Comparison of DSM-IV and DSM-5 criteria for alcohol use disorders in VA primary care patients with frequent heavy drinking enrolled in a trial

The Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5) criteria for alcohol use disorders (AUD), released in 2013 [1], represent a significant departure from previous criteria. For the previous 20 years, since the 4th edition of the DSM (DSM-IV), alcohol dependence and abuse had been considered mutually exclusive diagnoses that together made up alcohol use disorders [2]. The diagnosis of these distinct disorders was based on “a maladaptive pattern of alcohol use leading to clinically significant impairment or distress” as manifested by separate criteria, with DSM-IV dependence requiring at least 3 of 7 criteria and DSM-IV alcohol abuse requiring exclusion of DSM-IV dependence and at least 1 of 4 separate criteria (Fig. 1). In part because of recent studies calling into question the hierarchical distinction between abuse and dependence [3‐5], DSM-5 replaces these two diagnoses with a single spectrum of AUD with a continuum of severity. Removal of “abuse” from DSM-5 may also serve to reduce the stigma and negative judgment associated with such terminology [6]. The DSM-5 includes 11 criteria: 10 of the 11 combined DSM-IV abuse and dependence criteria (excluding legal problems) and a new criterion for craving (Fig. 1) [7]. DSM-5 further specifies AUD severity as mild, moderate or severe based on the number of diagnostic criteria endorsed, with at least 2 of 11 criteria required for a diagnosis (mild = 2–3, moderate = 4–5, severe ≥6). DSM-5 AUD was designed to reduce the number of “diagnostic orphans” that occurred with DSM-IV, whereby patients with 1 or 2 dependence criteria (and no abuse criteria) did not meet diagnostic criteria for abuse or dependence [8, 9]. Diagnostic orphans have an increased risk of developing a DSM-IV alcohol use disorder compared to those with no AUD symptoms in certain populations [e.g. young adults] and are therefore an important group to identify [10]. However, those who had DSM-IV abuse based on 1 criterion will not meet criteria for DSM-5 AUD [11]. The design of the DSM-5 AUD criteria is such that all people who met criteria for DSM-IV dependence but only some people with DSM-IV abuse will have an AUD based on DSM-5 criteria, while additional patients with only 2 symptoms of dependence will meet criteria.

DSM-5 criteria were intended to result in an overall prevalence of AUD similar to the prevalence of AUD determined from DSM-IV criteria [8]. A recent review found 12 studies that compared the prevalence of AUD based on DSM-IV and DSM-5 criteria [12]. Seven of the studies showed an increase in prevalence of AUD based on DSM-5 compared to DSM-IV, and the authors of the review concluded that DSM-5 “inflated” rates of AUD [12]. However, only one of the studies reviewed included a medical sample—patients seeking care in an emergency department. In contrast, the prevalence of AUD using DSM-5 criteria was found to be lower compared to DSM-IV criteria among high-risk Swiss young men [13] and in a large cross-national sample from the World Health Organization’s World Mental Health Survey Initiative [14]. None of these studies were restricted to primary care patients and clinical samples could differ (e.g. have more severe AUD) which could decrease the impact of the shift from DSM-IV to DSM-5.

Even if a difference in the prevalence of DSM-IV and DSM-5 AUD is not observed in other settings, patients identified by the two diagnostic criteria could have different demographic or clinical profiles. Little research has focused on demographic and clinical differences between patients who meet DSM-IV and DSM-5 criteria for AUD in clinical settings. Both the change in number of diagnostic criteria required, and the change from “legal problems” to “craving,” could result in differences in the clinical characteristics of the populations of patients diagnosed with DSM-5 AUD and DSM-IV AUD.

The purpose of the present study was to compare the characteristics of patients diagnosed with AUD according to DSM-IV and DSM-5 criteria in a clinical sample of US Veterans who received general medical care (e.g. “primary care”) in Veterans Affairs (VA) clinics and reported frequent heavy drinking and were recruited into a trial of alcohol-related care management. We first describe the prevalence of DSM-IV and DSM-5 AUD, and categorize patients into four mutually exclusive groups: those who met criteria for: (1) neither DSM-IV nor DSM-5 AUD; (2) DSM-IV AUD alone; (3) DSM-5 AUD alone; or (4) both DSM-IV and DSM-5 AUD. We then compare sociodemographic characteristics, alcohol and drug related problem severity, mental health comorbidity, and readiness to change drinking across the four mutually exclusive diagnostic groups.

