1 Consuming a Varied Diet can Prevent Diabetes But Can You Afford the Added Cost? Annalijn Conklin 18 January 2017, Vancouver, Canada
2 Overview The problem of type 2 diabetes What is diet diversity / food variety and how is it measured? The role of diet diversity for type 2 diabetes Cost implications of greater diet diversity Summary and relevance
3 Common problem rising fast concern for prevention Type 2 diabetes (T2D): a common chronic condition 1 in 12 people globally (46% undiagnosed); 1 in 4 Canadians Rates doubled in past decade, expected to rise to 1 in 3 by 2020 Diabetes rates are 3 5 times higher in First Nations Serious concern for prevention Leading cause of morbidity & early death (1/10) Major healthcare cost (C$11.7 bn), 80% from complications Widening health inequities IDF diabetes Atlas 6 th ed., Canadian Diabetes Association
4 Good diets are key to managing & preventing T2D Up to 80% of T2D can be prevented by changing diet, PA, etc. Dietary interventions most effective in preventing diabetes complications & disease progression Specific dietary components are protective Low-fat fermented dairy products reduce risk of T2D, by 28% Both quantity & variety of fruits & vegetables lower risk of T2D 12 different F&V items a week reduced T2D incidence by 39% O Connor et al 2014; Cooper et al 2012
5 Food variety is critical for healthy eating Long-standing concept in guidelines Recommended for Nutritional adequacy Balance of important nutrients Known health implications, e.g. Mortality (all-cause, cause-specific) Hypertension, T2D Overall health status; physical functioning Key limitations Mixed nomenclature (variety = diversity) Multiple ways of assessing variety Kouris-Blazos et al 2005; Clausen et al 2005; Kant et al 1993, 1995; Miller et al 1992.
6 Extending the concept: diversity between & within food groups Focus on food groups 1. Circumvents limits of nutrient databases 2. Closer to people s choices & more useful 3. Income effects different for various groups 4. True diet heterogeneity (5 major, 18 minor) Diversity scores derived from Food Frequency Questionnaire Habitual diet over 1 year Frequency of 131 food items consumed Consumed at least twice per week Total count of 5 food groups & subtypes in each
7 EPIC-Norfolk cohort: large health dataset Norfolk component of a large European collaborative community study of risk factors of chronic conditions (initially diet & cancer) Diet data (self-reported) assessed by validated semi-quantitative FFQ 10 y follow-up, incident T2D (Hospital records, ONS, disease registry)
N=25,639 participants attended a baseline health examination between 1993 and 1997 55% 8 N=24,784 participants without prevalent diabetes followed up for incident diabetes status N= 892 verified incident diabetes cases N= 23,892 participants without diabetes at 31 st July 2006 N= 822 diabetes cases included for this analysis with data on potential confounders N= 22,416 participants included for analysis with data on potential confounders N=23,238 final analytic sample
9 Measuring diet diversity & costs (1) Assigned FFQ items to food groups using WHO/FAO guidance Created 6 diversity scores counting food groups/subtypes 1. Total (0 5): dairy, fruit, vegetable, grain, meat/alternative 2. Dairy (0 3): milk; cheese; yoghurt 3. Fruit (0 3): vitamin A-rich; citrus & berry; other 4. Vegetable (0 4): vitamin A-rich; dark green leafy, starchy tubers; other 5. Grain/bread (0 2): wholegrains; non-wholegrains 6. Meat & alternative (0 6): red flesh meat; poultry flesh meat; organ meat; fish & seafood; eggs; pulses
10 Measuring diet diversity & costs (2) Estimated monetary cost Created food price per 100 g edible portion Adjusted for preparation & waste Derived daily retail cost of each cohort participant s whole diet ( /d)
Regular intake = 2+ times/week 11 i.e. FFQ response categories 4 9
12 Analytic approach 1. Prospective association of diversity scores and incident T2D Multivariable Cox regression (n=23,238, 892 new cases) Series of models adjusted for age, sex, BMI, total energy intake, lifestyles factors, family history, and SES (education & social class) Mutually adjusted for other diversity scores Some scores regrouped due to low numbers (e.g. total (0 3, 4, 5)) 2. Cross-sectional association of diversity and diet cost Multivariable linear regression, adjusted for age, sex & total energy intake Post-estimation of adjusted means (95% CIs)
