Participants had to have consumed alcohol the prior year but not ever fit criteria for alcohol dependence in DSMIII or DSM-III-R , with individuals excluded for histories of bipolar disorder, schizophrenia, or physical problems that precluded alcohol challenges. Recruitment began with a participant who reported a father with alcohol dependence, with subsequent selection of a family-history–negative comparison individual with similar demography, substance use history, and past-year drinking pattern. Potential participants were evaluated using in-person interviews based on a precursor of the follow-up instrument described below . Laboratory-based challenges with 0.75 ml/kg of absolute alcohol established their intensity of response to alcohol using the Subjective High Assessment Scale , changes in body sway, and changes in hormones and electrophysiological measures, depending on the specific protocol . Data from different measures and across years were combined using z scores into one overall alcohol-challenge LR value in which lower scores reflected lower LRs per drink.Assessments began in 1988 and then were conducted every 5 years using a modification of the Semi-Structured Assessment for the Genetics of Alcoholism interview, with validities and reliabilities greater than .Age 30, 35, and 40 evaluations were face-to-face, with a parallel interview about the proband performed with a spouse or close friend. Reflecting fifinancial restrictions, age 50 and 55 follow-ups were limited to phone interviews of probands. At age 35, probands completed the then recently developed Self-Report of the Effects of Alcohol retrospective questionnaire regarding the average drinks required for up to four effects actually experienced during three life epochs: the first five times drinking , most recent 3 months, and their heaviest drinking period. Total scores reflected average drinks for effects across all three epochs,vertical grow rack as the sum of the number drinks required for up to four effects divided by the number of the effects reported, and SRE-5 scores reflected the average drinks required during the first 5 times of drinking only .

Higher SRE scores indicate more drinks needed for effects, or lower LRs per drink. The SRE Cronbach’s + is greater than .Age 35 follow-ups also evaluated novelty seeking , sensation seeking , and impulsivity . By age 50, 11 of the original 453 probands had died, leaving 442 eligible for follow-up . Among these, 397 participated in all follow-ups from ages 30 through 50 , 165 of whom had developed a DSMIV AUD at any assessment . With the emphasis on predictors of the course of AUDs and their prevalence during the sixth life decade, these men with prior AUDs were the focus of the current analyses.Comparisons across the four outcome groups were first evaluated using chi-square for categorical data and analysis of variance for continuous variables. We next evaluated which variables from the age 30–50 follow-ups significantly differentiated across the four age 50–55 outcome groups when considered along with other significant variables. There are three types of data used in the relevant analyses in Table 3: single assessment items ; drinking variables in which the number represents the maximum value reported across age 30–50 interviews , or a simple count of occurrences ; and dichotomous variables indicating that a subject fulfilled that item at any age 30–50 interview. For the regression analyses in Table 4, we considered a simultaneous-entry multi-nomial logistic regression analysis but rejected this approach because of our modest sample size and our desire to identify variables that predicted each outcome rather than evaluating predictors of only three groups with the fourth used as a reference group. Therefore, this final analytic step used four binary logistic regression analyses predicting each outcome independently.The characteristics reported at age 30–50 follow-ups and their relationships to age 50–55 outcome groups are presented in Table 3. For most alcohol-related variables, men who reported low-risk drinking needed the lowest number of drinks for effects , as well as lowest alcohol quantities, frequencies, problems, and treatment exposure at age 30–50 follow-ups. The pattern for alcohol-related variables generally increased across Groups 1–4, with the highest number of drinks for effects , quantities, frequencies, problems, and treatment exposure for the abstinent Group 4, followed by men in Group 3 who fulfilled criteria for DSM-5 AUDs at ages 50–55. Patterns across Groups 3 and 4 included an earlier AUD onset and greater experience with alcohol-related treatment and/or self-help groups for the abstinent Group 4.

High-risk drinking was associated with LR values and alcohol histories between Group 1 and Groups 3 and 4. The lower portion of Table 3 lists the age 20–50 patterns across groups regarding drug-related items. Most men had experience with illicit drugs during those follow-ups, but the only significant group difference was for the prevalence of cannabis use disorders, with the lowest values for Group 1 and the highest for Group 4. The analyses next turned to an evaluation of how the nine variables from the age 30–50 follow-ups that significantly differentiated across the groups in Table 3 related to age 50–55 outcome groups when considered in the same analysis. Because the two SRE measures correlated at .66, and the number of DSM-IV AUD criterion items endorsed and DSM-IV dependence diagnoses correlated at .69, to minimize multi-collinearity among the four variables only SRE-T and numbers of AUD items were used in the regression analyses. Five of the remaining seven variables from Table 3 contributed significantly to any regression analysis related to age 50–55 outcomes. Low-risk drinking was related to prior lower drinking frequencies; high-risk drinking was related to needing fewer drinks for effects and an older AUD onset; DSM-5 AUDs were related to higher prior drinking frequencies and the absence of prior treatment and/or self-help group participation; and abstinence was related to needing the most drinks for effects , higher odds ratios for prior treatment or self-help participation, and prior cannabis use disorders. Thus, the most consistent predictors of age 50–55 outcomes were LR, prior drinking frequencies, and having received prior help for drinking problems, each of which contributed significantly to two of the four regression analyses in Table 4. It is worth noting that although the odds ratio for the numbers of DSM items endorsed for Group 4 was not significant, the value is actually greater than 1 but appears lower in the table because of a suppressor effect in the regression analysis.This article presents the age 50–55 outcomes for 156 men who developed an AUD during the initial 30 years of the SDPS, as well as predictors of those outcomes. Contrary to Hypothesis 1, only 10% of these men were abstinent from alcohol during ages 50–55, and 16% reported low-risk drinking. The remaining 74% either fulfilled DSM-5 AUD criteria or reported risky drinking. Although these probands had impressive educations and incomes, the data document the tenacity of alcohol-related problems when individuals enter their sixth life decade. These results and the study by Vaillant demonstrate that many individuals with AUDs do not fifit the erroneous stereotype that they are likely to be unemployed and live on the street or in public housing.

