Changes over time in endorsement of 11 DSM-IV alcohol use disorder (AUD) criteria in young adults with persistent or recurrent AUD in the Collaborative Study on the Genetics of Alcoholism

Abstract

Background:

Endorsement of specific DSM-IV alcohol use disorder (AUD) criteria items changed significantly over time in men in their thirties with persistent or recurrent AUD. Few studies have documented whether endorsement of AUD items change over time in younger individuals or in women. The prospective analyses presented here evaluate changes in endorsement of AUD criteria in 377 men and women with persistent or recurrent AUD during their twenties.

Methods:

AUD item endorsement over time between average ages of 20 and 25 in 223 men and 154 women with persistent or recurrent AUD in at least three interviews were available from participants in the Collaborative Study on the Genetics of Alcoholism. Statistical significance of endorsement changes over time were evaluated using related-samples Cochran’s Q for the full sample and for men and women separately. Additional analyses evaluated potential sex differences in the patterns of change.

Results:

In the full sample, the predominant pattern was for significant increased rates of endorsements for six of the seven alcohol dependence criteria, but not in the four abuse items. A similar pattern was seen within men, but women demonstrated significant changes in only three of the seven dependence criteria.

Conclusions:

Endorsement of the seven alcohol dependence criteria during the twenties generally increased in individuals with persistent or recurrent AUD, but few changes were observed regarding rates of endorsement of the four abuse items. The paper discusses how the results might reflect on the nature of AUD and on the DSM criteria.

Introduction

Criteria in the Diagnostic and Statistical Manuals (DSMs) of the American Psychiatric Association were created to help clinicians make data-based decisions regarding whether treatment is needed for a patient and to find the therapeutic approach with the best asset to liability ratio (Goodwin and Guze, 1996; Hasin et al., 2013; Schuckit et al., 1995). Those criteria for a disorder include multiple items that are written in general terms rather than only a few detailed and precise descriptions of the conditions involved (Hasin et al., 2013; Schuckit, 2013; Schuckit and Saunders, 2006). This more generic approach in DSM allows clinicians the flexibility required for the application of diagnostic criteria to patients of different ages, across the sexes, and from different cultural backgrounds. In the field of substance use disorders, this generic approach to diagnostic criteria facilitated the development of a relatively easy to remember single set of substance use disorder criteria to be applied across 10 different categories of substances, including alcohol (i.e., individuals with alcohol use disorder [AUD]) (Grant et al., 2017; Kendler et al., 2015; Nolen-Hoeksma and Hilt 2006; Salvatore et al., 2017; Selgem et al., 2016). Rates of item endorsement of specific AUD criteria items have been shown to change with age in older men with AUD (Schuckit and Smith, 2021). The analyses of prospective data presented below evaluate changes over time in diagnostic item endorsement in a large group of men and women in their twenties with persistent or recurrent DSM-IV AUD (American Psychiatric Association, 1994).

Regarding age, as originally predicted by Jellinek (1952), retrospective comparisons across older and younger individuals with AUD demonstrated different ages of onset for different alcohol related problems (e.g., Schuckit et al., 1993, 1995). These studies indicated higher levels of endorsement of tolerance to alcohol and of using alcohol in hazardous situations in younger versus older drinkers, and higher rates of experiencing alcohol withdrawal syndromes in older versus younger individuals with alcohol problems (Buu et al., 2012; Harford et al., 2005; Marmet et al., 2019). However, most such studies were retrospective and compared younger individuals with older individuals, while only a few papers have reported data from prospective follow-ups that documented changes in problem patterns over time in the same individuals (Jacob et al., 2009; Sloan et al., 2011; Verges et al., 2021).

Regarding differences in problem endorsement across the sexes, men are more likely than women to report experiencing withdrawal symptoms (Deshmulsh et al., 2003), but women might have higher rates of some alcohol-related physiological and psychological problems (Ceylan-Isik et al., 2010; Edens et al., 2008; Grant et al., 2017; Karastergiou et al., 2012; Mann et al., 1992; Schuckit et al., 2016). In addition, women metabolize alcohol more slowly, have higher blood alcohol concentrations (BACs) per drink, are more sensitive to the effects of alcohol per drink and are likely to have fewer years between the onset of heavier drinking and entering treatment (Eng et al 1999; Kapoor et al., 2017; Schuckit et al., 2019). Despite those differences, the outcomes of AUD following treatment in the two sexes appear to be similar (Green et al., 2002; Timko et al., 2002). Regarding sex or gender, most studies cited in this literature review did not distinguish between biological sex (male or female) and gender roles (men or women), but rather relied on whether a person listed themselves as a man or a woman. Therefore, in this review and in the original data described below, we use the terms man or woman generically throughout the manuscript without precision related to biological sex or gender roles.

