0

The full content of Annals is available to subscribers

Subscribe/Learn More  >
Original Research |

Prevention of Diabetes With Mediterranean Diets: A Subgroup Analysis of a Randomized Trial

Jordi Salas-Salvadó, MD, PhD*; Mònica Bulló, PhD; Ramón Estruch, MD, PhD; Emilio Ros, MD, PhD; Maria-Isabel Covas, DPharm; Núria Ibarrola-Jurado, RD, PhD; Dolores Corella, DPharm, PhD; Fernando Arós, MD, PhD; Enrique Gómez-Gracia, MD, PhD; Valentina Ruiz-Gutiérrez, PhD; Dora Romaguera, MD, PhD; José Lapetra, MD, PhD; Rosa Maria Lamuela-Raventós, DPharm, PhD; Lluís Serra-Majem, MD, PhD; Xavier Pintó, MD, PhD; Josep Basora, MD, PhD; Miguel Angel Muñoz, MD, PhD; José V. Sorlí, MD, PhD; and Miguel A. Martínez-González, MD, PhD*
[+] Article and Author Information

* Drs. Salas-Salvadó and Martínez-González contributed equally to this work.


From the Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, and PREDIMED Network, Instituto de Salud Carlos III, Madrid, Spain; Hospital Universitari Sant Joan, Institut d’Investigació Sanitaria Pere Virgili, and Universitat Rovira i Virgili, Reus, Spain; Institut d’Investigacions Biomèdiques August Pi i Sunyer, Hospital Clinic, School of Pharmacy, Xarxa de Referència en Tecnologia dels Aliments, and Instituto de Investigación en Nutrición y Seguridad Alimentaria, University of Barcelona, Cardiovascular and Nutrition Research Group, Institut de Recerca Hospital del Mar, Hospital Universitario de Bellvitge, and Hospitalet de Llobregat, Barcelona, Spain; University of Valencia and Valencia Institute of Health, Valencia, Spain; University Hospital of Alava, Vitoria, Spain; University of Malaga, Malaga, Spain; Instituto de la Grasa, Consejo Superior de Investigaciones Cientificas, and San Pablo Health Center, Seville, Spain; Research Unit, University Hospital Son Espases, Balearic Islands, Spain; School of Public Health, Imperial College London, United Kingdom; University of Las Palmas de Gran Canaria, Las Palmas, Spain; University of Navarra, Pamplona, Spain; and Catalan Institute of Health, Institut d’Investigació en Atenció Primària Jordi Gol, Tarragona-Reus and Barcelona, Spain.

Grant Support: By the official funding agency for biomedical research of the Spanish government, Instituto de Salud Carlos III, through grants provided to research networks specifically developed for the trial (RTIC G03/140 to Dr. Estruch; RTIC RD 06/0045 to Dr. Martínez-González) and through Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición and by grants from Centro Nacional de Investigaciones Cardiovasculares (CNIC 06/2007), Fondo de Investigación Sanitaria–Fondo Europeo de Desarrollo Regional (PI04-2239, PI 05/2584, CP06/00100, PI07/0240, PI07/1138, PI07/0954, PI 07/0473, PI10/01407, PI10/02658, PI11/01647, and P11/02505), Ministerio de Ciencia e Innovación (AGL-2009-13906-C02, AGL2010-22319-C03, and AGL2011-23430), Fundación Mapfre 2010, Consejería de Salud de la Junta de Andalucía (PI0105/2007), Public Health Division of the Department of Health of the Autonomous Government of Catalonia, Generalitat Valenciana (ACOMP06109, GVACOMP2010-181, GVACOMP2011-151, ACOMP2012-190, ACOMP2013-159, ACOMP2013-165, CS2010-AP-111), Agencia Canaria de Investigación, Innovación y Sociedad de la Información-EU FEDER (PI 2007/050 and CS2011-AP-042), and Regional Government of Navarra (P27/2011). The Fundación Patrimonio Comunal Olivarero and Hojiblanca (Malaga, Spain), California Walnut Commission (Sacramento, California), Borges (Reus, Spain), and Morella Nuts (Reus, Spain) donated the olive oil, walnuts, almonds, and hazelnuts, respectively, used in the study.

Potential Conflicts of Interest: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M13-1725.

Reproducible Research Statement: Study protocol: Available from Dr. Salas-Salvadó (e-mail, jordi.salas@urv.cat). Statistical code and data set: Not available.