The Considering Healthier Drinking Options in primary CarE (aka “CHOICE”) trial was a randomized controlled trial of a 12 months nurse-led care management intervention for patients with frequent heavy drinking who were receiving medical care in a general practice clinic for US Veterans of military service. Analyses presented here are from the baseline interviews (December 2011–September 2014) that assessed both DSM-IV and DSM-5 alcohol use disorders. The study was approved by both the VA Puget Sound and the Group Health Institutional Review Boards.

In this sample of VA primary care patients at high risk for AUD due to frequent binge drinking, DSM-5 AUD criteria identified about 13% more patients with AUD than DSM-IV criteria. Almost three quarters of patients met both DSM-IV and DSM-5 criteria for AUD and 14% met criteria for neither, but only one patient out of 304 met criteria for DSM-IV AUD but not DSM-5 AUD. Moreover, in this sample of primary care patients recruited for a trial of care management for patients at high risk for AUD, there were marked differences in the clinical characteristics across the four groups. Those who met both DSM-IV and DSM-5 AUD criteria not only had more negative consequences due to drinking, but also more mental health and drug use symptoms and greater reported readiness to change compared with those who met either DSM-IV or DSM-5 AUD alone, and those who met neither DSM-IV nor DSM-5 criteria for AUD.

A unique strength of this study is that it compared DSM-IV and DSM-5 criteria in a clinical sample of heavy drinkers. This is an important population to study because it is these patients in whom assessments for AUD are used clinically. It is not possible to determine the extent to which the higher prevalence of AUD using DSM-5 criteria (85.9%) compared to DSM-IV criteria (73.3%) is due to the specific characteristics of the sample we studied—primary care patients who reported frequent binge drinking—or if this result would be found in more general clinical populations. However, this study’s finding that a sizable group of patients (13%) met criteria for DSM-5 AUD alone is consistent with 7 of 12 recent studies comparing the prevalence of AUD based on DSM-IV and DSM-5 criteria in mostly non-clinical settings [12]. The fact that only one patient met criteria for DSM-IV AUD alone might be somewhat surprising. One might have expected more than 1 patient with frequent binge drinking to meet the DSM-IV AUD criteria with just 1 DSM-IV abuse criterion or 2 DSM-IV criteria that included legal consequences, which was excluded from DSM-5. However, drinking 5 or more drinks on an occasion is strongly associated with AUD symptoms [3], and thus restriction of our sample to frequent binge drinking could account for this finding. Further research is needed on other clinical samples of patients with frequent binge drinking to understand if this unexpected finding was unique to our study sample.

Patients meeting criteria for both DSM-IV and DSM-5 AUD had relatively high rates of mental health symptoms, drug use disorders, and recent negative consequences due to drinking. Moreover, almost three quarters of these patients who met both DSM-IV and DSM-5 criteria were somewhat or very interested in changing their drinking. In contrast, patients who met criteria for DSM-5 AUD alone, who by definition met criteria for mild DSM-5 AUD, tended to report fewer negative consequences due to drinking, less tobacco use and drug use disorders, and fewer symptoms of anxiety and depression. This is an expected finding since DSM-5 was designed to better capture less severe forms of AUD and to have dimensionality such that participants with fewer DSM criteria have a milder AUD [8‐10]. Further, patients who met criteria for DSM-5 AUD alone had less readiness to change compared to patients who met both DSM-IV and DSM-5 criteria, despite having similar demographic characteristics. This association between alcohol-related symptom burden and readiness to change has been previously described [28], but the finding that DSM-5 identifies more of these patients with milder alcohol-related symptoms who report less readiness to change may have clinical implications for primary care. Given the lower burden of mental health symptoms reported by patients with DSM-5 AUD alone, and their greater confidence in their ability to change their drinking, patients meeting DSM-5 criteria alone may be more likely to respond to brief opportunistic interventions in primary care compared to those who meet criteria for both DSM-IV and DSM-5 AUD. On the other hand, these patients tended to report less treatment-seeking and placing less importance on changing their drinking than those meeting both DSM-IV and DSM-5 AUD criteria and they were less likely to report that they were ready to change. These characteristics may make patients meeting criteria for DSM-5 alone more successful at self-change, but more challenging to engage in opportunistic primary care interventions—like the CHOICE trial intervention–which seek to engage patients repeatedly over time and/or interest them in medications for AUD [29, 30]. Future research will be needed to assess whether the effectiveness of the CHOICE intervention differed for those who met criteria for DSM-5 AUD alone vs both DSM-IV and DSM-5 AUD.

