Food biodiversity and total and cause-specific mortality in 9 European countries: An analysis of a prospective cohort study – PLoS Blogs

PLOS Medicine publishes research and commentary of general interest with clear implications for patient care, public policy or clinical research agendas.
Get Started
Loading metrics
Open Access
Peer-reviewed
Research Article
Contributed equally to this work with: Giles T. Hanley-Cook, Inge Huybrechts
Roles Conceptualization, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Visualization, Writing – original draft, Writing – review & editing
Affiliation Department of Food Technology, Safety and Health, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium
https://orcid.org/0000-0001-9907-594X
Contributed equally to this work with: Giles T. Hanley-Cook, Inge Huybrechts
Roles Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Writing – review & editing
[email protected] (IH); [email protected] (CL)
Affiliation Nutritional Epidemiology Group, Nutrition and Metabolism Section, International Agency for Research on Cancer, World Health Organization, Lyon, France
https://orcid.org/0000-0003-3838-855X
Roles Data curation, Formal analysis, Methodology, Software, Validation, Visualization, Writing – review & editing
Affiliation Nutritional Methodology and Biostatistics Group, Nutrition and Metabolism Section, International Agency for Research on Cancer, World Health Organization, Lyon, France
https://orcid.org/0000-0003-3950-0929
Roles Writing – review & editing
Affiliations Bioversity International, Heverlee, Belgium, Department of Agrotechnology and Food Sciences, Wageningen University & Research, Wageningen, the Netherlands
https://orcid.org/0000-0003-3659-8529
Roles Writing – review & editing
Affiliation Global Alliance for Improved Nutrition (GAIN), Washington, DC, United States of America
Roles Writing – review & editing
Affiliation Sorbonne Paris Nord University, Inserm U1153, Inrae U1125, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center—University of Paris (CRESS), Bobigny, France
https://orcid.org/0000-0002-3359-420X
Roles Writing – review & editing
Affiliations MRC Centre for Global Infectious Disease Analysis, School of Public Health, Imperial College London, London, United Kingdom, MRC Unit The Gambia at London School of Hygiene & Tropical Medicine, Fajara, Banjul, The Gambia
Roles Writing – review & editing
Affiliation Sorbonne Paris Nord University, Inserm U1153, Inrae U1125, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center—University of Paris (CRESS), Bobigny, France
https://orcid.org/0000-0002-8322-8857
Roles Writing – review & editing
Affiliation Department of Community Medicine, Faculty of Health Sciences, UiT, The Arctic University of Norway, Tromsø, Norway
https://orcid.org/0000-0003-2476-4251
Roles Writing – review & editing
Affiliation Sorbonne Paris Nord University, Inserm U1153, Inrae U1125, Cnam, Nutritional Epidemiology Research Team (EREN), Epidemiology and Statistics Research Center—University of Paris (CRESS), Bobigny, France
https://orcid.org/0000-0002-9715-3534
Roles Writing – review & editing
Affiliations Department of Food Technology, Safety and Health, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium, Department of Population and Family Health, Institute of Health, Jimma University, Jimma, Ethiopia
https://orcid.org/0000-0003-3982-5408
Roles Data curation, Formal analysis, Software, Validation, Writing – review & editing
Affiliation Nutritional Methodology and Biostatistics Group, Nutrition and Metabolism Section, International Agency for Research on Cancer, World Health Organization, Lyon, France
Roles Data curation, Formal analysis, Software, Validation
Affiliation Biomarkers Group, Nutrition and Metabolism Section, International Agency for Research on Cancer, World Health Organization, Lyon, France
https://orcid.org/0000-0001-7871-7149
Roles Writing – review & editing
Affiliation Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London, London, United Kingdom
Roles Writing – review & editing
Affiliation Public Health Policy Evaluation Unit, School of Public Health, Imperial College London, London, United Kingdom
https://orcid.org/0000-0002-0793-9884
Roles Writing – review & editing
Affiliation International Agency for Research on Cancer, World Health Organization, Lyon, France
https://orcid.org/0000-0003-2237-0128
Roles Data curation, Writing – review & editing
Affiliation Nutritional Methodology and Biostatistics Group, Nutrition and Metabolism Section, International Agency for Research on Cancer, World Health Organization, Lyon, France
Roles Writing – review & editing
Affiliation Department of Public Health, Aarhus University, Aarhus, Denmark
https://orcid.org/0000-0003-0481-2893
Roles Writing – review & editing
Affiliation Department for Determinants of Chronic Diseases (DCD), National Institute for Public Health and the Environment (RIVM), Bilthoven, the Netherlands
Roles Writing – review & editing
Affiliation Department of Community Medicine, Faculty of Health Sciences, UiT, The Arctic University of Norway, Tromsø, Norway
Roles Writing – review & editing
Affiliations Department of Public Health, Aarhus University, Aarhus, Denmark, Institute of Environmental Medicine, Karolinska Institute, Stockholm, Sweden
https://orcid.org/0000-0002-7038-4770
Roles Writing – review & editing
Affiliation Nutritional Methodology and Biostatistics Group, Nutrition and Metabolism Section, International Agency for Research on Cancer, World Health Organization, Lyon, France
https://orcid.org/0000-0001-8648-4998
Roles Writing – review & editing
Affiliation Department of Clinical Sciences, Lund University, Malmö, Sweden
https://orcid.org/0000-0003-1291-4494
Roles Writing – review & editing
Affiliations Department of Molecular Epidemiology, German Institute of Human Nutrition Potsdam-Rehbruecke, Nuthetal, Germany, German Center for Diabetes Research (DZD), München-Neuherberg, Germany, NutriAct—Competence Cluster Nutrition Research Berlin-Potsdam, Nuthetal, Germany
https://orcid.org/0000-0003-3478-4758
Roles Writing – review & editing
Affiliation Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Utrecht, the Netherlands
https://orcid.org/0000-0002-4605-435X
Roles Writing – review & editing
Affiliations Department of Molecular Epidemiology, German Institute of Human Nutrition Potsdam-Rehbruecke, Nuthetal, Germany, Institute of Nutritional Sciences, University of Potsdam, Nuthetal, Germany
https://orcid.org/0000-0002-0830-5277
Roles Writing – review & editing
Affiliations Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London, London, United Kingdom, Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina, Greece
Roles Writing – review & editing