Requests for Single Reprints: Jordi Salas-Salvadó, MD, PhD, Human Nutrition Unit, Faculty of Medicine and Healthy Sciences, Universitat Rovira i Virgili, C/ Sant Llorenç, 21, 43201 Reus, Spain; e-mail, jordi.salas@urv.cat.

Current Author Addresses: Drs. Salas-Salvadó, Bulló, and Ibarrola-Jurado: Human Nutrition Unit, Faculty of Medicine and Healthy Sciences, Universitat Rovira i Virgili, C/ Sant Llorenç, 21, 43201 Reus, Spain.

Dr. Estruch: Department of Internal Medicine, Institut d’Investigacions Biomèdiques August Pi i Sunyer, Hospital Clinic, University of Barcelona, Barcelona, C/ Villarroel, 170, 08036 Barcelona, Spain.

Dr. Ros: Lipid Clinic, Department of Endocrinology and Nutrition, Institut d’Investigacions Biomèdiques August Pi i Sunyer, Hospital Clinic, Barcelona, C/ Villarroel, 170, 08036 Barcelona, Spain.

Dr. Covas: Cardiovascular and Nutrition Research Group, Institut de Recerca Hospital del Mar, Barcelona, C/ Doctor Aiguader, 88, 08003 Barcelona, Spain.

Dr. Corella: Genetic and Molecular Epidemiology Unit, School of Medicine, Department of Preventive Medicine, University of Valencia, Avda Blasco Ibañez, 15, 46010 Valencia, Spain.

Dr. Arós: Department of Cardiology, University Hospital of Alava, C/ Jose Atxotegi, S/N 01009 Vitoria-Gasteiz, Spain.

Dr. Gómez-Gracia: Department of Preventive Medicine, Faculty of Medicine, University of Malaga, Bulevard Luis Pastor, 32, 29071 Malaga, Spain.

Dr. Ruiz-Gutiérrez: Profesora de Investigación del Consejo Superior de Investigaciones Cientíificas, Nutrition and Lipids Metabolism, Instituto de la Grasa, Avenida Padre García Tejero, 4, 41012 Sevilla, Spain.

Dr. Romaguera: Research Unit, University Hospital Son Espases, Balearic Islands, Spain and School of Public Health, Carretera de Valldemossa, 79, 07120 Palma de Mallorca, Illes Balears, Spain.

Dr. Lapetra: Department of Family Medicine, Primary Care Division of Seville, San Pablo Health Center, Sevilla, Jerusalem, S/N 41007 Sevilla, Spain.

Dr. Lamuela-Raventos: Department of Nutrition and Food Science, School of Pharmacy, University of Barcelona, Barcelona, Avinguda de Joan XXIII, 31, 08028 Barcelona, Spain.

Dr. Serra-Majem: Department of Clinical Sciences, University of Las Palmas de Gran Canaria, Edificio Ciencias de la Salud, Campus de San Cristóbal, 35016 Las Palmas de Gran Canaria, Spain.

Dr. Pintó: Lipids and Vascular Risk Unit, Internal Medicine, Hospital Universitario de Bellvitge, Avd. Feixa Llarga, S/N, 08907 L’Hospitalet de Llobregat, Barcelona, Spain.

Dr. Basora: Primary Care Division, Catalan Institute of Health, Institut d’Investigació en Atenció Primària Jordi Gol, Tarragona-Reus, Universitat Rovira i Virgili, C/ Sant Llorenç, 21, 43201 Reus, Spain.

Dr. Muñoz: Primary Care Division, Catalan Institute of Health, Institut d’Investigació en Atenció Primària Jordi Gol, Barcelona, C/ Sardenya 375, Entlo, 08025 Barcelona, Spain.

Dr. Sorlíi: Primary Care Division, Valencia Institute of Health, Valencia, Faculty of Medicine, Av Vicente Blasco Ibáñez, 26, 46010 Valencia, Spain.

Dr. Martíinez-González: Department of Preventive Medicine and Public Health, Faculty of Medicine, University of Navarra, C/ Irunlarrea 1, 31008, Pamplona, Spain.

Author Contributions: Conception and design: J. Salas-Salvadó, E. Ros, R. Estruch, M.I. Covas, D. Corella, V. Ruiz-Gutiérrez, J. Lapetra, R.M. Lamuela-Raventos, L. Serra-Majem, M.A. Martíinez-González.

