Preliminary evidence for perceived rapid alcohol tolerance growth as a risk factor for problematic alcohol use

Abstract

Background:

Pre-clinical research with animals suggests that a propensity to more rapidly adapt to the effects of alcohol could be an important neurobiological difference underlying alcohol use disorder (AUD). Research translating this work to humans is scarce. We tested whether adults’ self-perceptions of rapidly developing alcohol tolerance following the onset of regular drinking was associated with increased risk of problematic alcohol use (using scores that excluded tolerance items), while controlling for several potentially important confounders.

Methods:

Adults who reported current heavy drinking completed an online survey (N=160; M(SD)=32.24(9.30) years old; 56.3% Black, 43.8% White; 55% Female, 45% Male). Participants reported the extent to which “my tolerance to alcohol seemed to build up very quickly once I started drinking frequently.” Initial subjective response was measured as the number of drinks to experience alcohol effects during the first five drinking experiences and acquired tolerance as the difference between current and initial subjective response. Participants reported on alcohol consequences and dependence symptoms using modified DSM-5 criteria (i.e., problematic alcohol use), past year alcohol consumption, and demographic characteristics.

Results:

Hierarchical linear regression showed that perceived rapid tolerance growth was uniquely associated with problematic alcohol use (excluding tolerance symptoms) after controlling for initial subjective response, acquired tolerance, DSM-5-related tolerance, alcohol consumption, race, sex, and age.

Conclusions:

Individuals who reported experiencing rapid tolerance growth when they started regularly drinking showed increased risk for problematic alcohol use. Future prospective research on this topic is warranted and could help uncover a novel risk factor for AUD.

Short Summary

Individuals who reported experiencing rapid tolerance growth when they started regularly drinking showed increased risk for problematic alcohol use after controlling for initial subjective response, alcohol tolerance and consumption, and demographic factors. Findings provide preliminary support for future studies, which could uncover a novel risk factor for alcohol use disorder.

Introduction

Tolerance to alcohol is important in the study of alcohol use disorder (AUD) due to its implications for escalating alcohol consumption, AUD diagnosis, and treatment. Unfortunately, this critical risk factor for AUD has been understudied in the last several decades (Elvig et al., 2021). Chronic tolerance occurs when an individual experiences a lessened effect from the same alcohol dose, or needs more alcohol to produce the same effect, after repeated alcohol exposure (American Psychiatric Association, 2013). In addition to being a diagnostic criterion for AUD after crossing a certain severity threshold, chronic tolerance (hereafter referred to as “tolerance”) is a developmental process that progresses with repeated alcohol exposure (American Psychiatric Association, 2013; Elvig et al., 2021). Note that chronic tolerance is distinct from acute tolerance, which refers to adaptation within the same drinking session, and from rapid tolerance, which refers to adaptation after one drinking session.

Interestingly, pre-clinical research using animal models suggests that more quickly developing tolerance following the onset of drinking may be an important neurobiological difference underlying AUD (e.g., Lê & Kiianmaa, 1988; Nikander & Pekkanen, 1977). Theoretically, a propensity to more rapidly adapt to the effects of alcohol could lead to heavier and more frequent alcohol intake, which could in turn cause and potentiate alcohol-related problems and physical dependence symptoms. Unfortunately, few studies have tested whether quickly developing tolerance following drinking onset is a risk factor for AUD in humans. In this study, we used cross-sectional data from humans to test whether individuals’ perceptions of rapidly developing alcohol tolerance were associated with problematic alcohol use after controlling for relevant covariates. As we recognize the limited nature of cross-sectional data, the primary goal of this investigation was to provide preliminary data that could help inform and motivate future research on this understudied topic.