Affiliations Danish Cancer Society Research Center, Copenhagen, Denmark, Department of Public Health, University of Copenhagen, Copenhagen, Denmark
https://orcid.org/0000-0003-4385-2097
Roles Writing – review & editing
Affiliations Navarra Public Health Institute, Pamplona, Spain, IdiSNA, Navarra Institute for Health Research, Pamplona, Spain, Biomedical Research Networking Center for Epidemiology and Public Health (CIBERESP), Madrid, Spain
Roles Writing – review & editing
Affiliation Department of Radiation Sciences, Oncology, Umeå University, Umeå, Sweden
https://orcid.org/0000-0002-8958-975X
Roles Writing – review & editing
Affiliations Biomedical Research Networking Center for Epidemiology and Public Health (CIBERESP), Madrid, Spain, Department of Epidemiology, Murcia Regional Health Council—IMIB-Arrixaca, Murcia, Spain, Department of Health and Social Sciences, University of Murcia, Murcia, Spain
Roles Writing – review & editing
Affiliation Epidemiology and Prevention Unit, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy
https://orcid.org/0000-0002-0813-3022
Roles Writing – review & editing
Affiliation Danish Cancer Society Research Center, Copenhagen, Denmark
Roles Writing – review & editing
Affiliations Unit of Nutrition and Cancer, Cancer Epidemiology Research Program, Catalan Institute of Oncology (ICO-IDIBELL), L’Hospitalet de Llobregat, Barcelona, Spain, Blanquerna School of Health Sciences, Ramon Llull University, Barcelona, Spain
https://orcid.org/0000-0003-0672-8847
Roles Writing – review & editing
Affiliation School of Dentistry, Cariology, Department of Odontology, Umeå University, Umeå, Sweden
https://orcid.org/0000-0002-9227-8434
Roles Writing – review & editing
Affiliation Division of Cancer Epidemiology, German Cancer Research Center (DKFZ), Heidelberg, Germany
https://orcid.org/0000-0002-6509-6555
Roles Writing – review & editing
Affiliation Cancer Risk Factors and Life-Style Epidemiology Unit, Institute for Cancer Research, Prevention and Clinical Network (ISPRO), Florence, Italy
https://orcid.org/0000-0002-5758-9069
Roles Writing – review & editing
Affiliation Dipartimento di Medicina Clinica e Chirurgia, Federico II University, Naples, Italy
Roles Writing – review & editing
Affiliations Biomedical Research Networking Center for Epidemiology and Public Health (CIBERESP), Madrid, Spain, Andalusian School of Public Health (EASP), Granada, Spain, Instituto de Investigación Biosanitaria de Granada (ibs.GRANADA), Universidad de Granada, Granada, Spain
https://orcid.org/0000-0002-9972-9779
Roles Writing – review & editing
Affiliation Unit of Cancer Epidemiology, Città della Salute e della Scienza University-Hospital and Centre for Cancer Prevention (CPO), Turin, Italy
https://orcid.org/0000-0002-8008-5096
Roles Writing – review & editing
Affiliation Division of Cancer Epidemiology, German Cancer Research Center (DKFZ), Heidelberg, Germany
https://orcid.org/0000-0002-1277-3380
Roles Writing – review & editing
Affiliation Cancer Registry and Histopathology Department, Azienda Sanitaria Provinciale Ragusa (ASP 7), Ragusa, Italy
https://orcid.org/0000-0003-2666-414X
Roles Writing – review & editing
Affiliation Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London, London, United Kingdom
https://orcid.org/0000-0001-6795-6080
Roles Project administration, Resources, Software, Writing – review & editing
Affiliation Nutritional Epidemiology Group, Nutrition and Metabolism Section, International Agency for Research on Cancer, World Health Organization, Lyon, France
Roles Supervision, Writing – review & editing
Affiliation Department of Nutritional Sciences, School of Public Health, University of Michigan, Ann Arbor, Michigan, United States of America
Roles Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Writing – review & editing
[email protected] (IH); [email protected] (CL)
Affiliation Department of Food Technology, Safety and Health, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium
https://orcid.org/0000-0002-1389-8855
Food biodiversity, encompassing the variety of plants, animals, and other organisms consumed as food and drink, has intrinsic potential to underpin diverse, nutritious diets and improve Earth system resilience. Dietary species richness (DSR), which is recommended as a crosscutting measure of food biodiversity, has been positively associated with the micronutrient adequacy of diets in women and young children in low- and middle-income countries (LMICs). However, the relationships between DSR and major health outcomes have yet to be assessed in any population.
We examined the associations between DSR and subsequent total and cause-specific mortality among 451,390 adults enrolled in the European Prospective Investigation into Cancer and Nutrition (EPIC) study (1992 to 2014, median follow-up: 17 years), free of cancer, diabetes, heart attack, or stroke at baseline. Usual dietary intakes were assessed at recruitment with country-specific dietary questionnaires (DQs). DSR of an individual’s yearly diet was calculated based on the absolute number of unique biological species in each (composite) food and drink. Associations were assessed by fitting multivariable-adjusted Cox proportional hazards regression models. In the EPIC cohort, 2 crops (common wheat and potato) and 2 animal species (cow and pig) accounted for approximately 45% of self-reported total dietary energy intake [median (P10P90): 68 (40 to 83) species consumed per year]. Overall, higher DSR was inversely associated with all-cause mortality rate. Hazard ratios (HRs) and 95% confidence intervals (CIs) comparing total mortality in the second, third, fourth, and fifth (highest) quintiles (Qs) of DSR to the first (lowest) Q indicate significant inverse associations, after stratification by sex, age, and study center and adjustment for smoking status, educational level, marital status, physical activity, alcohol intake, and total energy intake, Mediterranean diet score, red and processed meat intake, and fiber intake [HR (95% CI): 0.91 (0.88 to 0.94), 0.80 (0.76 to 0.83), 0.69 (0.66 to 0.72), and 0.63 (0.59 to 0.66), respectively; PWald < 0.001 for trend]. Absolute death rates among participants in the highest and lowest fifth of DSR were 65.4 and 69.3 cases/10,000 person-years, respectively. Significant inverse associations were also observed between DSR and deaths due to cancer, heart disease, digestive disease, and respiratory disease. An important study limitation is that our findings were based on an observational cohort using self-reported dietary data obtained through single baseline food frequency questionnaires (FFQs); thus, exposure misclassification and residual confounding cannot be ruled out.