Analysis and interpretation of the data: J. Salas-Salvadó, M. Bulló, R. Estruch, E. Ros, M.I. Covas, N. Ibarrola-Jurado, F. Arós, R.M. Lamuela-Raventos, L. Serra-Majem, M.A. Muñoz, M.A. Martíinez-González.

Drafting of the article: J. Salas-Salvadó, M.A. Martíinez-González.

Critical revision of the article for important intellectual content: J. Salas-Salvadó, M. Bulló, R. Estruch, E. Ros, N. Ibarrola-Jurado, D. Corella, E. Gómez-Gracia, V. Ruiz-Gutiérrez, D. Romaguera, J. Lapetra, R.M. Lamuela-Raventos, L. Serra-Majem, X. Pintó, J. Basora, M.A. Muñoz, M.A. Martíinez-González.

Final approval of the article: J. Salas-Salvadó, M. Bulló, E. Ros, M.I. Covas, N. Ibarrola-Jurado, R. Estruch, D. Corella, F. Arós, E. Gómez-Gracia, V. Ruiz-Gutiérrez, J. Lapetra, R.M. Lamuela-Raventos, L. Serra-Majem, X. Pintó, J. Basora, M.A. Muñoz, M.A. Martíinez-González.

Provision of study materials or patients: J. Salas-Salvadó, R. Estruch, D. Corella, E. Gómez-Gracia, J. Lapetra, L. Serra-Majem, X. Pintó, J. Basora, M.A. Muñoz, J.V. Sorlíi, M.A. Martíinez-González.

Statistical expertise: J. Salas-Salvadó, N. Ibarrola-Jurado, D. Corella, M.A. Martíinez-González.

Obtaining of funding: J. Salas-Salvadó, M. Bulló, R. Estruch, E. Ros, D. Corella, F. Arós, E. Gómez-Gracia, V. Ruiz-Gutiérrez, J. Lapetra, M.A. Martíinez-González.

Administrative, technical, or logistic support: J. Salas-Salvadó, N. Ibarrola-Jurado, R. Estruch, E. Gómez-Gracia, J. Lapetra, L. Serra-Majem, X. Pintó, J. Basora, M.A. Martíinez-González.

Collection and assembly of data: J. Salas-Salvadó, M. Bulló, R. Estruch, N. Ibarrola-Jurado, D. Corella, F. Arós, E. Gómez-Gracia, J. Lapetra, L. Serra-Majem, X. Pintó, J. Basora, M.A. Muñoz, J.V. Sorlíi, M.A. Martíinez-González.


Ann Intern Med. 2014;160(1):1-10. doi:10.7326/M13-1725
Text Size: A A A

Background: Interventions promoting weight loss can reduce the incidence of type 2 diabetes mellitus. Whether dietary changes without calorie restriction also protect from diabetes has not been evaluated.

Objective: To assess the efficacy of Mediterranean diets for the primary prevention of diabetes in the Prevención con Dieta Mediterránea trial, from October 2003 to December 2010 (median follow-up, 4.1 years).

Design: Subgroup analysis of a multicenter, randomized trial. (Current Controlled Trials: ISRCTN35739639)

Setting: Primary care centers in Spain.

Participants: Men and women without diabetes (3541 patients aged 55 to 80 years) at high cardiovascular risk.

Intervention: Participants were randomly assigned and stratified by site, sex, and age but not diabetes status to receive 1 of 3 diets: Mediterranean diet supplemented with extra-virgin olive oil (EVOO), Mediterranean diet supplemented with nuts, or a control diet (advice on a low-fat diet). No intervention to increase physical activity or lose weight was included.

Measurements: Incidence of new-onset type 2 diabetes mellitus (prespecified secondary outcome).

Results: During follow-up, 80, 92, and 101 new-onset cases of diabetes occurred in the Mediterranean diet supplemented with EVOO, Mediterranean diet supplemented with mixed nuts, and control diet groups, respectively, corresponding to rates of 16.0, 18.7, and 23.6 cases per 1000 person-years. Multivariate-adjusted hazard ratios were 0.60 (95% CI, 0.43 to 0.85) for the Mediterranean diet supplemented with EVOO and 0.82 (CI, 0.61 to 1.10) for the Mediterranean diet supplemented with nuts compared with the control diet.