Animal models have long been studied to elucidate the biological underpinnings of AUD. Differences observed in alcohol-preferring relative to non-alcohol-preferring animals (i.e., who have been selectively bred for this behavior) may represent key neurobiologically or genetically driven AUD risk factors (Crowley et al., 2019). Interestingly, research showed that more quickly developing chronic and rapid tolerance to alcohol were more characteristic of alcohol-preferring relative to non-alcohol-preferring rodents, even when alcohol doses were equal across groups. This effect was established across various rodent lines (e.g., AA vs. ANA; P vs.NP; LAP vs. HAP), for a variety of alcohol effects (motor impairing, hypothermic, sedative/hypnotic), at moderate alcohol doses, and with and without intoxicated practice (Nikander and Pekkanen, 1977; Waller et al., 1983; Gatto et al., 1987; Lê and Kiianmaa, 1988; Murphy et al., 1990; Stewart et al., 1992; Kurtz et al., 1996; Froehlich and Wand, 1997; Bell et al., 2001). Notably, the alcohol-preferring P line of rats satisfies all criteria for an animal model of AUD, suggesting that observed phenotypic differences in tolerance development may reflect a neurobiological difference underlying AUD (McBride and Li, 1998).

Translating this work to humans may be important but has some challenges. For instance, the most rigorous method to assess this phenomenon would be to conduct an experimental alcohol challenge to identify those who develop chronic tolerance more quickly, but there are clear ethical implications of such a protocol (as opposed to acute tolerance, which has been examined in prior research; Anthenelli et al., 2021). Based on the pre-clinical evidence, it seems worthwhile to explore alternative ways of measuring individual differences in the rapidity of adaptation to alcohol. Indeed, because the propensity to develop tolerance rapidly could have a neurobiological basis, further translational research on this topic could inform novel treatment targets, prevention, and biological mechanisms of AUD.

Some observational human studies have examined similar empirical questions, although none were designed specifically to examine the propensity for more quickly developing tolerance. For instance, three studies investigated how retrospective reports of the number of drinks needed to experience certain alcohol effects (i.e., subjective alcohol effects) changed from individuals’ initial drinking experiences (i.e., first five experiences) to their current adulthood drinking experiences (Schuckit and Smith, 1996; Morean and Corbin, 2008; Corbin et al., 2013). Differences in the number of drinks needed from initial to current drinking experiences have been conceptualized as one way to capture alcohol tolerance, termed by some as “acquired tolerance.” These studies found that risk for alcohol outcomes was heightened amongst individuals who reported greater acquired tolerance over time, that is, who needed more drinks to feel the same alcohol effects from their initial to their current drinking experiences. The effect of acquired tolerance was also independent from one’s initial, or “innate” subjective response. Despite the potential conceptual overlap between acquired tolerance and rapid tolerance growth, the extent to which acquired tolerance actually reflects rapidly developing alcohol tolerance is unknown. Indeed, greater acquired tolerance may have simply reflected heavier drinking or a longer history of drinking exposure, which were not consistently controlled in these studies. Nonetheless, findings from these studies are at least consistent with the notion that individuals who more rapidly develop alcohol tolerance may be at higher risk for AUD.

Another line of research analyzed how binary DSM-related reports of alcohol tolerance predicted alcohol-related outcomes. O’Neill and Sher (2000) found that college students who endorsed tolerance in their freshman year showed levels of alcohol involvement that were above-average for the sample throughout seven years of follow up assessments relative to those who did not endorse tolerance. Tolerance was defined in this study as needing larger amounts to feel the effect, no longer able to get high or drunk on the same amount, or drinking 7 or more drinks daily for two weeks. Similarly, two other studies found that young adults who endorsed DSM-related alcohol tolerance were more likely to experience later AUD problems and symptoms (Schuckit et al., 2008; O’Dean et al., 2022). This association persisted even after controlling for maximum lifetime drinks, suggesting that tolerance was not just a proxy for drinking quantity (Schuckit et al., 2008). All studies were prospective, providing greater confidence that it was the increased risk of tolerance itself that predicted worsened alcohol-related outcomes. Because these studies used binary measures of DSM-related tolerance, they do not directly inform the potential impact of rapidly developing alcohol tolerance. Nonetheless, findings are likewise consistent with the notion that rapidly developing tolerance could be important in the development of AUD.

One way to capture this phenomenon that, to our knowledge, has not yet been explored is to query individuals’ self-perceptions of whether their tolerance to alcohol quickly developed once they started drinking regularly. Self-perceptions of tolerance, while potentially more prone to biases than some other methods, may be one way to gain insights into rapid adaptation to alcohol above what is possible using methods common in extant research. Indeed, alcohol tolerance that develops quite rapidly once regular drinking has begun might be a noticeable, noteworthy experience for individuals. Thus, this study tested the hypothesis that rapid tolerance growth from the onset of regular drinking is a unique risk factor for problematic alcohol use by asking individuals the extent to which: “My tolerance to alcohol seemed to build up very quickly once I started drinking frequently.”

