year craving prevalence among a nationally representative sample of 27 percent among adults who had used cocaine in the past year [30] (Table 2.33).
Prevalence of DSM-5 Cocaine Use Disorder Craving Criterion from Available Studies.
Data from NSDUH indicate that there will be an increase in the prevalence of stimulant use disorder under DSM-5 criteria. The annual average weighted prevalence of past year DSM-IV–defined stimulant use disorder was 0.2 percent (Table 2.34). Under DSM-5 criteria, excluding the craving criterion, the prevalence of stimulant use disorder was similar to DSM-IV and 0.1 percent of respondents (weighted) endorsed only one criterion; therefore, the risk of from the missing craving criterion is up to 0.1 percent. Withdrawal (0.1 percent) was the most frequently endorsed single criterion (Table 2.35).
Stimulant Use Disorder among People Aged 12 or Older under DSM-IV and DSM-5 Criteria, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Respondents Who Endorsed Only One Stimulant Use Disorder Criterion among People Aged 12 or Older, by Criterion and Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
The data for cocaine use disorder differed from that of stimulants. The prevalence for past year DSM-IV was slightly lower than DSM-5 cocaine use disorder, excluding the craving criterion (0.5 percent and 0.6 percent, respectively; Table 2.36). Up to 0.2 percent of respondents were at risk for misclassification (primarily from withdrawal endorsement, Table 2.37). Combining NSDUH criteria data for stimulant and cocaine, to closer approximate the DSM-5 combined category indicated a slight increase in past year prevalence between DSM-IV and DSM-5 (0.6 to 0.7, Table 2.38) and a 0.3 prevalence of single criterion endorsement among respondents. The most frequently endorsed single criterion for combined stimulants and cocaine was withdrawal (Table 2.39).
Cocaine Use Disorder among People Aged 12 or Older under DSM-IV and DSM-5 Criteria, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Respondents Who Endorsed Only One Cocaine Use Disorder Criterion among People Aged 12 or Older, by Criterion and Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Stimulant/Cocaine Use Disorder among People Aged 12 or Older under DSM-IV and DSM-5 Criteria, by Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Respondents Who Endorsed Only One Stimulant/Cocaine Use Disorder Criterion among People Aged 12 or Older, by Criterion and Demographic Characteristic: Weighted Percentages, Annual Averages Based on 2002–2012 NSDUHs.
Significant changes have occurred to diagnostic criteria for nicotine or tobacco use disorders. DSM-IV–only assessed nicotine dependence and alternate scales, such as the Fagerström Nicotine Dependence Scale [12], were frequently used to assess nicotine dependence in some studies [37]. However, based on evidence that abuse criteria, such as use in hazardous situations (e.g., smoking in bed) and continued use despite problems (e.g., inability to work in nonsmoking positions), are increasingly relevant because of an increase in laws prohibiting smoking in public places, the DSM-5 has brought criteria for nicotine addiction in line with other SUDs. The DSM-5 refers to nicotine dependence as tobacco use disorder to acknowledge the departure from the former dependence diagnosis.
NSDUH currently uses two nicotine dependence scales among respondents who smoked in the past month—the Fagerström Nicotine Dependence Scale [12] and the Nicotine Dependence Syndrome Scale [11]. These scales, although correlated with DSM-IV–defined dependence, do not match DSM-IV criteria precisely and make it difficult to determine the impact of criteria changes on NSDUH. Tests of the expanded DSM-5 criteria indicate a higher prevalence of tobacco use disorder than nicotine dependence as defined by the DSM-IV; therefore, without changes, NSDUH will likely underestimate the prevalence of tobacco use disorder in the general population [37]. The amount of misclassification is difficult to quantify and may not be equivalent across population groups. A small study of adolescents and young adults who had smoked cigarettes in the past year found that the prevalence of tobacco use disorder was double (68.7 percent) the prevalence of DSM-IV nicotine dependence (33.0 percent) among adolescents (aged 14 to 18) and the prevalence of tobacco use disorder (86.0 percent) was over a third greater than DSM-IV nicotine dependence (59.6 percent) among young adults (aged 18 to 35) [49]. To collect DSM-5 nicotine use disorder, NSDUH would need to add the three “abuse” criteria (i.e., hazardous use, failure to fulfill obligations due to use, and continued use despite social problems cause by use), modify the existing questions to better align with DSM-5 “dependence” symptoms (e.g., spending a great deal of time obtaining nicotine), and assess substance “abuse and dependence” of other forms of nicotine (e.g., chew, snuff).
