effect of current dietary recommendations on weight loss ...special focus issue: cardiovascular...
TRANSCRIPT
Listen to this manuscript’s
audio summary by
JACC Editor-in-Chief
Dr. Valentin Fuster.
J O U R N A L O F T H E AM E R I C A N C O L L E G E O F C A R D I O L O G Y V O L . 6 9 , N O . 9 , 2 0 1 7
ª 2 0 1 7 T H E A U T HO R S . P U B L I S H E D B Y E L S E V I E R O N B E H A L F O F T H E A M E R I C A N
C O L L E G E O F C A R D I O L O G Y F OU N D A T I O N . T H I S I S A N O P E N A C C E S S A R T I C L E U N D E R
T H E C C B Y - N C - N D L I C E N S E ( h t t p : / / c r e a t i v e c o mm o n s . o r g / l i c e n s e s / b y - n c - n d / 4 . 0 / ) .
I S S N 0 7 3 5 - 1 0 9 7
h t t p : / / d x . d o i . o r g / 1 0 . 1 0 1 6 / j . j a c c . 2 0 1 6 . 1 0 . 0 8 9
SPECIAL FOCUS ISSUE: CARDIOVASCULAR HEALTH PROMOTION
ORIGINAL INVESTIGATIONS
Effect of Current DietaryRecommendations on Weight Loss andCardiovascular Risk Factors
David J.A. Jenkins, MD,a,b,c,d Beatrice A. Boucher, MHSC,a,e Fredrick D. Ashbury, PHD,f,g,h,i Margaret Sloan, BA,ePatrick Brown, PHD,e,f Ahmed El-Sohemy, PHD,a Anthony J. Hanley, PHD,a Walter Willett, MD,j
Melanie Paquette, MSC,a,c Russell J. de Souza, SD,k Christopher Ireland, BSC,a,c Natalie Kwan, MSC,a,c
Amy Jenkins, MSC,c,l Sathish C. Pichika, MSC,a,c Nancy Kreiger, PHDa,e,f
ABSTRACT
Fro
Kn
To
ve
To
On
Fa
Sch
Im
BACKGROUND Dietary recommendations emphasize increased consumption of fruit, vegetables, and whole grain
cereals for prevention of chronic disease.
OBJECTIVES This study assessed the effect of dietary advice and/or food provision on body weight and cardiovascular
disease risk factors.
METHODS Healthy overweight men (n¼ 209) and women (n¼ 710), mean age 44.7 years, body mass index [BMI]
32.4 kg/m2, were randomized between November 2005 and August 2009 to receive Health Canada’s food guide (control,
n ¼ 486) or 1 of 3 interventions: dietary advice consistent with both Dietary Approaches to Stop Hypertension (DASH) and
dietary portfolio principles (n¼ 145); weekly food provision reflecting this advice (n¼ 148); or food delivery plus advice
(n¼ 140). Interventions lasted 6 months with 12-month follow-up. Semiquantitative food frequency questionnaires and
fasting blood, anthropometric and blood pressure measurements were obtained at baseline, 6 months, and 18 months.
RESULTS Participant retention at 6 and 18 months was 91% and 81%, respectively, after food provision compared to
67% and 57% when no food was provided (p < 0.0001). Test and control treatments showed small reductions in
body weight (�0.8 to �1.2 kg), waist circumference (�1.1 to �1.9 cm), and mean arterial pressure (0.0 to �1.1 mm Hg) at
6 months and Framingham coronary heart disease risk score at 18 months (�0.19 to �0.42%), which were significant
overall. Outcomes did not differ among test and control groups.
CONCLUSIONS Provision of foods increased retention but only modestly increased intake of recommended foods.
Current dietary recommendations showed small overall benefits in coronary heart disease risk factors. Additional dietary
strategies to maximize these benefits are required. (Fruits, Vegetables, and Whole Grains: A Community-based
Intervention; NCT00516620) (J Am Coll Cardiol 2017;69:1103–12) © 2017 The Authors. Published by Elsevier on behalf
of the American College of Cardiology Foundation. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
m the aDepartment of Nutritional Sciences, Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada; bLi Ka Shing
owledge Institute, St. Michael’s Hospital, Toronto, Ontario, Canada; cRisk Factor Modification Centre, St. Michael’s Hospital,
ronto, Ontario, Canada; dDivision of Endocrinology and Metabolism, St. Michael’s Hospital, Toronto, Ontario, Canada; ePre-
ntion and Cancer Control, Cancer Care Ontario, Toronto, Ontario, Canada; fDalla Lana School of Public Health, University of
ronto, Toronto, Ontario, Canada; gInstitute of Health Policy, Management and Evaluation, University of Toronto, Toronto,
tario, Canada; hDivision of Preventive Oncology, University of Calgary, Calgary, Alberta, Canada; iDepartment of Psychology,
culty of Social Sciences, University of Wollongong, Wollongong, New South Wales, Australia; jNutrition Department, Harvard
ool of Public Health, Harvard University, Boston, Massachusetts; kDepartment of Health Research Methods, Evidence, and
pact, McMaster University, Ontario, Canada; and the lDepartment of Food Science, University of Guelph, Guelph, Ontario, Canada.
