Medical school-based teaching kitchen improvesHbA1c, blood pressure, and cholesterol for patientswith type 2 diabetes: Results from a novelrandomized controlled trial

Dominique J. Monlezun a,b,*, Eric Kasprowicz a, Katherine W. Tosh a,Jenni Nix a, Pedro Urday a, Daphne Tice a, Leah Sarris a, Timothy S. Harlan a

aThe Goldring Center for Culinary Medicine, Tulane University, 1430 Tulane Avenue No. 8541, New Orleans,

LA 70112, USAbDepartment of Global Health Systems & Development, Tulane University School of Public Health & Tropical

Medicine, 1440 Canal Street, Suite 1900, Box TB-46, New Orleans, LA 70112, USA

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 1 0 9 ( 2 0 1 5 ) 4 2 0 – 4 2 6

a r t i c l e i n f o

Article history:

Received 3 December 2014

Received in revised form

18 March 2015

Accepted 2 May 2015

Available online 12 May 2015

Keywords:

Diabetes nutrition education

Culinary medicine

Health disparities

a b s t r a c t

Aims: A medical school-based teaching kitchen sought to establish proof-of-principle for its

hands-on Mediterranean diet (MD)-based cooking and nutrition curriculum for patients

with type 2 diabetes (T2D).

Methods: This pilot randomized controlled trial (RCT) allocated 27 patients with T2D be-

tween the control and GCCM arms. Mixed effects linear regression with repeated measures

was used to investigate differences from baseline to 6 months. The primary and secondary

endpoints were HbA1c!0.3% (!27 mmol/mol) and diastolic blood pressure (DBP)!10 mmHg

and a 25% improved responses in dietary habits and attitudes and competencies in healthy

nutrition.

Results: Compared to the control group, the GCCM group had superior HbA1c reduction

(!0.4% vs. !0.3%, p = 0.575) that was not statistically significant. There were significantly

greater reductions in the GCCM vs. control group for DBP (!4 vs. 7 mmHg, p = 0.037) and total

cholesterol (!14 vs. 17 mg/dL, p = 0.044). There was a greater proportion increase though not

significant of GCCM subjects compared to controls who mostly believed they could eat

correct portions (18% vs. !11%, p = 0.124), and who used nutrition panels to make food

choices (34% vs. 0%, p = 0.745).

Conclusion: This is the first known RCT demonstrating improved biometrics using a novel

MD-based hands on cooking and nutrition curriculum for patients with T2D. These results

suggest subsequent clinical trials are warranted on the grounds of documented feasibility

and clinical efficacy.

# 2015 Elsevier Ireland Ltd. All rights reserved.

* Corresponding author at: The Goldring Center for Culinary Medicine at Tulane University, 1430 Tulane Avenue, New Orleans, LA 8541,USA. Tel.: +1 337 329 0724; fax: +1 504 988 2687.

E-mail address: dmonlezu@tulane.edu (D.J. Monlezun).

Contents available at ScienceDirect

Diabetes Researchand Clinical Practice

journal homepage: www.elsevier .com/locate/diabres

http://dx.doi.org/10.1016/j.diabres.2015.05.0070168-8227/# 2015 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

Despite the critical link between obesity and the global diabetes

epidemic amid widening health disparities, medical profes-sionals are inadequately prepared to respond with financiallyand clinically effective nutrition lifestyle counseling for patients[1–3]. These hurdles have significant human and financial costsas diabetes is the seventh leading cause of mortality in theUnited States [4], resulting in $245 billion annually or 1 of every 5health care dollars [5]. This epidemic is worsenedby the negativeimpact of reduced access to healthy foods and underlying healthdisparities in the lower income minority food deserts, as notedby the American Diabetes Association’s 2013 Scientific State-ment [6]. Delays in eliminating such racial and ethnic health

inequities results in an additional $308 billion annual costs [7].Improved diet is a promising approach that has been shown toresult in significant reductions in disease incidence and burden,chiefly with the Mediterranean diet (MD) [8–11]. Currently, firstline therapy for 90% of diabetes cases [12,13] features dietmodification aimed at reducing patients’ serum glucose tooptimal levels [14]. The standard for nutrition education forpatients with documented patient diet improvements is regis-tered dietitian (RD)-led medical nutrition therapy (MNT); [15,16]however, MNT programs have several key challenges includingattrition rates up to 79%, for the sub-set of patients who follow

their primary care physicians’ referral for RD counseling [17].An alternative approach is pioneered by the Goldring

