Association between the Prognostic Nutritional Index and Dietary Intake in Community-Dwelling Older Adults with Heart Failure: Findings from NHANES III

The objective of this study was to examine the association between nutritional status and dietary intake in community-dwelling older adults with heart failure (HF). A cross-sectional analysis of NHANES III data was conducted. The analytic sample was comprised of n = 445 individuals aged 50+ years with congestive HF (54.4% male, 22.9% non-Hispanic Black, 43.8% low-income). Nutritional status was measured using the Prognostic Nutritional Index (PNI). Participants were classified by PNI quintiles with lower PNI scores indicating lower nutritional status. Participants in quintile 5 showed significantly greater intakes of energy, protein, vegetables, magnesium, zinc, copper, potassium, red meat, saturated fat, and sodium. In multivariate analyses, increased intake of red meat (β = 0.253, p = 0.040) and vegetables (β = 0.255, p = 0.038) was associated with significantly better nutritional status. In the absence of comprehensive nutritional guidance for HF patients, it appears that small increases in energy, protein (red meat), and vegetable consumption are associated with improved nutritional status.


Introduction
Heart failure (HF) is a serious public health concern in the United States. More than six million Americans currently live with HF [1], and it is estimated that over eight million Americans will be diagnosed with HF by 2030 [2]. In older adults aged 65+ years, HF incidence is as high as 21 per 1000 [3], and prevalence rates increase from 6% in adults 60-79 years of age to 14% in adults 80+ years of age [1]. The progressive nature of HF deterioration causes functional limitations such as orthopnea, paroxysmal nocturnal dyspnea, fatigue and/or lethargy [4], all of which significantly reduce patient quality of life [5]. In addition to the human cost, total medical costs for HF were estimated to be $30.7 billion in 2012, about 80% of which were related to hospitalizations [2]. Older Americans are triglyceride, and other clinical marker levels have been observed and linked to objectively assessed, validated nutritional risk in community-dwelling older adults with HF, such as the Mini Nutritional Assessment (MNA) [33].
Given the strong impact of poor nutritional status on HF prognosis and outcomes, and the potential for low-cost, dietary intervention for the early prevention of poor outcomes in community-dwelling older adults with HF, it is essential to better understand the relationship between nutritional status and dietary intake in this population. The objective of this study was, therefore, to examine the association between nutritional status and dietary intake in community-dwelling older adults with HF.

Study Design and Population
This study represents a secondary data analysis of survey data from the National Health and Nutrition Examination Survey (NHANES) III using a cross-sectional design. NHANES III was conducted by the National Center for Health Statistics between 1988 and 1994, and uses a multistate, stratified sampling design to include a nationally representative sample of non-institutionalized community-dwelling U.S. adults. The survey design and sampling methods have been described more in detail elsewhere [34].

Analytic Sample
The analytic sample included older adults 50+ years of age who self-reported "yes" to a query about HF ("Has a doctor ever told you that you had congestive heart failure?"). A total of n = 546 older adults with self-reported congestive HF completed NHANES interviews and Mobile Examination Center (MEC) visits. Individuals in the "other" racial/ethnic groups (n = 20) were excluded following the recommendation from the NHANES III Analytic and Reporting Guidelines [35]. Subjects with missing data on dietary intake (n = 23) and clinical parameters (serum albumin and lymphocyte levels) used for nutritional assessment (n = 54), and implausible reported energy intakes (<1st and >99th percentiles of energy intake per day, n = 31) were excluded from the analysis. The final sample included n = 445 older adults with self-reported congestive HF (54.4% male, 22.9% non-Hispanic Black, 59.8% currently married, 43.8% low-income).

Nutritional Status
The Prognostic Nutritional Index assesses nutritional status based on clinical marker values by using the following Equation (1) [29][30][31][32]: Relatively lower PNI scores indicate poorer nutritional status and were associated with an increased risk of rehospitalization and mortality in previous studies [29][30][31][32]. The specimen collection and processing procedures have been described elsewhere [36]. Levels of serum albumin may indicate protein reserve levels [37,38], and low total lymphocyte count may be associated with low immune function caused by undernutrition [39][40][41]. Participants in the analytic sample were classified by PNI quintiles, with participants in quintile 1 (lowest PNI scores) indicating a relatively higher risk of malnutrition whereas those in quintile 5 (highest PNI scores) indicating a relatively lower risk of malnutrition.