In this large Pan-European cohort, higher DSR was inversely associated with total and cause-specific mortality, independent of sociodemographic, lifestyle, and other known dietary risk factors. Our findings support the potential of food (species) biodiversity as a guiding principle of sustainable dietary recommendations and food-based dietary guidelines.
Citation: Hanley-Cook GT, Huybrechts I, Biessy C, Remans R, Kennedy G, Deschasaux-Tanguy M, et al. (2021) Food biodiversity and total and cause-specific mortality in 9 European countries: An analysis of a prospective cohort study. PLoS Med 18(10): e1003834. https://doi.org/10.1371/journal.pmed.1003834
Academic Editor: Jean Adams, University of Cambridge, UNITED KINGDOM
Received: February 24, 2021; Accepted: October 4, 2021; Published: October 18, 2021
Copyright: © 2021 Hanley-Cook et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: EPIC data and biospecimens are available to investigators in the context of research projects that are consistent with the legal and ethical standard practices of IARC/WHO and the EPIC Centers. The use of a random sample of anonymized data from the EPIC study can be requested by contacting [email protected]. For information on EPIC data access policy and on how to submit an application for gaining access to EPIC data and/or biospecimens, please follow the instructions at http://epic.iarc.fr/access/index.php.
Funding: The coordination of EPIC is financially supported by the European Commission (DG-SANCO) and IARC. The national cohorts are supported by Danish Cancer Society (Denmark); Ligue Contre le Cancer, Institut Gustave Roussy, Mutuelle Générale de l’Education Nationale, and Institut National de la Santé et de la Recherche Médicale (Inserm) (France); Deutsche Krebshilfe, Deutsches Krebsforschungszentrum, and Federal Ministry of Education and Research (Germany); the Hellenic Health Foundation (Greece); Associazione Italiana per la Ricerca sul Cancro-IARC-Italy and National Research Council (Italy); Dutch Ministry of Public Health, Welfare, and Sports (VWS), Netherlands Cancer Registry (NKR), LK Research Funds, Dutch Prevention Funds, Dutch ZON (Zorg Onderzoek Nederland), World Cancer Research Fund (WCRF), and Statistics Netherlands (the Netherlands); Health Research Fund (FIS), PI13/00061 to Granada, Regional Governments of Andalusía, Asturias, Basque Country, Murcia (no. 6236) and Navarra, and ISCIII RETIC (RD06/0020) (Spain); Swedish Cancer Society, Swedish Scientific Council and County Councils of Skåne and Västerbotten (Sweden); Cancer Research UK (14136 to EPIC-Norfolk; C570/A16491 and C8221/A19170 to EPIC-Oxford) (United Kingdom). Funding for grant number IIG_FULL_2020_034 was obtained from Wereld Kanker Onderzoek Fonds (WKOF), as part of the World Cancer Research Fund International grant programme (PI: IH, Co-I: GH-C, MD-T, KM, EK-G, MT, PV, CL). Researchers were independent from the funders. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: BMI, body mass index; CHD, coronary heart disease; CI, confidence interval; CVD, cardiovascular disease; DQ, dietary questionnaire; DSR, dietary species richness; EPIC, European Prospective Investigation into Cancer and Nutrition; FFQ, food frequency questionnaire; HR, hazard ratio; ICD-10, 10th revision of the International Statistical Classification of Diseases, Injuries, and Causes of Death; LMIC, low- and middle-income country; Q, quintile; SDG, Sustainable Development Goal; STROBE, STrengthening the Reporting of OBservational Studies in Epidemiology; WCRF, World Cancer Research Fund
Diets inextricably link human and environmental health [1]. The global food system is the primary driver of unprecedented Earth system biodiversity loss (e.g., monocropping, land degradation, and deforestation for agriculture) [2,3], while low-quality, nondiverse diets are responsible for the greatest burden of disease worldwide, affecting countries and populations at all levels of socioeconomic development [4]. The short- and long-term consequences of accelerated biodiversity collapse [5,6] and the triple burden of malnutrition [7,8] restrain sustainable and inclusive global development and convey unacceptable human consequences [9,10].
At present, rapid socioeconomic, demographic, and technological transitions, coupled with agricultural policies skewed toward a narrow range of staple crops, crop varieties, and animal species [11], are driving a progressive homogeneity of human diets [12,13]. In parallel, the associated global food systems, which are mainly focused on cheap calories, rather than nutrients, are redirected toward more resource-intensive, energy-dense, and nutrient-poor food species [14]. The Sustainable Development Goals (SDGs), the United Nations Decade of Action on Nutrition 2016–2025, and the Convention on Biological Diversity’s forthcoming post-2020 biodiversity agenda provide global and national stimuli to fast-track transition from business-as-usual to win-win scenarios for human and environmental health in the Anthropocene epoch [15].
Food biodiversity, defined as the variety of plants, animals, and other organisms (e.g., fungi and yeast cultures) that are used for food and drink, both cultivated and from the wild, has intrinsic potential to underpin diverse, nutritious diets and conserve (neglected and underutilized) finite genetic resources (i.e., biodiversity) [16,17]. Thus, the concept of food biodiversity potentially offers a unique and novel entry point to guide the development of sustainable food-based dietary guidelines (and interventions) cutting across human and planetary health [18,19]. Dietary diversity, which is conventionally measured as consumption between, rather than within nutrient-dense food groups, is a widely acknowledged and established public health recommendation to promote healthy, nutritionally adequate diets [20]. Furthermore, diets based on a wide diversity of (locally available, nonthreatened) biological species exert lower pressure on single species, hence increasing Earth system stability, resilience, ecosystem services, and enhanced productivity of natural and agricultural systems [5,21]. Furthermore, observational studies have indicated consistently positive, but small associations between agricultural biodiversity [22] and forest patterns [23] with dietary diversity in low- and middle-income countries (LMICs). Nevertheless, environmental sustainability criteria, including biodiversity preservation in, e.g., Brazil, the Netherlands, and Sweden, are only explicitly included in 8 (quasi) official sustainable food-based dietary guidelines worldwide [24]. Moreover, globally, the potential dual benefits of at-scale adoption of national food-based dietary guidelines on human health and environmental impacts can be substantially improved [25].