Center for Culinary Medicine (GCCM) at Tulane UniversitySchool of Medicine, to our knowledge the world’s first medicalschool-based research-oriented teaching kitchen. GCCM con-currently educates patients and physicians and trainees innutrition for immediate and long-term social capital develop-ment by hands-on cooking and nutrition education [18–21].Such operationalization of the MD as a scalable educationmodel for medical schools nationally is needed as this diet hasbeen extensively proven for health improvements in multiplechronic diseases [22,23]. Yet little evidence exists for inter-

ventions that equip people with the dietary habits, attitudes,and competencies needed to have sustainable healthimprovements even after the study period.

This pilot study, Cooking for Health Optimization withPatients (CHOP)-Diabetes Randomized Controlled Trial inves-tigated feasibility and clinical activity of a novel, scalable MD-education model for patients with type 2 diabetes (T2D)including those from food desert areas affected by healthinequities. This phase 1 trial sought to provide proof-of-principle for later phases of a multi-center RCT and compara-tive effectiveness trial. This current study is thus necessary for

eventual testing of the null hypothesis that GCCM hands-oncooking and nutrition curriculum led by chefs, physicians, andmedical students is equivalent to RD-led MNT for T2Dnutrition management as a model replicable at medicalschools nationally.

2. Subjects

Inclusion criterion was patients with diagnosed T2D. Exclu-sion criterion was pre-existing enrollment in another study

involving interventions for diabetes. Institutional ReviewBoard approval was granted through Tulane University forthis study, and informed patient consent was collected.

3. Materials and methods

3.1. Design

This study uses a RCT design to compare RD-led MNT withchef, physician, and medical student-led hands-on cookingand nutrition classes. The six-module cooking and nutritioncurriculum translates the MD for culture-specific kitchensacross different socioeconomic levels. The control groupreceived the standard of nutrition education, RD-led MNT,

consisting of a one-time RD counseling visit with a referralopportunity to an American Diabetes Association-certifieddiabetes education class. The treatment group participated inthe GCCM modules over 1.5 months as part of an evidence-based GCCM curriculum [18]. Each two-hour cooking classconsisted of 30 min of didactic lessons and 90 min of cookingtime. This pilot study and analysis was designed after a recentRCT provided evidence of MD improving HbA1c for patientswith T2D [24].

3.2. Data

Biometric data were collected through chart reviews frombaseline to 6 months after participation in the intervention orcontrol groups according to regularly scheduled primary careclinic visits. Reviewers had to have knowledge of subjects’study groups in order to collect biometric data at the correct 6month time point following their completion of the lastcooking class or their MNT session. Biometric data pointsincluded: HbA1c, systolic and diastolic blood pressures (SBPand DBP), total cholesterol, triglycerides, low density lipopro-teins (LDL), high density lipoproteins (HDL), heart rate (HR),body mass index (BMI), and hypoglycemic agents and insulin.

Psychometric data were collected through a survey developedfrom validated tools [1,2,20] at baseline and again at 1.5months following the MNT session for control patients or thelast GCCM class for intervention patients. These data pointsincluded: dietary habits, attitudes, and competencies (DACs)for healthy shopping, meal-preparation, eating, and storage.Food desert residence was defined by the United States Foodand Drug Administration (USDA) as a subject claiming apermanent home address in a low-income census track wherea significant number or share of residents are more than 1 mile(urban) or 10 miles (rural) from the nearest supermarket [25].

