1. Background

Alcohol Use Disorders (AUD) contribute significantly to the burden of mental health disorders [1]. Studies have also shown that individuals diagnosed with AUD commonly suffer from comorbid mental disorders [2–5], further adding to the overall burden. Disability-Adjusted Life Years (DALYs) attributed to alcohol have increased by more than 25% between 1990 and 2016, accounting for 4.2% and 5.2% of the total DALYs and deaths, respectively [6]. While AUD is more common in developed regions in the world, the potential harm is much higher in developing countries such as the Eastern Mediterranean region (EMR), where national alcohol harm reduction policies remain either absent or poorly enforced [7].

Alcohol drinking often begins in adolescence, as early as 13 years of age [8, 9]. Early initiation is particularly concerning because it has been linked to a higher likelihood of developing AUD later in life [10–12]. While rates of youth alcohol use seem to have stabilized in the U.S. [13, 14], rates continue to rise in developing countries [9]. This increase in alcohol use and its associated harms are closely linked to the globalization of economic markets with transnational alcohol companies emerging in poorer countries [15, 16].

For almost two decades (1993–2014), the Diagnostic and Statistical Manual of Mental Disorders 4^th Edition (DSM-IV) [17, 18] had helped clinicians to classify and screen for alcohol abuse and alcohol dependence and had helped researchers systematically screen for alcohol-related problems in the community [19–23]. The release of DSM-5 in 2013 brought some important changes to AUD terminology and classification criteria. Mainly, the diagnosis shifted from dichotomized measures of alcohol abuse (one out of four symptom criteria required) and alcohol dependence (three out of seven symptom criteria required) in DSM-IV, to a spectrum of DSM-5 AUD ranging from mild (2 to 3 out of eleven criteria endorsed) to severe (6 or more criteria); presence of two or more criteria indicates presence of any DSM-5 AUD (Table 1) [17, 18]. Changes were also made to the criteria: “legal problems due to alcohol use” was dropped, and “craving” was added [17, 18]. In a recent systematic review, the authors concluded that DSM-5 generally inflates the AUD prevalence when compared to DSM-IV [24], although some studies did report a lower DSM-5 AUD prevalence rate compared to the combined DSM-IV alcohol abuse/dependence rate [25, 26]. This evidence is largely based on adult surveys [24], with only three studies focused on college students or youth, which concluded that AUD prevalence seem to be higher using DSM-5 diagnostic criteria [27–29].

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Criteria used to screen for DSM-IV abuse/dependence and DSM-5 AUD.

https://doi.org/10.1371/journal.pone.0233657.t001

In addition to generating inconsistent prevalence estimates, the change from DSM-IV to DSM-5 has not fully addressed the issue of “diagnostic orphans.” The term ‘diagnostic orphans’ (DO) was coined by Kaczynski & Martin (1995) to describe a group of individuals with one or two DSM-IV alcohol dependence criteria and no alcohol abuse criteria, thus receiving no formal diagnosis [30–32]. DOs have therefore constituted a hidden and overlooked group of problem alcohol drinkers (i.e. drinkers with health or social problems related to their alcohol use) who are at a high risk of progressing to AUD [30]. In a large cross-national community study of the World Mental Health Surveys in nine countries, of which Iraq was the only Arab country, DSM-IV DOs made up 5.3% of the total sample of 44,341 respondents aged between 18 and 97 years [26]. Among young people, previous studies have shown considerable proportions of DSM-IV DOs ranging between 10% and 30% [28, 32–35].

By re-classifying some of the DSM-IV DOs as having “mild AUD,” the DSM-5 revisions partially improve detection of problem alcohol drinkers. Yet, individuals who fulfill only one of the eleven DSM-5 criteria continue being undetected and thus can be labeled as ‘DOs' [30]. This subgroup may still include problem drinkers who, despite not fulfilling DSM criteria for an AUD [30], may eventually inflict harm on themselves and/or others, if left undetected [35]. In the U.S., DSM-5 criteria DOs were shown to consume more alcohol and have more substance-related problems than those who met none of the DSM-5 criteria [28].

A clear research gap still exists in understanding how DSM changes have affected AUD prevalence in young people outside of the Americas, and in understanding the implications of these changes on measuring DOs specifically. AUD among college youth is of particular public health concern since it can have detrimental effects, including poor academic performance and dropout, unintentional injuries, health problems, heightened risk of engaging in other risky practices, legal problems, increased risk of other substance use disorders in adulthood, and death [36]. Therefore, accurate screening of AUD in college samples is critical for more adequate planning and implementation of early, evidence-based interventions to prevent morbidity and mortality [37].

