Couples’ Alcohol Use in Middle and Later Life: Stability and Mutual Influence

Abstract

Objective:

Wives and husbands may influence one another’s alcohol use, but little is known about within-couple patterns of alcohol consumption over time during midlife and later life. Drawing from a nationally representative U.S. sample of middle-aged and older married couples, we examined individual stability and partner influence in alcohol use across a 16-year period.

Method:

The analytic sample comprised 1,257 married couples age 40 and older who completed nine waves of the Health and Retirement Study (1996–2012). Dyadic multilevel models were estimated to simultaneously examine the stability of one’s own alcohol use (number of drinks per occasion and per week) over time as well as whether partner drinking predicted subsequent alcohol use within couples. Models controlled for marital duration.

Results:

Wives’ and husbands’ own prior alcohol use positively predicted alcohol use across waves, demonstrating individual stability in drinking patterns. Partners’ alcohol use also positively predicted wives’ and husbands’ subsequent alcohol use, revealing partner influence in drinking patterns. Both stability and influence effects were stronger for husbands than for wives.

Conclusions:

This study provides evidence of stability and mutual influence in long-term patterns of alcohol use within mid- and late-life married couples. Findings highlight the interdependence between spouses and indicate that partners’ drinking should be considered when examining men’s and women’s alcohol use over time.

Scholars have long recognized within-couple concordance in health habits. For example, spouses tend to report similar diets, physical activity, and smoking (Falba & Sindelar, 2008; Li et al., 2013; Meyler et al., 2007). Likewise, couples are often concordant in their alcohol use (Birditt et al., 2016; Demers et al., 1999; Falba & Sindelar, 2008; Graham & Braun, 1999; Leonard & Mudar, 2004; Li et al., 2013; McLeod, 1993; Reczek et al., 2016). Yet previous research is largely cross-sectional and/or specific to heavy drinking. Most studies have also focused on young couples. Given the health risks associated with alcohol use among aging adults (Ferreira & Weems, 2008; Wilson et al., 2014), it is imperative to understand how couples’ overall drinking patterns unfold later in life.

This study explored individual stability and partner influence in alcohol use among middle-aged and older married couples. Drawing from a nationally representative U.S. sample, we examined the effects of own and partner’s alcohol use on subsequent use across 16 years. We also considered gender differences in these associations.

Theoretical and empirical background

According to concordance theories, spouses may demonstrate convergence in their drinking that is shaped by several factors (Meyler et al., 2007). People with similar drinking habits may be more likely to marry one another (Leonard & Das Eiden, 1999). Spouses can later influence their partner’s drinking through modeling and social reinforcement (Graham & Braun, 1999; Leonard & Mudar, 2004). Spouses also share a social network that may promote and maintain a couple’s drinking patterns (Meyler et al., 2007; Orford et al., 2002). Complementing these perspectives, interdependence theory holds that spouses mutually affect each other’s thoughts, feelings, and behaviors (Rusbult & Van Lange, 2008). Hence, wives and husbands may influence one another’s alcohol use over time.

Within-couple drinking patterns among aging spouses are crucial to consider for several reasons. First, moderate to heavy drinking in midlife and late life is linked to adverse health consequences including hypertension, stroke, and cognitive impairment (Handing et al., 2015; Hillbom et al., 2011; Jones et al., 2015; Mukamal et al., 2005). Older adults metabolize alcohol more slowly than young adults, and so they may develop health problems from relatively low levels of use (Ferreira & Weems, 2008; Wilson et al., 2014). Second, problematic drinking by older adults is a growing public health issue. U.S. adults age 65 and older are the most frequent binge (i.e., five or more drinks per occasion for men or four or more drinks per occasion for women) drinkers (Centers for Disease Control and Prevention, 2012), and a substantial proportion of middle-aged and older adults drink in a manner that is hazardous to their health (Blazer & Wu, 2009; Breslow et al., 2017; Wilson et al., 2014). Strikingly, more than three quarters (77.8%) of U.S. drinkers age 65 and older regularly use medications that interact with alcohol (Breslow et al., 2015), which poses serious threats such as increased risk of falls, gastrointestinal bleeding, and liver damage (Moore et al., 2007; National Institute on Alcohol Abuse and Alcoholism, 2014). Third, spouses may have stronger influences on one another’s drinking in middle and later life. Older couples have likely been married for a number of years, which may heighten their concordance in drinking (Meyler et al., 2007). Moreover, convergence in alcohol use may arise as spouses age because they spend more time together owing to retirement and/or health problems (Graham & Braun, 1999; Reczek, 2012).