Survey responses used a multi-point Lickert scale fordietary habits with different foods (9 point scale ranging fromnever per month to over 3 times daily), attitudes (5 pointsranging from very unconfident to very confident), andcompetencies as frequencies of healthy eating actions (5points ranging from never to always). Survey responses weredichotomized to aid in interpretation into: most times fordietary habits (over 4 times weekly versus less), mostlyconfident for attitudes (confident and very confident versusneither), and competencies (most of the times or alwaysversus neither).

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 1 0 9 ( 2 0 1 5 ) 4 2 0 – 4 2 6 421

3.3. Endpoints

A permuted block design randomization scheme was used.The primary endpoint over six months of follow-up was

HbA1c reduction of !0.3% (!27 mmol/mol) from baseline tofollow-up within each group, with secondary endpointsincluding DBP reduction of 10 points and a 25% improvedresponses in the DAC sections from baseline to follow-up,based on the existing literature for such nutrition educationinterventions for this patient population [15,16,22,23,27].

3.4. Statistical analysis

This intention-to-treat statistical analysis used the meanbiometric values with 95% confidence intervals, which were

calculated after verifying normal distribution with theShapiro–Walk, Shapiro–Francia, and skewness and kurtosistests for normality. Mixed effects linear regression withrepeated time measures were used to investigate differencesfrom baseline to 6 months for the GCCM compared to controlgroups [26]. An increase in a metric over the study period wasrepresented as a positive value, and a decrease was repre-sented as a negative value. This regression was used in place ofANOVA with repeated measures due this method producingcorrect standard errors for each effect automatically whilehandling unequal observations within subjects. This irregu-

larity was contributed to by the sample including lowersocioeconomic patients with less reliable transportation andcommunication capacities, with their subsequent irregularattendance at scheduled clinic visits or response to study staffattempts to contact them. Analyses were conducted usingSTATA 12.0 (StataCorp LP, College Station, TX, USA). A p-value<0.05 was considered statistically significant.

4. Results

4.1. Biometric results

In this study, 18 (67%) participated in the GCCM interventionwith an average age of 62 years, 75% being African American,67% female, and 46% residing in a USDA-defined food desert, allcomparable to the control group. In contrast to the controlgroup, the GCCM group had superior mean HbA1c reductionfrom baseline to 6 months, !0.4% (!28 mmol/mol) vs. !0.3%(!27 mmol/mol) p = 0.575, that was not statistically significant(Table 1). There were significantly greater reductions in theGCCM vs. control group for DBP (!4 vs. 7 mmHg, p = 0.037) andtotal cholesterol (!14 vs. 17 mg/dL, p = 0.044). The control group

had a transient negative HbA1c improvement one month afterthe intervention; whereas, the improvement was sustained at 6months for the GCCM group after the intervention lasted 1.5months. GCCM subjects either had greater reductions or aslower rise in all negative risk factor biometrics compared to thecontrol group, which instead worsened in these biometrics(SBP, DBP, total cholesterol, and LDL) with the exception oftriglycerides and HR (Fig. 1). Both the GCCM and controls groupshad similar pre-post BMI measurements.

The largest biometric difference for both groups was totalcholesterol as the GCCM group decreased by 14 mg/dL

(!0.4 mmol/L), compared to the control group increasing by17 mg/dL (0.4 mmol/L). This inverse relationship is similar forLDL, as the GCCM group decreased by 6 mg/dL (!0.2 mmol/L)while the control group increased by 13 mg/dL (0.3 mmol/L).

The control group did have a !20 mg/dL (!0.2 mmol/L) greaterdrop in triglycerides than the GCCM group and 15 mg/dL(0.4 mmol/L) greater increase in HDL. Lipid lowering medica-tions and insulin were similar for both groups at baseline,though there were two GCCM patients who ceased use ofhypoglycemic agents, compared to no control patients comingoff their medications during the study period. Metformin1000 mg and Lantus (15–50 units) respectively were the mostcommon hypoglycemic agent and insulin used for both groups.