When examining whether self-perceptions of rapid tolerance development were associated with problematic alcohol use, it was important to rule out confounding factors. For example, acquired tolerance, initial subjective response, and endorsements of tolerance based on DSM operationalizations could together capture rapidly developing tolerance on some level. It was important to understand whether self-perceptions of rapid tolerance growth would provide additional information in the prediction of problematic alcohol use beyond these established indicators of tolerance. Additionally, we included several other theoretically important covariates, including age, assigned sex at birth (sex), racial identity (race), and various indicators of alcohol consumption. Characteristics like sex, race, and age could influence the rapidity of tolerance growth based on metabolic, body size, environmental, or cultural differences (Zapolski et al., 2014; Erol and Karpyak, 2015). Rapid tolerance growth may also simply reflect more frequent or heavier drinking, making alcohol consumption an important covariate (Schuckit et al., 2008; Corbin et al., 2013).

In the current study, we examined whether individuals’ perceived rapid tolerance growth was uniquely associated with problematic alcohol use (operationalized as the sum of alcohol consequences and dependence symptoms that excluded tolerance items) over and above important covariates. We hypothesized that perceived rapid tolerance growth would be uniquely associated with problematic alcohol use over and above acquired tolerance, initial subjective response, DSM-5-related tolerance, alcohol consumption and heavy episodic drinking, sex, race, and age. Our preliminary examination of this topic will serve as one of the first studies to our knowledge to inform whether future pursuits of this understudied topic are worthwhile, with implications for uncovering an AUD risk factor that has the potential to enhance treatment and prevention.

Method

Participants

Participants were recruited as part of a larger study focused on impaired control over alcohol use. Recruitment occurred through flyers posted in the community, the Pitt+Me Research Registry, the NIH’s Research Match Registry, BuildClinical, and other social media platforms. Participants were eligible if they were 18–50 years old, identified their racial identity as either White or Black (or if Biracial/Multiracial, identifying with one of those identities more closely), had greater than 8^th grade education, consumed 5+ (assigned male) or 4+ (assigned female) drinks in one sitting at least once/week for three months within the past year, did not show psychotic symptoms interfering with ability to respond, and were able to verify their identities to research staff. Inclusion criteria related to race ensured adequate power to test differences by race, in line with the original aims of the larger study. To ensure that participants represented a spectrum of alcohol use-related severity, ~30% of participants (N=77) were required to endorse at least one of two questions (“In the past two years, did you ever have trouble controlling your drinking?” or “In the past two years, have you ever had an irresistible urge to continue drinking once you’d started?”). Participants received $25 for completion of the survey. The University of Pittsburgh IRB approved this study.

Eligible participants were sent an electronic Qualtrics survey link that retained their anonymity (N=246). Partway through the larger study, a measure of tolerance, including the item measuring perceived rapid tolerance growth, was added to the survey. Participants were included in the present study if they were administered the tolerance item (N=164) and if they had data on other study variables (N=160). Compared to excluded participants (Table 1), included participants reported having more drinks on a typical day in the past year, t(244)=−2.35, p=0.01. There was also a higher proportion of Black individuals in the included than the excluded subsample (χ²(1) = 26.45, p<0.001).

Table 1.

Descriptive Statistics and Differences Between Included and Excluded Participants

	Included		Excluded
	N	M(SD)	N	M(SD)	t(df)
Rapid Tolerance Growth	160	1.98(1.41)	4	1.25(1.50)	−1.03(162)
Problematic Alcohol Use	160	15.28(11.38)	86	15.19(10.51)	−0.06(244)
Age	160	32.24(9.30)	86	32.80(7.33)	0.52(211.14)
Initial Subjective Response	160	0.03(0.93)	82	−0.12(0.64)	−1.33(240)
Acquired Tolerance	160	0.00(0.82)	81	0.00(0.59)	−0.03(210.35)
DSM-related Tolerance	160	1.83(1.51)	86	1.87(1.59)	0.20(244)
Drinking Frequency	160	3.23(0.75)	86	3.23(0.70)	0.08(244)
Drinking Quantity	160	1.95(1.15)	86	1.59(1.10)	−2.35(244)*
Heavy Episodic Drinking	160	2.32(1.13)	86	2.21(1.04)	−0.74(244)
	N	%	N	%	χ2 (df)
Race	160	56.3% Black	86	22.1% Black	26.5(1)**
		43.8% White		77.9% White
Sex at birth	160	55% Female	85	64.7% Female	2.15(1)
		45% Male		35.3% Male

Note.

