Mediterranean Diet, Kidney Function, and Mortality in Men with CKD

Summary

Background and objectives

Adherence to a Mediterranean diet may link to a better preserved kidney function in the community as well as a favorable cardiometabolic profile and reduced mortality risk in individuals with manifest CKD.

Design, setting, participants, & measurements

Dietary habits were determined by 7-day dietary records in a population-based cohort of 1110 Swedish men (age 70 years) from 1991 to 1995, 506 of whom were considered to have CKD because of a GFR<60 ml/min per 1.73 m². A Mediterranean Diet Score was calculated, and participants were categorized as having low, medium, or high adherence. Adequate dietary reporters were identified with Goldberg cutoffs (n=597). Deaths were registered during a median follow-up of 9.9 years.

Results

Compared with low adherents, medium and high adherents were 23% and 42% less likely to have CKD, respectively (adjusted odds ratio [95% confidence interval]=0.77 [0.57 to 1.05] and 0.58 [0.38 to 0.87], respectively, P for trend=0.04). Among those individuals with CKD, phosphate intake and net endogenous acid production were progressively lower across increasing adherence groups. No differences were observed regarding other cardiometabolic risk factors across adherence groups. As many as 168 (33%) CKD individuals died during follow-up. Compared with low adherents, proportional hazards regression associated medium and high adherents to a 25% and 23% lower mortality risk, respectively (adjusted hazard ratio [95% confidence interval]=0.75 [0.52 to 1.06] and 0.77 [0.44 to 1.36], respectively, P for trend=0.10). Sensitivity analyses showed significant and stronger associations when only adequate dietary reporters were considered.

Conclusions

Adherence to a Mediterranean dietary pattern is associated with lower likelihood of CKD in elderly men. A greater adherence to this diet independently predicted survival in those patients with manifest CKD. Clinical trials are warranted to test the hypothesis that following such a diet could improve outcomes (independent of other healthy lifestyles) in CKD patients.

Introduction

Patients with CKD frequently experience high risk of comorbidities and mortality. Altered food intake (i.e., insufficient intake and/or suboptimal diet quality) is prevalent in patients with CKD (1) and probably contributes to this excess risk (2–4). Previous research on dietary practice in the CKD population mainly focuses on single nutrients. Because of the complexity and intercorrelation of food components (we eat meals rather than isolated nutrients), dietary patterns may better capture interrelationships among nutrients and facilitate the interpretation of risk implications (5).

The traditional Mediterranean dietary pattern is characterized by a high intake of vegetables, legumes, fruits, nuts, cereals, and olive oil; a moderately high intake of fish; a low-to-moderate intake of dairy products; a low intake of saturated fats, meat, and poultry; and a regular but moderate intake of wine during meals (6). Strong evidence exists on the beneficial effects of a Mediterranean diet for incidence of several chronic diseases and reduced risk of mortality (7–19). Mechanisms underlying these relationships include benefits of this dietary pattern on the cardiometabolic profile (e.g., adiposity, BP, blood lipids, serum fasting glucose, insulin resistance, and inflammation [20,21]) as well as on metabolic acidosis (22,23). Because such risk factors are highly prevalent in CKD and contribute to its onset, it is plausible to hypothesize that adoption of this healthy dietary lifestyle may link to better preserved renal function. Limited research offering mixed results has, however, investigated this issue (24–26), and errors inherent to dietary assessment methods could have contributed to these divergent findings (27). We tested this hypothesis in a Swedish population-based cohort of elderly men of similar age, devoting particular efforts to minimize reporting biases in dietary recalls. Furthermore, we also tested the hypothesis that, like in the general population, a greater adherence to this healthy diet associates with a more favorable risk profile and a lower mortality risk among individuals with manifest CKD.

Materials and Methods

Participants

This study is based on the Uppsala Longitudinal Study of Adult Men cohort (ULSAM; described in detail at http://www2.pubcare.uu.se/ULSAM/). All 50-year-old men born from 1920 to 1924 and living in Uppsala were invited to participate in the ULSAM cohort study between 1970 and 1974. Participants returned for subsequent examinations 10 and 20 years later. The present study is based on the third examination cycle of the ULSAM cohort, when dietary records were collected for the first time. In this occasion (during 1991–1995), 73% of eligible men agreed to participate (n=1221) and were approximately 70 years of age. Exclusion criteria for our analysis were unavailable data on 7-day dietary records and/or serum cystatin C (n=116) and extreme values of reported energy intake (<3200 or >18,000 kJ/d; n=5). The present analysis, therefore, comprises 1110 men. The study was approved by the ethics committee at Uppsala University, and all participants gave informed consent to participate.