4.2. Psychometric results

GCCM patients’ psychometrics improved but without statisti-cal significance compared to the control group in theirattitudes and competencies in healthy food shopping andeating (Fig. 2). The two largest differences in DAC changesbetween the groups were in competencies. There was a greaterproportion increase of GCCM subjects compared to controlswho mostly believed they could eat correct portions (18% vs.!11%, p = 0.124), and who used nutrition panels to make foodchoices (34% vs. 0%, p = 0.745) (Table 1). The GCCM grouplagged behind the control group in the proportion increasing

their vegetable consumption (16% vs. 23%, p = 0.736) and fruits(0% vs. 22%, p = 0.255). Though both the GCCM and controlsubjects met the primary endpoint for HbA1c and missed thesecondary endpoint for DBP, only the GCCM group met the 25%endpoint for DAC improvement, specifically in competenciesfor using nutrition panels in food choices.

5. Discussion

5.1. Contribution

This is the first randomized controlled trial to our knowledgethat demonstrates improved blood pressure and lipid panelsusing a novel Mediterranean diet (MD)-based hands oncooking and nutrition curriculum for patients with T2D,including subjects from lower income food deserts. We alsodemonstrate non-significant HbA1c and healthy eating com-petency improvements for the cooking vs. control subjects.This intervention may thus provide evidence for a novelclinical education model able to operationalize the validatedMD into a scalable competency and attitude modificationmodel, particularly for patients burdened by healthy food

disparities. These results suggest that subsequent clinical trialphases are warranted on the grounds of documentedfeasibility and clinical efficacy to confirm these findings.Global health system implications for this model include apotentially rapid diffusion of this education innovationthrough medical schools and hospitals to improve patients’health in the short-term as 13 medical schools in the past twoyears have partnered with GCCM to provide this curriculumfor their communities. The long-term implication is buildingworkforce capacity of medical students and physicians bettertrained in nutrition counseling as they teach these classes for

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 1 0 9 ( 2 0 1 5 ) 4 2 0 – 4 2 6422

Table 1 – GCCM vs. control changes in mixed effects linear regression with repeated measures.

Control mean, 95% CI n = 9, 33% Baselineto 6

months

GCCM mean, 95% CI n = 18, 67% Baselineto 6

months

GCCM vs.controlchanges

P-value*

Baseline 6 Months post Baseline 6 Months post

BiometricHbA1c* % (mmol/mol) 8.1 (65), 6.8–9.4 (51–79) 7.8 (62), 6.6–9.0 (49–75) !0.3 (!27) 7.6 (60), 7.1–8.1 (54–65) 7.2 (55), 6.8–7.6 (51–60) !0.4 (!28) !0.1 (!25) 0.575BP mmHg

SBP 127, 117–136 140, 131–150 13 129, 124–135 132, 125–139 3 !10 0.110DBP 74, 69–80 81, 72–91 7 81, 77–85 77, 69–84 !4 !11 0.037

Lipids mg/dL (mmol/L)Total chol 128 (1.5), 119–137 (1.3–1.5) 145 (1.6), 113–177 (1.3–2.0) 17 (0.4) 185 (2.1), 150–220 (1.7–2.5) 171 (1.9), 134–208 (1.5–2.4) !14 (!0.4) !31 (!0.8) 0.044Trig 188 (2.1), 47–330 (0.5–3.7) 130 (1.5), 37–223 (0.4–2.5) !58 (!0.7) 170 (1.9), 105–234 (1.2–2.6) 132 (1.5), 79–184 (0.9–2.1) !38 (!0.4) 20 (0.2) 0.789LDL 66 (0.7), 60–72 (0.7–0.8) 79 (0.9), 60–97 (0.7–1.1) 13 (0.3) 110 (1.2), 83–138 (0.9–1.6) 104 (1.2), 78–130 (0.9–1.5) !6 (!0.2) !19 (!0.5) 0.149HDL 36 (0.4), 21–51 (0.2–0.6) 56 (0.6), 34–79 (0.4–0.9) 20 (0.5) 49 (0.6), 43–55 (0.5–0.6) 54 (0.6), 44–64 (0.5–0.7) 5 (0.1) !15 (!0.4) 0.043

