Letters

tion, the effect of a drug on conduction and refractoriness is highly influenced by diseased states.4 In our patient, who had significant comorbidities and who received high doses of intravenous phenytoin in addition to a wide variety of other drugs, the effect of phenytoin toxicity is not likely to be as straightforward as outpatient oral phenytoin overdose. Algren and Christian also argue that the free phenytoin level was not markedly elevated. However, the calculated free phenytoin level can dramatically underestimate actual freely available drug and should not be used to rule out toxicity in such complex patients. Most laboratories use the Winter-Tozer equation to “correct” for changes in albumin levels. However, when compared with the actual free phenytoin measurements in patients with hypoalbuminemia, this mathematical correction significantly underestimates the actual free phenytoin level.5 It has also been shown to underestimate the true level by at least 25% in patients with renal failure.5 Our patient exhibited hypoalbuminemia and renal failure; the mathematically corrected level therefore underestimates the true free phenytoin level. For these reasons, the correction should not be used clinically to normalize phenytoin concentrations in complex patients. Therefore, as we originally described, we believed the changes on this patient’s ECG were caused by polypharmacy rather than phenytoin alone. As we originally stated, the combination of fluconazole and phenytoin—both of which have been shown to have effects on potassium channels—likely resulted in the abnormalities seen. Furthermore, the most compelling evidence of phenytoin involvement was the observed dose-response normalization in the ECG that corresponded with decreasing levels of phenytoin. Therefore, we believe that polypharmacy that included the use of phenytoin resulted in the ECG changes found in our patient.

Colleen J. Johnson, MD, MS Mintu P. Turakhia, MD, MAS Author Affiliations: Department of Medicine (Heart and Vascular Institute), Tulane University, New Orleans, Louisiana (Johnson); Veterans Affairs Palo Alto Health Care System, Palo Alto, California (Turakhia); Department of Medicine (Cardiovascular Medicine), Stanford University, Stanford, California (Turakhia). Corresponding Author: Mintu Turakhia, MD, MAS, Palo Alto VA Health Care System, Stanford University, 3801 Miranda Ave, 111C, Palo Alto, CA 94304 ([email protected]). Conflict of Interest Disclosures: None reported. Disclaimer: The content and opinions expressed are solely the responsibility of the authors and do not necessarily represent the views or policies of the Department of Veterans Affairs. 1. Johnson CJ, Scheinman MA, Turakhia MP. Bizarre and wide QRS after liver transplant—quiz case. JAMA Intern Med. 2013;173(11):953-955. 2. Evers ML, Izhar A, Aqil A. Cardiac monitoring after phenytoin overdose. Heart Lung. 1997;26(4):325-328. 3. Wyte CD, Berk WA. Severe oral phenytoin overdose does not cause cardiovascular morbidity. Ann Emerg Med. 1991;20(5):508-512. 4. Josephson ME. Clinical Cardiac Electrophysiology: Techniques and Interpretations. Philadelphia, PA: Lippincott Williams & Wilkins; 2008. 5. Mauro LS, Mauro VF, Bachmann KA, Higgins JT. Accuracy of two equations in determining normalized phenytoin concentrations. DICP. 1989;23(1):64-68. 168

Lack of Adjustment for Body Mass Index To the Editor In a recent issue of the Journal, Orlich et al1 described in an observational cohort study a relation between dietary patterns and all-cause mortality. Compared with nonvegetarians, vegans had a hazard ratio of 0.85 (95% CI, 0.73-1.01) for all-cause mortality. The regression models were adjusted for known covariates, such as age, smoking, physical activity, alcohol consumption, and socioeconomic status. We would like to emphasize some elements that could influence the observed relationship between dietary pattern and all-cause mortality. The regression models have not been adjusted for adiposity, usually characterized by the proxy body mass index (BMI). Adiposity can be seen as a possible confounder because it is related to the exposition (ie, dietary pattern) and to the outcome (ie, all-cause mortality), and adiposity is not part of the causal pathway between dietary pattern and all-cause mortality. It is clear from Table 2 of their study that vegans, with the lowest all-cause mortality, had a mean (SD) BMI of 24.1 (4.7) (calculated as weight in kilograms divided by height in meters squared), compared with nonvegetarians (28.3 [6.1]). The high standard deviation for nonvegetarians indicates a higher prevalence of obesity in this category. Compared with male subjects with a BMI between 23.5 to 25.0, subjects with a BMI of 30.0 to 35.0 and 35.0 or greater would have an all-cause mortality 24% and 94% higher after 10 years of follow-up.2 Second, the observed effect on mortality seems to be more important in men compared with women. It would be worth presenting the descriptive data in Table 2 stratified by sex, so that readers could understand whether sex-specific differences in physical exercise or BMI could explain study results. Lastly, 11 956 potential participants had a prior diagnosis of cancer or of cardiovascular disease, ie, 12% of the total cohort. Bearing in mind that the average age at baseline was approximately 58 years, the proportion of subjects with a serious condition was high. This high proportion was probably due to the way the cohort was assembled, with a selection of subjects less healthy than the average US population of same age. In any case, we do not understand why subjects with prior disease were excluded from a prospective study having mortality as an end point. Also, what was the distribution of these subjects in the various dietary pattern groups? Unbalanced distribution could point at variability in health status at cohort inception, suggesting that, because of the way subjects selfselected themselves for being part of the cohort, survival patterns of the dietary pattern groups could have been different from the study start. In conclusion, we believe that those points need clarification before drawing conclusions on the relationship between dietary pattern and all-cause mortality. Patrick Mullie, PhD Philippe Autier, MD, PhD Author Affiliations: International Prevention Research Institute (iPRI), Lyon, France (Mullie, Autier).