There may be gender differences in within-couple drinking patterns. Compared with women, men drink greater quantities of alcohol, engage in more frequent or problematic drinking, and encounter more social pressure to drink (Erol & Karpyak, 2015). Thus, stability in alcohol use may be more evident for husbands. Regarding partner influence, however, wives may be more affected. In a cross-sectional study of older couples, frequency of drinking with one’s spouse was more strongly related to own frequency of drinking for wives (Graham & Braun, 1999). Similarly, partners’ heavy drinking in aging couples predicted increased drinking for wives but not husbands (Reczek et al., 2016). These findings complement studies of younger couples suggesting that wives may increase their alcohol use to match husbands’ consumption (Leonard & Mudar, 2003; Roberts & Leonard, 1998).

Study aims and hypotheses

This study had two major aims. First, we evaluated the effects of own previous alcohol use on wives’ and husbands’ later use. We hypothesized the following. Hypothesis 1: Wives’ and husbands’ prior number of drinks per drinking occasion and number of drinks per week will be positively associated with their own number of drinks per occasion and per week over time. Hypothesis 2: The effects of one’s own prior alcohol use on subsequent use will be significantly stronger for husbands than for wives.

Second, we considered the effects of partners’ previous alcohol use on wives’ and husbands’ later use with two additional hypotheses. Hypothesis 3: Partners’ prior number of drinks per drinking occasion and number of drinks per week will be positively associated with wives’ and husbands’ own number of drinks per occasion and per week over time. Hypothesis 4: The effects of one’s partner’s prior alcohol use on subsequent use will be significantly stronger for wives than for husbands.

Method

Participants and sample

The U.S. sample was drawn from nine waves of the nationally representative Health and Retirement Study (1996–2012). Ethical approval was not required for this study because we used publicly available secondary data with no individual identifiers.

The Health and Retirement Study has collected data biennially since 1992 with a response rate of more than 80% at each wave. Because of variations in drinking variables in 1992 and 1994, we used data from nine waves from 1996 to 2012. There were 2 years in between each study wave. In total, 3,142 participants (1,571 heterosexual couples) reported that they were married to the same spouse from 1996 to 2012. To examine drinking patterns in mid- and late-life, we selected 1,551 couples in which both spouses were age 40 or older at baseline (1996) and participated in each wave. We removed 231 couples in which neither spouse reported drinking over the last 3 months in any of the waves. We then removed 63 couples missing data on one or more study variables. Thus, we focused on 1,257 consistently married couples (i.e., spouses who remained married to one another across waves included in the present study) who had been married for an average of 31.45 years (SD = 11.39, range: 0.50–55.90) at baseline (Table 1).

Table 1.

Baseline characteristics and scores on key variables for wives and husbands

Characteristic	Wives M (SD)	Husbands M (SD)
Age, in years	56.91 (5.93)	60.48 (5.28)
Education, in years	12.84 (2.86)	13.04 (3.24)
No. of chronic health conditions	1.08 (1.01)	1.19 (1.00)
No. of living children	3.40 (1.90)	3.40 (1.90)
No. of drinks per occasion	0.66 (1.10)	1.32 (1.66)
No. of drinks per week	1.97 (4.51)	4.59 (7.69)
	Proportions
Race/ethnicity (non-Hispanic White)	.91	.90
Employment status (works for pay)	.56	.69
Marital order (first marriage)	.76	.73
Prescription drug use	.71	.51
Lifetime history of alcohol problems	.05	.15
Total household income
≤$24,999	.14
$25,000–$49,999	.30
$50,000–$74,999	.24
$75,000–$99,999	.12
≥$100,000	.21

Notes: N = 1,257 married couples. No. = number.

Compared with the 294 couples who were removed, the 1,257 couples in this study were younger, t(3,100) = -2.00, p = .05; had been married for fewer years, t(3,090) = -3.02, p = .003; and had more years of education, t(3,100) = 7.40, p < .001 at baseline. Couples in this study also reported more baseline drinks per occasion, t(828.98) = 10.30, p < .001; and per week, t(756.30) = 6.56, p < .001.

Measures

Alcohol use.