Previous cross-sectional research indicated that higher dietary species richness (DSR), a recommended measure of food biodiversity, which captures both inter- and intra-food group diversity, was associated with increased micronutrient adequacy of diets in approximately 2,200 women of reproductive age and approximately 4,000 children aged 6 to 23 months in LMICs [26]. Furthermore, several prospective studies have reported lower mortality and chronic disease rates among participants with higher between food group diversity [27,28], within specific food group richness (e.g., fruit and vegetables) [29], and food item variety [30]. To date, however, the evidence base for relationships between food (species) biodiversity of whole dietary patterns and major human health outcomes is missing. In this study, we address the knowledge gap by assessing associations between DSR and total and cause-specific mortality in a large and diverse Pan-European cohort.
Our research was reported using the STrengthening the Reporting of OBservational Studies in Epidemiology (STROBE)-nut checklist [31].
The European Prospective Investigation into Cancer and Nutrition (EPIC) study (http://epic.iarc.fr/) is an ongoing multicenter, prospective cohort study investigating metabolic, dietary, lifestyle, and environmental factors in relation to cancer and other chronic diseases. Between 1992 and 2000, more than 500,000 volunteers (25 to 70 years) were recruited from 10 European countries (23 administrative centers): Denmark, France, Germany, Greece, Italy, the Netherlands, Norway, Spain, Sweden, and the United Kingdom. Most of the participants were recruited from the general population residing in a given geographic area, town, or province. Exceptions were the cohorts of France (female members of a health insurance scheme for school employees), Utrecht (breast cancer screening attendees), Ragusa (blood donors and their spouses), and Oxford (mainly vegetarian and healthy eaters). All participants gave written informed consent and completed questionnaires on their diet, lifestyle, and medical history. The study was approved by the local ethics committees and by the Internal Review Board of the International Agency for Research on Cancer. Details of the study design, recruitment, and data collection have been previously published [3234].
Of the 521,324 participants enrolled, 451,390 were included in the analyses, with 46,627 recorded deaths between 1992 and 2014. We excluded participants with missing lifestyle or dietary information, those with an extreme ratio of energy intake to energy requirement (top and bottom 1%, as these values were considered physiologically implausible [34]), volunteers with null follow-up, those with prevalent disease at baseline (history of cancer, cardiovascular diseases [CVDs], and diabetes), and all participants from the EPIC-Greece cohort, due to administrative constraints (S1 Fig).
An extensive and standardized phenotypic characterization was performed for each participant upon enrollment. Questionnaires were used to collect sociodemographic information, educational level (standardized for the whole cohort), personal and familial history of diseases, lifestyle (e.g., smoking, alcohol use, and physical activity), and menstrual and reproductive history of women. Anthropometric measurements (e.g., height, weight, waist, and hip circumferences) were performed in all centers (except France, Oxford, and Norway: self-reported data) [35]. Height and weight were complemented with available self-reported values or imputed with center-, age-, and gender-specific average values when missing. Body mass index (BMI) was calculated as weight divided by height squared (kg/m2).
Usual dietary intake was assessed for each individual at recruitment using country- or center-specific validated dietary questionnaires (DQs) developed to capture the geographical specificity of an individual’s diet over the preceding year. The type of DQ used differed according to study centers and included self- or interviewer-administered semiquantitative food frequency questionnaires (FFQs) with an estimation of individual average portions or with the same standard portion assigned to all participants or diet history questionnaires combining an FFQ and 7-day dietary records. In most centers, DQs were self-administered, with the exception of Ragusa, Naples, and Spain, where face-to-face interviews were conducted. Extensive quantitative DQs were used in northern Italy, the Netherlands, and Germany, which were structured by meals in Spain, France, and Ragusa. Semiquantitative FFQs were used in Denmark, Norway, Naples, Umeå, and the UK, while an FFQ was combined with a 7-day record on hot meals in Malmö [33]. Post-harmonization of DQ data was conducted, following standardized procedures (e.g., decomposing recipes into ingredients), to obtain a standardized food list for which the level of detail is comparable between countries. The EPIC food composition database comprises more than 11,000 food and beverage items reflecting the specificities of each country [36].
(Bio)diversity can be partitioned into 3 components: richness, evenness, and disparity (Fig 1) [37]. Nevertheless, our study focuses only on DSR, previously recommended as the most appropriate measure of food biodiversity for dietary intake studies, as we aim to inform food-based interventions and policy based on a simple, crosscutting indicator of human and planetary health [26]. We argue that species evenness, which is defined as a perfectly equal distribution of food and drinks in the diet, is neither desired from a nutritional [38] nor environmental protection perspective [39]. Hence, dietary evenness requires an arbitrary a priori selection of a relative abundance unit (e.g., energy, nutrients, weight, volume, and frequency) and “healthfulness” weighting factors [40,41]. Moreover, unlike for species richness, there is currently no consensus on the measurement of species evenness (e.g., Shannon entropy, Berry–Simpson, and Pielou’s index) [37]. Dietary disparity is defined for our research purposes as the consumption of foods with distinct human health [42] and ecosystem attributes [43], rather than the more narrow, but well-established measures of nutritional food group diversity [44]. Similarly to evenness, ecological metrics of species dissimilarity are based on an inconsistent selection (and number) of phylogenetic, functional, and/or morphological traits (e.g., Rao’s quadratic diversity and Jaccard index) [37].