13 Characteristics of total diversity in EPIC participants 3 food groups 4 food groups 5 food groups Obs. 1,028 5,104 17,838 Women, N. (%) 538 (52%) 2,697 (53%) 9,960 (56%) Mean age at recruitment 58 (9) 58 (9) 59 (9) A-level (up to 18 y) or degree educated 489 (48%) 2,593 (51%) 9,696 (54%) Highest social classes (I & II) 395 (39%) 2,069 (42%) 7,844 (45%) Moderately active/ Active 374 (36%) 2,017 (39%) 7,520 (42%) Never smoker 393 (39%) 2,204 (44%) 8,456 (48%) Mean BMI (kg/m 2 ) 26.1 (3.8) 26.2 (3.9) 26.3 (3.8) Waist circumference (cm) 88.2 (12.3) 88.0 (12.3) 88.0 (12.3) Total energy intake (kcal/d) 1544 (508) 1815 (518) 2145 (591) Total alcohol intake (g/d) 137 (277) 143 (275) 124 (224)
14 Total dietary diversity lowers risk of T2D Ref: 0 3 major groups Ref: 0 3 major groups 1.25 1 p-trend=0.024 1.5 1.25 0.75 0.70 1 0.75 0.5 0.5 0.25 0.25 0 4 groups 5 groups 0 4 groups 5 groups Hazard ratios (95% Cis) adjusted for age, sex, BMI, total energy, smoking, alcohol, PA, family history, education & social class Hazard ratios (95% Cis) adjusted for age, sex, BMI, lifestyle factors, family history, SES & within-group diversity scores
15 Each dairy subgroup lowers risk of diabetes 1.25 1 Ref: 0 dairy groups p-trend=0.002 1.25 1 Ref: 0 dairy groups p-trend=0.008 0.75 0.72 0.75 0.61 0.75 0.73 0.78 0.64 0.5 0.5 0.25 0.25 0 1 group 2 groups 3 groups 0 1 group 2 groups 3 groups Hazard ratios (95% Cis) adjusted for age, sex, BMI, total energy, smoking, alcohol, PA, family history, education & social class Hazard ratios (95% Cis) adjusted for age, sex, BMI, lifestyle factors, family history, SES & within-group diversity scores
16 Fruit diversity lowers risk of diabetes 1.25 Ref: 0 fruit groups p-trend=0.005 1.25 Ref: 0 fruit groups p-trend=0.02 1 1 0.75 0.69 0.75 0.75 0.5 0.5 0.25 0.25 0 1 group 2 groups 3 groups 0 1 group 2 groups 3 groups Hazard ratios (95% Cis) adjusted for age, sex, BMI, total energy, smoking, alcohol, PA, family history, education & social class Hazard ratios (95% Cis) adjusted for age, sex, BMI, lifestyle factors, family history, SES & within-group diversity scores
17 Vegetable diversity lowers risk of diabetes 1.25 Ref: 0 1 vegetable groups p-trend=0.001 1.25 Ref: 0 1 vegetable groups p-trend=0.004 1 1 0.75 0.69 0.67 0.75 0.72 0.71 0.5 0.5 0.25 0.25 0 2 groups 3 groups 4 groups 0 2 groups 3 groups 4 groups Hazard ratios (95% Cis) adjusted for age, sex, BMI, total energy, smoking, alcohol, PA, family history, education & social class Hazard ratios (95% Cis) adjusted for age, sex, BMI, lifestyle factors, family history, SES & within-group diversity scores
18 Total diet cost increases with greater diversity between & within food groups per day 5 p-trend < 0.001 4.5 4.15 (C$6.72) 4 3.5 3 3.53 (C$5.71) Overall (0-3, 4, 5) dairy (0, 1, 2, 3) fruit (0, 1, 2, 3) vegetable (0-1, 2, 3, 4) # of food groups Mean cost adjusted for age, sex & total energy
19 Summary A varied diet is recommended for healthy eating, but diversity between and within food groups rarely studied together Consuming a combination of five major food groups is associated with a 30% reduced incidence of T2D, and 18% added cost Independent effects on lower T2D incidence of greatest diversity in dairy products (36%), fruits (25%) and vegetables (29%) Each additional subtype consumed also increased daily diet cost Conklin AI, et al. PLOS Medicine, 2016; 13(7)
20 Relevance Canada endorsed the UN Declaration on preventing and controlling noncommunicable diseases (2011) Global 5-a-day campaigns emphasise variety of FVs Added clarity on need for diversity of fruits, separate from diversity of vegetables Rising cost of foods, especially fruits & vegetables Comprehensive food pricing strategy
Relevance 21
22 Acknowledgements Co-authors: Pablo Monsivais, Kay-Tee Khaw, Nick Wareham & Nita Forouhi This work was supported by the Gates Cambridge Trust, by the Canadian Institute of Health Research (CIHR) Postdoctoral Award (MFE-135520), and by the WORLD Policy Analysis Center. Work was mostly undertaken by the Centre for Diet and Activity Research (CEDAR), a UKCRC Public Health Research Centre of Excellence, funded by: the British Heart Foundation, Economic and Social Research Council, Medical Research Council, the National Institute for Health Research, and the Wellcome Trust. EPIC-Norfolk is supported by programme grants from the Medical Research Council and Cancer Research UK.