Dietary and Food Group Intake
Dietary intake was assessed via a single 24-hour dietary recall, using an automated, microcomputer-based dietary interview [42]. The food composition data for NHANES III was based on the U.S. Department of Agriculture (USDA) Survey Nutrient Database and the University of Minnesota's Nutrition Coordinating Center.
Food group intake was quantified using the Pyramid Servings Database (PSDB) for NHANES III, developed by the National Cancer Institute [43]. The PSDB facilitates serving size calculations based on the USDA Food Guide Pyramid's food groups and includes 4542 different food item codes that correspond to all foods reported in the 24-hour dietary recalls. Major food groups in the USDA Food Guide Pyramid recommendations include dairy, fruit, vegetables, grains, and meat/other proteins [44]. Food group intake of fruit, vegetables, grains, and dairy was described using serving sizes, and intake of meat and other proteins was described using ounces or ounce lean meat equivalents.

Covariates
Socio-demographic and economic characteristics collected included age, gender, race/ethnicity (non-Hispanic White, non-Hispanic Black, and Mexican-American), education (<high school, high school, and >high school), working status, marital status, and poverty income ratio (PIR) (<1.300, 1.301-3.500, and ≥3.501). Health-related covariates included body mass index (BMI; kg/m 2 ), systolic and diastolic blood pressure (mmHg), number of comorbidities (0-2, 3-4, 5+), number of hospitalizations in the past 12 months (never, 1, 2+), self-reported length of time since HF diagnosis (0-2, 3-5, 6-10, 11+ years), and clinical markers not included in the PNI score (levels of total cholesterol, triglycerides, HDL-cholesterol, and C-reactive protein (CRP)). In addition, we considered self-reported HF medication use, including diuretics (e.g., loop, thiazide, potassium-sparing diuretics), angiotensin-converting-enzyme inhibitors (ACEI), beta-blockers, digoxin, and nitrates (e.g., isosorbide mononitrate). Smoking status was defined by two questions ("Have you smoked at least 100 cigarettes during your entire life?" and "Do you smoke cigarettes now?"), and grouped into three categories (never, former, current smoker). Habitual drinking status was defined by two questions ("In your entire life, have you had at least 12 drinks of any kind of alcoholic beverage?" and "In the past 12 months did you have at least 12 drinks of any kind of alcoholic beverage?"), and if both answers were affirmative, a follow-up question asked about the number of drinks/ day on the days of alcohol consumption. According to the definition from the Substance Abuse and Mental Health Services Administration, if individuals drank 5+ drinks (men) or 4+ drinks (women) in one sitting, they were categorized as binge drinkers [45]. Physical activity (PA) levels were classified as active, insufficiently active, and inactive [46]. Individuals who reported 5+ episodes of moderate-intensity activity or 3+ episodes per week of vigorous-intensity activity in the past month were classified as active [47]. Those who reported some engagement of physical activity but did not meet the recommended levels in the previous month were classified as insufficiently active.

Statistical Analysis
All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA). The survey weights, which account for the complex survey sampling design of NHANES III, were not applied to the analyses because of the small size of the HF target population (e.g., larger variance). Participant characteristics were summarized using means and standard deviations (SD) for continuous variables and percentages for categorical variables. Univariate analysis of differences in sample characteristics by nutritional status was examined by t-tests or Mann-Whitney tests for continuous variables and by chi-squared tests for categorical variables. Linear regression analyses were conducted to generate beta coefficients and 95% confidence intervals (CIs) to quantify the association between nutritional status and sample characteristics while controlling for potential confounders. Data points influencing the estimated slope coefficient were excluded if Cook's distance values were greater than 1.0. Multivariate analyses were used to assess the association between nutritional status and food group intake while adjusting for potentially confounding factors, as informed by univariate analysis using simple linear regressions. p values of < 0.05 were considered statistically significant.