Clinical Examinations

All investigations were performed under standardized conditions, including anthropometric measurements, BP, blood sampling, and questionnaires regarding medical history, smoking habits, education, and physical activity level as described elsewhere (28). Body mass index (BMI) was calculated as the ratio of the body weight (kilograms) to the height (meters squared). Waist circumference was measured midway between the lowest rib and the iliac crest. Smoking status was defined as current smoking versus nonsmoking. Regular physical activity was defined as the reporting of regular or athletic leisure time exercise habits according to four physical activity categories (sedentary, moderate, regular, and athletic). Educational level was recorded as low (elementary school), medium (secondary school), and high (university studies). Supine systolic and diastolic BPs were measured two times in the right arm after 10 minutes of rest, and means were calculated. Hypertension was defined as systolic BP≥140 mmHg, diastolic BP≥90 mmHg, or use of antihypertensive medications. Hyperlipidemia was defined as serum cholesterol>250 mg/dl and/or serum triglycerides>200 mg/dl and/or treatment with lipid-lowering medications. Diabetes mellitus was defined as fasting plasma glucose≥126 mg/dl, 2-hour postload glucose levels≥200 mg/dl, or use of oral hypoglycemic agents or insulin.

Venous blood samples were drawn after an overnight fast and stored at −70°C until analyses. The assays were performed at the Department of Clinical Chemistry, University Hospital, Uppsala, which is accredited according to the Swedish Board for Accreditation and Conformity Assessment (Swedac) standard ISO/IEC 17025. Serum triglyceride, total cholesterol, and HDL cholesterol concentrations were assayed by enzymatic techniques. Fasting blood glucose concentration was determined by an oxidase method. Fasting plasma insulin was assayed using an enzymatic–immunologic assay (Enzymmun; Boehringer Mannheim, Mannheim, Germany) performed in an ES300 automatic analyzer (mU/L; Boehringer Mannheim). High-sensitivity C-reactive protein measurements were performed by latex-enhanced reagent (Dade Behring, Deerfield, IL) using a Behring BN ProSpec analyzer (Dade Behring). Serum albumin concentration was measured with spectrophotometry using bromine cresol green and reagents from Boehringer Mannheim (Hitachi 717 or 911; Hitachi, Japan). Serum cystatin C was measured by latex-enhanced reagent (N Latex Cystatin C; Dade Behring, Deerfield, IL) with a Behring BN ProSpec analyzer (Dade Behring). Estimated GFR (eGFR) was calculated from serum cystatin C concentrations (mg/L) by the following formula: y=77.24×x^−1.2623, which has been shown to be closely correlated with iohexol clearance (29). Individuals with eGFR<60 ml/min per 1.73 m² were considered to have manifest CKD according to the current Kidney Disease Outcomes Quality Initiative definition (30).

Assessment of the Mediterranean Dietary Pattern

Dietary habits were evaluated with an optically readable form of a 7-day dietary record based on a validated precoded menu book (31), which was prepared and previously used by the Swedish National Food Administration (32). The participants were given oral instructions by a dietitian on how to perform the dietary registration, and the amounts consumed were reported in household measurements or specified as portion sizes. The daily intake of energy and various food items was calculated by using a database from the Swedish National Food Administration. Dietary acid load was estimated with net endogenous acid production (NEAP; mEq/d) using the following algorithm (33): 54.5×protein intake (g/d)÷potassium intake (mEq/d)−10.2. Higher NEAP values reflect an acid-forming potential.

To minimize systematic errors in dietary records, the Goldberg cutoff was applied to identify adequate reporters of energy intake (27). In this procedure, an acceptable range of energy intake is determined for each participant in relation to estimated energy expenditure taking the level of physical activity and calculated basal metabolic rate into account (i.e., yielding a 95% confidence interval [CI] for energy intake required to maintain body weight). Individuals with reported energy intake within the 95% CI were considered as adequate dietary reporters, rendering a subpopulation of 597 individuals, 250 of which were considered to have manifest CKD. As shown in Supplemental Table 1, nonadequate reporters had higher BMI, lower educational level, and more comorbities than adequate reporters.