^*p<0.05. Levene’s test for the homogeneity of variance was calculated for each t test. For those instances in which equal variances were not assumed, a t-test with a Satterthwaite approximation for the degrees of freedom is presented.

Measures

Perceived rapid tolerance growth following the onset of regular drinking.

Participants rated the extent to which, “My tolerance to alcohol seemed to build up very quickly once I started drinking frequently” (0=Not at all accurate [endorsed by 19.4% of participants]; 1=A little bit accurate [20%]; 2=Somewhat Accurate [25%]; 3=Mostly Accurate [14.4%]; 4=Very accurate [21.3%]). Endorsement of each response option was relatively evenly spread across participants.

Problematic alcohol use.

The 11 DSM-5-TR AUD symptoms (American Psychiatric Association, 2013) were adapted to be in first-person and to no longer be double- or triple-barreled by splitting them as necessary. Participants reported on twenty-two statements about their drinking in the last 12 months (0=No, 1=Somewhat, 2=Yes). Twenty items captured alcohol use consequences and dependence symptoms other than tolerance and were summed. Tolerance items were excluded from this score as they were included separately as covariates (see below). Cronbach’s alpha (α) was 0.95.

DSM-5-related alcohol tolerance symptoms.

Two items adapted from the DSM-5-TR AUD symptoms captured tolerance (“I needed greater amounts of alcohol than I used to in order to feel intoxicated OR to get a desired effect” and “I got much less of an effect by using the same amount of alcohol as in the past”). Participants rated items (0=No, 1=Somewhat, 2=Yes) based on the past 12 months. Cronbach’s α was 0.81, justifying summing the items to create an overall score representing DSM-5-related alcohol tolerance symptoms.

Initial and current subjective response to alcohol.

The Self-Rating of the Effects of Alcohol (SRE) asks participants to report whether they ever experienced four alcohol effects: (1) feel different, (2) feel a bit dizzy or slur your speech, (3) stumbled or walked in an uncoordinated manner, or (4) passed out or fell asleep when they did not want to. If they ever experienced an alcohol effect, they reported on the number of standard drinks to begin to feel each one. (Schuckit, Smith, and Tipp, 1997). Participants reported on each alcohol effect during their first five drinking episodes and most recent period of drinking. As recommended, we computed SRE scores using standardized person-mean imputation using only those alcohol effects that participants ever experienced (Lee et al., 2015). SRE items were converted to z-scores, which were averaged for ‘first 5’ (α=0.88) and ‘recent’ drinking experiences (α=0.90), respectively.

As all participants were required to have a period of three months in the past year during which they drank at least once a week, the SRE items pertaining to most recent use can be considered a measure of current (past year) subjective alcohol response. Acquired tolerance was measured by subtracting initial from current subjective response, consistent with previous studies (Morean and Corbin, 2008; Corbin et al., 2013).

Alcohol consumption.

The 10-item Alcohol Use Disorder Identification Test (AUDIT; Saunders et al., 1993) was adapted to ask about drinking behavior from the past year. We used as covariates three separate questions from the AUDIT: (1) “How often did you have a drink containing alcohol in the past year?” (i.e., drinking frequency), (2) “How many drinks containing alcohol did you have on a typical day when you were drinking in the past year?” (i.e., drinking quantity), (3) “How often during the last year did you have six or more drinks on one occasion?” (i.e., heavy episodic drinking). The response options for each item ranged from 0–4, with higher ratings corresponding to greater consumption.

Demographic covariates.