ABBR EV I A T I ON S
AND ACRONYMS
BMI = body mass index
CHD = coronary heart disease
CVD = cardiovascular disease
HDL = high-density lipoprotein
LDL = low-density lipopro
This resear
Loblaw Cos
items for th
Foods, Ear
Products, a
Innovation
Barilla, Alm
Procter & G
ciation of N
Internation
Canada, Ca
Research F
from Almon
Danone, Sa
Research a
mental for
Hospital, A
Pioneer Hi
Material Ne
the Canadi
Nutrition F
chael’s Hos
Society of N
has receive
Internation
and Metab
Research/H
employee o
have no re
Manuscript
Jenkins et al. J A C C V O L . 6 9 , N O . 9 , 2 0 1 7
Effect of Dietary Recommendations on Risk Factors M A R C H 7 , 2 0 1 7 : 1 1 0 3 – 1 2
1104
D ata from major cohort studies sug-gest that 82% of cardiovasculardisease and 91% of diabetes risk
may be prevented by changes in diet andlifestyle (1,2). To address the growing burdenof chronic disease, dietary guidelines for thegeneral public, including the Dietary Refer-ence Intakes for the United States and Can-
ada, focus on chronic disease reduction in additionto simple nutritional adequacy. The most recent sci-entific recommendations of the U.S. Dietary Guide-lines Advisory committee now advocate 3 dietarypatterns to prevent chronic disease: the healthyAmerican diet, the Mediterranean diet, and for the
tein
SEE PAGE 1113
first time, a vegetarian diet (3). All of these dietarypatterns incorporate traditional advice to eat morefruit, vegetables, and whole grain cereals, plus morerecent advice to eat more cholesterol-lowering “func-tional” foods such as oats, barley, nuts, and plant pro-tein foods (e.g., soy and other legumes). This dietaryapproach fits with the recent report of the workshopconvened by the World Heart Federation (4). Thera-peutic diets such as the Dietary Approaches to StopHypertension (DASH) recommended in the AHA/ACCguidelines (5) and the dietary portfolio recommended
ch was supported by Institute of Nutrition, Metabolism, Diabetes
. Ltd. Dr. Jenkins was funded by the government of Canada thr
e food basket were contributed by Loblaw Cos. Ltd., Alberta Barle
th’s Own Food Co. Inc., International Nut Council, Kellogg Co., M
nd So Soyaþ Inc. Dr. Jenkins has received research grants from Sa
Program through the Pulse Research Network, Advanced Foods
ond Board of CA, Agriculture and Agri-food Canada, Pulse Ca
amble Technical Centre Ltd., Bayer Consumer Care, Pepsi/Quake
orth America, Coca-Cola Co. (investigator-initiated, unrestricted
al Tree Nut Council Nutrition Research and Education Founda
nola Council of Canada, Canadian Institutes of Health Researc
und; and has served on the speaker’s panel and scientific advisory
d Board of California, Canadian Agriculture Policy Institute, Lobla
skatchewan Pulse Growers, Sanitarium Co., Orafti, American Pean
nd Education Foundation, Peanut Institute, Herbalife Internatio
Health, Metagenics, Bayer Consumer Care, Unilever Canada an
bbott Laboratories, Dean Foods, California Strawberry Commi
-Bred International, DuPont Nutrition and Health, Spherix Consu
twork, Canola and Flax Councils of Canada, Nutritional Fundame
an Agri-Food Policy Institute, Pulse Canada, Saskatchewan Pulse
oundation of Italy, Nutra-Source Diagnostics, McDougall Progra
pital), Canadian College of Naturopathic Medicine, Hospital for S
utrition, Arizona State University, Paolo Sorbini Foundation and
d honoraria from U.S. Department of Agriculture; and has received
al Nut and Dried Fruit Council; and has received funding and tra
olism. Dr. de Souza has received personal fees from World H
ealth Canada; and grants from Dietitians of Canada and Canadia
f Vineland Research and Innovation Centre, a not-for-profit org
lationships relevant to the contents of this paper to disclose.
received August 27, 2016; revised manuscript received Septemb
in the Canadian Cardiovascular Society guidelines (6)to lower cholesterol also emphasize these principlesand consistently result in large reductions in bloodpressure and lipids when taken under metabolicallycontrolled conditions (7–9). Yet despite efforts toencourage the general public to increase plant foodconsumption, the response has been slow (10).
We determined whether we could increase adher-ence to the DASH-type and dietary portfolio eatingpatterns and improve health outcomes by providingboth advice and food to healthy overweightindividuals.
METHODS
PARTICIPANTS. Participants were residents of thecity of Toronto, 18 years or older, English speaking, andhad body mass index (BMI) >25 kg/m2. Individuals orfamilies were recruited if at least 1 family member hadBMI >25 kg/m2, and blood pressure and thyroid medi-cations (thyroxin) dosages, if taken, were stable for atleast 1 month prior to starting the study. Exclusioncriteria included pregnancy or breastfeeding; activelyfollowing a special diet or weight-loss program; majorsurgery or a cardiovascular event in the previous 6months; diabetes, liver disease, renal failure, cancer(except nonmelanoma skin cancer), inflammatorybowel disease or major chronic inflammatory diseases;
of the Canadian Institute for Health Research, and
ough the Canada Research Chair Endowment. Food
y Commission, Almond Board of CA, Barilla, ConAgra
ars Canada, Pepsi-QTG Canada, Puresource Natural
skatchewan Pulse Growers, Agricultural Bioproducts
and Material Network, Loblaw Cos. Ltd., Unilever,
nada, Kellogg’s Co., Canada, Quaker Oats, Canada,
r, International Nut & Dried Fruit, Soy Foods Asso-
grant), Solae, Haine Celestial, Sanitarium Co., Orafti,
tion, Peanut Institute, Canola and Flax Councils of
h, Canada Foundation for Innovation and Ontario
board and has received travel support and honoraria
w Cos. Ltd., Griffin Hospital, Coca-Cola Co., Epicure,
ut Council, International Tree Nut Council Nutrition
nal, Pacific Health Laboratories, Nutritional Funda-
d Netherlands, Solae, Kellogg, Quaker Oats, Griffin
ssion, Haine Celestial, PepsiCo, Alpro Foundation,
lting and WhiteWave Foods, Advanced Foods and
ntals for Health, Agri-Culture and Agri-Food Canada,
Growers, Soy Foods Association of North America,
m, Toronto Knowledge Translation Group (St. Mi-
ick Children, Canadian Nutrition Society, American
Institute of Nutrition, Metabolism, and Diabetes; and
the 2013 Award for Excellence in Research from the
vel support from Canadian Society of Endocrinology
ealth Organization, Canadian Institutes for Health
n Institutes for Health Research. Amy Jenkins is an
anization. All other authors have reported that they
er 17, 2016, accepted October 18, 2016.
J A C C V O L . 6 9 , N O . 9 , 2 0 1 7 Jenkins et al.M A R C H 7 , 2 0 1 7 : 1 1 0 3 – 1 2 Effect of Dietary Recommendations on Risk Factors
1105
acute or chronic infections, irritable bowel syndrome;peanut or nut allergy; or a blood pressure >145/95mm Hg on more than one occasion.
PROTOCOL. Eligible participants recruited by adver-tisements between October 2005 and July 2009 weresent questionnaires including a semiquantitativefood frequency questionnaire (11). Completed ques-tionnaires, fasting blood, anthropometric, and bloodpressure measurements were obtained at baselineand at subsequent clinic attendances at 6 and 18months at St. Michael’s Hospital, a University ofToronto Teaching Hospital. Research Ethics boards ofthe University of Toronto and St. Michael’s Hospitalapproved the protocol. All participants providedsigned consent. The trial was registered withClinicalTrials.gov (NCT00516620).