To help expand our understanding of differences in endorsement of different problems with age and across men and women, our group recently published within-subjects prospective data regarding changes in rates of endorsement of specific AUD criteria in individuals with persistent or recurrent AUD (Schuckit and Smith, 2021). These data were extracted from the every five-year prospective evaluation of two generations of participants in the San Diego Prospective Study. That protocol began in 1978 with 453 drinking men who did not meet AUD criteria at an average age of about 20 and who were followed with personal interviews every 5 years up to an average age of about 60 (e.g., Schuckit and Smith, 2021). Follow-up data revealed 106 men who fulfilled AUD criteria during at least three of the five-year periods between average ages of 31 and 43. Consistent with DSM-IV guidelines, the diagnosis of dependence required endorsing problems associated in three or more of up to seven areas of life as described below. In the absence of dependence, abuse required endorsement of one or more of the four repetitive abuse items to the point of life impairment (American Psychiatric Association, 1994). In that study, during the thirties to early forties significant decreases over time were reported for tolerance (DSM-IV dependence item 1, or D1), withdrawal (D2), alcohol-related failure to meet obligations (abuse item 1, or A1), and use of alcohol in hazardous situations (A2). During that same period these men reported increases over time in endorsement of drinking higher amounts or for longer periods than intended (D3), spending a great deal of time involved with alcohol (D5) and continuing to use alcohol despite social or interpersonal problems (A4). Rates of endorsement did not change significantly over time for criteria related to an inability to stop or control alcohol intake (D4), giving up activities related to alcohol (D6), continuing to use alcohol despite physical or psychological problems (D7), or repetitive legal problems related to drinking (A3). In the San Diego study protocol no data were available on the one criterion found in DSM-5 but not DSM-IV, craving, and, thus, this criterion could not be tested in those analyses (Hasin et al., 2013).

San Diego Prospective Study data were also available from a small sample of 68 offspring (71% men) of probands who fulfilled criteria for an AUD during at least two evaluations between average ages of 21 and 27 (Schuckit and Smith, 2021). The small sample size and availability of only two follow-up data points contributed to lower statistical power for the analyses in this younger generation. Despite that limitation, during their twenties the offspring with persistent or recurrent AUDs demonstrated statistically significant increasing rates of endorsement for criterion D5 (spending much time regarding alcohol) and D6 (giving up activities because of alcohol). Non-significant patterns were also seen for decreases from 54.4% to 41.2% over time for the endorsement of tolerance (D1) (p=.12) as well as increases from 19.1% to 29.4% in impaired ability to decrease or control drinking (D4) (p=.15). Rates of endorsement did not change significantly for any abuse item for these participants. The relatively small sample hindered comparisons of changes in item endorsement over time across the sexes.

The primary goal of the current analyses was to evaluate prospectively measured levels of endorsement of specific AUD criteria early in the course of persistent or recurrent AUD in a relatively large sample of men and women. To accomplish this, we evaluated changes in endorsement of DSM-IV AUD criteria items across three timepoints during their twenties for 377 (59.1% men) Collaborative Study on the Genetics of Alcoholism (COGA) Prospective Study participants. The analyses tested four hypotheses. Hypothesis 1 predicted that the data will demonstrate significant increases over time in the two dependence items that increased significantly in the smaller sample of San Diego offspring: namely criteria D5 (spending a great deal of time regarding alcohol) and D6 (giving up important activities because of alcohol). The second hypothesis was that in their twenties the pattern of endorsement of the DSM-IV abuse items will not change significantly over time. Third, we predicted that in the current larger sample with three timepoints, these COGA Prospective Study subjects will demonstrate significant changes in endorsement of the two criteria with the most robust nonsignificant levels of change in the smaller San Diego sample, namely increases in impaired ability to decrease or control of drinking (D4) and decreases in endorsement of tolerance (D1). Fourth, based on findings from an earlier retrospective study of similar ages of onset of alcohol problems in men and women (Schuckit et al., 1995), Hypothesis 4 predicted that similar changes in endorsement of DSM criteria items will be seen for men and women.

Methods

Original COGA Recruitment and Evaluations

Beginning in 1989 and after receiving Human Subjects’ Protections Committees approvals, the original subjects (probands) were enrolled across COGA sites as individuals in treatment for AUD who also had multiple AUD relatives (Bucholz et al., 1994, 2017; Hesselbrock et al., 1999). Using different approaches across the centers, comparison families were recruited from dental and medical clinics, drivers’ license facilities, and university students. Once selected, all available biological relatives of the original probands and comparison families were interviewed.