Distinct species are indicated by their color. Richness is the absolute number of species: In both dietary patterns, it is equal to 5. Evenness is the equitability of the species abundance distribution in the diet: In dietary pattern A, all species are present in an equal abundance (e.g., frequency) and so it is perfectly even, while dietary pattern B is very uneven since it is dominated by the yellow [Zea mays (maize)] species. Disparity is the level of similarity between species: For example, red [Bos taurus (cow)] and pink [Gallus gallus (chicken)] species are more similar to each other, e.g., nutritionally and taxonomically, than the purple [Solanum melongena (aubergine)] species.
https://doi.org/10.1371/journal.pmed.1003834.g001
Therefore, for the 451,390 participants included in our analysis, food biodiversity in an individual’s diet was calculated based on the absolute number of unique biological species in each (composite) food, drink, and recipe, using the European Food Safety Authority’s FoodEx2 food classification and description system [45] in combination with the detailed EPIC food classification system (NCLASS). Food items consumed “never or less than once per month” (on average) were recalled under one category; accordingly, these species did not count toward DSR. Moreover, quantities (g/day) were disregarded for overall DSR computation, since our interest is the sum of distinct species consumed per year (i.e., DQs recalled dietary intake over the preceding 12 months). However, relative quantities were considered during sensitivity analyses, which excluded species consumed in trivial amounts (see below). Furthermore, although a species can be consumed multiple times per year, potentially from diverse functional food groups (e.g., chicken meat and eggs, which are nutritionally disparate), through a “biodiversity conservation” lens, it contributes only one species to an individual’s DSR in all scenarios (taxonomically identical: Gallus gallus). Here, we consider that high food biodiversity is thus a combination of multiple species that act synergistically and complimentary for both human and environmental health (e.g., ecological and net nutritional benefits from the Mesoamerican combination of corn, beans, and squash, known as the “three sisters”) [21,46].
For all countries, composite dishes were decomposed into their ingredients (species) using standard recipes. Therefore, herbs and spices and other ingredients potentially used in small amounts, for which we cannot be certain if they were added to the recipe by each EPIC participant, might bias/inflate the true value of an individual’s DSR. To assess the impact of food and drink species consumed in relatively small quantities, we calculated 3 different scenarios of DSR, namely the following:
Data on vital status and cause and date of death were obtained using record linkages with population-based cancer registries, boards of health, health insurance registries, pathology registries and mortality registries (Denmark, Italy, the Netherlands, Norway, Spain, Sweden, and the UK), or through active follow-up and next of kin (France and Germany). Germany identified deceased individuals from undelivered follow-up mailings and subsequent enquiries to municipality registries, regional health department, physicians, or hospitals. In France, information on deceased participants was obtained using the database of health insurance for school employees and national death index. The end of follow-up/closure dates of the study period varied between 2009 and 2014 depending on the countries.
Cause-specific mortality data were coded according to the 10th revision of the International Statistical Classification of Diseases, Injuries, and Causes of Death (ICD-10) [47]. Causes of death assessed include heart disease [i.e., coronary heart disease (CHD) (ICD-10 codes: I20 to I25) and CVD other than CHD (I00 to I99, excluding I20 to I25)], cancer (C00 to D48), diseases of the respiratory system (J00 to J99), and diseases of the digestive system (K00 to K93). Total mortality (main outcome) was defined as mortality from all causes, except external causes of death (S00 to T98 and V01 to Y98).
Statistical analyses were preplanned and followed the plan detailed in the project proposal that was submitted to and approved (March 2019) by the EPIC Steering Committee (see S1 Text). Unadjusted absolute death rates were calculated as the number of cases per 10,000 person-years in Q5 and Q1 of DSR, respectively. Associations between food biodiversity [DSR per year; count variable and quintiles (Qs)] and total and cause-specific mortality were characterized [hazard ratio (HR) and 95% confidence interval (CI)] using multivariable-adjusted Cox proportional hazard regression models with age as the primary underlying time variable. The Breslow method was adopted for handling ties. Examination of the Schoenfeld residuals, according to follow-up time (years) for Qs of DSR, confirmed that the assumptions of proportionality were satisfied. Overall survival curves by Q of DSR were generated using the Kaplan–Meier method (S2 Fig). Participants contributed person-time to the model until their date of death, their date of emigration/loss to follow-up, or end-of-follow-up, whichever occurred first. P-values for linear trend were calculated with the use of the Wald test of a pseudo-continuous score variable, based on the median number of species consumed per year for each Q of DSR. Nonlinear associations between DSR and total mortality were examined nonparametrically with restricted cubic splines [48]. P-value for nonlinear trend was calculated with the use of the likelihood ratio test, comparing the model with only the linear term to the model with the linear and the cubic spline terms.
Pooled cohort models were stratified by sex, age at recruitment (1-year intervals), and study center (“strata” option in proc phreg, SAS) and multivariable-adjusted for confounding factors using a 5% change-in-estimate criterion for β-coefficients (applied to all variables reported in Table 1, due to limited knowledge on factors relating to DSR): smoking status (current, 1 to 15 cigarettes/day; current, 16 to 25 cigarettes/day; current, 26+ cigarettes/day; current, pipe/cigar/occasional; current/former, missing; former, quit 11 to 20 years; former, quit 20+ years; former, quit ≤10 years; never; unknown), educational level, as a proxy variable for socioeconomic status [longer education (including university degree, technical, or professional school); secondary school; primary school completed; not specified], marital status (single, divorced, separated, or widowed; married or living together; unknown), physical activity (Cambridge index: active; moderately active; moderately inactive; inactive; missing), alcohol intake at recruitment (g/day), total energy intake (kcal/day), the 18-point relative Mediterranean diet score, as an indicator for an overall healthy diet [49], the consumption of red and processed meat (g/day) [50], and fiber intake (g/day; i.e., to reflect carbohydrate quality [51], such as whole grains). Possible multicollinearity of (dietary) variables included in our models were assessed by “collin” and “vif” options in proc phreg, SAS (all condition indices <30 and variance inflation factors <3, respectively). When data on categorical covariates were missing, a “missing class” was introduced to the model.