DIETARY INTERVENTIONS. All participants receiveda copy of Health Canada’s Food Guide. No furtheradvice was given to the control group. The firsttreatment group received additional dietary adviceweekly for the first month and monthly for thefollowing 5 months as 20- to 30-min telephone in-terviews with individual participants or the families’primary shopper or cook. The advice addressed ben-efits, strategies for change, and barriers to change foreach participating family member. Participants wereencouraged to increase intake of fruit, vegetables,whole grain cereals, to reduce meat and sweets, andto increase consumption of cholesterol-loweringfunctional foods including soy foods, nuts, andviscous fiber sources such as oats and barley. A sec-ond treatment group received a weekly food basket(Food Share, Toronto, Ontario) for 6 months, reflect-ing the advice given to the first treatment group butdid not receive dietary advice. A third treatmentgroup received both the weekly food basket and di-etary advice. All members of the same family wereexpected to follow the same treatment. Exercisepatterns were recorded but no additional advice wasgiven (Figure 1). The effect of advice to avoid sugar-sweetened beverages was also assessed using afactorial design with the addition of a treatmentgroup that involved sugar sweetened beverage advicealone. The effect of sugar-sweetened beverage advicewill be reported separately.
BIOCHEMICAL AND DIETARY ANALYSES. Biochem-ical analyses were performed in the St. Michael’sHospital (Toronto) routine laboratory. Participants’body weight was measured without shoes in lightindoor clothing on the same beam balance at eachclinic visit. Blood pressure was the mean of 3 readingsusing an automatic sphygmomanometer (Omron HEM907 XL, Omron Healthcare Inc., Burlington, Ontario,
Canada). The original validated semiquantitativefood frequency questionnaire of Willett et al. (11) wasexpanded to 184 food items to better capture wholegrains and viscous fiber products, including oats andbarley. Questionnaires were processed in the Nutri-tion Department, Harvard School of Public Health.
SAMPLE SIZE. The power calculation was based onhouseholds, and randomization was planned toassess dietary change. Sample size was calculatedusing Student’s t-test for independent observationsthat required 360 controls, 120 for dietary advice,120 for food delivery, and 120 for food delivery anddietary advice (Online Methods 1).
RANDOMIZATION. Randomization took place be-tween November 2005 and August 2009 to control or1 of the 3 dietary treatment groups to achieve anapproximately 3:1:1:1 ratio for the control and the 3dietary interventions, respectively. Participants wererandomized as households (assuming an average of 2family members per household), and all members ofthe same household received the same treatment. Astatistician not involved in the day-to-day operationof the interventions created blocks of random as-signments (n ¼ 39). Assignments were sealed in or-dered, numbered, opaque envelopes. Upon consentand eligibility confirmation for the individual orhousehold, the coordinator opened each envelope insequence and assigned the participant to the treat-ment group it contained. To allow for assessment ofthe effect of advice to reduce consumption of sugar-sweetened beverages (to be reported separately), afactorial design was used with the addition of an ex-tra cell of 60 households given only this advice. Theirdata are reported as part of the control group in thepresent study.
STATISTICS. Means and 95% confidence intervals(CIs) are provided. Body weight change was the pri-mary outcome. To model change, the dependentvariable was the end-of-study value with treatment(categorical) and the baseline value as independentvariables in a mixed model ANCOVA (PROC MIXED),SAS version 9.4 software (Cary, North Carolina). ATukey correction was applied for multiple compari-sons where significance was found. We used an ITT(intention-to-treat) analysis that accounted for dropouts using multiple imputation (5 sets) and a mono-tone predictive mean matching method (PROC MI)modeled with visit, age, sex, baseline weight andheight, smoking status, income, education, andbaseline outcome as covariates (Online Methods 2and 3). Results from the 5 imputations were pooledusing PROC MIANALYZE. To address correlated ob-servations within families, we added a random effect
FIGURE 1 Participant Flow Diagram
Advertising4,634 calls received
919 Randomized
Control*
n=486
6 monthsn=334
6 monthsn=89
6 monthsn=136
6 monthsn=126
108 dropout34 lost contact7 withdrawn†
3 no visit‡
40 dropout13 lost contact2 withdrawn†
1 no visit‡
6 dropout2 lost contact1 withdrawn†
3 no visit‡
7 dropout4 lost contact2 withdrawn†
1 no visit‡
10 dropout2 lost contact0 withdrawn†
10 dropout4 lost contact1 withdrawn†
10 dropout3 lost contact4 withdrawn†
35 dropout19 lost contact0 withdrawn†
18 months*
n=28318 months*
n=7818 months*
n=12418 months*
n=110
Advice*
n=145Food*
n=148Food & Advice*
n=140
*Households at baseline ¼ 441 control, 127 advice, 126 food, 125 food and advice; at 18 months ¼ 260 control, 67 advice, 105 food, 100 food and advice.
†Withdrawn due to medical reasons, pregnancy, or participating in another study. ‡Did not attend the 6-month intervention visit but attended the
18-month follow-up visit. Participants who missed their 6-month visit (n ¼ 8) were added to the 18-month visit.
Jenkins et al. J A C C V O L . 6 9 , N O . 9 , 2 0 1 7
Effect of Dietary Recommendations on Risk Factors M A R C H 7 , 2 0 1 7 : 1 1 0 3 – 1 2
1106
term to the statistical model to account for familyunit. Framingham coronary heart disease (CHD) riskscores were calculated (12) using participants’ age atbaseline. Dietary data were analyzed using ANCOVA,using all available data. Two further assessments ofthe data were also undertaken. First to assess theoutcomes using household as the unit of analysis, themean values of each household were taken and sec-ondly those participants whose BMI was <25 kg/m2
(n ¼ 36) were removed from the analysis, and againthe data were reanalyzed.
RESULTS
SIX-MONTH OUTCOMES. A total of 919 participantswere randomized (209 men, 710 women; mean age:44.7 years, and BMI: 32.4 kg/m2) (Online Table 1).A total of 722 were randomized as single-memberhouseholds, 94 as 2-member households (n ¼ 188),and 3 as 3-member households (n ¼ 9). A total of
685 participants completed 6 months (170 men, 515women), equating to a 75% retention rate (Figure 1).