Limitations: Randomization was not stratified by diabetes status. Withdrawals were greater in the control group.

Conclusion: A Mediterranean diet enriched with EVOO but without energy restrictions reduced diabetes risk among persons with high cardiovascular risk.

Primary Funding Source: Instituto de Salud Carlos III.

Figures

Grahic Jump Location
Figure 1.

Study flow diagram.

EVOO = extra-virgin olive oil; MedDiet = Mediterranean diet.

Grahic Jump Location
Grahic Jump Location
Figure 2.

Cumulative incidence of diabetes (or either diabetes or death).

Nelson-Aalen curves are shown with the outcome of new-onset diabetes (top) or either diabetes or death (bottom), by exposure to each MedDiet intervention vs. the control diet. EVOO = extra-virgin olive oil; HR = hazard ratio; MedDiet = Mediterranean diet.

Grahic Jump Location

Tables

References

Letters

NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Comments

Submit a Comment
Methodological Flaws
Posted on January 7, 2014
Eran Kopel
Personal opinion
Conflict of Interest: None Declared
Patient withdrawals from the study during follow-up were significantly greater in the control group than in the intervention groups. These patients, a substantial 10% of the entire control group, had "worse cardiovascular risk profile at baseline", which importantly imply on the overall higher baseline cardiovascular risk of the control group, than in the intervention groups, a major concern that has been raised recently [Kopel E, Sidi Y, Kivity S. Mediterranean diet for primary prevention of cardiovascular disease. N Engl J Med. 2013;369:672. link: http://www.nejm.org/doi/full/10.1056/NEJMc1306659].

Therefore, the current post-hoc analysis on diabetes outcome would require to model the data differently than was actually done. A competing risks analysis that accounts for any cardiovascular morbidity specific-outcome that occurred before the diabetes outcome should have been used. As competing cardiovascular morbidity events were not accounted for, also the Kaplan-Meier curves are not applicable and might mislead.
Reponse to Methodological Flaws
Posted on February 20, 2014
Jordi Salas-Salvadó, Ramon Estruch, Miguel Ángel Martínez-González
Human Nutrition Unit, Rovira i Virgili University, Reus, Spain. Centro de Investigación Biomédica en Red de Fisiopatología de la Obesidad y Nutrición, Barcelona, Spain (CIBERObn);
Conflict of Interest: None Declared
Kopel et al. mistakenly pointed out that there were “significantly higher percentages of men (+5.7%),... in the control group than in the intervention group assigned to a Mediterranean diet supplemented with nuts" (1) in our PREDIMED trial. The truth was exactly in the opposite direction: the control group included a lower percentage of men (40% men) than the Mediterranean Diet (MeDiet) + nuts group (46% men). As male sex was a strong predictor of a significantly higher risk for the primary cardiovascular end-point, under the null hypothesis of no effect of the dietary intervention, a higher risk would be expected in the MeDiet+nuts group. Lesser degrees of imbalance for some other risk factors were clinically irrelevant and they were not as strong predictors of cardiovascular events as sex was. Therefore, the minor imbalance in sex distribution, in any case, will operate in support of our hypothesis because, despite this imbalance in sex, we found a significant protection by a MeDiet+nuts. Furthermore, we adjusted our estimates for sex and cardiovascular risk factors.
To address the new concern raised by Kopel in his letter about “methodological flaws” in our article with diabetes as outcome (2), we have rerun our models using the approach to tackle competing risks proposed by Fine and Gray. They introduced modifications in the proportional hazards Cox model to allow for the presence of competing risks (3). The modification consisted in keeping the competing risks observations (cardiovascular events in our case) in the risk set with a diminishing weight.
The estimates for hazard ratios in models fitted with the Fine and Gray’s method were very similar to those fitted with ordinary Cox models (<5% change in all cases). Therefore, the potential amount of bias introduced by competing risks can be assumed to be almost negligible. For example, the Hazard ratios (95% confidence intervals) in sex-, age-, body mass-index- and center-adjusted models were:
MeDiet+Extra-virgin olive oil vs. control: 0.65 (0.48-0.87), meaning a 4.4% change.
MeDiet+nuts vs. control: 0.85 (0.64-1.13), meaning a 3.7% change.
MeDiet (both groups merged together) vs. control: 0.74 (0.58-0.95), meaning a 1.3% change.
In summary, there is no methodological problem in our published analyses.