Using a college sample from Greater Beirut, Lebanon, this study aims: (1) to compare the past-year DSM-IV and DSM-5 prevalence in a college sample of past-year drinkers; and (2) to examine the extent of DOs, and to compare their substance use profiles (i) to drinkers with no DSM-5 AUD criteria endorsed and (ii) to drinkers with 2 or more AUD criteria endorsed (i.e. having any AUD). The findings will add to the limited literature on this topic among college youth and address the gap on this issue within the developing world.

2. Methods

3. Results

3.2 AUD prevalence estimates as per DSM- IV and DSM-5 criteria

The 13 questions (mapped onto the 12 AUD criteria) are presented in Table 1. Among past-year drinkers, 62% (n = 329) were screened to have DSM-IV alcohol abuse (≥1 criteria), and 65% (n = 377) to have alcohol abuse or dependence. Using DSM-5 criteria, 35% (n = 203) of the past-year drinkers were screened to have any AUD: 20% had mild AUD, 7% moderate AUD and 7% severe AUD. Overall percent agreement was 68% (kappa = 0.41) indicating only a fair agreement between DSM-IV alcohol abuse/dependence and DSM-5 AUD.

Table 2 presents the results of a cross-tabulation between DSM-IV and DSM-5 categories of AUD; the percentage of students who fulfilled DSM-IV categories of AUD (no abuse/no dependence, abuse/no dependence; abuse or dependence) within DSM-5 AUD categories are described. We note that of the 203 students who scored positive on DSM-5 AUD, the overwhelming majority (97.04%) also screened positive for DSM-IV abuse or dependence. Within the 379 past-year drinkers who were screened negative on DSM-5 defined AUD, almost half (47%) were screened positive for DSM-IV abuse or dependence (driven by the abuse classification) (Table 2- Part A). Past-year drinkers with no DSM-5 AUD may have endorsed zero or one DSM-5 criteria for AUD. As shown in Table 2 (Part B), among those with zero-endorsed DSM-5 criteria, about 99% also did not fulfill DSM-IV-defined abuse or dependence; the exception was two students that endorsed the DSM-IV legal problems criterion, which is no longer a criterion in DSM-5. However, of the 191 alcohol-drinking students who endorsed only one of the eleven DSM-5 AUD criteria (i.e. DSM-5 DOs), 93% (n = 178) screened positive for DSM-IV abuse or dependence (driven by alcohol abuse).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Comparing estimates of DSM-IV abuse/dependence and DSM-5 AUD among 582 past-year alcohol drinking university students from Greater Beirut, Lebanon.

https://doi.org/10.1371/journal.pone.0233657.t002

3.3 Characteristics of the DSM-5 Diagnostic Orphans (DOs) using DSM-5 criteria

Table 3 describes the distribution of each of the 11 DSM-5 AUD criteria in the total sample of past-year drinkers and within subcategories of DSM-5 AUD: zero-endorsed DSM-5 criteria (AUD = 0), DOs with only one DSM-5 criterion endorsed (i.e. AUD = 1), and past-year drinkers who screened positive for DSM-5 AUD (i.e. AUD = 2+). Focusing on the DOs (i.e. AUD = 1), we note that the most commonly reported DSM-5 AUD criterion was ‘physically hazardous alcohol use’ (91.62%, n = 175), which reflects “recurrent alcohol use in situations in which it is physically hazardous even though actual harm might not have occurred.” Only five (2.62%) of DSM-5 DOs met the DSM-IV criterion of ‘legal problems’, which was dropped in DSM-5.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Distribution of DSM-5 AUD criteria in the total sample of 582 past-year alcohol drinkers from Greater Beirut, Lebanon, and within subcategories of DSM-5 defined AUD.

https://doi.org/10.1371/journal.pone.0233657.t003

Table 4 compares the socio-demographic and substance use characteristics of drinkers with only one DSM-5 criterion endorsed (AUD = 1) against those of drinkers with zero-endorsed DSM-5 criteria (AUD = 0) and two or more endorsed DSM-5 criteria (AUD = 2+). Compared to past-year drinkers with zero-endorsed DSM-5 AUD criteria, DOs were more likely to be males, aged 21 years or above, and less likely to perceive their SES as poorer than vs. the same as others. DOs were also more likely to report drinking alcohol 1–2 times/week rather than 1–3 times/month or less, and to report past-year cigarette smoking, compared to past-year drinkers with zero-endorsed DSM-5 criteria. Compared to past-year drinkers with any DSM-5 AUD (i.e. AUD = 2+), DOs were more likely to be aged 21 years or above, and to report living with their parents for most of the time. DSM-5 DOs, as compared to those with any DSM-5 AUD, were less likely to drink frequently (3–4 times per week or more vs. 1–3 times per month or less) and to report waterpipe smoking (Table 4).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Distribution of DSM-5 AUD and AUD categories by socio-demographic and substance use characteristics of 582 past-year alcohol drinking university students from Greater Beirut, Lebanon.