The Mediterranean Diet Score (MDS) (8) was applied with slight modifications (Supplemental Table 2, footnotes) (9). MDS is a population-based quality score using dietary intake medians as cutoffs for eight food components of a typical Mediterranean diet. Before scoring, the food variables were adjusted for total energy intake by the residual method (g/d) (34). An intake on the favorable side of the median was coded as one; otherwise, the intake was coded zero. Hence, the modified MDS could take a value from 0 to 8 points, and individuals were categorized as low- (≤2 points), medium- (3–5 points), and high-adherent (≥6 points) individuals (9).

Follow-Up

Follow-up for 10-year all-cause mortality was done from the examination date until death or December 31, 2003 as specified a priori. The Swedish National Registry recording for cause of death was used to define end points. The register includes all Swedish citizens, with no loss to follow-up.

Statistical Analyses

Values were expressed as mean ± SD, median (interquartile range [IQR]), or percentage of total as appropriate. Because dietary data used to compose the MDS did not substantially differ among adequate and nonadequate reporters, the same MDS, based on the whole cohort, was used through all analyses. Comparisons between CKD and non-CKD men were evaluated by the unpaired t test, nonparametric Mann–Whitney test, or χ² test as appropriate. Linear change of proportions of CKD across groups with different degrees of adherence to a Mediterranean diet (low, medium, and high adherence groups) was evaluated, and P value for trend was reported. Crude and multiple adjusted logistic regression models were fitted to evaluate the association of MDS with the presence of CKD. Covariance in the adjusted models included BMI, physical activity, smoking status, education, hypertension, hyperlipidemia, and diabetes. As a sensitivity analysis, the logistic regression models were performed repeatedly in the subpopulation of adequate dietary reporters. Data are presented as odds ratios (ORs) and 95% CIs.

In individuals with CKD, linear changes of cardiometabolic and other dietary risk factors across ordinal MDS groups (low, medium, and high adherence groups) were evaluated. Kaplan–Meier curves were used to describe mortality characteristics in low-, medium-, and high-adherent individuals. Proportional hazards regression models were fitted to determine the predictive value of MDS on all-cause mortality. Proportional hazard assumptions were confirmed by the Schoenfeldt test. Potential confounders (i.e., BMI, physical activity, smoking status, education, hypertension, hyperlipidemia, and diabetes) were included as covariance in the adjusted models. As a sensitivity analysis, the proportional hazards regression models were performed in the subpopulation of adequate dietary reporters. Data are expressed as hazard ratios (HRs) and 95% CIs.

All tests were two-tailed, and P<0.05 was considered significant. Statistical analyses were performed using STATA version 12 (Stata Corporation, College Station, TX).

Results

General Characteristics

Comparison of general characteristics between CKD and non-CKD populations included in the present study is shown in Table 1. CKD individuals, compared with those individuals without CKD, had higher BMI and were more likely to be hypertensive. In addition, individuals with CKD presented a lower adherence to a Mediterranean dietary pattern. No differences were observed for smoking status, physical activity, education, hyperlipidemia, and diabetes between these two groups.

Table 1.

General characteristics of men included in the study and stratified by the presence of CKD

Characteristic	CKD (n=506)	Non-CKD (n=604)	P Value
Estimated GFR, ml/min per 1.73 m2	51.9 (46.3, 56.6)	69.3 (64.0, 77.0)
Body mass index, kg/m2	26.5±3.5	26.0±3.3	0.01
Smoking, n (%)	111 (22)	105 (18)	0.07
Physical activity, n (%)			0.10
Sedentary	23 (5)	19 (3)
Moderate	177 (36)	191 (33)
Regular	267 (55)	327 (56)
Athletic	21 (4)	42 (7)
Education, n (%)			0.44
Elementary school	289 (57)	338 (56)
Secondary school	151 (30)	171 (28)
University or equivalent	66 (13)	95 (16)
Hypertension, n (%)	189 (37)	159 (26)	<0.001
Hyperlipidemia, n (%)	185 (37)	208 (34)	0.46
Diabetes, n (%)	67 (13)	88 (15)	0.52
Mediterranean diet score	3.6±1.6	3.8±1.5	0.03

Data are expressed as mean ± SD, median (interquartile range), or number (percentage) as appropriate. Conversion factor for units is ×0.01667 for estimated GFR (ml/min per 1.73 m² to ml/s per 1.73 m²).