https://doi.org/10.1371/journal.pmed.1003834.t001
We performed several sensitivity analyses to test the robustness of our all-cause mortality results. First, we stratified our multivariable-adjusted models to estimate associations by sex and country (owing to the varying detail of food and drink items captured by DQs) separately. Furthermore, we removed Mediterranean diet score, red and processed meat, fiber, and total energy intake from the models to examine their role as mediators, rather than confounders. To assess the potential for residual confounding, we carried out subgroup analyses according to major potential categorical effect modifiers: educational level, smoking status, marital status, and physical activity. Furthermore, we used a “complete cases” approach, excluding participants with missing/unknown data on covariates. Not all food and drink items received a specific FoodEx2 species code, but rather kept a generic NCLASS classification (e.g., “other root vegetables,” which counted as one species toward an individual’s overall DSR). Therefore, we reran our models dropping these generic food and drink items. Analyses were also conducted including DSR without the lowest 5% and 10% of species intakes for each EPIC food group (see methodological reasoning above). In addition, we repeated our prospective analyses for species richness within each main EPIC food group adjusted for overall DSR (minus itself) to investigate whether one or more food groups were responsible for the observed associations [52]. Moreover, from the 46,627 fatal events, we excluded deaths within the first 3 (n = 2,969) and 6 years (n = 7,928) of follow-up to allow sufficient delay between baseline dietary assessment and mortality, thereby limiting reverse causality of subclinical disease. Findings from sensitivity analyses, which are not different (i.e., stable direction, strength, and trend of association) from those using the entire EPIC cohort, are not shown. To assess potential residual confounding from unmeasured or uncontrolled confounders, E-values were used [53,54].
All statistical tests were 2 sided, and P < 0.05 was considered statistically significant. P-values were adjusted for multiple testing of hypothesis using the Benjamini–Hochberg method. SAS software, version 9.4 (SAS Institute) was used for the analyses.
Initial characteristics from 451,390 eligible participants (71% female; median age 51 years) according to Qs of DSR are shown in Table 1. After a median follow-up of 17 years (7,506,482 person-years), 44,892 deaths from nonexternal causes occurred, among which 19,284 from cancer, 11,353 from diseases of the circulatory system, 2,479 from diseases of the respiratory system, and 1,386 from diseases of the digestive system. From the 11,858 items included in the EPIC food list, 80% were assigned FoodEx2 species codes (248 unique values; 78% of total kcal/day; see S1 Table), whereas 16% received a generic NCLASS code (100 unique values, e.g., “other citrus fruits”), and 4% were classified as “not applicable” (e.g., added salt and water). In the whole cohort, participant’s [median (P10P90)] DSR was 68 (40 to 83) species per year. Bos taurus (cow), Triticum aestivum (common wheat grain), Sus scrofa (domestic pig), and Solanum tuberosum (potato) contributed most to self-reported total dietary energy intake (i.e., approximately 45%) with [mean % (SD)] 19% (8), 16% (8), 4% (3), and 4% (3) kcal/day, respectively. When comparing the fifth Q of DSR (highest; largely represented by the predominately vegetarian, “health-conscious” EPIC-Oxford (30%) and “omnivorous” German (35%) cohorts) against the first (lowest) Q, our findings indicate large differences in median dietary vegetable richness (22 versus 10 species), fruit, nuts, and seed richness (11 versus 5 species) and condiment richness (7 versus 2 species). In France, increased DSR across Qs was observed due to a significant positive gradient in vegetables richness only (7 versus 24 species).
Pooled multivariable analysis indicated that average DSR consumption was inversely associated with total mortality (PWald < 0.001 for trend), in that participants with low DSR (Q1; <48 species per year) had notably higher mortality rates than individuals with moderate (Q3; 64 to 72 species per year) or high DSR (Q5; ≥81 species per year) (see S2 Table). The corresponding pooled HRs (95% CIs) were 0·80 (0.76 to 0.83) for moderate DSR and 0.63 (0.59 to 0.66) for high DSR in comparison with low DSR (Fig 2). Absolute mortality rates among participants in the highest and lowest fifth of DSR were 65.4 and 69.3 cases/10,000 person-years, respectively (see S2 Table).

Multiadjusted models were stratified for center, age at recruitment (1-year intervals, timescale), and sex and adjusted for baseline alcohol intake (g/day), physical activity (Cambridge index: active; moderately active; moderately inactive; inactive; missing), marital status (single, divorced, separated, or widowed; married or living together; unknown), smoking status and intensity of smoking (current, 1 to 15 cigarettes/day; current, 16 to 25 cigarettes/day; current, 26+ cigarettes/day; current, pipe/cigar/occasional; current/former, missing; former, quit 11 to 20 years; former, quit 20+ years; former, quit ≤10 years; never; unknown), educational level [longer education (including university degree, technical or professional school); secondary school; primary school completed; not specified], baseline energy intake (kcal/day), baseline fiber intake (g/day), baseline red and processed meat consumption (g/day), and an 18-point Mediterranean diet score [49]. P-values remained statistically significant after adjustment for multiple testing using the Benjamini–Hochberg method. CI, confidence interval; DSR, dietary species richness; EPIC, European Prospective Investigation into Cancer and Nutrition; HR, hazard ratio; Q, quintile.
https://doi.org/10.1371/journal.pmed.1003834.g002
Our findings indicate slightly stronger relationships among males (see S3 Table). To illustrate, a 10-species increment in DSR was associated (95% CI) with a 14% to 17% and 6% to 8% reduction in all-cause mortality rates among males and females during approximately 20 years of follow-up, respectively. Overall, results for mortality rates from nonexternal causes were consistent across 8 countries (P-value and PWald both <0·05; see S4 Table). In the UK, we observed overall higher DSR and a subsequent smaller contrast between participants with lower and higher scores (Q1; <71, Q5; ≥82 species per year). The associated protective effect of DSR was not substantially changed when removing potential dietary mediators, among major subgroups (although the lowest HRs were reported among current smokers and participants with secondary education) or complete cases, when dropping generic food and drink codes, or exclusion of the lowest 5% (Q1; <47, Q5; ≥80 species per year) and 10% species intake (Q1; <46, Q5; ≥78 species per year) from each EPIC food (sub)group. Furthermore, our observed associations were not explained by species richness within one single food group, suggesting a positive cumulative effect of overall DSR (see S5 Table). Limited graphical, but statistically significant (PLR<0.001), evidence of nonlinearity was observed for all-cause mortality (S3 Fig).
