The Search for Optimal Diets: A Progress Report
Walter C. Willett, MD, DrPH Department of Nutrition
Harvard School of Public Health
June 20, 2014
Disclosure Statement
• I have nothing to disclose.
10-Year Coronary Incidence Per 10,000 Men
Keys, 1980
Incidence
0
1000
2000
3000
0 5 10 15 20 25
% Diet Calories from Saturated Fat
Velika Krsna
Zrenjanin
Belgrade Montegiorgio
Crevalcore
Corfu
Crete
Slavonia
Zutphen
west Finland
east Finland
Ushibuka
Tanushimaru
y=77+78x r=0.73
9.006
(Keys 1980)
0
10
20
30
40
50
0 40 80 120 160 200 240 280 320
Per Capita Daily Meat Consumption (grams)
Co
lon
Can
cer
Inci
den
ce /
100,
000
Wo
men
(Armstrong & Doll, 1975)
ROM
CHI
COL
YUG
JAM
JAP
NIG
ISR NOR
PR POL
HUN
FIN
DEN
SWE NET FDR
ICE DDR
NZ
USA
CAN
UK
Meat & Colon Cancer Incidence
2.003
Carroll, 1975
Taiwan Japan
Ceylon
El Salvador Thailand
Panama
Philippines Mexico
Columbia Spain
Greece
Poland
Puerto Rico
Chile Venezuela
Hong Kong
Romania Yugoslavia
Czechoslovakia Italy
Portugal
Bulgaria
Australia
Norway France
Finland Hungary
Belgium Switzerland
Sweden Germany Austria
Canada UK Netherlands
USA Ireland
New Zealand Denmark
0.001
0
5
10
15
20
25
0 20 40 60 80 100 120 140 160
Animal Fat Intake (g/day)
Bre
ast
Can
cer
Dea
ths
/ 100
,000
po
p
Animal Fat and Breast Cancer Mortality
21.004
To reduce your fat intake:
Eat more fruits, vegetables, and their juices. Most are naturally low in fat…and high in vitamins and minerals.
Here are some other ways to reduce fat in your food.
TRY INSTEAD OF
Butter-flavored granules Butter or margarine Nonfat yogurt Regular yogurt Nonfat salad dressings Regular salad dressings Angel food cake Devil’s food cake Fat-free cookies and crackers High-fat cookies and crackers
1: Use Nonfat Products
9.071
21.031
21.038
9.010
8.063
Effect of Trans and Saturated Fat (10% E) on Blood Lipids (vs Monounsaturated fat)
(Mensink & Katan, 1990)
Trans fat Saturated fat
Total cholesterol +6% +12%
LDL cholesterol +14% +18%
HDL cholesterol -12% 0%
LDL/HDL ratio +29% +18%
9.110
Blood lipids
Blood pressure
Thrombotic tendency
Insulin resistance
Oxidation
Homocysteine
Inflammation/endothelial dysfunction
Ventricular irritability & arrhythmia
Diet CHD
9.105R
Age-Adjusted Plasma CRP by Quintiles of Trans Fatty Acid Intake in the Nurses’ Health Study
0
0.5
1
1.5
2
Q1 Q2 Q3 Q4 Q5
CR
P (
mg
/L)
Trans Fat Intake (Quintiles)
(Lopez-Garcia 2005)
(P, trend = <0.001)
8.097
Nurses’ Health Study (n=121,700)
Health Professionals Follow-up Study (n=52,000)
Nurses’ Health Study II (n=116,000)
Investigators: Frank Hu, Frank Speizer, Bernie Rosner, Meir Stampfer, Graham Colditz, David Hunter, JoAnn Manson, Sue Hankinson, Eric Rimm, Edward Giovannucci, Alberto Ascherio, Gary Curhan, Charlie Fuchs, Fran Grodstein, Michelle Holmes, Donna Spiegelman
1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2010
Diet Ocs Smoking Weight/Ht Med. Hist.