When “food provided” was compared to “no foodprovided,” a major difference was seen in retentionfavoring food provision, with 91% retention for thosewith food provision versus 67% (p < 0.0001) forthose without food provision completing the 6-monthactive intervention.
SIX-MONTH CHANGE IN DIETARY INTAKE. At 6 months,only small increases were seen in intake of fruit (0.3 to1.1 servings/day), vegetables (0.4 to 1.3 servings/day),and whole grains (0.0 to 1.0 servings/day) (Figure 2,Online Table 2). The increases in intake in the treat-ment groups compared to the control were evensmaller, with only food provision plus advice showingconsistent increases in fruit, vegetables, and wholegrain cereals (Online Table 3). The differences were0.8, 0.9, and 0.9 servings/day, respectively, for the 3treatments (Figure 2). Other treatment differences,where they occurred, were small (Online Table 3).
FIGURE 2 Changes in Servings of Fruit, Vegetable, and Whole Grains Intake by
Treatment at 6 Months
Fruit
Serv
ings
6 Month Changes in Diet Components
Vegetable
*
*
*
*
*
*
Control Advice only Food only Food and advice
Whole Grains
1.4
1.2
0.8
0.6
0.4
0.2
0
1
*Significantly different from control group (p < 0.05).
J A C C V O L . 6 9 , N O . 9 , 2 0 1 7 Jenkins et al.M A R C H 7 , 2 0 1 7 : 1 1 0 3 – 1 2 Effect of Dietary Recommendations on Risk Factors
1107
The same pattern of very small increases was seenwith the so-called functional, cholesterol-lowering,foods (nuts, viscous fiber, and soy protein foods).Thus the maximum increase in nut intake was only6.5 g/day (under one-quarter of an ounce) followingfood delivery plus advice (Online Table 2).
SIX-MONTH OUTCOMES: BODY WEIGHT AND CHD
RISK FACTORS. Overall, at 6 months, the cohortshowed small reductions in body weight (�1.0 kg),waist circumference (�1.4 cm), and blood pressure(diastolic: �0.8 mm Hg and mean arterialpressure: �0.7 mm Hg) (Table 1). Similar reductionswere seen on the control and individual test treat-ments for body weight (�0.8 to �1.2 kg), waistcircumference (�1.1 to �1.9 cm), and blood pressure(diastolic: �0.1 to �1.0 mm Hg, and mean arterialpressure: 0.0 to �1.1 mm Hg) (Table 2). The withintreatment differences in body weight and waistcircumference were significant for the control andtest treatments (Central Illustration).
The only other significant reductions were for bothdiastolic blood pressure and mean arterial bloodpressure on control (�1.0 mm Hg, p ¼ 0.002, and �0.9mm Hg, p ¼ 0.042, respectively) (Table 2). However,there were no significant treatment differences for anyof the outcome measurements between any of the4 groups (control and 3 treatments) (Online Table 4).
DIFFERENCES AND “LEGACY EFFECTS” AT 18
MONTHS. At 18 months, the overall retention was65%, including 81% for those provided with food and57% for those without food provision. Only small in-creases from baseline remained for fruit (0.4 to 0.6servings/day), vegetables (0.3 to 0.6 servings/day),and whole grain cereals (0 to 0.6 servings/day). Theseincreases were significantly reduced from the alreadymodest 6-month increases (Online Figure 1, OnlineTable 5). The same pattern was seen with thecholesterol-lowering functional foods, for example,for nuts, the increase was reduced at 18 months tobetween 3.3 and 4.7 g/day on the test treatments,and when compared to the control increase of3.6 g/day, these differences became negligible(�0.8 to 1.3 g/day) (Online Table 3).
For the whole cohort, the reductions in bodyweight, BMI and waist circumference that were seenat 6 months were maintained (Online Table 6). High-density lipoprotein cholesterol (HDL-C) rose between6 and 18 months (0.05 mmol/l, p < 0.0001) and thetotal-to-HDL-C ratio was reduced. Reduction in bloodpressure at 6 months was further reduced signifi-cantly between 6 and 18 months, with the reductionsin the Framingham CHD risk score becoming signifi-cant at 18 months (Online Table 6). These differences
were also reflected in the individual treatment re-sponses (Online Figure 2), with no significant treat-ment differences either between the 18-monthchanges from baseline or between the 6 to 18 monthsdifferences (Online Tables 4 and 7).
ADDITIONAL ANALYSES. No outcome differences at6 and 18 months were uncovered by assessing thedata by using household as the unit of analysis or byeliminating those family members who had bodyweights in the normal range (Online Tables 8 and 9).
ADVERSE EVENTS. There were 24 serious adverseevents according to the Code of Federal Regulations(13) (7 control, 11 food basket and dietary advice,4 food basket only, and 2 dietary advice only) (OnlineTable 10). After adjustment for length of participanttime in the study the difference between the pooledtest and control treatments failed to reach signifi-cance (p ¼ 0.068). However, separate assessment ofgastrointestinal-related events indicated significantlymore events among test participants thancontrol participants: 7 versus 0. Adverse eventswere cholecystectomy (n ¼ 3), appendectomy (n ¼ 2),hiatal hernia repair (n ¼ 1), and diverticulitis (n ¼ 1),p ¼ 0.005 (Online Table 10).