COGA interviews used the Semi-Structured Assessment for the Genetics of Alcoholism (SSAGA) instrument (Bucholz et al., 1994; Hesselbrock et al., 1999). The data gathered included demographic characteristics, lifetime histories of alcohol and other drug use and related problems, additional mental health diagnoses, and family AUD histories. In all versions of this interview, a standard drink is defined as containing 10 to 12 grams of ethanol.

The COGA Next Generations Prospective Study

Beginning in 2004 and following additional Human Subjects’ Protections Committees approvals, the COGA Prospective Study began by enrolling 12- to 23-year-old grandchildren, nieces, nephews, and offspring of original subjects (Bucholz et al., 2017, de Viteri et al., 2019). These younger participants were reinterviewed about every two years using a SSAGA version that included gathering details regarding the 11 DSM-IV AUD criteria as well as the single new DSM-5 criterion of craving (American Psychiatric Association, 1994, 2013). DSM AUD questions first asked if a criterion item (described the same way as listed in DSM-IV) had ever occurred, followed by the ages of first and most recent experience with the item. Problems that occurred in the 24 months prior to the interview were noted as recent in determining which items were endorsed in each follow-up period.

Analyses were limited to individuals who met full DSM-IV AUD criteria for abuse or dependence during at least three evaluations separated by at least two years (defined as persistent or recurrent AUD). Time 1 was the first interview a participant with persistent or recurrent AUD met abuse or dependence criteria, which started that person’s “clock” for the second follow-up on average about 2 years later; then the next follow-up an average of about 2 years after that; and so on.

Ages at Time 1 for individuals in the current analyses ranged from 18 to 28 for the full sample, with 97.9% occurring between ages 18 and 25; Time 2 ages ranged from 19 to 31, with 95.2% between 19 and 27; and at Time 3 ages ranged from 21 to 36 with 94.7% age 30 or less. The three interviews occurred over three consecutive follow-ups in 81.4%; an additional 14.1% met full AUD diagnosis in three out of four consecutive interviews; with the remaining about 4% spread over five timepoints. Time 1 dependence was seen in 29.4% of these men and women, including 24.7% who demonstrated dependence during at least one additional follow up and 4.7% who demonstrated abuse during all follow-ups (totaling 29.4%). Time 1 abuse was seen in 70.6%, including 34.3% who remained abuse in both subsequent follow-ups and 36.3% who went on to dependence.

Time 1 data in Table 1 were extracted from the participants’ SSAGAs and AUD histories in fathers or mothers were determined from interviews with those parents, or, if a parent had not been interviewed, were extracted from reports about the paarents from other family members. Tables 2–4 list rates of endorsement for each DSM-IV AUD criteria over time. COGA also recorded rates of endorsements over time for the single DSM-5 AUD criterion not included in DSM-IV, craving. In relevant tables, data are reported for the full sample of 377 individuals with persistent or recurrent AUD and then separately for subsets of 223 men and 154 women with persistent or recurrent AUD.

Table 1.

Time 1 Demography and Alcohol Related Variables for 377 COGA Prospective Study Participants with Persistent or Recurrent DSM-IV AUD at Their First Interview with AUD

Time 1 Variables	All Ss % or Mean (SD)	Men % or Mean (SD)	Women % or Mean (SD)	X2 or F-test
	N=377	N=223	N=154
Demography
Men %	59.2	na	na
Age at First AUD Interview	20.3 (1.95)	20.2 (1.95)	20.4 (1.94)	1.31
Ethnicity %
European-American %	70.8	70.9	70.8	0.32
African-American	11.9	12.6	11.0
Hispanic	14.1	13.5	14.9
Other	3.2	3.1	3.2
Education (years)	12.7 (1.68)	12.6 (1.54)	12.8 (1.87)	1.02
Have a Parent with AUD %	76.9	74.4	80.5	1.90
Baseline Alcohol Related Variables
Alcohol Dependence % (2 year)	29.4	26.9	33.1	1.69
Max Drinks/Occasion (6 month)	11.7 (5.59)	13.2 (5.44)	9.5 (5.06)	29.58c
Usual Drinks/Occasion (6 month)	3.8 (3.88)	4.0 (4.00)	3.6 (3.70)	1.15
Number DSM-IV Criteria Endorsed (2 year)	3.4 (1.84)	3.2 (1.77)	3.6 (1.92)	3.91a

Abbreviations: COGA = Collaborative Studies on the Genetics of Alcoholism; AUD = Alcohol Use Disorder; “persistent or recurrent” defined as at least 3 interviews when met past 2-year AUD criteria; DSM-IV = Fourth Diagnostic and Statistical Manual of American Psychiatric Association; Standard deviations are in ( ).

Superscripts a: p < .05; b: c: p < .001

Table 2.

