Addressing common biases in the evaluation of lifetime alcohol consumption patterns and dementia risk: the EPIC-Spain Dementia cohort

José María Huerta^1,2*Sandra M Colorado-Yohar^1,2,3*M. Encarnación Andreu-Reinón^2,4Olatz Mokoroa^5,6Mikel Tainta^5,7Marcela Guevara^1,8,9Alba Gasque^1,8,9Jesús Castilla⁸Dafina Petrova^1,10,11,12Marta Crous-Bou^13,14Raul Zamora-Ros^13,14María José Sánchez^1,10,11,12María Dolores Chirlaque^1,15,2

• ¹Centro de Investigacion Biomedica en Red de Epidemiologia y Salud Publica, Madrid, Spain
• ²Instituto Murciano de Investigacion Biosanitaria, El Palmar, Spain
• ³Universidad de Antioquia, Medellín, Colombia
• ⁴Hospital Clinico Universitario Virgen de la Arrixaca, El Palmar, Spain
• ⁵Biogipuzkoa Instituto de Investigacion Sanitaria, San Sebastián, Spain
• ⁶Gobierno Vasco Departamento de Salud Direccion General, Vitoria-Gasteiz, Spain
• ⁷Osakidetza Servicio Vasco de Salud, Vitoria-Gasteiz, Spain
• ⁸Instituto de Salud Publica y Laboral de Navarra, Pamplona, Spain
• ⁹Instituto de Investigacion Sanitaria de Navarra, Pamplona, Spain
• ¹⁰Escuela Andaluza de Salud Publica, Granada, Spain
• ¹¹Instituto de Investigacion Biosanitaria de Granada, Granada, Spain
• ¹²Hospital Universitario Virgen de las Nieves, Granada, Spain
• ¹³Institut Catala d'Oncologia, L'Hospitalet de Llobregat, Spain
• ¹⁴Institut d'Investigacio Biomedica de Bellvitge, Barcelona, Spain
• ¹⁵Universidad de Murcia, Murcia, Spain

ABSTRACT Background: Alcohol consumption has been described to exhibit a J-shaped relationship with dementia risk, but previous observations may be partly biased due to 'sick-quitters' and competing risks of death. Objective: To examine the association between baseline and lifetime alcohol consumption and the risk of dementia and subtypes in a large Mediterranean cohort, accounting for lifetime drinking patterns, potential confounding, and competing risks of death. Methods: Prospective study of 30,211 participants, 29–69 years at recruitment (1992–1996), from the EPIC-Spain Dementia cohort. Alcohol intake was assessed using a validated dietary history and retrospective questionnaires covering ages 20, 30, and 40 years. Dementia cases (n = 1,114) were ascertained through linkage with healthcare and mortality databases and individual medical record review over a mean follow-up of 22.8 years. Multivariate competing risk models were used to estimate sub-hazard ratios (sHRs) for dementia by categories of baseline and lifetime alcohol consumption, using lifetime abstainers as the reference group. Results: Mean lifetime alcohol consumption was 41.9 and 4.4 g/d in men and women, respectively. No significant associations were found between baseline or lifetime alcohol consumption and risk of overall dementia (sHRcurrent vs. never = 0.96, 95% CI: 0.82, 1.13; sHRever vs. never = 0.96, 95% CI: 0.82, 1.11), Alzheimer's disease, or non-Alzheimer subtypes. These null findings remained consistent across strata of sex, BMI or smoking categories, and by beverage type. Sensitivity analyses excluding mis-reporters of energy intake or low-quality diagnoses yielded similar results. Conclusions: In this large prospective cohort with over 1,100 dementia cases and long-term follow-up, alcohol consumption was not significantly associated with dementia risk. These findings challenge the notion of a protective effect of moderate drinking and warrant continued investigation using methodologically rigorous approaches to clarify the role of alcohol dose, timing and pattern on dementia risk.