Spouses shape each other’s sleep, drinking and social habits in later life

When one partner in a middle-aged or older Chinese couple develops an unhealthy habit, the other often follows, and the influence runs in both directions. That is the central finding of a longitudinal analysis of more than 6,500 married couples drawn from the China Health and Retirement Longitudinal Study, known as CHARLS, a nationally representative survey of adults aged 45 and older. Researchers tracked seven health behaviors across four survey waves between 2011 and 2018 and found that spouses were significantly similar in every behavior examined, and that for several of them, one partner’s behavior at an earlier wave predicted the other partner’s behavior at the next.

The study, published in the Journal of Behavioral Medicine, analyzed 6,579 couples, or 13,158 individuals, who were married to another CHARLS participant in the same household at each wave analyzed. The behaviors included obesity or overweight, physical inactivity, social inactivity, short or long sleep, extended napping, current smoking, and heavy drinking. Each was treated as a binary indicator, so a person was classified as, for example, a short or long sleeper if their nocturnal sleep fell outside age-specific recommendations, or an extended napper if their post-lunch nap exceeded 60 minutes. The team used generalized estimating equations to measure cross-sectional concordance between spouses and cross-lagged panel models to test whether a spouse’s earlier behavior predicted a partner’s later behavior, adjusting for age, education, household registration status, work status, falls, difficulties with daily activities, multimorbidity, and self-rated health.

Cross-sectionally, the associations were substantial. Heavy drinking showed the largest adjusted spousal association, with an odds ratio of 5.60, meaning a husband who drank heavily was more than five times as likely to have a wife who drank heavily, and vice versa. Extended napping followed with an odds ratio of 3.19, physical inactivity with 3.56, and obesity showed the smallest association at 1.33. All seven behaviors were significantly associated between spouses both before and after adjustment for covariates, indicating that concordance was not simply an artifact of the demographic characteristics the couples shared.

The longitudinal results were more differentiated. Four behaviors showed genuinely bidirectional influence: short or long sleep, with a wife-to-husband odds ratio of 1.17 and a husband-to-wife odds ratio of 1.19; heavy drinking, with wife-to-husband 2.78 and husband-to-wife 2.29; social inactivity, with wife-to-husband 1.14 and husband-to-wife 1.11; and extended napping, with wife-to-husband 1.55 and husband-to-wife 1.28. Obesity showed a significant husband-to-wife path, at an odds ratio of 1.30, while the reverse path from wife to husband was not statistically significant. Physical inactivity and current smoking showed no significant longitudinal influence in either direction, with all relevant tests returning p-values above 0.19.

A formal test of symmetry found no statistically significant difference between the wife-to-husband and husband-to-wife paths for any of the seven behaviors after correction for multiple comparisons. The smallest unadjusted p-value appeared for extended napping, where the wife-to-husband effect of 1.55 was numerically larger than the husband-to-wife effect of 1.28, but this did not survive false discovery rate correction. In practical terms, the data support a picture of mutual influence rather than one in which husbands or wives systematically drive their partner’s behavior more strongly.

The researchers also examined residual concordance, asking whether spouses’ current behaviors remained correlated after accounting for each partner’s own prior behavior and the other partner’s prior behavior. Physical inactivity, social inactivity, short or long sleep, and extended napping retained significant residual correlations at every wave-transition, ranging from 0.104 to 0.253. Obesity, current smoking, and heavy drinking showed near-zero and largely non-significant residual correlations. This pattern suggests that for routine-dependent behaviors such as sleep, napping, social participation, and physical activity, some of the similarity between spouses arises from contemporaneous sharing of daily schedules, joint activities, and the household environment, whereas for drinking and obesity, concordance appears to accumulate more through partner influence over time.

Sleep stands out as an inherently dyadic behavior. Unlike eating or exercising, which can be done separately, sleep is constrained by a shared bedroom, synchronized sleep-wake routines, and the physical presence of another person in the bed. The authors suggest that when one partner experiences nighttime awakening, the other may be disturbed, and when one partner initiates earlier sleeping with less awakening, the other may follow with synchronized patterns. Extended napping, a culturally prominent post-lunch practice in China, showed the same bidirectional pattern, with the wife-to-husband effect notably larger than the reverse, though the formal symmetry test did not confirm a statistically significant difference.

Heavy drinking showed the largest effect sizes of any behavior in either direction, which the authors interpret in light of shared access to alcohol within the household and the social nature of drinking in Chinese culture, where drinking together during meals is a common ritual. Social inactivity, defined as no participation in any of the activities CHARLS asked about, including friend interactions, games or mahjong, clubs, community organizations, volunteering, unpaid caregiving, and educational courses, also showed bidirectional influence. The authors note that middle-aged and older Chinese couples predominantly socialize together rather than independently, visiting relatives, attending community events, or engaging in joint leisure such as square dancing or card games, creating a natural context for mutual influence.

The absence of longitudinal effects for physical inactivity and smoking is instructive. Physical inactivity showed the strongest cross-sectional concordance of any behavior, yet no directional influence over time, which the authors attribute to assortative mating, in which individuals with similar activity levels marry each other, and to shared environmental factors, rather than to one spouse directly changing the other’s behavior. Smoking, meanwhile, was the only behavior with neither a significant spousal association nor reciprocal influence, likely because smoking among Chinese women in this cohort has historically been socially stigmatized and rare, so a husband’s smoking did not meaningfully predict his wife’s status.

The authors conclude that the couple should be considered a fundamental unit for health promotion, and that couple-based behavioral change interventions may be more effective than individual-focused approaches for behaviors that are highly shared and transmissible within couples. They propose dyadic screening, in which clinicians assess the partner of a patient presenting with an unhealthy behavior, and behavior-specific intervention targets: couple-level approaches for bidirectional sleep and social behaviors, engagement of husbands for obesity’s trending husband-to-wife pattern, and attention to shared environmental and motivational barriers for physical inactivity. The study’s limitations include reliance on self-reported behaviors, the possibility of unmeasured confounding from factors such as social support, neighborhood environment, or pre-marriage behaviors, and a sample restricted to middle-aged and older Chinese adults, which may limit generalizability to younger or non-Chinese populations.

Subject of Research: Longitudinal spousal concordance and bidirectional influence in seven health behaviors among middle-aged and older Chinese couples

Article Title: Health behavior concordance in middle-aged and older Chinese couples: longitudinal analysis from the China Health and Retirement Longitudinal Study (CHARLS)

Article References: Zhong, J., Dai, Y., Qi, X., Liang, W., Zhao, W., & Cao, J. (2026). Health behavior concordance in middle-aged and older Chinese couples: longitudinal analysis from the China Health and Retirement Longitudinal Study (CHARLS). Journal of Behavioral Medicine. https://doi.org/10.1007/s10865-026-00721-y

Image Credits: AI Generated

DOI: 10.1007/s10865-026-00721-y

Keywords: spousal concordance, health behaviors, CHARLS, cross-lagged panel models, sleep, napping, heavy drinking, social inactivity, obesity, physical inactivity, smoking, Chinese couples