At each wave, participants reported, on average (a) how many days per week they drank in the last 3 months and (b) how many drinks they had on drinking days. Drinks per day were multiplied by drinking days each week to determine drinks per week. Post hoc tests examined abstinence status across waves from participants’ reports on whether they drank alcohol in the last 3 months (1 = yes, 0 = no).

Covariates.

Baseline marital duration in years was considered because within-couple drinking patterns may vary by length of marriage (Leonard & Mudar, 2004).

Post hoc models controlled for sociodemographic and health correlates of drinking among older individuals and couples (Birditt et al., 2016; Breslow et al., 2015; Reczek et al., 2016). Time-invariant covariates included baseline age, education in years, and race/ethnicity (1 = non-Hispanic White, -1 = racial/ethnic minority). Time-variant covariates (same and previous wave reports) included own and partner self-reported number of diagnosed chronic health conditions (arthritis, cancer, diabetes, heart disease, hypertension, lung disease, stroke) and whether prescription drugs were regularly taken to treat these conditions, a psychiatric condition, or other unspecified conditions (1 = yes, -1 = no).

Couples’ history of alcohol problems was assessed with the four-item CAGE questionnaire (Ewing, 1984; Maisto & Saitz, 2003; Mayfield et al., 1974). Participants reported whether they had ever experienced (a) feeling they should cut down on drinking, (b) being annoyed by people criticizing their drinking, (c) feeling bad or guilty about drinking, and (d) drinking first thing in the morning to steady nerves or relieve a hangover. Summed scores of 2 or higher reflect a clinically significant history, which was reported by one or both spouses in 231 couples (18.4%).

Statistical analysis

Stability and influence models were estimated using the MIXED procedure in IBM SPSS Statistics for Windows, Version 24 (IBM Corp., Armonk, NY). This multilevel approach is a version of the actor-partner interdependence model (Kenny et al., 2006) that enables the modeling of own and partner influence from one time point to the next (West, 2013). Models included the recommended two levels for longitudinal dyadic data, with the lower level representing variability attributable to within-person repeated measures for wives and husbands and the upper level representing between-couple variability across wives and husbands (Kenny et al., 2006). Correlated errors between wives and husbands were allowed within a given wave using a heterogeneous compound symmetry (CSH) error structure. In this study, actor effects represent wives’ and husbands’ own stability in drinking across waves (e.g., wives’ drinks per week predicting their own subsequent drinks per week). Partner effects represent the influence of partners’ drinking on wives’ and husbands’ own drinking across waves (e.g., husbands’ drinks per week predicting wives’ subsequent drinks per week). Maximum likelihood estimation was used to permit model comparison.

Separate models were conducted for drinks per occasion and per week to consider stability and influence in how much alcohol wives and husbands consumed when they drank as well as in their weekly levels of alcohol use. Before the analysis, tests of distinguishability were conducted to determine whether dyads were distinguishable by gender (Kenny et al., 2006). Compared with an indistinguishable model, the model fit significantly improved when including gender effects for drinks per occasion, χ²(4) = 7,398.53, p < .001, and drinks per week, χ²(4) = 9,178.83, p < .001. Thus, separate intercepts and slopes were estimated for wives and husbands using spouse gender (1 = wife, -1 = husband) as a distinguishing variable (Kenny et al., 2006). Intercepts and slopes were permitted to vary between and within couples. TEST statements were used in the syntax to evaluate significant gender differences for each effect.

Marital duration was included as a covariate in the first step of the models. Own and partner lagged drinks per occasion/week were entered as predictors in the second step to examine how couples’ alcohol use in the previous wave was associated with their current alcohol use over time. Baseline continuous predictors and covariates were grand-mean centered, whereas time-varying continuous predictors and covariates were centered on the person-level grand mean (i.e., the individual’s overall mean level across the study period) (Kenny et al., 2006). Significant differences between the fit of the full model relative to the covariate-only model were assessed by subtracting their -2 log likelihood estimations and examining differences on a chi-square distribution with degrees of freedom equaling the change in number of model parameters (Singer & Willett, 2003).

Post hoc tests used generalized estimating equation models to assess stability and influence in wives’ and husbands’ abstinence status. This approach is recommended for analyzing dyadic data with categorical outcomes (Loeys et al., 2014).

Results

Paired t tests were performed to examine baseline gender differences. Husbands drank more per occasion, t(1,256) = -13.48, p < .001, and week, t(1,256) = -11.88, p < .001, than wives. Spouses’ reports were positively correlated (r = .27, p < .001 for drinks per occasion and week).