(1) Kopel E, Sidi Y, Kivity S. Letter. N Engl J Med. 2013;369:672.
(2) Salas Salvado J, Bullo M, Estruch R, et al. Prevention of diabetes with Mediterranean diets. Ann Intern Med 2014;160:1-10.
(2) Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496–509.
Response by E. Kopel
Posted on March 4, 2014
Eran Kopel
Tel Aviv, Israel
Conflict of Interest: None Declared
We are well aware of this technical erratum and have already asked the NEJM to correct it. I believe it will appear on-line soon (The word "men" will be replaced by the word "women").

If you carefully read our original critique on your study, however, we did not point on "male gender" as the specific imbalanced risk factor but on the surprising imbalance itself between the RCT groups.

This technicality has nothing to do with our valid methodological concern regarding the imbalance itself in key-baseline characteristics of the PREDIMED RCT, such as gender, by which the randomization was supposed to be explicitly stratified.

A significant evident imbalance such as was previously noticed by us, in gender and other important baseline characteristics as well, will be accompanied by significant unmeasured confounding.

Even extensive adjustment, in the analysis phase, will not resolve such a substantial "residual" confounding, and this distortion in the very infrastructure of the RCT should have been ideally avoided at the randomization phase, as expected from any properly valid RCT.

I would also remind on this opportunity that female sex can also be associated with cardiovascular disease, when it is accompanied by diabetes, for example. A higher percentage of diabetic women in the "low-fat diet" control group is a suspected imbalance that its existence might be supported by the fact that several diabetes-associated drug groups (i.e. oral hypoglycemic use) were also imbalanced significantly in the PREDIMED study at baseline.

Is this the case with the PREDIMED study? Is the baseline CVS risk (i.e. by Framingham's risk score) evenly represented across the groups and between genders in each study group?

Additionally, in the more recent analysis of the study, as published in the Annals, when you excluded patients with diabetes at baseline, the baseline variables were indeed more balanced. However, in this study, a competing-risk analysis between diabetes and CVS morbidity outcomes should have been performed, as I noted earlier in the aforementioned correspondence, since you still had a significant portion of drop-out in the control group for CVS reasons, which implies on the overall higher CVS risk in the control group from the beginning of this study.

In your response you indicate that such additional analysis was done without a significant change of the HR results; however, these derived sub-distribution HRs based on Fine–Gray competing-risk models are difficult to interpret as a standard epidemiological covariate rate:risk association [1] and this analysis should have actually aided to correctly draw the biased naïve Kaplan-Meier survival curves, as noted in my original correspondence, and which I hope could still be presented along with the published form of my original correspondence and your response, in the forthcoming Annals issue.

Best regards,
Eran Kopel, MD MPH

Reference:
1. Andersen PK, Geskus RB, de Witte T, Putter H. Competing risks in epidemiology:
possibilities and pitfalls. Int J Epidemiol. 2012;41(3):861-70.
Revised Comment
Posted on April 22, 2014
The Editor's
Annals of Internal Medicine
Conflict of Interest: None Declared
The editors asked Dr. Kopel to consolidate the remarks from the two letters that he had posted earlier and to focus his criticisms on the paper published in Annals. Dr. Kopel's letter posted 4/22/2014 is that consolidation."

Revised Author's Response
Posted on April 22, 2014
Martinez-Gonzalez MA, Estruch R, Corella D, Ros E, Salas-Salvado J
University of Navarra, Instituto de Salud Carlos III , University of Barcelona, University of Valencia, Universitat Rovira i Virgili
Conflict of Interest: None Declared