MDS and the Presence of CKD

Figure 1 illustrates lower proportions of individuals with CKD across increasing MDS groups in both the whole and the adequate reporter populations. In crude logistic models, every two-point increase in MDS significantly associated with lower odds of having CKD (Table 2), which was somewhat alleviated after adjustment for confounders. We found independently lower odds of having CKD across groups with a greater adherence to a Mediterranean dietary pattern (adjusted OR [95% CI]=0.77 [0.57 to 1.05] for medium adherents and adjusted OR [95% CI]=0.58 [0.38 to 0.87] for high adherents, P for trend=0.04, compared with low adherents). As a sensitivity analysis, similar associations were observed in adequate reporters.

Figure 1.

Proportions of men with CKD across groups according to adherence to a Mediterranean diet in both all men and adequate reporters only. Individuals with Mediterranean Diet Scores of ≤2, 3–5, and ≥6 points were considered low, medium, and high adherents, respectively.

Table 2.

Logistic regression models for the presence of CKD

Exposure	Crude Odds Ratio (95% CI)	Adjusted Odds Ratio (95% CI)
All participants (n=1110)
Two-point increase in Mediterranean diet score	0.84 (0.72 to 0.98)	0.88 (0.75 to 1.02)
Adherence groups
Low (Mediterranean diet score=1–2)	Reference	Reference
Medium (Mediterranean diet score=3–5)	0.72 (0.54 to 0.97)	0.77 (0.57 to 1.05)
High (Mediterranean diet score=6–8)	0.54 (0.36 to 0.81)	0.58 (0.38 to 0.87)
P for trend	0.01	0.04
Adequate reporters (n=597)
Two-point increase in Mediterranean diet score	0.86 (0.69 to 1.07)	0.89 (0.71 to 1.12)
Adherence groups
Low (Mediterranean diet score=1–2)	Reference	Reference
Medium (Mediterranean diet score=3–5)	0.73 (0.48 to 1.11)	0.78 (0.50 to 1.20)
High (Mediterranean diet score=6–8)	0.47 (0.26 to 0.85)	0.49 (0.27 to 0.91)
P for trend	0.04	0.08

Data are presented as odds ratio (95% confidence interval [95% CI]). Covariance in the adjusted models includes body mass index, physical activity, smoking status, education, hypertension, hyperlipidemia, and diabetes.

MDS and Risk Factors in Individuals with CKD

In individuals with CKD, we studied the phenotype associated to a higher adherence to a Mediterranean diet (Table 3). Across increasing adherence, no differences were observed regarding BMI, waist circumference, BP, serum lipoproteins, fasting glucose, insulin, C-reactive protein, serum albumin, or sodium intake (all P values for linear trend>0.05). Phosphate intake and NEAP were progressively lower across increasing adherence groups. These results were confirmed in adequate reporters (data not shown).

Table 3.

Cardiometabolic risk factors in men with CKD stratified by their degree of adherence to a Mediterranean dietary pattern

Cardiometabolic Risk Factor	Low Adherence (n=131)	Medium Adherence (n=319)	High Adherence (n=56)	P for Trend
Body mass index, kg/m2	26.9±3.3	26.3±3.4	27.1±3.9	0.48
Waist circumference, cm	97±10	95±10	97±11	0.44
Systolic BP, mmHg	149±19	148±19	148±19	0.80
Diastolic BP, mmHg	85±10	85±10	84±9	0.59
Triglycerides, mg/dl	125 (89, 169)	112 (84, 152)	119 (77, 170)	0.12
Total cholesterol, mg/dl	220±41	222±38	228±37	0.25
HDL cholesterol, mg/dl	47±14	49±14	48±14	0.23
Glucose, mg/dl	103±20	101±21	106±28	0.60
Insulin, μU/ml	9 (5, 12)	7 (5, 10)	8 (6, 13)	0.41
C-reactive protein, mg/L	2.4 (1.2, 5.7)	2.2 (1.2, 5.0)	2.1 (1.4, 3.6)	0.20
Albumin, g/dl	4.3±0.3	4.3±0.3	4.3±0.3	0.65
Sodium intake, g/d	2.4±0.7	2.6±0.8	2.5±0.6	0.09
Phosphate intake, g/d	1.6 (1.2, 2.8)	1.5 (1.1, 2.5)	1.3 (1.0, 1.8)	0.003
Net endogenous acid production, mEq/d	44±8	41±8	38±7	<0.001

Data are expressed as mean ± SD or median (interquartile range) as appropriate. Conversion factors for units are ×0.01129 for triglycerides (mg/dl to mmol/L), ×0.02586 for total and HDL cholesterol (mg/dl to mmol/L), ×0.05551 for glucose (mg/dl to mmol/L), and ×6 for insulin (μU/ml to pmol/L).