In this sample of primary care patients who reported frequent binge drinking and enrolled in a trial, 14% did not meet criteria for DSM-IV or DSM-5 AUD. To our knowledge, little research has focused on this group. These patients had similar rates of mental health co-morbidity compared to those with an AUD based on DSM-5 criteria alone. However, they tended to be older and reported a mean of 4 of 15 negative consequences due to drinking in their lifetimes, suggesting that they may have had prior AUD now in remission. Patients with frequent binge drinking who do not meet DSM criteria for AUD on average respond to opportunistic brief interventions [31], but little is known about whether older patients who have previously experienced negative consequences due to drinking and may have AUD in remission respond to brief interventions. Future research is also needed to evaluate whether alcohol medications can decrease drinking and future negative consequences related to drinking for this group.

This study has several important limitations. First, the study sample reflects VA primary care patients with frequent binge drinking who were willing to enroll in a trial in which they would be offered alcohol-related services, limiting generalizability. The prevalence of AUD based on DSM-IV and DSM-5 is likely to be markedly lower in general primary care populations, and may differ in younger and non-research populations. It is also possible that those patients meeting criteria for DSM-IV AUD alone (1 DSM-IV abuse criterion or 2 DSM-IV criteria including legal consequences) were less likely to participate in a research study on alcohol and therefore were not represented in our sample. For example, this could disproportionately impact racial minorities who may be more likely to have legal consequences of drinking [32], but less likely to participate in trials. The recruited sample was also 90% men, who tend have higher rates of AUD than women, and the sample had a relatively high prevalence of mental health symptoms and drug use disorders.

Several limitations of this study’s measures should also be noted. The DSM-IV skip pattern limited our ability to measure DSM-5 AUD severity in those with DSM-IV alcohol dependence, because the interview skip pattern that did not assess the 4 questions about abuse in patients with dependence. In addition, we used a craving question with a lifetime timeframe. However, sensitivity analyses determined that the impact of the “ever” time frame of the craving question had minimal impact on our findings (Additional file 1, Additional file 2). This descriptive study also lacked a priori hypotheses about differences across the four mutually exclusive groups and did not have power to test multiple comparisons. Additional studies of the prevalence and severity of DSM-5 AUD and associated mental health comorbidity and readiness to change in primary care patients with frequent binge drinking will be important (Additional file 3).

Despite the above limitations, these findings may be pertinent to patients with frequent binge drinking in the VA, the largest integrated health system in the US with over 900 clinical sites nationwide. As above, this is also the first study to compare DSM-IV and DSM-5 criteria in a primary care population to our knowledge. Additionally, primary care patients with frequent binge drinking are an important clinical sample in which to study DSM diagnostic criteria. Most brief alcohol screens assess frequent or recent heavy drinking and a follow-up diagnostic assessment for AUD can aid treatment decisions (i.e. brief intervention vs. medication) [33]. Further, results of this study highlight the spectrum and complexity of patients with frequent binge drinking who might be open to receiving alcohol-related care in their primary care clinic.

In sum, in this study of VA primary care patients with frequent binge drinking, 13% of patients met DSM-5 AUD criteria but did not meet DSM-IV AUD criteria. These patients appeared to have fewer negative consequences due to drinking and less other substance use and mental health comorbidity, than patients who met criteria for both DSM-IV and DSM-5 AUD. If findings are corroborated by future primary care studies, developing and testing optimal approaches to managing these primary care patients with less complex AUD may be important.