We write to address Dr. Kopel’s criticisms about potential “imbalances” in baseline characteristics of PREDIMED trial participants and competing risk analyses (1). Sex distribution was the only potentially relevant “imbalance” in our trial. The control group included a slightly higher proportion of women than the intervention group. Female sex was associated with a lower risk of both new-onset diabetes (age-adjusted Hazard Ratio [HR]: 0.81, 95% confidence interval 0.63-1.04) and cardiovascular events in the overall cohort (HR: 0.45, 95% CI: 0.35-0.57). It was also associated with a lower risk of cardiovascular disease among diabetic participants (age-adjusted HR: 0.52; 95% CI: 0.39-0.70). Only minor changes in the estimates for the intervention effect were seen after adjusting for sex (0.8% change for MeDiet+EVOO and 2.1% change for MeDiet+nuts). After removal of the potential confounding effect of sex, the apparent protection from the MeDiet intervention was even stronger. Other statistically “significant” minor imbalances in the overall PREDIMED cohort showed P values close to 0.05 (with the exception of age and body mass index). The actual observed differences varied in direction in groups, and were small and clinically meaningless. Adjusted estimates, regardless of the factors that we adjusted for, were essentially unchanged (2).
We built new Kaplan-Meier survival curves with either diabetes or cardiovascular events, whichever occurred first, and using inverse probability weighting to account for potential confounding due to minor imbalances in baseline factors (Figure 1print only). When using either cardiovascular disease events or diabetes as outcomes among participants initially free of diabetes to take into account the possibility of competing risks between cardiovascular disease and diabetes, the age- and sex-adjusted HRs versus control were 0.68 (95% CI: 0.52-0.87) for MeDiet+EVOO and 0.76 (0.59-0.98) for nuts. The overall higher cardiovascular risk in the control group led to stronger inverse associations. When we also included all-cause death in this composite outcome, the age- and sex-adjusted HRs versus control were 0.72 (95% CI: 0.57-0.91) for MeDiet+EVOO and 0.80 (0.64-1.01) for MeDiet+nuts.
From a practical point of view, the relevant issue for residual confounding is not statistical significance, but the magnitude of the absolute between-group differences and their clinical relevance, which would be translated into changes between crude and adjusted estimates (3). Therefore, substantial residual confounding was not an issue in the PREDIMED. We have already published (in supplements and technical annexes (2,4,5) various competing-risk and multiple imputation analyses to account for potential biases due to drop-outs. These ancillary analyses also support our main conclusions.
Andersen et al., (6) have dealt in depth with the issue of competing risks. They recommended the use of the Fine and Gray method (7) for competing risk analysis, as we did (8). Our results were essentially unchanged when we used the Fine and Gray methodology (8). While Andersen et al. acknowledge the potential difficulties in interpretation of this approach, they also state that “both rates and risks for all competing events remain useful and tend to supplement each other when studying models for competing risks”. In this context, the use of a composite outcome including all competing events (i.e., the combination of new-onset diabetes, cardiovascular events or death) also reinforces our main conclusions.


References:


1. Kopel E, Sidi Y, Kivity S. Prevention of Diabetes With Mediterranean Diets: A Subgroup Analysis of a Randomized Trial. Ann Intern Med 2014 (In press).

2. Estruch R, Ros E, Salas-Salvadó J, Covas MI, Corella D, Arós F, Gómez-Gracia E, Ruiz-Gutiérrez V, Fiol M, Lapetra J, Lamuela-Raventos RM, Serra-Majem L, Pintó X, Basora J, Muñoz MA, Sorlí JV, Martínez JA, Martínez-González MA; PREDIMED Study Investigators. Primary prevention of cardiovascular disease with a Mediterranean diet. N Engl J Med 2013;368:1279-90. Erratum in: N Engl J Med 2014;370:886.

3. Greenland S, Rothman KJ. Introduction to stratified analysis. In. Rothman KJ, Greenland S, Lash T (ed.). Modern Epidemiology, 3rd ed. Philadelphia: Lippicott Williams and Wilkins, 2008; 262.

4. Salas-Salvadó J, Bulló M, Estruch R, Ros E, Covas MI, Ibarrola-Jurado N, Corella D, Arós F, Gómez-Gracia E, Ruiz-Gutiérrez V, Romaguera D, Lapetra J, Lamuela-Raventós RM, Serra-Majem L, Pintó X, Basora J, Muñoz MA, Sorlí JV, Martínez-González MA. Prevention of diabetes with Mediterranean diets: a subgroup analysis of a randomized trial. Ann Intern Med 2014;160:1-10.

5. Ruiz-Canela M, Estruch R, Corella D, Salas-Salvadó J, Martínez-González MA. Association of Mediterranean diet with peripheral artery disease: the PREDIMED randomized trial. JAMA 2014;311:415-7.

6. Andersen PK, Geskus RB, de Witte T, Putter H. Competing risks in epidemiology: possibilities and pitfalls. Int J Epidemiol 2012;41:861-70.

7. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496–509.

8. Salas-Salvado J, Estruch R, Martinez-Gonzalez MA. Response to methodological flaws. Ann Intern Med 2014 (e-letter). Epub Feb 20, 2014.