Assigned sex, race, and age were used as covariates. We used assigned sex (and not gender identity) as a covariate due to known differences across sex on alcohol metabolism (Baraona et al., 2001). Indeed, the percentage of individuals reporting a gender diverse identity was too small to be analyzed separately (5%). Participants were asked to self-report their race: (a) Biracial, (b) Black/African American, (c) Multiracial, (d) White/European American, or (e) Not Listed. Participants who selected Biracial or Multiracial were prompted to choose which racial group they most closely identified with, which was the response used in analyses. Age was reported in years. Ethnicity was not included as a covariate due to low variability (6.3% identified as being “of Spanish or Hispanic origin, such as Latin American, Mexican, Puerto Rican, or Cuban”)

Data Analytic Plan

Zero-order correlations were examined among all study variables as a preliminary test of whether any should be combined or excluded (if r>0.80). We conducted a hierarchical linear regression using SPSS version 29. The normality of residuals, linearity, and homoscedasticity assumptions were checked using a normal P-P plot and a scatterplot of the standardized predicted value with the standardized residual. We calculated the variance inflation factor (VIF) for all predictor variables to determine the impact of collinearity, or linear dependence, among the predictors in the regression analysis on the precision of estimation. Any variable whose VIF exceeded 5 was considered for exclusion due to multicollinearity (Cohen et al., 2013). Because only four of the included participants did not have complete data, we used listwise deletion. The variables were entered stepwise in the regression predicting problematic alcohol use in the following order: (1) age, race, and sex, (2) initial subjective response to alcohol, (3) acquired tolerance, (4) DSM-5-related tolerance symptoms, (5) the three AUDIT consumption items, and (6) perceived rapid tolerance growth. This approach was chosen to allow an understanding of how the inclusion of different groups of variables influenced associations with problematic alcohol use.

Results

Zero-order correlations

No correlation coefficients among predictor variables exceeded (+/−) 0.67 (Table 2). Perceived rapid tolerance growth was moderately-to-highly and significantly correlated with DSM tolerance and frequency of heavy episodic drinking (r=0.42–0.67); weakly and significantly correlated with acquired tolerance, drinking frequency and quantity, and age (r=0.17–0.36); and not significantly correlated with race, sex at birth or initial subjective response. Higher problematic alcohol use was moderately and significantly correlated with higher levels of perceived rapid tolerance growth, DSM tolerance, drinking quantity, and frequency of heavy episodic drinking (r=0.51–0.54); weakly and significantly correlated with older age, greater acquired tolerance, and greater drinking frequency (r=0.16–.38); and not significantly correlated with race, sex, or initial subjective response.

Table 2.

Zero-order correlations among all study variables

	Rapid tolerance growth	Age	Race	Sex at birth	Initial subjective response	Acquired tolerance	DSM tolerance	Drinking frequency	Drinking quantity	Heavy episodic drinking	Problem Alcohol Use
Rapid tolerance growth	1
Age	0.17*	1
Race	−0.01	0.32**	1
Sex at birth	−0.06	−0.15	0.14	1
Initial subjective response	0.00	−0.06	0.08	−0.04	1
Acquired tolerance	0.25**	0.21**	−0.25**	−0.26**	−0.40**	1
DSM tolerance	0.67**	0.13	−0.10	−0.08	−0.02	0.31**	1
Drinking frequency	0.25**	0.22**	0.01	−0.11	0.04	0.32**	0.32**	1
Drinking quantity	0.36**	0.09	−0.02	−0.16*	0.19*	0.26**	0.37**	0.27**	1
Heavy episodic drinking	0.42**	0.13	−0.18*	−0.21**	0.19*	0.34**	0.44**	0.43**	0.63**	1
Problematic Alcohol Use	0.54**	0.21**	0.10	−0.06	0.08	0.16*	0.53**	0.38**	0.53**	0.51**	1

Note. N=160.

^*p<0.05;

^**p<0.01.

0=White, 1=Black. 0=Male, 1=Female.

Hierarchical Linear Regression Results

All predictor variables were retained in the final analysis as all VIFs were less than 5, indicating that dependence among predictor variables was moderate and likely not to have an appreciable impact on the precision of estimation. In the initial regression model, the normality of residuals and linearity assumptions were met. However, the scatterplot of the standardized predicted value with the standardized residual was not scattered uniformly and randomly around zero, but cone shaped, suggesting that homoscedasticity was not met. Thus, we performed a square root transformation on the outcome variable, resulting in all assumptions being met upon visual inspection (Cohen et al., 2013). Regression analyses were performed with this transformed variable.