The SDPS began with students and nonacademic staff at UCSD and their earlier high functioning predicted impressive achievements despite their AUDs. Yet, as shown in Table 3, between ages 20 and 50 these men had clearly filled AUD criteria, reporting 13–16 maximum drinks per occasion and experiencing 4–6 of the 11 DSM-IV AUD items. The inaccurate AUD stereotype is often shared by health care deliverers who might be reluctant to gather alcohol or drug problem histories from affluent and well-educated patients and to intervene when appropriate. These data support the contention that—regardless of social status, income, and age—all patients should be screened for substance intake patterns and related problems. Regarding Hypothesis 2, only 16% demonstrated sustained “controlled drinking” over the 5 years with alcohol quantities in the low-risk range . Although short-term, low-risk drinking is common , our findings are consistent with other studies that found less than 20% of individuals with alcohol dependence maintained controlled drinking over extended periods . As predicted, such non-problematic outcomes are most likely in men who were more sensitive to alcohol and had lower past drinking quantities, frequencies, and alcohol problems . This profile might help identify individuals for whom long-term controlled drinking might be an appropriate option. As predicted in Hypothesis 3, high-risk drinkers reflected earlier alcohol-related characteristics that were between those that predicted low-risk drinking and the higher quantities and problems that related to continued AUDs in Group 3 and abstinence in Group 4. The Nagelkerke’s Pseudo R2s in Table 4 indicate that the age 20–50 independent variables predicted Group 2 less well than the other outcomes,cannabis grow racks raising the question of whether some alcohol problems went unreported by this group. Even in the absence of multiple alcohol problems, heavier drinking is not optimal because it carries elevated risks for cardiovascular disease, stroke, cancers, and other adverse health outcomes that are likely to contribute to a shortened life span . Persistent abstinence was not only relatively rare but was also consistent with Hypothesis 4 and several other studies . These men had the lowest LR to alcohol and reported the highest alcohol quantities, problems, and rates of alcohol dependence. More research is needed, but one possible explanation for this finding is that experiencing greater alcohol-related problems, perhaps as a consequence of a lower LR to alcohol, may have contributed to men in Group 4 seeking help, and their exposure to treatments may have made abstinence especially acceptable to members of Group 4. Their cannabis-related problems may have also increased the likelihood of entering substance-related programs. Table 4 used a binary logistic regression analysis to evaluate how the predictors of group membership operated when considered in the same analyses. Focusing primarily on statistically significant findings in Table 4, a lower number of drinks needed for effects on the SRE was associated with a lower likelihood of being categorized as a high-risk drinker at age 50–55, and the need for more drinks for effects was associated with a greater likelihood of being abstinent during the age 50–55 follow- up. Although not statistically significant, the pattern in Table 4 indicated the possibility that the need for more drinks for effects might have increased the chances of meeting criteria for an AUD on follow up but decreased the chances of falling into the low-risk drinking category.

Drinking frequency also contributed significantly to two regression analyses, with prior lower frequencies robustly predicting low-risk drinking and prior higher frequencies indicating a higher risk for meeting DSM-5 AUD criteria after age 50. Prior frequencies might be a measure of the importance, or salience, of alcohol to probands in these outcome groups. Past experience with formal treatment or self-help groups was significantly associated with later abstinence and was less likely to be seen in participants with active DSM-5 AUDs during the age 50–55 follow-up. Later onset AUDs predicted only high-risk drinking in Table 4 and may have related to the development of an AUD at a time with greater maturity and life experiences, which, in turn, might be associated with less likelihood of alcohol-related life problems even in the context of continued risky drinking. Cannabis use disorder histories also significantly predicted one outcome, abstinence, perhaps reflecting cannabis related interference with cognitive functioning in AUDs that might contribute to more severe alcohol problems, as proposed in a recent article , and/or may have contributed to the probability of seeking help. More research will be needed with additional populations of older individuals with AUDs to determine whether these findings are replicable and apply to other samples of older individuals with histories of AUDs. The current data also generate some thoughts regarding DSM AUD approaches over the years. Regarding criteria for Group 3, we are aware that since 1987 , remission could only be diagnosed in the absence of endorsement of any DSM criterion items. Thus, earlier analyses attempted to define continued AUDs as endorsement of more than one DSM criterion item, with the result that Group 3 constituted 56% of the outcomes. Although those analyses identified the same predictors of outcome groups reported here, we were concerned that such a severe restriction regarding what was called remission might not fit the preferences of many current clinicians , and we decided to use the DSM-5–based less demanding definition requiring two criterion problems for an active diagnosis . There are no perfect and universally accepted definitions for remission versus active AUDs, but we feel the current approach is a reasonable compromise among possibilities.