 

 

 

Revised Comment
Posted on April 22, 2014
Erin Kopel, MD
Tel Aviv, Israel
Conflict of Interest: None Declared

The subgroup analysis (1) of the PREDIMED trial (Prevención con Dieta Mediterránea) that found that a Mediterranean diet enriched with extra-virgin olive oil reduced diabetes risk is methodologically flawed, in our opinion. During the study, more participants were lost to follow-up in the low-fat diet control group (10.5%) than in the Mediterranean diet groups supplemented with either olive oil (4.1%) or mixed nuts (6.9%). As stated by the authors, participants in the control group “who withdrew had a worse cardiovascular risk profile at baseline than those who remained in the study.” This fact indicates a potentially higher baseline cardiovascular risk profile of the entire control group in the subgroup analysis, when compared to the intervention groups.

A higher baseline cardiovascular risk profile in the control group could coexist with higher likelihood of baseline prediabetes and hyperinsulinemia states, particularly in women (2). Although the original trial stratified randomization by gender, both the trial and the subgroup analysis showed more women in the control group than in the Mediterranean diet groups (3). Therefore, the control group could bear a higher risk for incident diabetes during follow-up.

To sort through these issues, we think it is important to examine whether the baseline cardiovascular risk (e.g. by mean Framingham's 10-year cardiovascular risk score) was evenly represented across the 3 comparison groups and by genders in the subgroup analysis.

Also of note is that important between-group differences at baseline (such as the gender difference) are often associated with differential distribution among groups of other, unmeasured, baseline factors that are related to outcomes, and therefore may lead to significant confounding. Even though the authors of the subgroup analysis did extensive multivariable adjustment in their analyses, we do not believe that such analyses would resolve substantial "residual" confounding.

Rather, we recommend that the authors perform a competing-risk analysis (4) that accounts for competing cardiovascular morbidity outcomes that could occur prior to the diabetes outcome. We believe that the results of that analysis should be presented and discussed, and that the simple Kaplan-Meier curves in the article should be redrawn to reflect the new analyses (5).


Eran Kopel, MD MPH
Yechezkel Sidi, MD
Shaye Kivity, MD
Chaim Sheba Medical Center, Tel Hashomer, Israel
eran.kopel@mail.huji.ac.il

Potential Conflicts of Interest: None Disclosed.

References
1. Salas-Salvadó J, Bulló M, Estruch R, Ros E, Covas MI, Ibarrola-Jurado N, et al. Prevention of diabetes with Mediterranean diets: a subgroup analysis of a randomized trial. Ann Intern Med. 2014;160:1-10.
2. Hu FB, Stampfer MJ, Haffner SM, Solomon CG, Willett WC, Manson JE. Elevated risk of cardiovascular disease prior to clinical diagnosis of type 2 diabetes. Diabetes Care. 2002;25:1129-34.
3. Kopel E, Sidi Y, Kivity S. Mediterranean diet for primary prevention of cardiovascular disease. N Engl J Med. 2013;369:672.
4. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496-509.
5. Andersen PK, Geskus RB, de Witte T, Putter H. Competing risks in epidemiology: possibilities and pitfalls. Int J Epidemiol. 2012;41:861-70.
Submit a Comment

Summary for Patients

Does the Mediterranean Diet Prevent Diabetes?

The full report is titled “Prevention of Diabetes With Mediterranean Diets. A Subgroup Analysis of a Randomized Trial.” It is in the 7 January 2014 issue of Annals of Internal Medicine (volume 160, pages 1-10). The authors are J. Salas-Salvadó, M. Bulló, R. Estruch, E. Ros, M.I. Covas, N. Ibarrola-Jurado, D. Corella, F. Arós, E. Gómez-Gracia, V. Ruiz-Gutiérrez, D. Romaguera, J. Lapetra, R.M. Lamuela-Raventos, L. Serra-Majem, X. Pintó, J. Basora, M.A. Muñoz, J.V. Sorlí, and M.A. Martínez-González.

Read More...

Clinical Slide Sets

Terms of Use

The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.

Toolkit

Buy Now

to gain full access to the content and tools.

Want to Subscribe?

Learn more about subscription options

Advertisement
Related Articles
Related Point of Care
Topic Collections
Forgot your password?
Enter your username and email address. We'll send you a reminder to the email address on record.
(Required)
(Required)