https://doi.org/10.1371/journal.pone.0233657.t004

Multivariable regression analyses (Table 4) included all correlates that had a p-value < 0.05 at the bivariate level (i.e. gender, age, pursuing degree, living with parents most of the time, perceived SES, frequency of drinking, past-year cigarette smoking, past-year waterpipe smoking, and past-year marijuana use). Specifically, DOs (compared to past-year drinkers with zero-endorsed DSM-5 AUD criteria) were more likely to be aged 21 years or above vs. <21 [OR = 1.57(1.21–2.03)] and less likely to perceive their SES as poorer vs. same as others [OR = 0.17(0.07–0.43)]. Quite importantly, they were also more likely to drink alcohol 1–2 times/week vs. less frequently (≤3 times per month) [OR = 2.24(1.44–3.49)], and more likely to report past-year cigarette smoking [OR = 2.16(1.10–4.24)]. Compared to past-year drinkers with DSM-5 AUD, DOs were more likely to be pursuing a graduate or medical degree vs. undergraduate degree [2.06 (1.09–3.89)] more likely to be living with parents most of the time [OR = 2.68(1.14–6.31)], less likely to drink at a high frequency (3–4 times /week or more) than low (≤3 times per month) [OR = 0.15(0.05–0.48)], and less likely to report past-year waterpipe smoking [OR = 0.54(0.34–0.85)] but more likely to report past-year marijuana use [1.89(1.10–3.23)].

4. Discussion

In this college student sample of past-year drinkers from Lebanon, almost twice as many students met DSM-IV abuse or dependence criteria than DSM-5 AUD criteria (65% vs. 35%, respectively). This fair/moderate agreement (kappa = 0.41) was lower than that reported in other populations: 0.6 in college student samples [31], between 0.79 and 0.90 in clinical/institutional samples, and between 0.50 and 0.85 in non-clinical samples [24]. At face value, our findings do not corroborate recent conclusions of a systematic review that DSM-5 diagnostic criteria inflate prevalence rates of AUD compared with DSM-IV [24]. In the review, the authors explained the DSM-5 inflated estimates mainly by the higher number of DSM-IV ‘diagnostic orphans’ (i.e. 1 or 2 DSM-IV dependence criteria) compared to DSM-IV single-criterion alcohol abuse individuals [24]. In our study, the scenario was the opposite. Results showed a higher percentage of drinkers who met DSM-IV criteria for abuse only (62%) than those who screened positive for 1 or 2 DSM-IV dependence criteria only (i.e. DOs), which explains the inflated percentage of DSM-IV abuse or dependence in our study.

The percentage of DSM-5 DOs in our sample [mean age 21.36 ± 1.83 years] of past-year drinkers (33%) was slightly higher than that found (24%) in a U.S.- based college sample [mean age 20.11 ± 1.79 years] [28]. Compared to those with zero-endorsed DSM-5 criteria, DSM-5 DOs were more likely to report past-year daily cigarette use, any past-year marijuana use, and to drink alcohol more frequently. This finding is consistent with Hagman and colleagues’ representation of DOs as a ‘distinct group of drinkers’ who engage in risky levels of alcohol and other drug use [28], and also supports published literature on possible co-occurrence of risk [47, 48].

The discordance in estimates between DSM-IV and DSM-5 criteria in this college sample was found to be due to the high percentage of DSM-IV alcohol abuse, driven namely by the hazardous alcohol use sub-criterion: driving while impaired (i.e. ‘operated a car or other vehicle shortly after you had had several drinks of alcohol’). Thus, youth whose only alcohol-related problem is driving under alcohol influence would fulfil DSM-IV alcohol abuse criteria, but not meet the threshold for any DSM-5 AUD. Using DSM-5 AUD classification may therefore result in a “missed opportunity” to detect drinkers with a potentially serious alcohol-related problem. The extent of reported hazardous alcohol use has been linked to the level of enforcement of drink driving laws [49–51]. Therefore, this “missed opportunity” may be more common in countries such as Lebanon, where drink-driving regulations are loosely enforced [7]. Critiques of the DSM-IV drink driving criterion suggest it tends to potentially over-diagnose AUD; in one study, nearly 70% of surveyed adults met criteria for alcohol abuse solely on the basis of the drink-driving variable [52]. Indeed, further analyses of our data confirmed that the DSM-IV abuse/dependence prevalence would be reduced by half had those who scored positive on ‘hazardous use’ actually scored negative. Some researchers have questioned the validity of the AUD diagnosis that relies on the ‘hazardous use’ indicator given its high prevalence rate [50, 53]. Whether it is truly that prevalent or a methodological artifact caused by age-related misinterpretation of classification criteria has also been investigated, and authors have shown that the hazardous use (and tolerance criteria) are communicated adequately and understood as intended by young adults [54]. While Slade at al. (2013) demonstrated that young adults do understand and adequately interpret the interview questions designed to operationalize the hazardous use criterion, there still remains ongoing debate regarding the validity and utility of this criterion, and some researchers have proposed that this criterion be removed from the DSM-5 diagnostic criteria for AUD [49, 51]. Nonetheless, others have suggested that the criterion could reflect a psychiatric disorder rather than a sociocultural factor, and thus should be considered as an appropriate diagnostic criterion [55].