DISCUSSION
These data demonstrate the difficulty in effectivelypromoting fruit, vegetable, and whole grain cereals to
TABLE 1 Mean Baseline Values and Changes in Outcomes at 6 Months for All Participants*
N
Baseline 6 Months 6-Month Change
Mean 95% CI Mean 95% CI Mean 95% CI Pr > |t|†
Weight, kg 919 88.2 87.0 to 89.3 87.1 86.0 to 88.3 �1.0 �1.4 to �0.7 <0.0001
BMI, kg/m2 919 32.4 32.1 to 32.8 32.1 31.7 to 32.5 �0.4 �0.5 to �0.2 <0.0001
Waist, cm 919 101.4 100.5 to 102.3 100.0 99.0 to 101.0 �1.4 �1.9 to �0.9 <0.0001
Glucose, mmol/l 919 4.82 4.78 to 4.85 4.83 4.78 to 4.87 0.01 �0.03 to 0.04 0.648
Total cholesterol, mmol/l 919 5.06 5.00 to 5.12 5.01 4.94 to 5.08 �0.05 �0.10 to 0.01 0.097
LDL-C, mmol/l 915 3.23 3.18 to 3.28 3.19 3.13 to 3.24 �0.04 �0.08 to 0.00 0.062
HDL-C, mmol/l 919 1.28 1.26 to 1.31 1.27 1.25 to 1.29 �0.01 �0.03 to 0.00 0.108
Triglycerides, mmol/l 919 1.21 1.16 to 1.25 1.22 1.17 to 1.27 0.01 �0.03 to 0.05 0.593
Total cholesterol/HDL 919 4.16 4.09 to 4.23 4.15 4.08 to 4.23 �0.01 �0.05 to 0.04 0.830
Non-HDL cholesterol, mmol/l 919 3.77 3.72 to 3.83 3.74 3.67 to 3.80 �0.03 �0.08 to 0.01 0.112
Systolic blood pressure, mm Hg 919 114.6 113.8 to 115.4 114.0 113.1 to 114.8 �0.6 �1.3 to 0.1 0.085
Diastolic blood pressure, mm Hg 918 73.0 72.4 to 73.5 72.1 71.5 to 72.7 �0.8 �1.3 to �0.4 <0.001
Mean arterial pressure, mm Hg 918 86.8 86.2 to 87.4 86.1 85.4 to 86.8 �0.7 �1.3 to �0.1 0.019
10-yr CHD risk % 919 3.58 3.32 to 3.84 3.48 3.22 to 3.74 �0.10 �0.21 to 0.02 0.098
*To convert the values to milligrams per deciliter, divide cholesterol by 0.0259, triglycerides by 0.0113, and glucose by 0.0555. †p values in boldface indicate significance.
BMI ¼ body mass index; CHD ¼ coronary heart disease; HDL ¼ high-density lipoprotein; LDL ¼ low-density lipoprotein.
Jenkins et al. J A C C V O L . 6 9 , N O . 9 , 2 0 1 7
Effect of Dietary Recommendations on Risk Factors M A R C H 7 , 2 0 1 7 : 1 1 0 3 – 1 2
1108
the general population, using recommendations that,when followed, decrease risk factors for chronic dis-ease (Central Illustration). They indicate an urgentneed for innovative approaches to support theimplementation of current dietary advice (5,6,14,15).
Our findings are supported by a meta-analysis ofsmaller shorter studies, beginning in 1998, thatshowed no change in body weight compared to thecontrol treatment even when fruit and vegetableswere provided (16,17). However our results contrastwith trials in participants at higher risk (8). One trial of345 participants with hypercholesterolemia, many ofwhom had been given a statin “holiday,” showed low-density lipoprotein cholesterol (LDL-C) reductions of13% to 14% with corresponding 41% to 46% adherenceto a dietary portfolio of cholesterol-lowering foods (8),as opposed to only 12% adherence in the presentstudy. In this respect, a very recent large (543 partici-pants) comprehensive lifestyle peer group-basedintervention on cardiovascular risk factors also foundno significant effect on individual risk factors,although due to a major success in smoking cessation,a reduction in estimated cardiovascular risk was ach-ieved (18). In that trial 31% of participants smokedcompared to 7.8% in the current study. Smokingcessation, exercise, and stress management were notpart of our purely dietary intervention.
ADVERSE EVENTS. Those given advice plus the foodbasket suffered an excess of gastrointestinal side ef-fects. Although the diets prescribed are recom-mended for long-term gastrointestinal health, it ispossible that the acute increase in fiber from cereals,fruits, and vegetables may have caused some
abdominal discomfort and drawn attention to pre-existing conditions. The increase in fiber intakecompared to the control at 6 months was 8.2 g/dayextra for a 2,000 kcal diet. This fiber intake wasequivalent to 4 slices of whole wheat bread daily.However, these participants were already accus-tomed to a reasonable fiber intake at baseline(25.4 g/day), and problems were not anticipatedwith higher fiber diets.
FACTORS FAVORING LIFESTYLE CHANGE. The suc-cess of dietary advice may be influenced by theperception of immediate benefit from the interven-tion. Further emphasis is therefore required on thelonger term health advantages of sustaining a gooddiet for otherwise healthy people, as has beendemonstrated in cohort studies with long-termfollow-up (19,20). Usual diets may also representstable “habits” that are resistant to change withoutspecific personal and environmental supports (21).Thus more emphasis should be placed on overcomingbarriers related to methods of food preparation and inillustrating those situations in which the desiredfoods can be eaten (meals and snacks). Addressingthese issues in cafeteria trials where entire mealswere fed to healthy people addressed the issue of dietadherence and demonstrated that lower body weightand blood lipids can be achieved in the long term (22).Consumption of nuts has increased considerablysince the repeated demonstration of their cholesterol-lowering potential, their possible CHD risk reductionproperties in cohort studies, and recently, in a ran-domized controlled trial (23–25). However this seculartrend did not increase our success in promoting