JAMA Internal Medicine January 2014 Volume 174, Number 1

Copyright 2014 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 05/25/2015

jamainternalmedicine.com

Letters

Corresponding Author: Patrick Mullie, PhD, International Prevention Research Institute, 95 cours Lafayette, 69006 Lyon, France ([email protected]). Conflict of Interest Disclosures: None reported. 1. Orlich MJ, Singh PN, Sabaté J, et al. Vegetarian dietary patterns and mortality in Adventist Health Study 2. JAMA Intern Med. 2013;173(13):1230-1238. 2. Pischon T, Boeing H, Hoffmann K, et al. General and abdominal adiposity and risk of death in Europe. N Engl J Med. 2008;359(20):2105-2120.

In Reply We thank Mullie and Autier for their correspondence. Regarding adjustment for body mass index (BMI), we disagree with the assertion that adiposity is not part of the causal pathway relating dietary patterns to mortality. In our article,1 Table 2 suggests an association of dietary pattern with BMI. Analysis of covariance analysis demonstrates this to be significant (P < .001) after adjustment for age, sex, race, exercise, and energy intake. Dietary pattern may plausibly affect BMI causally, the association persists after adjustment, and reverse causation seems unlikely. It is generally believed that adiposity causally contributes to mortality. Thus, a potential causal pathway from dietary pattern to mortality with BMI as an intermediate is suggested. Adjustment for such a potential intermediate would remove part of the total “effect” of dietary patterns on mortality. Nevertheless, we reported results for a sensitivity analysis in which BMI was a covariate. This had only modest effects, which were described in our article.1 The odds ratio for all-cause mortality for men and women combined for all vegetarians vs nonvegetarians changed from 0.88 (95% CI, 0.80-0.97) to 0.90 (95% CI, 0.82-0.98) when BMI was included.1 Exclusion of subjects reporting prior major disease is often done in similar analyses to avoid bias. In the article by Pan et al2 on meat intake and mortality in the Nurses’ Health Study and Health Professionals Follow-up Study that we cited in our study, those with a history of cardiovascular disease or cancer at baseline were excluded. Our primary reason for having performed these exclusions is the concern that those with prior major disease may have changed their dietary patterns in the hope of treating their disease or lowering their risk of adverse health outcomes. For example, a person diagnosed as having coronary artery disease may have adopted a vegetarian diet in the hope of reducing his risk of further adverse outcomes; if so, his inclusion would tend to bias results of vegetarians toward increased mortality. Restriction to those without major baseline disease should also tend to control for any selfselection bias based on perceived disease status, as raised by Mullie and Autier. Mullie and Autier suggest that the number of participants excluded for prior disease is unusually high and that the cohort was likely a selection of less-healthy-than-

jamainternalmedicine.com

average persons. We think such conclusions are not very useful in characterizing the health of the cohort. In the article by Pan et al, 2 11.2% of male health professionals (mean age approximately 53 years) and 5.8% of female nurses (mean age approximately 47 years) were excluded for baseline cardiovascular disease or cancer.2 Our exclusion rates seem comparable, given the older age of our subjects (mean age, 57 years). In an article from the National Institutes of Health–AARP Diet and Health Study,3 9.0% of participants (mean age of approximately 62 years) were excluded based on self-reported prevalent cancer, comparable to our exclusion of 8.1% owing to prior self-reported cancer. Variability in rates of exclusion may relate to differences in the definitions of prior disease to be excluded and to differences in the accuracy of self-report of prior disease.

Michael J. Orlich, MD Gary E. Fraser, MBchB, PhD Author Affiliations: School of Public Health, Loma Linda University, Loma Linda, California (Orlich, Fraser). Corresponding Author: Michael J. Orlich, MD, Adventist Health Studies, School of Public Health, 24951 N Circle Dr, NH 2031, Loma Linda University, Loma Linda, CA 92350 ([email protected]). Conflict of Interest Disclosures: Dr Orlich reports receiving a small honorarium from the Northern California Conference of Seventh-day Adventists to partially defray travel expenses for a speaking engagement at which he gave an overview and update of Adventist Health Studies research and a small honorarium from the Southern California Conference of Seventh-day Adventists for a speaking engagement at which he lectured on lifestyle approaches for chronic disease prevention. Dr Orlich’s research fellowship was supported by a grant from the National Institute of Food and Agriculture (NIFA) 2010-38938-20924. Dr Fraser reports no relevant financial interests. 1. Orlich MJ, Singh PN, Sabaté J, et al. Vegetarian dietary patterns and mortality in Adventist Health Study 2. JAMA Intern Med. 2013;173(13):1230-1238. 2. Pan A, Sun Q, Bernstein AM, et al. Red meat consumption and mortality: results from 2 prospective cohort studies. Arch Intern Med. 2012;172(7): 555-563. doi:10.1001/archinternmed.2011.2287. 3. Schatzkin A, Mouw T, Park Y, et al. Dietary fiber and whole-grain consumption in relation to colorectal cancer in the NIH-AARP Diet and Health Study. Am J Clin Nutr. 2007;85(5):1353-1360.

CORRECTION Incorrect Axis Label in Figure: In the Research Letter titled “Overdiagnosis and Overtreatment: Evaluation of What Physicians Tell Their Patients About Screening Harms” published online on October 21, 2013, in JAMA Internal Medicine (doi: 10.1001/jamainternmed.2013.10363), the Figure’s y-axis label was incorrect. The y-axis label should have been given as “People Choosing to Be Screened, %.” This article was corrected online.

JAMA Internal Medicine January 2014 Volume 174, Number 1

Copyright 2014 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 05/25/2015

169

Lack of adjustment for body mass index.

Lack of adjustment for body mass index. - PDF Download Free
67KB Sizes 0 Downloads 0 Views