The findings of the present study should be interpreted in light of some limitations. The self-reported data may be subject to recall bias or under-reporting, since alcohol consumption in the Middle Eastern region is considered a sensitive topic; still, the anonymous nature of the survey is likely to have reduced such information bias. The prevalence estimates of DSM-IV or DSM-5 AUD results, or the profile of the DOs, may not be generalizable to young drinkers in other universities or even non-college youth drinkers given the convenient nature of the sampling. Still, the internal validity of the results is likely not to be affected given that our primary objective relied on comparing the responses of the same students on various criteria, and selection bias is unlikely to affect the comparative validity of student responses. Post-hoc adjustments for multiple comparisons were also not conducted as we believed that the process oversimplifies a complex issue in hypothesis testing; also, in the case of exploratory studies with clear pre-planned hypotheses, a strict adjustment for multiple comparisons is less critical [56].

Overall, this study adds to the scant literature comparing DSM-IV versus DSM-5 diagnoses of AUD among young college drinkers and is, to our knowledge, the only research evidence to date on youth from a developing country. The study illustrates the potential for high agreement between the two criteria (DSM-IV and DSM-5) if it were not for the ‘hazardous alcohol use’ criterion, which may be more commonly reported in contexts where drink driving is an issue due to lax laws and enforcement. Unlike in DSM-IV, where a young person can screen positive for alcohol abuse by endorsing only one alcohol-related problem (e.g., hazardous use), the DSM-5 AUD diagnosis based on the endorsement of two or more criteria could miss detecting drinkers with a DUI (drinking under the influence) problem. While DUI alone is not necessarily an indication of an AUD, it can be a warning sign of preventable harm to oneself and to others. One concern is that DSM-5 AUD classification might miss young people in high-risk groups, such as repeat DUI offenders, that DSM-IV alcohol abuse would have captured, which may result in fewer people receiving any attention or care for their recidivism. Targeted educational programs or other community prevention programs may not suffice for this group of recidivists [57]. While DOs would need early detection and management, population-based and evidence-informed public health prevention measures and adequately enforced legislations such as tighter drink-driving regulations remain key for preventing DUI [58, 59], and reducing the risk of “over-pathologizing” human behaviors [60]. From a population-based epidemiological perspective where the identification of high-risk groups does not necessarily mean their clinical management, our findings on the profile of DOs highlight the importance of primary and secondary preventive programs and the effective enforcement of drink-driving laws.

Separate from the important methodological analysis, the high prevalence of AUD among our convenient sample of young college drinkers from Greater Beirut, Lebanon is concerning especially coupled with our knowledge of the high alcohol outlet density in areas surrounding schools and universities [61]. The assessment of DSM-5 AUD in college studies is not common. The majority of studies have employed the Alcohol Use Disorders Identification Test (AUDIT), a screening tool that has detected prevalence estimates as high as 66.4% in Ireland, 68.9% in the US, or as low as 27.5% and 6.9% respectively among male and female college students in Tanzania [62, 63]. Only two studies have assessed DSM-5 AUD among college samples and reported lower past year prevalence estimates of 27.7% and 50% in the southeastern and northeastern parts of the U.S., respectively [28, 31].

5. Conclusion

Early detection of problem drinking among college students is a critical component of any national alcohol harm-reduction strategy. In a country like Lebanon, where laws governing DUI are loosely enforced, DSM-IV alcohol abuse/dependence may be higher than DSM-5 AUD prevalence driven by the high endorsement of the ‘hazardous alcohol use’ criterion. Still, the latter may provide a more nuanced picture of potential harm from alcohol use among young persons in countries where alcohol policies are loosely enforced. Young people screened as DOs may have been assessed during their journey from non-drinking/occasional drinking to drinking frequently/developing an AUD. Indeed, the DSM-5 DOs in our sample, the majority of whom had reported driving while impaired, were more likely to use other substances compared to drinkers with zero-endorsed DSM-5 criteria, and less likely compared to drinkers with any DSM-5 AUD. More generally, a national alcohol-harm reduction strategy that includes proper screening and evidence-informed primary/secondary prevention efforts is critically needed to protect young persons from alcohol-related harms and to ensure that additional alcohol-related problems do not develop among DOs.

Acknowledgments

Authors would like thank Dr. Shannon Lea Watkins for her thorough and valuable editorial review of the manuscript.