Discussion
To our knowledge, this is the first study to examine the relationship between nutritional status and dietary intake in community-dwelling older adults with congestive HF. Better nutritional status, as measured by albumin and lymphocyte levels, was associated with significantly higher intakes of energy, protein (specifically red meat), vegetables, and important macro-and micronutrients than in those with lower nutritional status in this population.
In hospitalized patients with HF, poor nutritional status, as measured by clinical markers using the PNI, has been associated with higher levels of prognostic cardiometabolic markers, including brain naturetic peptide (BNP) and N-terminal pro BNP, indicating poor HF status and significantly longer hospital stays, cardiovascular events, and mortality [29][30][31][32]48]. In community-dwelling older adults, in whom regular blood collection is uncommon and inconvenient, subjective measures of nutritional status, including the MNA, have been validated and linked to adverse health outcomes, higher healthcare use, and mortality [49][50][51]. While the MNA measures nutritional risk due to a decline in food intake, unexpected weight loss, mobility, stress and psychological problems, and low BMI, MNA-assessed malnutrition has also been associated with significantly lower levels of albumin, total cholesterol, triglycerides, and higher levels of CRP. Given the links between HF pathophysiology, clinical markers of nutritional risk, and HF progression, application of these markers to community-dwelling, older HF patients holds strong potential for the identification and targeting of at-risk older adults with HF even before complications arise and patients are hospitalized.
Our sample of community-dwelling HF patients showed PNI scores ranging from 32.2 to 73.0, with participants in quintile 1 showing a mean PNI score of 43.8 ± 3.3 (mean ± SD), and those in quintile 5 showing a mean PNI score of 60.0 ± 3.5. While no standards have been established to evaluate the relationship of PNI scores with negative outcomes in community-dwelling HF patients, diverse thresholds have been suggested for hospitalized HF patients, including pre-determined scores (e.g., normal: PNI > 38; moderate risk of malnutrition: 35 < PNI < 38; and severe risk of malnutrition: PNI < 35) [30,31], receiver operating characteristic curves [29], and PNI tertiles [32]. Cheng and colleagues reported that acute HF patients at nutritional risk, as defined by PNI scores ≤44.8, had increased in-hospital, 90-day, and 4-year mortality rates [32].
On average, we found better nutritional status scores in our sample compared to average scores of hospitalized HF patients in previous studies. This is in line with the expectation of a more stable, community-dwelling HF population. In absolute terms, mean values for albumin (3.6 ± 0.4 g/dL; normal range: 3.5-5.5 g/dL) and lymphocyte count (1638 ± 535 in 1 µL; normal range: 1000-4800 in 1 µL) were at the lower end of the range of normal values. In relative terms, we found a proportion of patients in PNI quintile 1 to be at nutritional risk according to the threshold established by Cheng et al. in hospitalized HF patients. In addition, we found patients in PNI quintile 1 significantly more likely than patients in PNI quintile 5 to be hospitalized one or more times within the past year, which serves as a proxy for disease severity and is associated with increased mortality risk in HF patients [52]. Although only a small proportion of our sample fit the pre-determined criterion of nutritional risk, identifying nutritional characteristics of low nutritional status above thresholds associated with detrimental outcomes is likely important for the prevention of such events.
Applicability of dietary guidance for healthy individuals to HF patients is an ongoing debate; fueled in particular by conflicting evidence regarding the relationships between sodium intake and negative HF outcomes [9], discussion about adequate energy intake levels for HF [53], and the obesity paradox [54]. In the absence of comprehensive nutritional guidance for community-dwelling older adults with HF, more research is needed to determine how to best address nutritional deficiencies within complex HF pathophysiology and adverse effects of essential pharmacotherapy. In our sample of community-dwelling HF patients, total daily energy intake ranged from 1356 kcal (PNI quintile 1) to 1629 kcal (PNI quintile 5), and while adequacy of such levels based on predictive formulas was not assessed in this study, they are likely inadequate for a majority of participants in a sample with over 50% males, and as described in the previous literature [14,55]. Energy intake differed significantly by nutritional status, and this relationship was mainly driven by a difference in protein intake (58.6 ± 33.9 g vs. 70.1 ± 41.0 g) which is confirmed by results from previous studies [14,55]. Moreover, we found higher mean intakes of vegetables and red meat of~1 serving and~1 oz, respectively, to be significantly and independently associated with better nutritional status. Through these increases in intake, participants met food group intake recommendations for healthy adults at the time (3 servings of vegetables and 5 oz of protein, respectively) [44]. Therefore, our findings support the translation of food group recommendations for healthy adults to HF patients. There is growing evidence that protein intake, in particular, should be a focus of dietary HF management. Recent work from the BIOSTA-CHF trial in Europe showed that HF patients with higher intakes of protein (70+ g/day vs. ≤40 g/day) had significantly lower rates of mortality (31 vs. 18%; p < 0.001) [56]. These levels compare to our findings of the relationship between nutritional status and protein intake, and are supported by the significant relationship between dietary protein and serum albumin levels [57]. Furthermore, nutritional risk markers based on albumin levels have negative outcome predictive properties [57]. Interestingly, our findings suggest that heightened red meat intake (+1 oz/day), despite associated higher