In the final model, older individuals showed greater problematic alcohol use, but there were no differences by sex or race (step 1). Although initial subjective response was not associated with problematic alcohol use in step 2, this effect became significant with the addition of acquired tolerance in step 3. Specifically, those who needed more drinks during their first 5 drinking episodes and whose acquired tolerance increased over time (i.e., greater increase in the number of drinks needed) showed greater problematic alcohol use. Both of those effects and age became non-significant after adding the DSM-5-related tolerance score (step 4), which was significantly associated with greater problematic alcohol use. In step 5, drinking quantity and heavy episodic drinking, but not drinking frequency, were associated with problematic alcohol use; all other effects remained the same. Finally, in the sixth step, higher levels of perceived rapid tolerance growth were associated with greater levels of problematic alcohol use over and above all other covariates, with all other results remaining the same.

Post-hoc Sensitivity Analysis

Although the main covariates were chosen based on direct empirical or theoretical links to tolerance, other covariates could indirectly influence rapid tolerance growth through risk for alcohol consumption. For example, impulsivity robustly predicts alcohol use and consumption (Coskunpinar et al., 2013). As impulsivity is trait-like and shows rank-order stability over development (Borghuis et al., 2017), it could also reflect levels of impulsivity when individuals started drinking regularly. Moreover, ethnic identity (i.e., Latine vs. not Latine), identifying as a sexual or gender minority (SGM), and level of education predict differential risk for alcohol use (Crosnoe and Riegle-Crumb, 2007; Keyes et al., 2015; Talley et al., 2016). Thus, we conducted sensitivity analyses to determine whether rapid tolerance growth continued to predict alcohol problems after additionally adjusting for these variables, in addition to the main covariates. The effect of rapid tolerance growth persisted (β=0.16, p=0.03) after adding as additional covariates to the final regression model: education (β=0.09, p=0.16), ethnicity (β=−0.02, p=0.73), identifying as SGM (β=0.08, p=0.20), and five facets of impulsivity from the UPPS-P scale (Cyders et al., 2007), including positive urgency (β=0.29, p<0.001), negative urgency (β=0.16, p=0.04), lack of premeditation (β=0.03, p=0.70), lack of perseverance (β=0.01; p=0.90), and sensation seeking (β=−0.07, p=0.01).

Discussion

This study tested the hypothesis that a propensity to quickly adapt to the effects of alcohol is a unique risk factor for alcohol-related problems in humans. We queried individuals’ perceptions of whether their tolerance increased rapidly once they started drinking regularly and examined whether this was uniquely associated with problematic alcohol use. We controlled for established indicators of tolerance that could be proxies for rapid tolerance growth as well as drinking patterns and demographic characteristics. We found that perceived rapid growth in tolerance was uniquely associated with problematic alcohol use even after controlling for these covariates.

This was the first study to our knowledge to directly assess the phenomenon of rapid tolerance growth following the onset of regular drinking in humans by any method, as well as its association with problematic alcohol use. Findings were in line with our central hypothesis that perceived rapid tolerance growth from the onset of regular drinking is an important risk factor for AUD and related outcomes. That this associated persisted after the inclusion of important covariates further suggests that individuals who perceived more rapid tolerance growth from the onset of regular drinking do not exhibit greater problematic alcohol use simply because they drink more heavily, report current alcohol tolerance, or have an “innate” low level of alcohol response.

Our results support prior pre-clinical research in which alcohol-naïve animals who were provided the same doses of ethanol showed different rates of adaptation to the effects of alcohol. This depended on whether they were alcohol-preferring (faster adaptation to alcohol) versus alcohol-avoiding (slower adaptation) animals (Nikander and Pekkanen, 1977; Gatto et al., 1987; Lê and Kiianmaa, 1988; Stewart et al., 1992; Kurtz et al., 1996; Bell et al., 2001). Importantly, these effects have been established in the alcohol-preferring P line relative to the alcohol-non-preferring NP line of rats (Waller et al., 1983; Murphy et al., 1990; Bell et al., 2001). Because the P line of rats satisfies all criteria for an animal model of AUD, the observed phenotypic differences in tolerance may reflect neurobiological differences underlying AUD (McBride and Li, 1998). This study is unique in translating these pre-clinical findings to human participants.