In multivariable analyses, a 10-species increment in DSR was inversely associated with the rate of death [HR (95% CI)] due to digestive disease [0.80 (0.76 to 0.86)], respiratory disease [0.84 (0.80 to 0.88)], heart disease [0.88 (0.86 to 0.90)], and cancer [0.93 (0.92 to 0.95); all PWald < 0.001 for trend; Table 2].

https://doi.org/10.1371/journal.pmed.1003834.t002
The large E-values for total and cause-specific mortality suggest that residual confounding is likely to be low, conditional on the measured covariates in our models (see S6 Table).
To our knowledge, this study is the first effort to investigate the relationships between food biodiversity and total and cause-specific mortality in a large epidemiological study. In the EPIC cohort, higher DSR was associated with reduced rates of total mortality and deaths due to cancer, heart disease, respiratory disease, and digestive disease, after accounting for sociodemographic, lifestyle, and other known dietary risk factors, which included relative Mediterranean diet score, red and processed meat, fiber, and total energy intake.
The mechanisms driving the observed relationships between DSR and human health may be largely due to 4 processes. The first is coined as the “sampling effect” and postulates that as one increases DSR, the greater the probability—simply by chance—of including (a diversity of) highly nutritious or health protective foods. In this regard, DSR might characterize both the substantial inter- and intrafood group variations, often not captured by other diet quality and diversity indicators [44], in the content and density of essential nutrients [26], bioactive nonnutrients, and anti-nutrients [55,56]. The second mechanism is known as the “complementary effect,” in which (chemical) interactions between species result in a function greater than expected by chance, i.e., each food or drink species might make an important contribution to diets, but none of these foods alone provide total “healthfulness” [46]. The third potential mechanism for the protective effect of higher food biodiversity encompasses “minimizing trade-offs,” which might occur from consuming too much of one single species (e.g., potential toxicity from overconsumption of certain fish species [57], cruciferous vegetables [58], Brazil nuts [59], and cassava [60]). Lastly, diet-induced variations in human microbial communities may contribute to metabolic health. To illustrate, the differences between the United States and Malawian or Amerindian gut microbiomes have been related to the differences in their diets, with a typical US diet being rich in (animal) protein, whereas diets in Malawi and Amerindian populations are dominated by corn and cassava [61].
In addition, to our knowledge, this was the first study to characterize usual DSR over an approximately 1-year time frame in a large multicountry cohort. There have been marked changes in the biodiversity landscape [62,63] and global food and agriculture supply/system in recent times [12]. To illustrate, retail level food availability data, rather than actual food intake assessments, estimated that in excess of half the global food energy (kcal/capita) is supplied by 4 staple crops: Oryza spp. (rice), S. tuberosum (potatoes), Triticum spp. (wheat), and Zea mays (maize) [64]. Our individual level self-reported dietary intake data from 9 diverse European populations, which includes consumer level waste and intra-household food distribution, suggest that animal species alone contributed over a quarter of total dietary energy, whereas the aforementioned staple crops also contributed a further 25% between 1992 and 2000 in the EPIC cohort. Our findings are alarming considering the growing realization that upstream agroforestry, aquatic, and other biosphere biodiversity loss, approximately 1 million species are now threatened with extinction [5], might have caused a further bottleneck of downstream consumer food choice [15] and thus have subsequent negative impacts on dietary (bio)diversity and food system sustainability [65].
The direct comparison between DSR and usual diet quality scores is neither straightforward nor warranted. Diet quality scores allocate points based on the consumption of specific complementary food items, food groups, or nutrients relevant for overall or specific chronic disease and mortality rates (e.g., Mediterranean diet score [66], WCRF/AICR adherence score [67], and Alternate Health Eating Index [68]), with the objective to add support to dietary recommendations and/or be a basis for food-based dietary guidelines. In contrast, DSR was not designed to find the best predictive score for total or cause-specific mortality rates; hence, our main analyses controlled for potential dietary confounders (i.e., established components of diet quality). Rather, we propose DSR as a simple crosscutting measure of 2 critical dimensions of sustainable development, i.e., human nutrition and biodiversity stewardship, which complements existing indicators for healthy and sustainable diets [26,69]. To maintain simplicity in DSR computation, we assigned an equal weight to each (rare or common) species consumed. Our approach thus fails to account for the relative abundance distribution of foods across a diet or species’ unique functional traits (see above). Similar to crude diet scores, DSR has inherent statistical limitations, including between and within food group species richness being considered as independent from one another (i.e., correlated structure of dietary components or substitution effects disregarded) and assumptions of linear additive effects [70]. No single intra-food group richness explained our main findings, which potentially clarifies the weaker associations in France, where only a strong positive gradient was observed across Qs for vegetable richness. Nevertheless, it remains unlikely that each species consumed made an equal contribution to the associated protective effect on mortality [27]. Thus, our objective was not to compare DSR to other existing dietary or food scores, as richness alone takes no account of the nutritional quality [71], degree of processing [72], and quantities of food and beverages consumed [73], but to specifically assess the relevance of the use of DSR in the framework of sustainable dietary recommendations and food-based dietary guidelines aiming to introduce “biodiversity/variety” into the European population [20]. Against the backdrop of anthropogenic species collapse [74] and rising dietary uniformity [12], our findings champion the relevance of food biodiversity, as a guiding principle of (inter)national food-based dietary guidelines, as explicitly included in, e.g., the Mediterranean Diet Pyramid [75], the New Nordic Diet [76], and Brazilian dietary recommendations [77].