TABLE 2 Mean Values and Changes (95% CI) in Outcomes at Baseline and 6 Months for All Participants, by Treatment Group*
Control (n ¼ 486) Food Only (n ¼ 148)
Baseline 6 Months 6 Months D Pr > |t|† Baseline 6 Months 6 Months D Pr > |t|†
Weight, kg 88 (86.5 to 89.5) 87 (85.5 to 88.5) �1 (�1.5 to �0.5) <0.001 89.2 (86.3 to 92.2) 88 (85.1 to 91) �1.2 (�1.9 to �0.4) 0.002
BMI, kg/m2 32.5 (32 to 33) 32.2 (31.7 to 32.7) �0.3 (�0.5 to �0.2) <0.001 32.6 (31.6 to 33.5) 32.2 (31.3 to 33.2) �0.4 (�0.6 to �0.1) 0.015
Waist, cm 101.2 (100 to 102.5) 100 (98.7 to 101.3) �1.2 (�1.9 to �0.6) <0.001 102.3 (99.8 to 104.8) 100.4 (97.9 to 102.9) �1.9 (�2.8 to �1) <0.0001
Glucose, mmol/l 4.81 (4.77 to 4.86) 4.83 (4.77 to 4.9) 0.02 (�0.04 to 0.08) 0.507 4.8 (4.71 to 4.89) 4.82 (4.72 to 4.91) 0.02 (�0.06 to 0.1) 0.657
Total cholesterol, mmol/l 5.09 (5 to 5.17) 5.03 (4.92 to 5.14) �0.05 (�0.14 to 0.03) 0.204 5 (4.87 to 5.14) 4.96 (4.82 to 5.11) �0.04 (�0.14 to 0.06) 0.400
LDL-C, mmol/l 3.27 (3.19 to 3.34) 3.22 (3.13 to 3.32) �0.04 (�0.12 to 0.04) 0.266 3.17 (3.06 to 3.29) 3.17 (3.04 to 3.29) 0 (�0.09 to 0.08) 0.913
HDL-C, mmol/l 1.27 (1.24 to 1.3) 1.26 (1.23 to 1.29) �0.01 (�0.03 to 0.01) 0.337 1.27 (1.21 to 1.32) 1.27 (1.21 to 1.32) 0 (�0.04 to 0.03) 0.969
Triglycerides, mmol/l 1.23 (1.16 to 1.29) 1.22 (1.14 to 1.3) �0.01 (�0.06 to 0.05) 0.819 1.21 (1.1 to 1.32) 1.2 (1.09 to 1.31) �0.01 (�0.11 to 0.08) 0.801
Total cholesterol/HDL 4.2 (4.1 to 4.3) 4.2 (4.09 to 4.3) 0 (�0.07 to 0.06) 0.934 4.19 (3.99 to 4.38) 4.17 (3.96 to 4.37) �0.02 (�0.12 to 0.08) 0.686
Non-HDL cholesterol, mmol/l 3.81 (3.73 to 3.9) 3.78 (3.69 to 3.87) �0.04 (�0.1 to 0.02) 0.230 3.73 (3.6 to 3.87) 3.7 (3.55 to 3.84) �0.04 (�0.13 to 0.05) 0.414
Systolic blood pressure, mm Hg 114.7 (113.6 to 115.8) 113.9 (112.7 to 115) �0.9 (�1.8 to 0) 0.063 114.6 (112.7 to 116.5) 114.8 (113 to 116.5) 0.2 (�1.3 to 1.7) 0.789
Diastolic blood pressure, mm Hg 73.3 (72.5 to 74) 72.2 (71.4 to 73.1) �1 (�1.7 to �0.4) 0.002 72.9 (71.6 to 74.2) 72.1 (70.8 to 73.4) �0.8 (�1.9 to 0.3) 0.155
Mean arterial pressure, mm Hg 87.1 (86.2 to 87.9) 86.2 (85.2 to 87.2) �0.9 (�1.7 to 0) 0.042 86.8 (85.4 to 88.2) 86.3 (84.9 to 87.7) �0.5 (�1.7 to 0.7) 0.383
10-yr CHD risk % 3.55 (3.21 to 3.89) 3.5 (3.15 to 3.85) �0.05 (�0.19 to 0.09) 0.480 3.82 (3.1 to 4.54) 3.66 (3 to 4.33) �0.16 (�0.38 to 0.07) 0.175
Advice Only (n ¼ 145) Food and Advice (n ¼ 140)
Baseline 6 Months 6 Months D Pr > |t| Baseline 6 Months 6 Months D Pr > |t|
Weight, kg 86.2 (83.3 to 89.1) 85 (82 to 88) �1.2 (�2.4 to 0) 0.044 89.6 (86.6 to 92.6) 88.7 (85.6 to 91.9) �0.8 (�1.7 to 0) 0.049
BMI, kg/m2 31.7 (30.8 to 32.7) 31.3 (30.3 to 32.2) �0.4 (�0.7 to �0.1) 0.004 32.7 (31.7 to 33.7) 32.4 (31.3 to 33.4) �0.3 (�0.7 to 0) 0.065
Waist, cm 100.6 (98.3 to 102.8) 98.7 (96.2 to 101.3) �1.8 (�3.2 to �0.5) 0.012 101.9 (99.5 to 104.3) 100.8 (98.3 to 103.4) �1.1 (�2.2 to 0) 0.047
Glucose, mmol/l 4.87 (4.77 to 4.97) 4.85 (4.74 to 4.96) �0.02 (�0.11 to 0.07) 0.632 4.81 (4.72 to 4.9) 4.8 (4.71 to 4.89) �0.01 (�0.08 to 0.07) 0.873
Total cholesterol, mmol/l 5.17 (5.02 to 5.32) 5.17 (4.98 to 5.37) 0 (�0.16 to 0.16) 1.000 4.89 (4.74 to 5.04) 4.82 (4.66 to 4.99) �0.07 (�0.17 to 0.04) 0.200
LDL-C, mmol/l 3.3 (3.17 to 3.43) 3.27 (3.13 to 3.4) �0.04 (�0.13 to 0.06) 0.458 3.07 (2.94 to 3.2) 3 (2.85 to 3.14) �0.08 (�0.17 to 0.02) 0.113
HDL-C, mmol/l 1.32 (1.26 to 1.39) 1.3 (1.24 to 1.36) �0.02 (�0.06 to 0.01) 0.243 1.3 (1.24 to 1.36) 1.28 (1.22 to 1.33) �0.02 (�0.06 to 0.02) 0.264
Triglycerides, mmol/l 1.2 (1.1 to 1.31) 1.25 (1.13 to 1.37) 0.05 (�0.05 to 0.14) 0.351 1.14 (1.03 to 1.26) 1.2 (1.09 to 1.32) 0.06 (�0.03 to 0.14) 0.184
Total cholesterol/HDL 4.16 (3.97 to 4.35) 4.19 (3.98 to 4.4) 0.03 (�0.12 to 0.17) 0.726 3.99 (3.8 to 4.18) 3.96 (3.76 to 4.16) �0.03 (�0.17 to 0.12) 0.696
Non-HDL cholesterol, mmol/l 3.85 (3.71 to 3.99) 3.85 (3.69 to 4) 0 (�0.1 to 0.1) 0.968 3.59 (3.45 to 3.74) 3.54 (3.38 to 3.7) �0.06 (�0.16 to 0.05) 0.299
Systolic blood pressure, mm Hg 114.8 (112.8 to 116.7) 113.4 (111.2 to 115.6) �1.3 (�3.2 to 0.5) 0.147 114 (111.8 to 116.1) 114 (111.8 to 116.2) 0.1 (�1.7 to 1.9) 0.940
Diastolic blood pressure, mm Hg 72.2 (70.8 to 73.6) 71.3 (69.6 to 73) �0.9 (�2.2 to 0.3) 0.145 72.7 (71.2 to 74.2) 72.5 (70.9 to 74.1) �0.1 (�1.3 to 1) 0.841
Mean arterial pressure, mm Hg 86.4 (84.9 to 87.9) 85.3 (83.5 to 87.1) �1.1 (�2.6 to 0.4) 0.136 86.5 (84.9 to 88.1) 86.4 (84.7 to 88.1) 0 (�1.3 to 1.3) 0.995
10-yr CHD risk % 4.03 (3.35 to 4.71) 3.85 (3.15 to 4.55) �0.19 (�0.58 to 0.21) 0.339 2.94 (2.29 to 3.59) 2.83 (2.26 to 3.41) �0.11 (�0.36 to 0.15) 0.420
*To convert the values to milligrams per deciliter, divide cholesterol by 0.0259, triglycerides by 0.0113 and glucose by 0.0555. †p Values in boldface indicate significance.