Diet Diet Diet Blood
Diet Diet
1986 1988 1990 1992 1994 1996 1998 2000 2010
Diet Diet Diet Blood
Diet
1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009
Diet Diet Diet
0.198a
Nails
Nails
Blood
100
80
60
40
20
0
-20
-40
1%E 2%E 3%E 4%E 5%E
Trans
Sat
Mono
Poly
% C
han
ge
in C
HD
Hu FB, et al. N Engl J Med 1997;337:1491-9 9.131
Type of Dietary Fat and Risk of Coronary Heart Disease The Nurses' Health Study
14-Year Follow-up
9.246
Kaplan-Meier Estimates of the Incidence of Outcome Events in the Total Predimed Study Population
(Estruch R et al. NEJM 2013)
Years
0
0.5
1
3.6 4.8 5.6 6.9
Multivariate Relative Risk of Sudden Death (Albert et al., 2002)
P = 0.007
Quartile of blood N-3 fatty acid (Mean, % of fatty acids)
9.129
Years after randomization
1 2 3 4 5
Cardiac Deaths and Non-Fatal AMI (Lyon Heart Study)
9.018
1.0
0.9
0.8
Control
Experimental
=
(De Lorgeril et al.)
(Howard et al. 2006)
9.152
Conclusions 1. CHD rates can be dramatically reduced by
nutritional means, but this will not be achieved by replacing saturated fat with carbohydrate.
2. We should abandon recommendations regarding % of energy from fat and avoid pejorative references to fat or “fatty foods”.
3. Advice about dietary fat should focus on replacement of saturated and trans fat with vegetable oil, including sources of N-3 fatty acids.
9.062
29.534
Conclusion from Chowdhury Abstract
“Current evidence does not clearly support cardiovascular guidelines that encourage high consumption of polyunsaturated fatty acids and low consumption of total saturated fats.”
(Chowdhury R. et al. Ann Intern Med 2014:160:398-406)
Saturated Fat
Trans Fat
Refined Starch, Sugar
Whole Grains
Unsaturated Vegetable Fats --High monounsaturated vegetable fats --High polyunsaturated vegetable fats
Carbohydrates
29.536
0.6
0.8
1
1.2
1.4
Q1 Q2 Q3 Q4 Q5
Saturated FatMonoPolyTrans
Multivariate RR’s of type 2 diabetes according to quintiles of specific types of dietary fat (mutually adjusted)
(Salmeron et al, 1999)
Quintiles of Fat Intake
Mu
ltiv
aria
te R
R
25.004
Fat & Postmenopausal Breast Cancer in NHS, 1980-2000 (3537 cases)
0
0.5
1
1.5
2
<=2020.1-25
25.1-30
30.1-35
35.1-40
40.1-45
45.1-50
>50
P, trend test 0.11
(Kim et al. 2006)
RR
of
Bre
ast C
ance
r
Cumulative Average Fat Intake (%E)
0.255
00.20.40.60.8
11.21.41.61.8
2
1 2 3 4 5
Animal Fat (p=0.002)
Vegetable Fat (p=0.71)
Quintile of Intake
RR of Breast Cancer according to quintile of cumulative averaged fat intake in the Nurses’ Health Study II (1991-1999)
(Cho et al., 2003)
0.222
(Dansinger et al. 2005)
29.167
Changes in Food and Beverage Consumption
and Weight Changes Every 4 Years
According to Study Cohort
(Mozaffarian D et al., NEJM 2011)
Food Beverages
23.059
30% fat
33% fat 39% fat
Mean BMI at baseline = 31
Shai et al NEJM 2008
Randomized Dietary Trials of Equal Intensity Intervention
26.101
Cancer P-value, test for trend=0.88
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
<1.5 1.5-2.9 3-4.9 5-5.9 6-7.9 8+
Rel
ativ
e R
isk