MDS and All-Cause Mortality in Individuals with CKD

During the follow-up period (median=9.9 years, IQR=8.7–11.0 years, 4648 person-years at risk), 168 (33%) CKD men died. In the subpopulation of adequate reporters, the corresponding figures were median=10.1 years, IQR=8.9–11.0 years (2308 person-years at risk), and 79 (32%) deaths.

Kaplan–Meier curves illustrating the cumulative proportions of surviving CKD individuals across different MDS adherence groups are presented in Figure 2. Survival rates were worse for low-adherent individuals than for individuals in the other two groups. In proportional hazards regression shown in Table 4, crude models revealed that higher MDS was associated with a lower total mortality risk. The relationship was partly alleviated and became nonsignificant after adjustment of potential confounders. On a continuous scale, every two-point increase in MDS was associated with lower total mortality risk (adjusted HR [95% CI]=0.82 [0.69 to 1.01]). CKD individuals with high and medium MDS adherence presented 23% and 25% lower risk of all-cause mortality, respectively (adjusted HR [95% CI]=0.75 [0.52 to 1.06] and 0.77 [0.44 to 1.36], respectively, P for trend=0.10), compared with CKD individuals showing low MDS adherence. After exclusion of nonadequate dietary reporters, a higher MDS strongly and independently predicted better survival rates.

Figure 2.

Kaplan–Meier survival curves in individuals with CKD (left panel) and the subpopulation of adequate reporters of dietary intake (right panel) stratified by their degree of adherence to a Mediterranean diet.

Table 4.

Proportional hazards regression models predicting for all-cause mortality in men with CKD

Exposure	Crude Hazard Ratio (95% CI)	Adjusted Hazard Ratio (95% CI)
All CKD men (n=506)
Two-point increase in Mediterranean diet score	0.77 (0.63 to 0.93)	0.82 (0.69 to 1.01)
Adherence groups
Low (Mediterranean diet score=1–2)	Reference	Reference
Medium (Mediterranean diet score=3–5)	0.69 (0.50 to 0.97)	0.75 (0.52 to 1.06)
High (Mediterranean diet score=6–8)	0.72 (0.42 to 1.24)	0.77 (0.44 to 1.36)
P for trend	0.03	0.10
Adequate reporters (n=250)
Two-point increase in Mediterranean diet score	0.73 (0.54 to 0.99)	0.66 (0.48 to 0.90)
Adherence groups
Low (Mediterranean diet score=1–2)	Reference	Reference
Medium (Mediterranean diet score=3–5)	0.56 (0.35 to 0.92)	0.48 (0.29 to 0.80)
High (Mediterranean diet score=6–8)	0.54 (0.24 to 1.26)	0.42 (0.18 to 1.02)
P for trend	0.02	0.005

Data are presented as hazard ratio (95% confidence interval [95% CI]). Covariance in the adjusted models includes body mass index, physical activity, smoking status, education, hypertension, hyperlipidemia, and diabetes.

Discussion

This study shows that the adoption of a Mediterranean-like diet was associated with better kidney function in a population-based cohort of elderly men. In addition, it also shows that a low adherence to a Mediterranean diet linked to worse survival rates among individuals with CKD. Our observed association between low MDS adherence and the presence of CKD expands a recent study in a younger cohort (24). Consistent with this concept, high polyunsaturated fatty acid intake, especially of marine origin, has been considered as renally protective (35,36), whereas high saturated fatty acid intake may deteriorate kidney function (37). Alkali-inducing fruits and vegetables, important components of a Mediterranean diet, may improve metabolic acidosis and attenuate kidney injury (22,23,38). The aforementioned beneficial effects of this dietary pattern on the cardiometabolic profile as well as albuminuria (25) are also believed to contribute at least subclinically, because they play an essential role in the pathogenesis and progression of renal diseases (39–41). Of note, a recent interventional study showed improvement of renal function after 1 year of following a Mediterranean-like dietary pattern in elderly individuals at high risk of coronary heart disease (26). This change in eGFR, however, did not significantly differ from the change in eGFR of the control group (26). Díaz-López et al. (26) speculate that study limitations, such as short follow-up duration and highly specific inclusion criteria, may prevent the putative effect of this dietary pattern. Clinical trials taking these aspects into account are warranted.