Past 2-Year Rates of Endorsements of the 11 DSM-IV Criteria Items (Plus Craving) and Alcohol Related Variables Over Time in 377 COGA Prospective Study Men and Women with Persistent or Recurrent AUD

DSM IV items as %	Time 1 % (SD) Age 20.1(1.95) Range 18-28	Time 2 % (SD) Age 22.9 (2.26) Range 19-31	Time 3 % (SD) Age 25.4 (2.65) Range 21-36	Statistics Overall	Time 1 vs.2	Time 1 vs. 3	Time 2 vs. 3
Dependence
D1 Tolerance	50.9	57.3	49.9	6.58a	1.97	0.33	−2.32
D2 Withdrawal	6.4	10.3	12.2	11.32b	2.24	3.29b	1.05
D3 Drink Higher Amounts or Longer	54.4	66.8	74.8	45.42c	4.08c	6.68c	2.60a
D4 Desire/Unable Decrease or Control	41.4	47.7	49.6	8.96a	2.21	2.85a	0.64
D5 Much Time to Get/Use/Recover	8.0	20.4	24.9	48.48c	4.94c	6.72c	1.78
D6 Decrease Activities Due to Alcohol	8.8	14.6	16.4	12.84c	2.60a	3.43b	0.83
D7 Use Despite Physical/Psychological Problems	19.4	23.1	24.1	3.97	na	na	na
D-X Craving	10.3	14.3	12.2	3.80	na	na	na
Abuse
A1 Failure to Fulfill Obligations	23.6	26.5	22.5	2.59	na	na	na
A2 Use in Hazardous Situations	70.8	75.6	76.7	5.32	na	na	na
A3 Recurrent Legal Problems	1.3	1.9	2.1	0.74	na	na	na
A4 Use Despite Social/Interpersonal Problems	50.7	52.0	47.5	2.04	na	na	na
Alcohol Related Variables
Maximum Drinks /Occasion (6 month)	11.7 (5.59)	12.1 (5.48)	11.3 (5.68)	3.58a	1.50	1.92	7.79b
Usual Drinks /Occasion (6 month)	3.8 (3.88)	3.2 (3.71)	2.8 (3.43)	4.86b	5.59a	8.54b	0.51
Number DSM-IV Criteria Endorsed (2 year)	3.4 (1.84)	4.0 (2.18)	4.0 (2.30)	17.49c	26.00c	23.07c	0.02

Definitions are the same as in Table 1; D1-D7 are the DSM-IV dependence criteria (D-X relates to craving) and A1-A4 are the abuse items in the order presented in that DSM; short definitions are used for each DSM-IV criterion; na = not applicable, since there was no overall (across 3 timepoints) significance; for overall statistics and pairwise comparisons changes in endorsement for each AUD criterion were evaluated by standard test statistic using Cochran’s Q Test.

Superscripts a: p < .05; b: p < .01; c: p < .001 [Note: within each item run, significance values were adjusted by the Bonferroni corrections]

Table 4.

Past 2-Year Rates of Endorsements of the 11 DSM-IV Criteria Items and Alcohol Related Variables Over Time In 154 COGA Prospective Study Women with Persistent or Recurrent AUD

	Time 1 % (SD) Age 20.4 (1.94) Range 18-28	Time 2 % (SD) Age 23.1 (2.31) Range 20-31	Time 3 % (SD) Age 25.7 (2.84) Range 22-36	Statistics Overall	Time 1 vs. 2	Time 1 vs. 3	Time 2 vs. 3
DSM IV items
Dependence
D1 Tolerance	50.6	53.2	35.1	13.76c	0.49	−2.94b	−3.43b
D2 Withdrawal	7.8	9.7	9.1	0.58	na	na	na
D3 Drink Higher Amounts or Longer	56.5	70.8	77.3	19.61c	2.98b	4.33c	1.35
D4 Desire/Unable Decrease or Control	47.4	49.4	48.7	0.19	na	na	na
D5 Much Time to Get/Use/Recover	9.1	19.5	23.4	15.22c	2.74a	3.77c	1.03
D6 Decrease Activities Due to Alcohol	7.8	13.6	14.3	4.44	na	na	na
D7 Use Despite Physical/Psychological Problems	24.7	21.4	21.4	0.98	na	na	na
D-X Craving	13.0	16.9	10.4	3.80	na	na	na
Abuse
A1 Failure to Fulfill Obligations	33.1	26.6	27.3	2.80	na	na	na
A2 Use in Hazardous Situations	64.9	67.5	70.8	1.71	na	na	na
A3 Recurrent Legal Problems	0.6	0.0	1.3	2.00	na	na	na
A4 Use Despite Social/Interpersonal Problems	55.2	48.7	44.2	4.92	na	na	na
Alcohol Related Variables
Maximum Drinks/Occasion (6 month)	9.5 (5.06)	10.3 (5.07)	9.3 (5.07)	2.35	na	na	na
Usual Drinks/Occasion (6 month)	3.6 (3.70)	2.9 (3.06)	2.1 (2.61)	5.06b	1.00	8.63b	4.39a
Number DSM-IV Criteria Endorsed (2 year)	3.6 (1.92)	3.8 (2.17)	3.7 (2.12)	0.56	na	na	na