Abbreviations as in Table 1.
JACC
VOL.69,NO.9,2017
Jenkinset
al.MARCH
7,2017:11
03–12
Effectof
Dietary
Recom
mendations
onRisk
Factors
1109
CENTRAL ILLUSTRATION Current Dietary Recommendations on Weight Loss and Cardiovascular Risk Factor
Jenkins, D.J.A. et al. J Am Coll Cardiol. 2017;69(9):1103–12.
Six-month changes in study outcomes. Mean change and 95% confidence limit. *Significant change from baseline p < 0.05.
Jenkins et al. J A C C V O L . 6 9 , N O . 9 , 2 0 1 7
Effect of Dietary Recommendations on Risk Factors M A R C H 7 , 2 0 1 7 : 1 1 0 3 – 1 2
1110
nut intake by our group. Research must thereforecontinue to be funded to clearly establish the benefitof dietary strategies on health outcomes that, afterpublication, can be taken up by the media and hence
the general public. The successful PREDIMED(PREvencion con DIeta MEDiterranea) study providednuts and olive oil freely to participants (23). Shouldhealthy foods also be subsidized for the general
PERSPECTIVES
COMPETENCY IN SYSTEMS-BASED PRACTICE: Although
fruit, vegetables, and whole grain cereals are part of dietary
advice given internationally, foods containing soy protein, nuts,
viscous fiber sources (oats, barley, psyllium), and plant sterols
(e.g., as in enriched margarine) have been approved by the U.S.
Food and Drug Administration for heart health claims since they
specifically lower serum cholesterol.
TRANSLATIONAL OUTLOOK: Dietary change is difficult in
relatively healthy populations and may take community-based
strategies to implement effectively, but even limited patient
instructions, including pamphlets and guidelines from national
public health agencies may lead to reductions in body weight and
coronary heart disease risk scores.
J A C C V O L . 6 9 , N O . 9 , 2 0 1 7 Jenkins et al.M A R C H 7 , 2 0 1 7 : 1 1 0 3 – 1 2 Effect of Dietary Recommendations on Risk Factors
1111
population? In the present study, provision of foodsvery significantly increased retention. Although thisapproach only resulted in modest increases in con-sumption of recommended foods or reductions instudy outcomes, retention itself may be an indicationof continued adherence.
Finally, should the ecological sustainability ad-vantages of increased plant food consumption beemphasized? Sustainability considerations areincreasingly alluded to in current dietary guide-lines and science advisory reports internationally(3,26), and may increase the appeal to younger andhealthier people, similar to our participants.
ADHERENCE IN THE PRESENT STUDY. Using theabove ideas, adherence in our study might havebeen improved if our participants had had compel-ling reasons to change their dietary habits,including raised LDL-C or elevated blood pressure(baseline LDL-C: 3.23 mmol/l, and blood pressure:115/73 mm Hg in the present study). Furthermore,provision of complete diets or use of workplacecafeteria to provide meals incorporating DASH andportfolio dietary principles, plus a comprehensiveeducational program on the value of food in themaintenance of good health, might also have beenhelpful.
STUDY LIMITATIONS. A significant limitation of thestudy was the differential dropout rate between thetreatment arms, with a 6-month retention of 91% inthe 2 food delivery arms versus 67% when no foodwas provided. Similar effects of food provision havebeen seen in longer diet trials such as the PREDIMEDtrial where provision of nuts or olive oil was associ-ated with enhanced retention (23), and high retentionrates commonly observed in drug trials may relate inpart to the free provision of the intervention. It isinteresting that in our study there also appeared to bea “legacy effect” in that when no food was provided,prior provision of food still resulted at 18 months inan 81% retention versus 57% where no food had beenprovided.
A second limitation was the provision of Canada’sFood Guide, which may have had beneficial effectsin increasing intake of the desired foods in thecontrol group; however, the changes observed weresmall. Furthermore, we did not recruit sufficientfamily members to allow the assessment of familyinfluences on dietary change. Third, as the partici-pants were generally healthy at baseline, with 18%of participants who belonged to 2- or 3-memberfamily units already at a healthy body weight (4% of
total group), changes in risk factors may have beenmore difficult to detect and so limited our ability tosee an effect.
STRENGTHS. One strength of our study is that it isthe first to encourage specific food consumption byproviding both dietary advice and a range of foodsthat included less familiar functional foods. It isalso the first to do so in the context of DASH andportfolio dietary approaches, and determine theeffects on body weight and risk factors for CHD. Inaddition it demonstrated that removal of certainpotential barriers to consumption including avail-ability and familiarity (e.g., soy products) wasnot enough to increase substantially the use ofdesired foods in overweight but otherwise healthypeople.
CONCLUSIONS
Increasing the intake of not only fruit, vegetables,and whole grain cereals but also functional foodsproved difficult (27), even when these foods wereprovided weekly and despite the known effectivenessof such foods in managing CHD risk factors. Healthyshifts in diet among generally well populations islikely to require a range of sustained approaches andmultiple forms of communication in a processmeasured in decades rather than months (10).
ADDRESS FOR CORRESPONDENCE: Dr. David J.A.Jenkins, Department of Nutritional Sciences, Univer-sity of Toronto, 150 College Street, Fitzgerald Build-ing, Toronto, Ontario M5S 3E2, Canada. E-mail:[email protected].