Strengths of this study include its prospective design, large sample size, long (and high rates of) follow-up, and the inclusion of disease-free participants from different European countries with standardized data collection, especially for habitual diet, offering a broad and detailed perspective on a crosscutting measure of food biodiversity (approximately 250 unique species) in European diets. However, some limitations should be acknowledged. First, caution is needed regarding the extrapolation of these results to the entire European population or to other populations or ethnicities worldwide since this study included middle-aged volunteers from 9 European countries involved in a long-term cohort study investigating the association between nutrition and health, with overall more health-conscious behaviors compared to the general population. Therefore, individuals with lower DSR may have been underrepresented in this study, which may have weakened the observed inverse associations by inducing a smaller contrast between high and low DSR (or a potential food biodiversity threshold reached in the UK cohorts). Furthermore, in our models, we included all the participants with available dietary intake data, but with potential missing data on other covariates replaced with a “missing” class or imputation. Although this may have induced some bias, a “complete cases” model alone might have led to a selection bias toward more adherent participants in an already health-conscious population. Yet, our sensitivity analyses with complete cases provided similar results. In addition, this study used a single assessment of self-reported dietary intakes at baseline. Although diets may change over time, it is usually hypothesized that this estimation reflects general eating behavior throughout middle-aged adult life [78]. Traditional diet measurement instruments are built to capture the usual dietary intakes of an individual, but are still subject to imprecision and inaccuracy [79]. EPIC DQs consider self-reported usual food and drink intakes over longer periods of time, not the absolute number of species consumed per day or season specific dietary patterns. Hence, food items potentially consumed “less than once per month” or excluded during DQ development could not be counted toward DSR, which is hypothetically a source of underestimation. In addition, insufficient taxonomic detail was available to subdivide food and drink species into subspecies [e.g., Triticum aestivum subsp. spelta (spelt wheat)] or their source (e.g., locally produced or imported). Furthermore, the number of items that DQs cover depends on the country/center, which required in-depth standardization procedures to guarantee the comparability between countries. For all countries, recipes were decomposed into their ingredients using standard recipes. Therefore, herbs and spices and other ingredients potentially used in trivial amounts might have inflated the true value of an individual’s DSR. To best address this methodological limitation, we calculated 3 different scenarios of DSR consumption, namely overall DSR, including all food and drinks consumed in our EPIC food list (thus, also ingredients derived from standard recipes) and DSR, excluding the lowest 5% and 10% species intake from each EPIC food (sub)group. These sensitivity analyses confirmed the main analyses using overall DSR. Finally, this study was based on an observational cohort. Thus, even though EPIC included a large range of covariates, residual confounding in our models cannot be entirely ruled out (e.g., underlying inflammatory or metabolic disorders) [80] or unmeasured mediating pathways examined (e.g., role of gut microbiome). However, large E-values support the robustness of our observed DSR and mortality associations, providing support for the relationships having a causal basis.
In conclusion, the results from our analysis of a prospective study performed on a large Pan-European cohort with diverse profiles and dietary habits suggest that higher DSR is associated with lower rates of total and cause-specific mortality, independent from other known components of diet quality. Overall, this adds support to the relevance of public health and conservation measures advocating “dietary (species) biodiversity” aiming to influence the healthfulness at national and potentially supranational level. Future comparative and environmental impact (e.g., greenhouse gas emissions, land use, and water use) [39,81,82] studies may be carried out if other simple species diversity indicators with similar characteristics and a corresponding score derived at the individual level (e.g., capturing “optimal” species richness per food group) are to be proposed. In particular, this would complement strategies, such as food-based dietary guidelines [25], setting the basis for a diversified, environmentally sustainable diet mixing distinct types of food, both between and within food groups, and by highlighting food species for which a sensible consumption should be preferred for public and planetary health.
EPIC, European Prospective Investigation into Cancer and Nutrition.
https://doi.org/10.1371/journal.pmed.1003834.s001
(PDF)
DSR, dietary species richness; EPIC, European Prospective Investigation into Cancer and Nutrition; Q, quintile.
https://doi.org/10.1371/journal.pmed.1003834.s002
(PDF)
EPIC, European Prospective Investigation into Cancer and Nutrition.
https://doi.org/10.1371/journal.pmed.1003834.s003
(PDF)
EPIC, European Prospective Investigation into Cancer and Nutrition.
https://doi.org/10.1371/journal.pmed.1003834.s004
(PDF)
EPIC, European Prospective Investigation into Cancer and Nutrition.
https://doi.org/10.1371/journal.pmed.1003834.s005
(PDF)
EPIC, European Prospective Investigation into Cancer and Nutrition.
https://doi.org/10.1371/journal.pmed.1003834.s006
(PDF)
EPIC, European Prospective Investigation into Cancer and Nutrition.
https://doi.org/10.1371/journal.pmed.1003834.s007
(PDF)
EPIC, European Prospective Investigation into Cancer and Nutrition.
https://doi.org/10.1371/journal.pmed.1003834.s008
(PDF)
CI, confidence interval; EPIC, European Prospective Investigation into Cancer and Nutrition; HR, hazard ratio.
https://doi.org/10.1371/journal.pmed.1003834.s009
(PDF)
STROBE, STrengthening the Reporting of OBservational Studies in Epidemiology.
https://doi.org/10.1371/journal.pmed.1003834.s010
(PDF)
EPIC, European Prospective Investigation into Cancer and Nutrition.
https://doi.org/10.1371/journal.pmed.1003834.s011
(PDF)
The authors thank all EPIC participants and staff for their outstanding contribution to the study.
Where authors are identified as personnel of IARC/WHO, the authors alone are responsible for the views expressed in this article, and they do not necessarily represent the decisions, policy, or views of IARC/WHO.
The present analysis was approved by the Ethics Committee of Ghent University Hospital (EC/2019/1015).
For more information about PLOS Subject Areas, click here.
Is the Subject Area « Diet » applicable to this article?
Thanks for your feedback.
Is the Subject Area « Food » applicable to this article?
Thanks for your feedback.
Is the Subject Area « Biodiversity » applicable to this article?
Thanks for your feedback.
Is the Subject Area « Death rates » applicable to this article?
Thanks for your feedback.
Is the Subject Area « Cancers and neoplasms » applicable to this article?
Thanks for your feedback.
Is the Subject Area « Nutrition » applicable to this article?
Thanks for your feedback.
Is the Subject Area « Species diversity » applicable to this article?
Thanks for your feedback.
Is the Subject Area « Europe » applicable to this article?
Thanks for your feedback.
PLOS PLOS is a nonprofit 501(c)(3) corporation, #C2354500, based in San Francisco, California, US

source

Vous aimerez aussi...