A novel and important finding in our study is that a low adherence to this dietary pattern may relate to worse survival among individuals with CKD. Such results are in line with recent pooled analysis of prospective community studies relating this dietary pattern with a reduced risk of mortality (7). Our study, thus, supports the mounting observational (8–14) and interventional (15–19) evidence on the benefits of this healthy diet and agrees with current dietary recommendations for the general population (42). Despite this mortality association in our analysis and contrary to our hypothesis, we did not observe clear associations with most potential explanatory risk factors, such as obesity, BP, lipoproteins, glucose, insulin, or inflammation. We speculate that the homogeneity of included individuals in this cohort may mask such findings. As reviewed by Babio et al. (20), the benefits of the Mediterranean diet may be mainly accounted for by individual nutrients, such as high fiber intake and high polyunsaturated fatty acid but low saturated fatty acid intake. In addition and in agreement with previous studies (22,23,33) as well as our analysis, a Mediterranean diet may provide relatively low contents of salt, phosphate, and acid load, which may further support the survival association observed in the CKD population (43). Previous studies in nondialyzed CKD, dialysis, and renal transplant patients associate some of these individual nutrients with important risk factors, such as systemic inflammation, oxidative stress, dyslipidemia, metabolic acidosis control, and incidence of metabolic syndrome (2–4,22,44–47). Nevertheless, it is also possible that individuals with low adherence to a Mediterranean diet are less compliant with other lifestyle and medical recommendations (residual confounding). Because the recent randomized controlled study of Estruch et al. (19) showed that a Mediterranean diet supplemented with olive oil/nuts reduces cardiovascular events in individuals at high cardiovascular risk, well designed interventional studies of this type in CKD patients would be of great interest.

There are several advantages and limitations in our study. Its prospective nature, the use of the well established MDS scoring method (9), and the mortality registry data with high reliability and no loss to follow-up are strengths that improve the validity of our results. The 7-day dietary record is the most preferred dietary assessment method, and the use of Goldberg cutoffs to control for the possibility of over-/under-reporting represents another strength (48). A corresponding limitation is, however, that approximately one half of the participants failed to give an adequate dietary recall, and this result may limit representativeness. Another limitation is that data on the use of supplements were not available, which could underestimate our findings. Moreover, we did not directly measure creatinine clearance rate, and we based our estimation of GFR on serum cystatin C concentrations. When GFR was estimated by serum creatinine and the Chronic Kidney Disease Epidemiology Collaboration equation, the magnitude of the findings was similar (data not shown), although no significance was observed. We speculate that statistical power was insufficient, because serum creatinine is less adequate for assessing GFR than serum cystatin C (49). Despite different statistical strategies as well as sensitivity analyses to confirm the robustness of findings, some of the P values did not reach statistical significance. Because P values were not adjusted for multiple testing, they have to be considered as descriptive. Finally, although the inclusion of men with similar age and geographical distribution reduced important confounding, it implies that our results only apply to elderly men with moderate CKD, which constitutes a very homogenous cohort that may prevent the identification of intermediates in this association.

In conclusion, our study shows an independent relationship between adherence to a Mediterranean diet and renal function in a population-based cohort of elderly men. Greater adherence to such a dietary pattern directly predicted 10-year survival in those individuals with manifest CKD. These findings altogether suggest that a dietary pattern characterized by a high consumption of vegetables, legumes, fruits, nuts, cereals, and olive oil and a low consumption of saturated fats, meat, and poultry may be protective for kidney function and improve clinical outcomes in individuals with CKD. Even if dietary modifications can be challenging (50), our observations may entice the initiation of interventional studies addressing this hypothesis.

Disclosures

B.L. is employed by Baxter Healthcare Corporation. None of the other authors declare any conflict of interest. The sponsors had no role in the design and conduct of the study.