Quitting or Cutting Alcohol After Breast Cancer Did Not Lower Death Risk in Older Survivors

For older women who have survived breast cancer, one of the most common pieces of lifestyle advice is to cut back on alcohol. A new long-term study now complicates that picture: among postmenopausal breast cancer survivors, quitting or reducing alcohol consumption was not associated with a lower risk of dying from any cause, and in one time window, quitting was actually linked to a higher risk of death. The only clear warning sign ran in the opposite direction—women who were already drinking at least one drink per day and then increased their intake before diagnosis had a notably higher mortality risk.

The findings come from the American Cancer Society’s Cancer Prevention Study-II Nutrition Cohort, a prospective study that began in 1992–1993 and collected repeated health questionnaires over more than two decades. The research team, led by Samantha Rees and including colleagues at the American Cancer Society, focused on 1,558 postmenopausal women diagnosed with non-metastatic invasive breast cancer. The women had a mean age at diagnosis of 71 years, and the sample was 98.5% White and 44.3% college educated. Because the cohort was overwhelmingly postmenopausal at baseline, the analysis could not address premenopausal or perimenopausal patients.

What sets this study apart from most previous work is its focus on change. Rather than comparing drinkers with non-drinkers at a single point in time—an approach that can be biased if former drinkers are lumped in with never-drinkers—the researchers tracked how each woman’s alcohol intake shifted across questionnaires administered in 1992, 1997, 1999, 2003, 2007, and 2013. Each drink was standardized to 14 grams of ethanol, and consumption at every timepoint was sorted into five categories ranging from zero to two or more drinks per day. Women whose category changed were classified as having quit, reduced, or increased their intake, while those who stayed in the same category served as the stable reference group.

The researchers examined change during three distinct intervals relative to the cancer diagnosis. The pre-diagnosis interval compared baseline intake with the last questionnaire completed at least one year before diagnosis, a median of two years earlier. The pre- to post-diagnosis interval spanned the diagnosis itself, comparing the last pre-diagnosis measure with the first measure taken afterward. The post-diagnosis interval compared the first two measures taken after diagnosis. Person-time for mortality follow-up began one year after the relevant measurement in each case, a deliberate design choice intended to reduce reverse causation—the possibility that impending illness, rather than alcohol itself, drove both the behavior change and the death.

Over a median follow-up of 12.6 years, 750 of the women died, including 130 from breast cancer. Using Cox proportional hazards regression adjusted for prior alcohol intake, age at diagnosis, tumor stage, estrogen receptor status, smoking pack-years, and comorbidities such as diabetes and cardiovascular disease, the team estimated hazard ratios for each change category against stable consumption. The results diverged sharply by timing. Quitting before diagnosis showed a nonsignificant hint of lower mortality (hazard ratio 0.91, 95% confidence interval 0.69–1.21). Quitting across the diagnosis period was associated with a 39% higher risk of all-cause mortality (hazard ratio 1.39, 95% confidence interval 1.12–1.73). Quitting after diagnosis showed no association at all (hazard ratio 0.98, 95% confidence interval 0.70–1.35). Reduction of intake was not associated with mortality in any interval.

The most striking signal emerged in a secondary analysis restricted to women drinking at least one drink per day before diagnosis. Among this group, an increase in consumption before diagnosis was associated with an 83% higher risk of all-cause mortality (hazard ratio 1.83, 95% confidence interval 1.02–3.31). This finding was echoed in analyses of women consuming one drink per day or less in the pre-diagnosis interval, where increased intake likewise carried elevated risk. In other words, the direction of change mattered: drinking more appeared harmful, but drinking less or stopping did not clearly appear protective within the timeframes studied.

The elevated mortality among women who quit around the time of diagnosis is the study’s most counterintuitive result, and the authors are careful about interpretation. They suggest it may reflect confounding by factors not captured in the data, such as overall health status, treatment side effects, or unmeasured comorbidities. Women who quit drinking were, in fact, different from stable consumers in ways that matter: they were less educated, more likely to have or develop diabetes or cardiovascular disease, and more likely to report poor or fair perceived health. Reducers consumed the most alcohol before diagnosis, averaging 22.2 grams per day, while quitters consumed the least, at 4.6 grams per day—meaning many quitters had little alcohol exposure to eliminate in the first place.

Sensitivity analyses reinforced the possibility that health status shapes the apparent associations. When the team restricted analyses to never smokers, women without major comorbidities, those reporting good or excellent health, or those with localized tumors, most associations disappeared or weakened. Among women with estrogen receptor-positive tumors, quitting before diagnosis was associated with increased mortality (hazard ratio 1.34, 95% confidence interval 1.01–1.77), a result the authors note with caution, given that the low prevalence of receptor-negative cases prevented planned stratification. When they examined sustained change—women who altered their drinking and maintained that change at the second post-diagnosis measurement, over a median of four years—the association between cessation and higher mortality was no longer statistically significant, hinting that a longer duration of sustained abstinence might be needed to observe any benefit.

The authors draw a parallel with smoking research, where studies consistently show that quitting earlier in life and sustaining cessation for a decade or more yield the clearest survival gains. Their cohort, with a mean age at diagnosis of 71 and no data on lifetime drinking history, may simply have been too old and followed too briefly after change to capture a benefit. It is also possible that some effects of alcohol on health are irreversible, in which case modifying behavior later in life would not improve survival. Alternatively, cessation in later life may genuinely have little impact on mortality for this population. The study cannot distinguish among these explanations.

Several limitations temper the conclusions. Alcohol intake was self-reported, and no valid objective biomarker of long-term consumption exists, though the authors argue any misclassification would be unrelated to mortality. The cohort had little heavy drinking—only about 11% of women consumed two or more drinks per day—limiting the ability to assess large changes in intake. No data on breast cancer treatment were available, though stage and receptor status were adjusted. Post-diagnosis analyses are vulnerable to survival bias, since women had to live long enough after diagnosis to provide multiple post-diagnosis measures. And because the sample was 98.5% White and entirely older and postmenopausal, the findings may not generalize to younger women or women of color. Breast cancer-specific mortality analyses, involving fewer than 200 deaths, are presented only in supplementary materials and should be interpreted cautiously. Despite these caveats, the study carries a practical observation: roughly 86% of survivors who drank before diagnosis continued drinking afterward, and about 12% increased their intake beyond pre-diagnosis levels. The authors call for future research harmonizing data across multiple cohorts and including younger populations to clarify when, and for whom, changing alcohol consumption after a breast cancer diagnosis might improve survival.

Subject of Research: Change in alcohol consumption and all-cause mortality among postmenopausal breast cancer survivors

Article Title: Change in alcohol consumption and all-cause mortality among postmenopausal women diagnosed with breast cancer

Article References: Rees, S., Bodelon, C., Chakraborty, S., Millen, A. E., Wang, Y., Faw, K., Teras, L. R., Shams-White, M. M., Patel, A. V., & Freudenheim, J. L. (2026). Change in alcohol consumption and all-cause mortality among postmenopausal women diagnosed with breast cancer. Breast Cancer Research and Treatment, 220(1), Article 14. https://doi.org/10.1007/s10549-026-08113-1

Image Credits: AI Generated

DOI: 10.1007/s10549-026-08113-1

Keywords: breast cancer, alcohol consumption, all-cause mortality, cancer survivors, postmenopausal women, Cox proportional hazards, Cancer Prevention Study-II, epidemiology, survivorship, health behavior change, alcohol cessation, Change