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
<1.5 1.5-2.9 3-4.9 5-5.9 6-7.9 8+
Rel
ativ
e R
isk
Cardiovascular Disease P-value, test for trend=0.0003
(Hung et al., 2004) 28.021 Fruit/Vegetable Intake (Serving/Day)
29.294
Milling of Grains
Whole Grain
Refined Grain
All parts milled
Endosperm is milled
Germ & Bran
Source: General Mills 29.295
0
0.5
1
1.5
Women ( 591cases ) Wolk et al., 1999
Men ( 734 cases ) Rimm et al., 1996
RR of
CHD
Cereal Fiber, Energy- Adjusted, g/day
0
0.5
1
1.5
2.2 3.1 4.9 3.8 7.7 2.2 3.7 5.0 6.8 9.7
8.064
Blood Glucose
Insulin
Easily Digested Carbohydrate
-
Blood Glucose
Insulin
Slowly Digested Carbohydrate
0
0 1 2 3 4 5 0 1 2 3 4 5
Time (hr) Time (hr)
25.027 (Willett EDBH, 2001)
Relative Risk of Type 2 Diabetes by Different Levels of Cereal Fiber and Glycemic Load
2.5 2.32.05
2.171.8
1.62
1.511.28
1
0
1
2
3
High Medium Low
High
Medium
LowRelative Risk
>165 165-143 <143 Glycemic Load
>5.8 g/day
2.5 -5.8 g/day
<2.5 g/day
(Salmeron et al,1997)
(ref)
WOMEN
9.038
Cereal Fiber
1.501.85
1.061.001.39 1.41
1.00 1.11
0.0
0.5
1.01.5
2.0
2.5
<1/mo 1-4/mo 2-6/wk >=1/dSugar-sweetened soft drink consumption
Rel
ativ
e R
isk
multivariate adjusted multivariate + BMI
P<0.001 for trend
Regular Soft Drinks and Type 2 Diabetes, NHS2
(Schulze et al. 2004 JAMA) 25.080
RR .626039 1 2 3
Combined
This image cannot currently be displayed.
This image cannot currently be displayed.
This image cannot currently be displayed.
This image cannot currently be displayed.
This image cannot currently be displayed.
This image cannot currently be displayed.
This image cannot currently be displayed.
This image cannot currently be displayed.
1.26 (1.12, 1.41)
Montonen, 2007
Paynter Men, 2006
Paynter Women, 2006
Schulze, 2004
Palmer, 2008
Bazzano, 2008
Odegaard, 2010
Nettleton, 2009
de Koning, 2010
Sugar-sweetened beverage consumption and risk of T2DM, comparing extreme categories of intake (random-effects estimate)
Fixed-effects estimate: RR 1.25 (1.17, 1.32) Omitting 3 studies that adjusted for BMI and total energy: Random effects: RR 1.28 (1.13, 1.45) Fixed effects: 1.25 (1.18, 1.34) (Malik et al. Diabetes Care 2010) 37.013
Liu et al., 2000
Body Mass Index (kg/M2)
Rel
ativ
e R
isk
9.072
1.16
2 2
0.94 1.19
1.81
1 1.1 1.42
0
0.5
1
1.5
2
2.5
<23 23-29 >29
Tertile 1 (lowest) Tertile 2
Tertile 3 (highest)
Relative Risk of Coronary Heart Disease
0.4
0.6
0.8
1
1.2
1.4
1.6
Reference 1.5 - 5 > 5 - 10 > 10 - 20 > 20-30 > 30
glasses of milk per week
pool
ed R
R (9
5% C
I)
Pooled Analysis of Categories of Milk Intake* & Hip Fractures in Women (All Studies)
*The reference categories in different studies ranged from rarely/never to 1 glass/day
(Bischoff-Ferrari et al.) 24.056
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
<2/wk 2-6/wk 1/d 2-3/d 4+/d
RR of hip fractures by frequency of milk consumption during teenage years among men aged 50+ years in HPFS and among postmenopausal women in NHS
Rel
ativ
e R
isks
(Teenage Milk Consumption, Glasses)