Moreover, individuals’ perceived rapid growth in tolerance were related to, but not the same as, their acquired tolerance (r = 0.25) and DSM-5-related tolerance scores (r = 0.67) and were essentially independent from initial subjective response (r = 0.003). Thus, individuals who perceived rapid tolerance growth starting from the onset of regular drinking did not necessarily perceive their current alcohol tolerance to be heightened, nor their initial subjective response to be particularly low or high. Indeed, the eventual development of tolerance could be caused by many other factors aside from rapidly developing tolerance, including drinking heavily to cope or high-risk drinking environments (Cooper et al., 1995; White et al., 2006). The fact that individuals’ reports of rapid tolerance growth did not simply reflect current alcohol tolerance or initial alcohol response suggests that studies of tolerance (as widely operationalized in the existing literature) will not necessarily inform this potentially important phenomenon.

Collectively, these results provide necessary preliminary evidence to show that future prospective research on this topic is warranted to validate measures of rapid tolerance growth and to better understand how rapid tolerance growth prospectively predicts problematic alcohol use. This research could help uncover novel biological mechanisms, treatment, and prevention targets for AUD. For example, individuals who rapidly develop alcohol tolerance when they start drinking may benefit from early identification based on this characteristic and early prevention. The identification of rapid tolerance development as an important risk factor could also spur research to uncover biological (e.g., genetic) mechanisms underlying it, which could inform novel treatment development for AUD. Individuals at risk for AUD may also benefit from psychoeducation about individual differences in the speed of tolerance growth and how this influences risky drinking patterns. Our work also answers recent calls in the literature for increased research attention on alcohol tolerance, which has been significantly understudied yet may be critical to understanding targets to attenuate hazardous drinking patterns (Elvig et al., 2021).

This study had limitations to consider. Our data was cross-sectional and some of our measures, including rapid tolerance growth and initial subjective response, were retrospectively reported and could be subject to recall bias. Unfortunately, we also did not have data on when participants started regularly drinking and could not control for the possibility that individuals who had been drinking regularly for a longer period could be more prone to recall bias for rapid tolerance growth. The measure of rapid tolerance growth has not been validated in prospective studies mapping growth in tolerance from the onset of regular drinking. Our measure of drinking patterns, an important covariate, was only assessed concurrently with the outcome variable. It would have been important to control for alcohol consumption during the onset of regular drinking to be concurrent with the perceived rapid tolerance growth variable and/or to include a statement in the item itself that allows participants to consider the amount of alcohol consumed. The tolerance measure was also administered partway through a larger study and resulted in a relatively small sample size. However, there were few differences in study variables between those who did and did not receive the measure of rapid tolerance growth. Finally, our sample was at heightened risk for AUD relative to the general population, which can limit external validity. Replicating these effects within other samples would be important.

Because of these limitations, it is difficult to conclude whether rapid tolerance as measured in this study is a heritable trait, acquired state, or simply a selection phenomenon. However, our preliminary results suggest that it will be worthwhile to explore this topic using more rigorous and suitable research designs. For example, longitudinal studies are needed that collect real time, intensive data (e.g., ecological momentary assessment) on alcohol consumption and response during drinking episodes in individuals starting to drink regularly. Longitudinal alcohol administration studies that track response to alcohol in addition to comprehensively assessing alcohol consumption patterns could also shed light on these processes.

Our study also had multiple strengths. Our sample was characterized by racial diversity, allowing an examination of associations with race. We included nine important and empirically justified covariates to test the robustness of the association between perceived rapid tolerance growth and problematic alcohol use. Relatedly, this allowed us to replicate prior work showing that both low subjective response to alcohol and greater acquired tolerance were associated with alcohol-related problems (Morean and Corbin, 2008; Corbin et al., 2013), although these effects did not survive controlling for other variables.