HR bpm 73, 66–80 74, 68–80 1 76, 69–84 82, 76–89 6 5 0.831BMI kg/m2 36, 30–42 34, 29–40 !2 37, 32–41 37, 31–42 0 2 0.184

PsychometricMostly believe can eatcorrect portions

44%, 8–81 33%, !1–68 !11% 63%, 37–88 81%, 61–102 18% 29% 0.260

Eat vegetables mostnights per week

33%, !1–68 56%, 19–92 23% 31%, 7–56 47%, 19–74 16% !8% 0.632

Eat fruits most nightsper week

11%, !12–34 33%, !1–68 22% 47%, 21–73 47%, 19–74 0% !24% 0.386

Mostly use nutritionpanel for food choices

22%, !8–53 22%, !8–53 0% 35%, 11–60 69%, 44–94 34% 34% 0.242

oSBP, systolic blood pressure; DBP, diastolic blood pressure; Tot chol, total cholesterol; Trig, triglycerides; HR, heart rate; bpm, beats per minute; BMI, body mass index.* P-value for GCCM vs. control change from baseline to 6 months post intervention by mixed effects linear regression with repeated measures.

dia

be

te

sr

es

ea

rc

ha

nd

cl

inic

al

pr

ac

tic

e1

09

(20

15

)4

20

–4

26

42

3

Fig. 1 – Biometric changes for GCCM vs. control subjects from baseline to 6 months.

Fig. 2 – Psychometric changes in GCCM vs. control subject proportions from baseline to 6 months.

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 1 0 9 ( 2 0 1 5 ) 4 2 0 – 4 2 6424

health disparity reduction and improved population healthmanagement.

The results should be considered against the studylimitations. First, this is a smaller single-site pilot study

investigating feasibility and early signs of clinical efficacy.Second, though rigorous 1:1 randomization was utilized withfollow-up communication connecting subjects to their re-spective treatment or control exposures, there were a lowernumber of control subjects compared to treatment whocompleted their study exposure. The potential greater appealof cooking classes or recruiter bias may have contributed tothis, though control subjects were informed they couldparticipate in the cooking classes six months after theirMNT session. Medication changes by treating physicians mayhave also contributed to biometric changes, though this

treatment bias was addressed by randomization, blinding oftreating physicians to study group assignment, and medica-tion comparison showing no notable differences betweentreatment and control groups.

The only systematic review and meta-analysis of differentdietary interventions on T2D glycemic levels, lipids, andweight representing 20 RCTs (n = 3073 subjects) failed to showany other RCT achieving improved health metrics with hands-on cooking and nutrition education [27]. Rather than simplymirroring previous trials included in this large meta-analysisassessing the impact of MD foods on patients’ biometric

health, this trial actually taught patients with T2D what to dowith the foods they enjoy. Further, this study demonstratedsuperior HbA1c reduction with its MD-based interventiongroup compared to the MD groups in the meta-analysis,!0.4%(!28 mmol/mol) vs.!0.1% (!25 mmol/mol) that was sustainedeven after the intervention, including blood pressure andcholesterol improvements. Larger randomized controlledtrials are needed to determine if the HbA1c impact isstatistically significant. Yet this current study still suggeststhat subjects can improve their competencies in healthyeating with the MD even without ongoing education orprovision of healthy MD foods. Superiority testing compared

to the standard of care is proper to later clinical trial phases,but it is noteworthy that the GCCM treatment effect superiori-ty may be an underestimate in diet-based supplementalmanagement for diabetes, considering the majority of primarycare physicians fail to provide any substantive nutritioncounseling to their patients, let alone an RD referral [28,29]. Asthe GCCM curriculum targets not only patients but also theircurrent and future physicians, this intervention providesevidence for a capacity-building, clinically effective interven-tion that is implementable at medical schools and healthcenters nationally. Future research is needed to test the

optimal dose and timing of GCCM education for diabetes andother nutrition-related chronic diseases, in addition to theoptimal diffusion of this GCCM innovation through patients’social networks to determine if such networks can helpsustain the health improvements.