Two overarching limitations are apparent in the currently available literature on the DSM-IV to DSM-5 conversion. First, there is very little information available on measurement issues related to the assessment of the new DSM-5–defined SUDs. Reliability is the ability of any measure to produce consistent results, and multiple studies of DSM-IV diagnoses found that measures of substance abuse were consistently less reliable than measures of substance dependence [50,51]. This suggests that the criteria for abuse were more difficult and less consistent to measure than the dependence criteria. Reliability can be worse if the criteria for a disorder do not clearly reflect the underlying construct. Therefore, combining the abuse and dependence diagnoses may have a bearing on instrument reliability. On one hand, reliability may be increased as a result of diagnostic criteria that more accurately measure the underlying construct. However, combining the abuse criteria with the dependence criteria and changing the threshold for diagnosis may introduce additional variability to the measure, which could result in lower instrument reliability. Currently, there is little research assessing this aspect of the impact of diagnostic changes with one exception. Data examining the reliability of AUD assessment suggests that reliability may be in between that of alcohol abuse and alcohol dependence; however, this study did not use fully standardized assessments, which may increase reliability substantially [52]. Reliability is important since an unreliable measure threatens the validity of estimates made using the measure. It is possible that previously established reliability of the NSDUH instrument may not be the same if modifications are made to increase concordance with DSM-5 criteria and the concern over the overall reliability of criteria may make it more difficult to establish reliability of the revised instrument.
The second limitation is that the majority of studies on the DSM-IV to DSM-5 conversion have been conducted among adults. Few studies were identified that considered the impact among adolescents, and no studies were identified that examined younger children. In a review of the limited literature, Winters concluded that merging abuse and dependence into a single disorder diagnosis and the elimination of the legal criterion would improve the validity of the diagnosis for adolescent substance use; however, changing the threshold for diagnostic classification may lead to a higher rate of false positives among this population because of the overlap of some criteria (e.g., tolerance) with a maturation effect [53]. Moreover, no studies were identified that evaluated how the diagnostic changes might affect prevalence estimates in this population [53].
The changes for the DSM-IV to the DSM-5 were designed to facilitate an immediate transition for clinicians and insurers. Phasing out the DSM-IV began immediately after the release of the DSM-5, with the American Psychiatric Association’s expectation for insurance industry conversion to be completed in early 2014 [54]. As the transition progresses, the field (insurers, clinicians, and researchers) will move away from DSM-IV diagnostic criteria. This transition will be further solidified by the federally mandated transition from ICD-9-CM to ICD-10-CM in 2015 because a DSM-5 to ICD-10 crosswalk has been included in DSM-5 (something not available in DSM-IV). The changes in the field coupled with the fact that revisions to DSM-5 are meant to reflect advancements in scientific knowledge, mean that continued use of DSM-IV diagnostic criteria will introduce questions regarding the face validity of NSDUH. This supports transitioning of NSDUH to better reflect DSM-5 criteria. However, modifications to the instrument should be considered in conjunction with their impact on trend estimates, respondent burden, questionnaire design issues, and stakeholder interests.
To balance these concerns, changes to NSDUH can be considered at two levels. Changes could be made at the analytic level (i.e., changes only during the estimation process with no changes to the questionnaire), which would not introduce context effects and would potentially enable estimation of estimates under differing definitions but would be insufficient to meet all of the DSM-5 revisions. Alternatively, changes could be made to the questionnaire, which introduces the possibility that it may no longer be able to be used to compute DSM-IV–based SUD estimates and therefore affect the ability to assess trends. To balance the needs for trend preservation with data validity, potential changes for alcohol and other illicit drug use modules are discussed below. No changes to the NSDUH tobacco use module are suggested, despite a deviation from DSM criteria, because the NSDUH already uses the most frequently used measure of cigarette dependence in the research field.
Several changes in alcohol and illicit drug use disorder algorithms could be made that would improve alignment with DSM-5, although these changes alone would not be sufficient to align NSDUH with DSM-5. These changes involve creating new algorithms to:
Changes to the NSDUH questionnaire designed to better align it with DSM-5 criteria may include:
In the current NSDUH questionnaire, the stimulants module includes questions on methamphetamine use. Beginning in 2015, questions on methamphetamine use will comprise a separate module in the NSDUH questionnaire.
A “prodrome” is “an early or premonitory symptom of a disease,” according to Stedman’s Medical Dictionary, 27th edition [14].
All material appearing in this report is in the public domain and may be reproduced or copied without permission from SAMHSA. Citation of the source is appreciated. However, this publication may not be reproduced or distributed for a fee without the specific, written authorization of the Office of Communications, SAMHSA, HHS.
Your browsing activity is empty.
Activity recording is turned off.
See more...