In summary, results from this study highlight the possibility that individuals predisposed towards rapidly increasing alcohol tolerance may have increased risk for problematic alcohol use. Most importantly, findings provide important and novel preliminary evidence needed to support future studies that prospectively measure tolerance development in humans and examine how these individual differences predict risk for problematic alcohol use. This work could uncover a novel AUD risk factor with potential neurobiological underpinnings, which could in turn advance research on the etiology, treatment, and prevention of AUD.

Table 3.

Hierarchical Linear Regression Results Predicting Problematic Alcohol Use

Step	Predictors	β(SE)	p	CI95	Tolerance	VIF
1	Age	0.20(0.08)*	0.01	0.04,0.35	0.86	1.17
	Race	0.03(0.08)	0.68	−0.13,0.20	0.86	1.16
	Sex at birth	−0.05(0.08)	0.56	−0.20,0.11	0.94	1.06
2	Age	0.21(0.08)*	0.01	0.05,0.36	0.85	1.18
	Race	0.02(0.08)	0.80	−0.14,0.19	0.85	1.18
	Sex at birth	−0.04(0.08)	0.62	−0.20,0.12	0.93	1.07
	Initial Subjective Response (SR)	0.09(0.07)	0.19	−0.05,0.23	0.98	1.02
3	Age	0.16(0.08)	0.05	0.00,0.32	0.79	1.26
	Race	0.07(0.09)	0.39	−0.10,0.24	0.79	1.27
	Sex at birth	0.00(0.08)	0.97	−0.16,0.16	0.88	1.13
	Initial SR	0.16(0.08)*	0.03	0.01,0.32	0.81	1.23
	Acquired tolerance	0.19(0.08)*	0.03	0.02,0.36	0.68	1.46
4	Age	0.11(0.07)	0.10	−0.02,0.25	0.79	1.27
	Race	0.11(0.07)	0.15	−0.04,0.26	0.78	1.28
	Sex at birth	−0.02(0.07)	0.80	−0.16,0.12	0.88	1.13
	Initial SR	0.11(0.07)	0.12	−0.03,0.24	0.80	1.25
	Acquired tolerance	0.03(0.08)	0.70	−0.12,0.18	0.63	1.59
	DSM-5-related tolerance	0.53(0.07)*	0.00	0.39,0.67	0.88	1.13
5	Age	0.09(0.06)	0.14	−0.03,0.22	0.77	1.30
	Race	0.11(0.07)	0.11	−0.03,0.25	0.75	1.34
	Sex at birth	0.01(0.06)	0.85	−0.11,0.14	0.87	1.14
	Initial SR	−0.05(0.07)	0.49	−0.17,0.08	0.69	1.46
	Acquired tolerance	−0.13(0.07)	0.08	−0.28,0.02	0.54	1.84
	DSM-5-related tolerance	0.36(0.07)*	0.00	0.22,0.50	0.74	1.35
	Drinking frequency	0.11(0.07)	0.12	−0.03,0.24	0.74	1.35
	Drinking quantity	0.22(0.08)*	0.01	0.07,0.38	0.56	1.78
	Heavy episodic drinking	0.23(0.09)*	0.01	0.06,0.40	0.45	2.21
6	Age	0.08(0.06)	0.18	−0.04,0.20	0.77	1.30
	Race	0.10(0.07)	0.16	−0.04,0.23	0.74	1.35
	Sex at birth	0.00(0.06)	0.94	−0.12,0.13	0.87	1.14
	Initial SR	−0.04(0.06)	0.50	−0.17,0.08	0.69	1.46
	Acquired tolerance	−0.14(0.07)	0.06	−0.28,0.01	0.54	1.84
	DSM-5-related tolerance	0.21(0.08)*	0.01	0.04,0.37	0.50	2.01
	Drinking frequency	0.11(0.07)	0.08	−0.01,0.24	0.74	1.35
	Drinking quantity	0.21(0.08)*	0.01	0.06,0.36	0.56	1.79
	Heavy episodic drinking	0.19(0.08)*	0.02	0.02,0.36	0.44	2.25
	Rapid tolerance growth	0.26(0.08)**	0.00	0.10,0.41	0.52	1.92

Note.

^*p<0.05;

^**p<0.001.

0=White, 1=Black. 0=Male, 1=Female.

Data Availability:

The data analyzed in this manuscript are available from the corresponding author [F.L.W.] upon reasonable request.