Model parameters are shown in Table 2 (drinks per occasion) and Table 3. (drinks per week). Compared with the covariate-only model, own and partner alcohol use accounted for significantly greater variance in drinking patterns over time for drinks per occasion, χ²(2) = 1,094.39, p < .001, and per week, χ²(2) = 1,889.66, p < .001.

Table 2.

Stability and influence model estimates for wives’ and husbands’ drinks per occasion across waves

Parameter	Wives’ drinks per occasion		Husbands’ drinks per occasion
	b	SE	b	SE
Step 1
Intercept	0.57***	0.02	1.09***	0.03
Baseline marital duration	-0.003*	0.001	-0.01***	0.002
Step 2
Actor lagged drinks per occasion	0.12***	0.01	0.17***	0.01
Partner lagged drinks per occasion	0.02***	0.01	0.08***	0.02
Pseudo R2	0.05		0.06
Δ -2 log likelihood	1,094.39***

Notes: Coefficients are presented from the full model (Step 2). Pseudo R² is the proportion of explained variance in the random effect of the full model relative to the covariate-only model (Step 1). Change in -2 log likelihood for Step 2 (57,007.35) is based on comparison with Step 1 (58,101.74). N = 1,257 married couples.

^*p ≤ .05;

^***p ≤ .001.

Table 3.

Stability and influence model estimates for wives’ and husbands’ drinks per week across waves

Parameter	Wives’ drinks per week		Husbands’ drinks per week
	b	SE	b	SE
Step 1
Intercept	1 92***	0.08	3.98***	0.13
Baseline marital duration	-0.01*	0.01	-0.05***	0.01
Step 2
Actor lagged drinks per week	0.15***	0.01	0.20***	0.01
Partner lagged drinks per week	0.02**	0.01	0.07**	0.02
Pseudo R2	0.06		0.06
Δ -2 log likelihood		1,889.66***

Notes: Coefficients are presented from the full model (Step 2). Pseudo R² is the proportion of explained variance in the random effect of the full model relative to the covariate-only model (Step 1). Change in -2 log likelihood for Step 2 (121,729.42) is based on comparison with Step 1 (123,619.08). N = 1,257 married couples.

^*p ≤ .05;

^**p ≤ .01;

^***p ≤ .001.

Does own alcohol use show stability over time?

Wives’ alcohol use.

Wives’ prior drinking positively predicted their own subsequent drinks per occasion (b = 0.12, p < .001) and per week (b = 0.15, p < .001).

Husbands’ alcohol use.

Husbands’ previous drinking positively predicted their own later drinks per occasion (b = 0.17, p < .001) and per week (b = 0.20, p < .001). The effects for stability in drinks per occasion (b = -0.05, p = .01) and per week (b = -0.06, p = .004) were significantly stronger for husbands than for wives.

Does partners ’ alcohol use influence own use over time?

Wives’ alcohol use.

Husbands’ prior drinking positively predicted wives’ subsequent drinking per occasion (b = 0.02, p = .001) and per week (b = 0.02, p = .002).

Husbands’ alcohol use.

Wives’ previous drinking positively predicted husbands’ later drinking per occasion (b = 0.08, p < .001) and per week (b = 0.07, p = .002). The effects for partner influence on drinks per occasion (b = -0.06, p = .002) and per week (b = -0.04, p = .05) were significantly stronger for husbands than for wives.

Post hoc tests

We estimated models controlling for baseline sociodemographics, own and partner time-variant chronic health conditions and prescription drug use, and couples’ history of alcohol problems in a reduced sample of 1,185 couples with complete data. The pattern of findings from the main analyses remained, with the exception of the loss of significant gender differences in the strength of partner effects for drinks per occasion and per week.

A model assessing stability and influence in abstinence status was also estimated, controlling for marital duration. For wives and husbands, abstinence across waves was more likely if they (b = 2.64, p < .001 and b = 2.48, p < .001, respectively) or their partner (b = 0.63, p < .001 and b = 0.49, p < .001, respectively) abstained from drinking in the previous wave. The stability effect was significantly stronger for wives than for husbands (b = 0.08, p = .02), but there was no significant gender difference in the influence effect.