Note: Definitions are given in Tables 1 and 2. For overall statistics and pairwise comparisons, changes in endorsement of each Alcohol Use Disorder criterion were evaluated by standardized test statistic using related samples Cochran’s Q Test.

a: p < .05; b: p < .01; c: p < .001 [Note: within each item run, significance values were adjusted by the Bonferroni corrections]

Data Analyses

Using the Statistical Package for the Social Sciences (IBM Corporation, 2021), statistical significance for differences across men and women in Table 1 were evaluated using the F-tests for continuous variables and chi-square for categorical items. Maximum likelihood procedures were used to address missing data (Collins et al., 2001), and Little’s MCAR test (Little, 1988) supported the conclusion that the data were missing completely at random (p = .27). Changes in rates of endorsement over time for each DSM-IV criterion item were first evaluated using related-samples Cochran’s Q test, which, if significant (p<.05), was followed by Cochran’s Q-based pairwise analyses. Each statistical value was adjusted by Bonferroni corrections. An additional analysis of sex effects in the full sample used a mixed 2 (sex) by 3 (time) design ANOVA with the number of items endorsed as the dependent variable. Also, because the data in Tables 3 and 4 revealed that the usual drinks per occasion over time decreased significantly in women but not in men, a subsequent analysis was carried out to see if any sex differences in those drinking quantities explained the sex differences over time regarding changes in rates of endorsement over time for criteria D1, D2, D4, D6, and D7 as well as A1.

Table 3.

Past 2-Year Rates of Endorsements of the 11 DSM-IV Criteria Items and Alcohol Related Variables Over Time in 223 COGA Prospective Study Men with Persistent or Recurrent AUD

	Time 1 % (SD) Age 20.2 (1.95) Range 18-28	Time 2 % (SD) Age 22.7 (2.21) Range 19-30	Time 3 % (SD) Age 25.2 (2.50) Range 21-34	Statistics Overall	Time 1 vs. 2	Time 1 vs. 3	Time 2 vs. 3
DSM IV items
Dependence
D1 Tolerance	51.1	60.1	60.1	6.72a	2.24	2.24	0.00
D2 Withdrawal	6.4	10.8	14.3	14.14c	2.24	3.74c	1.49
D3 Drink Higher Amounts or Longer	52.9	64.1	73.1	26.07c	2.83a	5.10c	2.26
D4 Desire/Unable Decrease or Control	37.2	46.6	50.2	12.94b	2.52a	3.48c	0.96
D5 Much Time to Get/Use/Recover	7.2	21.1	26.0	33.48c	4.12c	5.58c	1.46
D6 Decrease Activities Due to Alcohol	9.4	15.2	17.9	8.58a	1.96	2.86a	0.90
D7 Use Despite Physical/Psychological Problems	15.7	24.2	26.0	10.79b	2.54a	3.07b	0.54
D-X Craving	8.5	12.6	13.5	4.20	na	na	na
Abuse
A1 Failure to Fulfill Obligations	17.0	26.5	19.3	9.63b	2.97b	0.70	−2.26
A2 Use in Hazardous Situations	74.9	81.2	80.7	4.36	na	na	na
A3 Recurrent Legal Problems	1.8	3.1	2.7	0.88	na	na	na
A4 Use Despite Social/Interpersonal Problems	47.5	54.3	49.8	2.59	na	na	na
Alcohol Related Variables
Maximum Drinks/Occasion (6nmonth)	13.2 (5.44)	13.3 (5.42)	12.6 (5.70)	1.76	na	na	na
Usual Drinks/Occasion (6 month)	4.0 (4.0)	3.4 (4.10)	3.3 (3.82)	2.43	na	na	na
Number DSM-IV Criteria Endorsed (2 year)	3.2 (1.76)	4.1 (2.19)	4.2 (2.41)	24.91c	34.56c	35.05c	0.32

Note: Definitions are given in Tables 1 and 2. For overall statistics and pairwise comparisons, changes in endorsement of each Alcohol Use Disorder criterion were evaluated by standardized test statistic using related samples Cochran’s Q Test. na = not applicable, since no overall (all 3 time points) significance.