Jenkins et al. J A C C V O L . 6 9 , N O . 9 , 2 0 1 7
Effect of Dietary Recommendations on Risk Factors M A R C H 7 , 2 0 1 7 : 1 1 0 3 – 1 2
1112
RE F E RENCE S
1. Stampfer MJ, Hu FB, Manson JE, Rimm EB,Willett WC. Primary prevention of coronary heartdisease in women through diet and lifestyle.N Engl J Med 2000;343:16–22.
2. Hu FB, Manson JE, Stampfer MJ, et al. Diet,lifestyle, and the risk of type 2 diabetes mellitus inwomen. N Engl J Med 2001;345:790–7.
3. USDA Dietary Guidelines for Americans 2015.Department of Health and Human Services. UnitedStates Department of Agriculture. Available at:http://health.gov/dietaryguidelines/2015–scientific-report/. Accessed January 4, 2016.
4. Anand SS, Hawkes C, de Souza RJ, et al.Food consumption and its impact on cardio-vascular disease: importance of solutionsfocused on the globalized food system: a reportfrom the workshop convened by the WorldHeart Federation. J Am Coll Cardiol 2015;66:1590–614.
5. Eckel RH, Jakicic JM, Ard JD, et al. 2013 AHA/ACC guideline on lifestyle management to reducecardiovascular risk: a report of the American Col-lege of Cardiology/American Heart AssociationTask Force on Practice Guidelines. J Am Coll Car-diol 2014;63 25_Pt B.
6. Anderson TJ, Gregoire J, Hegele RA, et al.2012 update of the Canadian CardiovascularSociety guidelines for the diagnosis and treat-ment of dyslipidemia for the prevention of car-diovascular disease in the adult. Can J Cardiol2013;29:151–67.
7. Sacks FM, Carey VJ, Anderson CA, et al. Effectsof high vs low glycemic index of dietary carbo-hydrate on cardiovascular disease risk factors andinsulin sensitivity: the OmniCarb randomized clin-ical trial. JAMA 2014;312:2531–41.
8. Jenkins DJ, Jones PJ, Lamarche B, et al. Ef-fect of a dietary portfolio of cholesterol-lowering foods given at 2 levels of intensity ofdietary advice on serum lipids in hyperlipidemia:a randomized controlled trial. JAMA 2011;306:831–9.
9. Saneei P, Salehi-Abargouei A, Esmaillzadeh A,Azadbakht L. Influence of Dietary Approaches toStop Hypertension (DASH) diet on blood pressure:a systematic review and meta-analysis on
randomized controlled trials. Nutr Metab Car-diovasc Dis 2014;24:1253–61.
10. Wang DD, Leung CW, Li Y, et al. Trends in di-etary quality among adults in the United States,1999 through 2010. JAMA Intern Med 2014;174:1587–95.
11. Willett WC, Sampson L, Stampfer MJ, et al.Reproducibility and validity of a semiquantitativefood frequency questionnaire. Am J Epidemiol1985;122:51–65.
12. Anderson KM, Wilson PW, Odell PM,Kannel WB. An updated coronary risk profile. Astatement for health professionals. Circulation1991;83:356–62.
13. US Food and Drug Administration CFR Title 21.Available at: http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr¼312.32.Accessed November 30, 2015.
14. Stroes ES, Thompson PD, Corsini A, et al. Statin-associated muscle symptoms: impact on statintherapy-European Atherosclerosis Society consensuspanel statement on assessment, aetiology andmanagement. Eur Heart J 2015;36:1012–22.
15. Heber D. Vegetables, fruits and phytoes-trogens in the prevention of diseases. J PostgradMed 2004;50:145–9.
16. Cox DN, Anderson AS, Reynolds J, McKellar S,Lean ME, Mela DJ. Take Five, a nutrition educationintervention to increase fruit and vegetable in-takes: impact on consumer choice and nutrientintakes. Brit J Nutr 1998;80:123–31.
17. Kaiser KA, Brown AW, Bohan Brown MM,Shikany JM, Mattes RD, Allison DB. Increased fruitand vegetable intake has no discernible effect onweight loss: a systematic review and meta-anal-ysis. Am J Clin Nutri 2014;100:567–76.
18. Gomez-Pardo E, Fernandez-Alvira JM,Vilanova M, et al. A comprehensive lifestyle peergroup-based intervention on cardiovascular riskfactors: the Randomized Controlled Fifty-FiftyProgram. J Am Coll Cardiol 2016;67:476–85.
19. LinJS,O’ConnorE,WhitlockEP,Beil TL.Behavioralcounseling topromotephysical activity andahealthfuldiet to prevent cardiovascular disease in adults: a
systematic review for theU.S. PreventiveServicesTaskForce. Ann Intern Med 2010;153:736–50.
20. Greaves CJ, Sheppard KE, Abraham C, et al.Systematic review of reviews of interventioncomponents associated with increased effective-ness in dietary and physical activity interventions.BMC Public Health 2011;11:119.
21. Godin G, Amireault S, Belanger-Gravel A, et al.Prediction of daily fruit and vegetable consump-tion among overweight and obese individuals.Appetite 2010;54:480–4.
22. Shai I, Schwarzfuchs D, Henkin Y, et al.Weight loss with a low-carbohydrate, Mediter-ranean, or low-fat diet. N Engl J Med 2008;359:229–41.
23. Estruch R, Ros E, Salas-Salvado J, et al. Pri-mary prevention of cardiovascular disease with aMediterranean diet. N Engl J Med 2013;368:1279–90.
24. Kris-Etherton PM, Zhao G, Binkoski AE,Coval SM, Etherton TD. The effects of nuts oncoronary heart disease risk. Nutr Rev 2001;59:103–11.
25. Sabate J, Oda K, Ros E. Nut consumption andblood lipid levels: a pooled analysis of 25 inter-vention trials. Arch Intern Med 2010;170:821–7.
26. Scientific Advisory Committee on Nutrition.Annual Report 2014. Public Health England.Available at: https://www.gov.uk/government/publications/sacn-annual-report-2014. AccessedJanuary 4, 2016.
27. Krebs-Smith SM, Guenther PM, Subar AF,Kirkpatrick SI, Dodd KW. Americans do not meetfederal dietary recommendations. J Nutr 2010;140:1832–8.
KEY WORDS body weight, cardiovasculardisease, diet, dietary recommendations,risk factors
APPENDIX For an expanded Methods sectionand additional tables and figures, please see theonline version of this article.