Dietary protein intake and all-cause and cause-specific mortality: results from the Rotterdam Study and a meta-analysis of prospective cohort studies

Evidence for associations between long-term protein intake with mortality is not consistent. We aimed to examine associations of dietary protein from different sources with all-cause and cause-specific mortality. We followed 7786 participants from three sub-cohorts of the Rotterdam Study, a population-based cohort in the Netherlands. Dietary data were collected using food-frequency questionnaires at baseline (1989–1993, 2000–2001, 2006–2008). Deaths were followed until 2018. Associations were examined using Cox regression. Additionally, we performed a highest versus lowest meta-analysis and a dose–response meta-analysis to summarize results from the Rotterdam Study and previous prospective cohorts. During a median follow-up of 13.0 years, 3589 deaths were documented in the Rotterdam Study. In this cohort, after multivariable adjustment, higher total protein intake was associated with higher all-cause mortality [e.g. highest versus lowest quartile of total protein intake as percentage of energy (Q4 versus Q1), HR = 1.12 (1.01, 1.25)]; mainly explained by higher animal protein intake and CVD mortality [Q4 versus Q1, CVD mortality: 1.28 (1.03, 1.60)]. The association of animal protein intake and CVD was mainly contributed to by protein from meat and dairy. Total plant protein intake was not associated with all-cause or cause-specific mortality, mainly explained by null associations for protein from grains and potatoes; but higher intake of protein from legumes, nuts, vegetables, and fruits was associated with lower risk of all-cause and cause-specific mortality. Findings for total and animal protein intake were corroborated in a meta-analysis of eleven prospective cohort studies including the Rotterdam Study (total 64,306 deaths among 350,452 participants): higher total protein intake was associated with higher all-cause mortality [pooled RR for highest versus lowest quantile 1.05 (1.01, 1.10)]; and for dose–response per 5 energy percent (E%) increment, 1.02 (1.004, 1.04); again mainly driven by an association between animal protein and CVD mortality [highest versus lowest, 1.09 (1.01, 1.18); per 5 E% increment, 1.05 (1.02, 1.09)]. Furthermore, in the meta-analysis a higher plant protein intake was associated with lower all-cause and CVD mortality [e.g. for all-cause mortality, highest versus lowest, 0.93 (0.87, 0.99); per 5 E% increment, 0.87 (0.78, 0.98), for CVD mortality, highest versus lowest 0.86 (0.73, 1.00)]. Evidence from prospective cohort studies to date suggests that total protein intake is positively associated with all-cause mortality, mainly driven by a harmful association of animal protein with CVD mortality. Plant protein intake is inversely associated with all-cause and CVD mortality. Our findings support current dietary recommendations to increase intake of plant protein in place of animal protein. Clinical trial registry number and website NTR6831, https://www.trialregister.nl/trial/6645 ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1007/s10654-020-00607-6) contains supplementary material, which is available to authorized users.