Superscripts: a p < .05; b: c: p < .001

Some COGA families had more than one individual included in these analyses. For these, the approach of Muthen and Satorra (1995) was used to measure the design effect where the small average cluster size (1.18) generated a small design effect (<1.2), suggesting that clustering did not pose a problem for the current analyses.

Results

Table 1 presents Time 1 (baseline) demographic and alcohol related variables for the full sample of 377 young adults with persistent or recurrent AUD in the COGA Prospective Study. Data are also given separately for the 223 men and 154 women within this sample. Demography for the full sample included an average age of 20 years, where, while not shown in the table, no individual was younger than 18 and eight (2.1%) were over age 25. Among the 377 individuals, 71% reported European American ethnicity, they had an average of 12.7 (SD=1.68) years of education, and in this COGA sample that had been enriched for a family history of AUD, 77% reported having a parent with AUD.

In the six months prior to Time 1, these participants with persistent or recurrent AUD reported an average maximum consumption of almost 12 standard drinks per occasion (13.2 for men and 9.5 for women, p<.001), and in the prior 2 years had experienced an average of 3.4 of the 11 DSM-IV AUD criteria (3.2 for men and 3.6 for women, p<.05). At Time 1, about 30% met criteria for alcohol dependence (the remainder had alcohol abuse) and, while not shown in the table, 51.5% of those with alcohol abuse at Time 1 met criteria for dependence during at least one of the two additional every two-year evaluations (52.1% of men and 50.5% of women, x²=0.79, p=.81).

Table 2 presents the average rates of endorsement over the three timepoints for each DSM-IV AUD criteria item for the combined group of 377 men and women with persistent or recurrent AUD. Data regarding the age at Time 1 was already given in Table 1. As shown in Table 2, the average age at Time 2 (22.9 years) included two individuals (0.5%) under age 20 and 3 (0.08%) over age 29. The average age at Time 3 was 25.4, with one person (0.3%) less than 22 and 31 (8.2%) over age 29.

The first column in Table 2 names the seven DSM-IV criteria items for dependence (D) and four items for abuse (A). The first three data columns then give the average proportions of subjects who endorsed each criterion at each of the three timepoints between average ages of 20 to 25. The next column presents the results of related samples Cochran’s Q test analyses of potential changes across all timepoints regarding the proportions endorsing each criterion. If that overall statistic was significant, the remaining columns present the results of pairwise comparisons across Times 1, 2 and 3.

The overall analyses in Table 2 revealed significant changes for all DSM-IV dependence items except D7 (use despite physical or psychological problems associated with drinking). Note that rates of endorsement of the single diagnostic criterion included in DSM-5 but not DSM-IV, craving, did not change significantly over time. Relevant to Hypothesis 1, the data included significant increased endorsement of spending a great deal of time regarding alcohol (D5) and giving up important activities because of alcohol (D6). Supporting Hypothesis 2, no abuse item demonstrated a significant change in the overall rate of endorsement over time. The data also offered partial support for Hypothesis 3 in that rates of endorsement of a desire or inability to decrease or control drinking (D4) increased, but the predicted significant decrease in tolerance (D1) was not observed in the full sample.

Table 2 also gives the prior six month usual and maximum drinks per occasion as well as the average number of AUD items endorsed over the prior two years at each timepoint. Note that drinking maximums were between 11.3 and 12.1 drinks per occasion, figures that are consistent with the mean between the 3.4 and 4.0 DSM criteria endorsed at each session. These numbers indicate evidence of notable life impairment from drinking. Significant changes over time were seen for all three drinking related variables at the bottom of Table 2.

The upper portions of Tables 3 and 4 present data regarding Hypothesis 4 that predicted changes in endorsement of DSM-IV criteria items would be similar in men (Table 3) and women (Table 4) with persistent or recurrent AUD. Average ages were similar across the sexes. Patterns of endorsement over time were also similar across the sexes and resembled data in Table 2 for drinking higher amounts or for longer periods than intended (D3), spending a great deal of time to obtain, use or recover from drinking (D5), as well as regarding the absence of significant changes over time for using alcohol in hazardous situations (A2), experiencing recurrent legal problems (A3), and use of alcohol despite social or interpersonal problems (A4). However, the patterns of changes in endorsement over time for men and women differed in several ways. First, self-reported tolerance (D1) increased significantly over time in men but decreased significantly in women. Second, only men demonstrated increases over time for withdrawal (D2), inability to decrease or control use (D4), decreasing activities due to alcohol (D6), continuing use despite physical or psychological problems (D7) and alcohol related failure to fulfill obligations (A1). Thus, sex differences were seen for six of the 11 DSM-IV items. The 2 (sex) by 3 (time) mixed design ANOVA yielded a sex by time interaction (F=7.01, p<.001) regarding how the total number of DSM-IV AUD items endorsed changed over time.