Feskanich D et al., JAMA Pediatrics 2013)
Men
Women
24.081
Diet and Fatal Prostate Cancer in SDA Men, 1960-1980
(n = 99 cases) (Snowdon et al, 1984)
Milk
<1 glass/day 1.0
1-2 glasses/day 1.8 (1.0-3.0)
3+ glasses/day 2.4 (1.3-4.3)
p trend = 0.005
3.020
Calcium Intake & Risk of Prostate Cancer, 1986-2002 (Giovannucci et al. 2006)
2
1
0
<500 500-749
750-999
1000-1499
1500-1999
2000+
Rel
ativ
e R
isk
Calcium Intake mg/day 3.032
Fatal (p=0.01)
Non-advanced (p=0.55)
Risk of Hip Fractures Between Vitamin D (700-800 IU/d) & Control Groups (Bischoff-Ferrari 2005)
24.059b
Risk of Colon Cancer by 25-OHD levels
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
38 50 60 70 88
25-OHD (nmol/l)
RR
(95
% C
I)Risk of Colon Cancer by 25-Hydroxyvitamin D Levels
(Feskanich, 2004) 32.001
P, trend=0.02
Meta-analysis of Vitamin D and Colorectal Cancer
(Gorham et al., 2007) 32.047
Attributable Risk of Coronary Heart Disease Due to Modifiable Diet and Lifestyle Risk
Factors in the NHS (1980 to 1994) Low Risk: 1. Non smoker 2. BMI < 25 kg/m2
3. Exercise > ½ hr of brisk walking/day 4. Good diet (upper 2 quartiles of score based on low trans fat, high p/s ratio, low glycemic load, high cereal fiber, high fish, high total folate) 5. Alcohol 5+g/day • Proportion at low risk = 3.1% • Population Attributable Risk = 82% (95% CI = 58-93%)
Stampfer et al, 2000 9.092
Percentage of Type 2 Diabetes Potentially Preventable by Simultaneous Reduction of Five
Modifiable Risk Factors (NHS) (Hu et al.)
Low Risk 1. Nonsmoking 2. BMI < 25 3. Moderate to vigorous exercise 4. Diet score in upper 40% (low trans fat, high cereal fiber, low glycemic load, high P:S ratio) 5. Alcohol 5+ grams/day Percent in low risk group: 4.1% Population attributable risk (PAR): 92% (82-96)
25.026R
2-Year Randomized Trial of Mediterranean Diet in Patients with Insulin Resistance
Syndrome (Esposito E, et al. 2004)
Resolution of Syndrome
Med Diet 50/90
Control Diet 12/90
P < 0.001
29.131
Proportion of Colon Cancers that are Potentially Preventable by Simultaneous Reduction 6
Modifiable Risk Factors (HPFS) (Platz et al. 2000)
Low Risk 1. BMI ≤25 kg/m2
2. Physical activity ≤30 min/day of vigorous – moderate activity 3. Alcohol <15 g/day or 15-30 g/day with supplemental folic acid 4. Folic acid supplement of e100 µg/day 5. ≤3 pack – years of smoking 6. Red meat ≤2 servings/week Joint low risk group = 3.1% of population Population attributable risk (PAR): 71% (33-92)
2.116R
Lowfat products whenever possible; calcium supplements are and effective substitute for preventing fractures
Importance is well-documented; greens and dark orange vegetables should be included. Even more frequent servings may be desirable
Whole-grain, minimally processed products should be emphasized
Ignores critical differences in types of fat. Monos appear desirable
Misleading as 2-3 servings of meat/day is probably unhealthy
Support for generous intake is well documented
21.002
21.090
29.446
29.443
21.058
21.054
21.055
21.056