5.2. Conclusion

The results demonstrate the feasibility and signals oftreatment efficacy needed to launch Phase II. This subsequentstage will include similar inclusion and exclusion criteria as

Phase I, but will include: a third arm for no dietary change, costanalysis, an extended study period of 12 months, social networkanalysis to determine if health improvements from GCCMsubjects extend to their friends and families in superior

magnitude compared to the control group, and tertiaryendpoint of 30% subjects from food desert areas. Thirteenmedical schools and universities are currently participating inthe multi-site study for assessing the GCCM curriculum amonga 10,000+ medical student sample. Expansion of this diabetesstudy at those sites thus should allow a larger Phase II multi-sitetrial, in addition to one novel enough to bridge the gap betweenmeaningful lifestyle modifications in chronic management ofT2D and improved patient outcomes in a sustainable andscalable manner.Conflict of interest statement

The authors declare that they have no conflict of interest.

Acknowledgement

The authors would like to thank Vivian Fonseca, M.D., for hisexpert review.

r e f e r e n c e s

[1] Spencer EH, Frank E, Elon LK, et al. Predictors of nutritioncounseling behaviors and attitudes in US medical students.Am J Clin Nutr 2006;84:655–62.

[2] Walsh CO, Ziniel SI, Delichatsios HK, et al. Nutritionattitudes and knowledge in medical students aftercompletion of an integrated nutrition curriculum comparedto a dedicated nutrition curriculum: a quasi-GCCM study.BMC Med Educ 2011;11:58.

[3] Adams KM, Kohlmeier M, Zeisel SH. Nutrition education inU.S. medical schools: latest update of a national survey.Acad Med 2010;85(9):1537–42.

[4] Centers for Disease Control and Prevention. Nationaldiabetes fact sheet; 2011, hhttp://www.cdc.gov/diabetes/pubs/pdf/ndfs_2011.pdfi.

[5] American Diabetes Association. Economic costs of diabetesin the U.S. in 2012. Diabetes Care 2013;36(April (4)):1033–46.http://dx.doi.org/10.2337/dc12-2625.

[6] Hill JO, Galloway JM, Goley A, Marrero DG, Minners R,Montgomery B, et al. Scientific statement: socioecologicaldeterminants of prediabetes and type 2 diabetes. DiabetesCare 2013;36(8):2430–9.

[7] LaVeist TA, Gaskin D, Richard P. Estimating the economicburden of racial health inequalities in the United States. IntJ Health Serv 2011;41(2):231–8.

[8] Trichopoulou A, Costacou T, Bamia C, et al. Adherence to aMediterranean diet and survival in a Greek population. NEng J Med 2003;348(26):2599–608.

[9] Diabetes Prevention Program Research Group. Reduction inthe incidence of type 2 diabetes with lifestyle. N Engl J Med2002;346:393–403.

[10] Gardener H, Wright CB, Gu Y, et al. Mediterranean-stylediet and risk of ischemic stroke, myocardial infarction, andvascular death: the Northern Manhattan Study. Am J ClinNutr 2011;94(6):1458–64.

[11] Foroughi M, Akhavanzanjani M, Maghsoudi Z, et al. Strokeand nutrition: a review of studies. Int J Prev Med2013;4(Suppl. 2):S165–79.

[12] American Diabetes Association. Diagnosis andclassification of diabetes mellitus. Diabetes Care2012;35(Suppl. 1):S64–71.