Discussion

This study demonstrates that drinking patterns among middle-aged and older spouses are characterized by both individual stability and partner influence. We built on previous research with younger couples by examining longitudinal drinking patterns within mid- and late-life couples. Overall, we found that levels of alcohol consumption depended partly on one’s own prior drinking. Beyond these effects, partners’ drinking wielded a strong influence on subsequent alcohol use. These findings are in line with concordance and interdependence perspectives on marriage (Meyler et al., 2007; Rusbult & Van Lange, 2008) and highlight the reciprocal influences within couples. Given that almost 60% of middle-aged and older people in the United States are married (U.S. Census Bureau, 2013), this study has crucial implications for policymakers and clinicians working with aging adults at risk of health problems related to their drinking habits.

Individual stability of alcohol use

Wives’ and husbands’ own alcohol use predicted subsequent drinks per occasion and per week. These findings imply that middle-aged and older adults who consume alcohol typically continue these habits, which may pose considerable health risks. Any sustained level of drinking may be detrimental for aging adults (Wilson et al., 2014), and older people commonly are involved in stressful situations (e.g., retirement, caring for an ill relative) that may exacerbate their drinking habits (Blow & Barry, 2012; Glass et al., 1995; Satre et al., 2012). Furthermore, many older drinkers take prescription drugs that could interact with alcohol in a harmful and potentially fatal manner (Breslow et al., 2015). Consequently, stable drinking patterns in middle and later life are a public health concern, especially considering recent evidence that problematic alcohol use is on the rise for aging men and women alike (Breslow et al., 2017). In line with the hypothesis, husbands showed significantly greater stability in their own drinks per occasion and per week than did wives. Similarly, post hoc tests showed that wives were more likely than husbands to remain abstinent from using alcohol. These findings indicate that aging men may be particularly likely to maintain their alcohol use over time, regardless of their partner’s drinking patterns.

Partner influence on alcohol use

Over and above the individual stability in alcohol use, partners’ drinking influenced own drinks per occasion and per week for wives and husbands. These findings support research showing high concordance in couples’ drinking (e.g., Demers et al., 1999; Leonard & Mudar, 2004; Li et al., 2013). We extended this work by demonstrating that middle-age and older spouses mutually influence one another’s alcohol consumption when they drink. Counter to prediction, partners’ influence on drinks per occasion and per week was stronger for husbands than for wives. That is, wives’ alcohol use was found to influence husbands’ drinking more so than the reverse situation. This study contrasts with research showing that within-couple drinking patterns shift from wives influencing husbands as newlyweds to husbands influencing wives by the second year of marriage (Leonard & Mudar, 2004). Taken together, these findings raise the question of whether men’s drinking becomes more interdependent with their partner as they age. Whereas older women tend to maintain relationships outside of their marriage, aging men often rely on their spouse as a primary source of social support (Antonucci, 2001). Such reliance on spouses may render men more susceptible than women to partner influence in their alcohol use during middle and later life.

Although perhaps counterintuitive, partners’ drinking may provide some benefits. Previous research suggests that partners’ moderate to heavy drinking can enhance the affective climate of marriage (Derrick et al., 2010; Orford et al., 2002). People often are happier when they drink, which could lead to more positive social experiences (Geiger & MacKerron, 2016). Consuming alcohol may be linked to better mood and greater social bonding for men in particular (Fairbairn et al., 2015). Moreover, although the negative consequences of alcohol use tend to be delayed and less likely to affect drinking behavior, the positive aspects are immediate and therefore more likely to maintain couples’ drinking patterns (McCrady & Epstein, 2015). Spouses may share more rewarding interactions when they both frequently consume alcohol that could play a role in the long-term maintenance of within-couple drinking patterns.

Ultimately, however, spouses’ drinking may have negative health implications. Within older couples, a partner’s more frequent drinking has been found to predict a greater likelihood of one’s own high-risk drinking (Moos et al., 2010). Similarly, among midlife adults with a history of alcohol use disorder, more frequent drinking with one’s spouse is linked to problematic alcohol use (Blonigen et al., 2015). Moreover, older adults’ high-risk drinking patterns may contribute to greater conflict and reduced support within families (Brennan & Moos, 1990; Brennan et al., 1994). A lack of family support can present substantial challenges in later life because aging adults often rely on their families to provide health-related assistance (National Alliance on Caregiving & AARP Public Policy Institute, 2015). Thus, partner influence in alcohol use may be especially harmful when one or both spouses drink frequently. Post hoc models revealed that partners’ abstinence increased the likelihood that wives and husbands abstained from drinking across time. Notably, such partner influence was over and above the significant effects of individual stability in abstinence from alcohol use. These findings suggest that spouses are also influential in limiting their partner’s drinking, which may be medically advised as people age and develop greater health-related vulnerabilities to alcohol.