The bottom portions of Tables 3 and 4 present changes over time in alcohol-related characteristics for men and women. Note that although neither maximum nor usual drinking quantities changed significantly over time for men, the usual drinks per occasion decreased significantly for women. Therefore, a step was taken to evaluate if the significant decreases in usual alcohol quantities for women but not men might have contributed to the sex differences in rates of endorsement over time for criteria D1, D2, D4, D6, D7, and A1 in Tables 3 and 4. Here the correlations between the sex by change in usual alcohol quantities interaction term and changes in rates of endorsements over time for the five relevant DSM dependence criteria were non-significant, with values between .01 and .09 across those five criterion items. This indicated that the sex differences in item endorsements over time were not explained by the significant decrease in usual drinking quantities in women but not men. However, the correlation between the sex by change in usual alcohol quantities interaction term and changes in rates of endorsements over time correlation for A1 was r=.13, p=.01, indicating that the sex differences in the only abuse item to change significantly in men but not women might have been related to the decrease in usual drinks over time for women but not men.

Discussion

The major goal of the current analyses was to document whether, in a large group of men and women with persistent or recurrent DSM-IV AUD in their twenties, individuals exhibited significant changes over time in rates of endorsement of specific AUD criteria. This prospective study of 377 individuals across three timepoints adds to prior tentative data on 68 young adult men (Schuckit and Smith 2021) in an approach that offers greater statistical power and data regarding both sexes. The results document that the predominant pattern during the twenties was increases over time in endorsements of six of the seven DSM-IV dependence criteria, but stable rates for the more social and interpersonal problems inherent in the abuse criteria.

Consistent with Hypothesis 1, the two dependence criteria that increased significantly in the small prior study (spending a great deal of time regarding alcohol and decreasing other activities in order to drink, items D5 and D6) also increased significantly in the COGA participants with a similar age range. Endorsement of these two items had also increased in drinkers with persistent AUD between age 31 and 43 (Schuckit and Smith, 2021). Such increasing endorsements over time across different age groups for D5 and D6 might reflect increasing salience for alcohol over time in individuals with AUD, an outcome consistent with the hypothesized increasing salience for alcohol as a central component of addiction (Edwards and Gross, 1976; Schuckit and Saunders, 2006).

As predicted in Hypothesis 2, no significant increases over time during their twenties were seen regarding the proportions endorsing the four DSM-IV abuse items. That said, it is important to note that the future course of Time 1 abuse was not benign, with 36% of those with Time 1 abuse going on to dependence and 34% remaining abuse. The different changes in endorsement patterns over time across abuse and dependence criteria might relate to the fact that dependence items were originally proposed to reflect more physiological effects of alcohol such as the development of tolerance, withdrawal symptoms, and potentially longer times needed to recover from the effects of heavier drinking (Schuckit and Saunders, 2006). Those dependence problems might be expected to increase as one ages and the effects of alcohol on the body become more severe (e.g., Fat et al., 2020; Grant et al., 2017; Han et al., 2019). Abuse items, on the other hand, were originally selected to reflect more social- and interpersonal-based alcohol consequences including interference with work or school, driving while intoxicated, and legal problems (Schuckit and Saunders, 2006), items that might be more persistent during the twenties.

The findings that abuse and dependence items performed somewhat differently over time is consistent with prior studies that reported different prognostic implications for these two components of DSM-IV AUD (Hasin et al., 1997; Schuckit et al., 2000). In those studies, individuals with dependence were more likely than those with abuse to maintain dependence and more likely to meet criteria for AUD overall on follow-up. Thus, the different pattern of endorsement over time for abuse and dependence highlight a potential downside to the decision in DSM-5 to abandon the distinction between abuse and dependence in favor of the simple count of the number of criteria items endorsed as the procedure for diagnosing an AUD in DSM-5 (Hasin et al., 2013). Although the DSM-5 algorithm is easier for clinicians to use, it does not consider potential prognostic differences between abuse and dependence (e.g., Hasin et al., 1997; Schuckit et al., 2000).

The current data support the part of Hypothesis 3 that predicted increases over time in endorsement of difficulties decreasing or controlling drinking (D4), but results did not support the prior finding of decreases in endorsement of tolerance across the twenties. However, as discussed below, the changes in endorsement of tolerance differed across the sexes with decreases for women but increases for men. Thus, changes in endorsement of this criterion may be difficult to predict overall and sensitive to a range of influences, including sex.