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 1 0 9 ( 2 0 1 5 ) 4 2 0 – 4 2 6 425

[13] Tricco AD, Antony J, Soobia C, et al. Safety, effectiveness,and cost of dipeptidyl peptidase-4 inhibitors versusintermediate acting insulin for type 2 diabetes: protocol fora systematic review and network meta-analysis. Syst Rev2013;2:47.

[14] Testa M, Simonson D. Health economic benefits and qualityof life during improved glycemic control in patients withtype 2 diabetes mellitus: a randomized, controlled, double-blind trial. JAMA 1998;280(17):1490–6.

[15] Franz M, Powers M, Leontos C, et al. The evidence formedical nutrition therapy for type 1 and type 2 diabetes inadults. J Am Diet Assoc 2010;110(12):1852–89.

[16] Pastors J, Warshaw H, Daly A, et al. The evidence for theeffectiveness of medical nutrition therapy in diabetesmanagement. Diabetes Care 2002;25(3):608–13.

[17] Norris S, Nichols P, Caspersen C, et al. Increasing diabetesself-management education in community settings: asystematic review. Am J Prev Med 2002;22(4S):39–66.

[18] Birkhead A, Foote S, Monlezun DJ, Loyd J, Joo E, Leong B,et al. Medical student-led community cooking classes: anovel preventative medicine model easy to swallow. Am JPrev Med 2014;46(3):e41–2.

[19] Leong B, Ren D, Monlezun DJ, et al. Teaching 3rd & 4th yearmedical students how to cook: an innovative approach tobalance lifestyle modification and medication therapy inchronic disease management. Med Sci Educ 2014;24(1):43.

[20] Monlezun DJ, Tsai P, Sarris L, Harlan T. Recipe for cancereducation: a novel integrated cooking and nutritioneducation curriculum for medical students and physiciansin dietary preventive and supplemental treatment forpancreatic cancer. J Med Pers 2014;12(3):125–8. http://dx.doi.org/10.1007/s12682-014-0190-5.

[21] Kasprowicz E, Monlezun DJ, Harlan TS. Reducing sodiumintake to prevent stroke. Stroke 2014;45(6):e108.

[22] Shai I, Schwarzfuchs D, Henkin Y, Shahar DR, Witkow S,Greenberg I, et al. Weight loss with a low-carbohydrate,Mediterranean, or low-fat diet. N Engl J Med2008;359(3):229–41.

[23] Estruch R, Ros E, Salas-Salvado J, Covas MI, Corella D,Aros F, et al. Primary prevention of cardiovascular diseasewith a Mediterranean diet. N Engl J Med2013;368(14):1279–90.

[24] Itsiopoulos C, Brazionis L, Kaimakamis M, Cameron M, BestJD, O’Dea K, et al. Can the Mediterranean diet lower HbA1cin type 2 diabetes?. Results from a randomized cross-overstudy. Nutr Metab Cardiovasc Dis 2011;21(9):740–7.

[25] USDA. Food access research atlas. 2014. Accessible at:hhttp://www.ers.usda.gov/data-products/food-access-research-atlas/go-to-the-atlas.aspx#.VCMgplZZEdti.

[26] Kabat GC, Heo M, Allison MA, Hou L, Nassir R, Zaslavsky O,et al. Association of anthropometric measures andhemostatic factors in postmenopausal women: alongitudinal study. Nutr Metab Cardiovasc Dis2014;24(10):1120–7.

[27] Ajala O, English P, Pinkney J. Systematic review and metaanalysis of different dietary approaches to themanagement of type 2 diabetes. Am J Clin Nutr2013;97(3):505–16.

[28] Kushner R. Barriers to providing nutrition counseling byphysicians: a survey of primary care practitioners. PrevMed 1995;24:546–52.

[29] Galuska D, Will J, Serdula M, Ford E. Are health careprofessionals advising obese patients to lose weight? JAMA1999;282:1576–8.

d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 1 0 9 ( 2 0 1 5 ) 4 2 0 – 4 2 6426