Strengths, limitations, and future directions

Strengths of this study include our focus on drinking patterns in middle-aged and older couples, 16-year longitudinal data on alcohol use, and a nationally representative base sample. Our use of the dyad as the unit of analysis allowed the modeling of individual stability and partner influence in couples’ drinking over time. In addition, this study shows that stability and influence in couples’ alcohol use patterns is evident when considering their number of drinks per drinking occasion as well as their average weekly levels of alcohol consumption. The findings held even after accounting for marital duration, sociodemographics, own and partner health characteristics, and couples’ lifetime history of alcohol problems. Hence, the present study demonstrates that previous patterns of own and partner drinking are robust predictors of alcohol consumption in middle and later life.

We acknowledge several limitations. First, frequent drinking is more socially acceptable for men (Erol & Karpyak, 2015), and therefore alcohol use may have been overreported by husbands but underreported by wives. Second, the findings are specific to long-term married couples and may not translate to couples who end their relationship. Couples who had little influence on one another’s drinking, for instance, may have divorced. Third, the findings may not generalize to recently married older spouses, cohabiting partners, or same-sex couples, all of whom may differ in their drinking patterns. Fourth, most couples were non-Hispanic White, and so the findings may not apply to more diverse couples. Last, the sample reported a low incidence of lifetime alcohol problems, and data on current alcohol problems were unavailable. Furthermore, overall alcohol use was relatively low and the sample included participants who abstained from drinking in one or more waves, limiting conclusions about potential health impacts. This study nevertheless generates valuable insights into couples’ long-term drinking patterns as they enter late middle and older adulthood, when sustained drinking may be especially hazardous to their health (Blow & Barry, 2012; Ferreira & Weems, 2008; Wilson et al., 2014).

A key area for future research is to explore the mechanisms through which individual stability and partner influence in alcohol use may occur in midlife and later life. Determining factors that may mitigate or intensify these patterns would also be informative. Drinking habits may remain more stable, for instance, among spouses who have friends or relatives with similar alcohol use (Moos et al., 2010; Orford et al., 2002). Likewise, people who experience stressors may be more likely to increase their drinking when exposed to their partner’s drinking (Lemke et al., 2008). Such work would augment interventions to reduce at-risk alcohol use and may help to identify couples who are more or less vulnerable to these drinking patterns.

Finally, future research should consider how this study informs clinical work with aging adults at risk for problematic drinking. Relative to individual-based approaches, couple-based models such as Alcohol-Focused Behavioral Couple Therapy have been found to be as or more efficacious in decreasing alcohol use and improving partners’ relationship quality (McCrady et al., 2016). The present findings boost the rationale for couple-based strategies by showing that partner influence has an impact on alcohol use beyond one’s own previous drinking and the couples’ history of alcohol problems. Exposure to partners’ drinking may be a powerful antecedent to alcohol use and could involve positive consequences (e.g., sharing enjoyable activities) that maintain drinking behavior (McCrady & Epstein, 2015). Thus, spouses may inadvertently contribute to problematic drinking patterns. At the same time, spouses may be a valuable resource in reducing at-risk drinking. Notably, improvements in health behaviors by one spouse—including reduced drinking—can lead to corresponding improvements made by partners (Falba & Sindelar, 2008).

The current findings may also inform brief interventions in healthcare settings such as the evidence-based Screening, Brief Intervention, and Referral to Treatment approach (Agerwala & McCance-Katz, 2012), which includes a focus on identifying, reducing, and preventing the problematic use of alcohol. The assessment of within-couple alcohol use patterns could facilitate early identification of aging individuals who are at risk for problematic drinking. Additionally, this information may help to determine protective factors (e.g., a partner who abstains from drinking) that reduce or prevent heavy or risky alcohol use in this population.

In sum, this study provides evidence that alcohol use patterns remain stable and are shaped by mutual influences within mid- and late-life couples. These findings underscore the interdependence in spouses’ drinking patterns and demonstrate the need for future work to adopt a couple-focused lens when considering men’s and women’s long-term alcohol use.

Acknowledgments

The authors thank Angela Turkelson for her assistance with the preparation of the data.