In Hypothesis 4, we had predicted similar patterns of changes of AUD item endorsement over time in men and women, but this forecast was only partially supported by the results. Both men and women demonstrated significant increases over time for drinking higher amounts or for longer periods than intended (D3) and for spending a great deal of time involving alcohol (D5) and neither group demonstrated significant changes for use of alcohol in hazardous situations (A2), legal problems (A3) and use despite social or interpersonal problems (A4). However, the direction of change in tolerance for men and women were in the opposite direction and the changes in endorsement for the remaining four dependence items and criterion A1 were only significant for men.

Our analyses subsequently evaluated if sex differences in changes in the quantities of alcohol consumed at different timepoints might have contributed to some of the differences in patterns of AUD item endorsement over time. Running correlations between the sex by change in quantity interaction term and changes in item endorsements yielded no significant results for dependence items, suggesting that sex differences in changes in usual quantities did not account for the sex differences in endorsement over time for dependence items. However, the sex difference in endorsement over time for A1 was related to the interaction term of sex by changing usual quantity. This finding is also consistent with differences between abuse and dependence.

It is important to note the fluctuation over time regarding whether an individual fulfilled criteria for abuse or dependence. This finding harkens back to the information offered early in the Introduction regarding the generic wording for psychiatric conditions offered for many DSM diagnoses, a practice that facilitates application of the general guidelines for men and women, older and younger individuals and across different ethnic and cultural groups (Hasin et al., 2013; Schuckit and Saunders 2000). The fluctuation between abuse and dependence in the same individual also relates to the view of AUD as a chronic relapsing disorder similar to diabetes, hypertension, and most autoimmune diseases (McClellan et al., 2000). The course of such conditions is likely to fluctuate over time, with temporary phases that include periods of no problems, times when problems are less severe, and times when symptoms seem out of control (Dawson et al., 2008; Delucchi and Wiesner, 2010; Schuckit et al., 2005). In this approach, the overarching illness might still be present but with different aspects of the condition predominating and few aspects of the disorder appearing constantly throughout its course. This phenomenon is also likely to have contributed to the fact that some individuals had follow-ups where they did not meet criteria for either AUD subtype between periods when AUD symptoms were prominent.

The listing of 11 different manifestations of AUD in the DSMs and how patterns of specific items endorsed changed over time and across men and women underscore a potential asset of the DSM approach. This more inclusive approach to criteria stands in contrast to the more limited number of criteria listed by the Eleventh Edition of International Classification of Diseases System (Saunders et al., 2019). While the latter has the advantage of fewer items for the clinician or researcher to consider, it might not be as sensitive to identifying AUD in different populations. Future research will be needed to determine if the two systems function similarly regarding predicting the course of substance related problems.

Additional work will also be required to determine if the current results generalize to DSM-5. Ten of the 11 DSM-IV criteria items are also included in DSM-5, but that more recent DSM added the new item of craving and deleted the abuse item of legal problems (Hasin et al., 2013). Data regarding craving were available from the COGA-based interview, and while the current analyses did not directly test the DSM-5 algorithm we were able to document that endorsement of craving did not change significantly in the full sample or within man or women.

These analyses have multiple strengths including using within-subjects prospective data that observed the same individuals with persistent or recurrent AUD as they aged rather than comparing older and younger participants, a large number of participants, the use of a standardized interview to gather data, and inclusion of both men and women. However, the results must also be considered in the context of some caveats. First, the pattern of endorsement and the outcomes over time for participants with AUD might be different among individuals with less persistent disorders (Berkson, 1946). Second, the original COGA probands were selected from families with multiple AUD relatives, not random drinkers. Third, 70% of the subjects included in these analyses reported EA heritage and the small numbers of subjects with other ethnicities did not allow for adequate evaluation of how racial/ethnic characteristics might have related to the findings. Fourth, the participants were evaluated approximately every two-years, a schedule that did not focus on “current” events from the recent several months. Fifth, the terms men and women used here reflected how a person referred to themselves when they entered the COGA protocol and do not specifically distinguish between biological sex and gender roles. Sixth, results might be different if longer timeframes for follow-ups had been used. Finally, the variables used in these analyses reflected the specific interests of the authors, some variables of interest such as personality traits were not available for time-varying analyses, and not all potentially important variables were included.

In conclusion, the results demonstrated that the pattern of endorsement of specific AUD criteria items changed significantly over time in individuals in their twenties who have persistent or recurrent AUD. The predominant pattern involved increase in endorsement of dependence items but less change over time for endorsements of abuse items. However, endorsement of abuse items that relate more to social and interpersonal problems were less likely to escalate with time. The data could be useful to clinicians as they help patients recognize the enhanced dangers that occur with continued AUD and with increasing age. The data also underscore the heterogeneity and fluctuation in a wide span of diagnostic criteria items inherent in the DSM manuals which might be optimal for understanding the course of AUD.