intake levels of saturated fat and sodium, as found in this study, may be protective of malnutrition. This finding suggests that in HF patients, particularly in those with comorbid hypertension and hyperlipidemia, long-term vs. short-term risks may need to be carefully evaluated. In this regard, it is likely that nutritional guidance for the prevention of cardiovascular events may not translate to HF patients at risk of malnutrition and cardiac cachexia. While past cardiovascular disease prevention research and Dietary Guidelines for Americans have recommended reduced intake of red meat due to concerns about cholesterol, red meat is a high quality source of essential amino acids and micronutrients, including thiamine, riboflavin, niacin, pyridoxine, vitamin B 12 , zinc, iron, selenium, copper, and magnesium, all of which are important for the nutritional well-being of older adults [58]. Grossniklaus and colleagues demonstrated that HF patients with higher consumption of protein, including red meat, had higher intakes of these micronutrients [59]. As a good source of magnesium, red meat may counter the loss of magnesium due to use of loop and thiazide diuretics, and may, therefore, counter symptoms of fatigue and ventricular ectopy, and improve HF prognosis [60,61]. Red meat consumption in older adults with HF may further increase muscle protein anabolism and reduce the progressive loss of muscle mass [62]. Therefore, adequate protein intake likely plays a crucial role in addressing the disease state-induced hypercatabolic state and in the prevention of cardiac cachexia.
Increased consumption of vegetables has been associated with a lower risk of chronic diseases and mortality [63,64]. Vegetables are a well-known source of phytochemicals, antioxidants, vitamins, and minerals, which may play an important role in the prevention of malnutrition in HF patients [65]. The hyperadrenergic state, resulting from the compensation of decreased cardiac output experienced in HF [66], increases plasma free fatty acid (FFA) levels, which in turn, inhibit glycolysis and glucose uptake by the heart and skeletal muscle [67]. Increased plasma glucose promotes the formation of reactive oxygen species (ROS) which inhibit key enzymes for glycolysis [67]. In contrast, diets rich in phytochemicals and antioxidants may reduce levels of ROS as well as reduce levels of inflammatory markers, such as CRP [68], which we found elevated in those with lowest vegetable intake levels. Furthermore, there is evidence that increased lycopene intake from tomato products among HF patients is associated with longer cardiac event-free survival [69]. Our findings, therefore, support that vegetable consumption is important for older HF patients in the prevention of poor nutritional status.
Lastly, we found cigarette smoking to be positively associated with higher PNI values which is a controversial finding. Cigarette smoking is a well-established cardiovascular risk factor, and smoking status is a strong independent predictor of hospitalization and mortality in HF patients [70]. On the contrary, smoking cessation among HF patients significantly reduced the risks of hospitalization and mortality [70]. Despite these associations, clinical literature has documented a relationship between smoking status and elevated lymphocyte count [71] and higher energy intake [72]. Such evidence is in line with our findings and suggests smoking status as a potential confounding factor between dietary intake and PNI which we accounted for in our analyses.
No study is without limitations. First, NHANES III is a cross-sectional study and causal inferences cannot be drawn. It is important to acknowledge that our findings reflect associations and may not be interpreted as causal relationships. In this context, it is possible that poor nutritional status, as shown by low PNI scores, may reflect more pronounced disease severity which may or may not be prevented by improved dietary intake. Furthermore, all self-reported information is subject to recall bias. Dietary information was collected by a single 24-hour dietary recall and may not represent habitual dietary intake of study participants. In addition, this study did not apply survey weights that adjust for biases arising from the complex sampling design used in NHANES III; therefore, the interpretation of the results should account for potential selection bias. Lastly, this study was limited in the adjustment for potentially confounding factors which may influence the association between dietary intake and clinically assessed nutritional status. For example, information on NYHA functional classification and other clinical characteristics of the HF disease state (e.g., reduced or preserved ejection fraction of left ventricular) were not available from the dataset. Moreover, we cannot exclude the possibility that any unmeasured factors may have led to improved survival in those with significantly better nutritional status, as we observed that individuals with better nutritional status had been diagnosed with HF for a longer amount of time. Nevertheless, this study provided valuable information for clinicians and patients on the relationship between nutritional status and dietary intake of energy and food groups. In addition, secondary data analysis of the NHANES III dataset had the advantage of using clinical and self-reported variables not available in other datasets (e.g., serum albumin and lymphocyte levels).

Conclusions
In the absence of comprehensive nutritional guidance for community-dwelling older adults with HF, it appears that small increases in energy, protein (red meat), and vegetable consumption are associated with an improved nutritional status which may prevent adverse events in this population. Controlled intervention studies are needed to determine if nutritional risk in HF patients can be minimized by the intake of specific dietary patterns. Funding: This work was funded by the National Center for Advancing Translational Sciences of the National Institutes of Health under Award Numbers UL1TR002378 and KLTR002381. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Conflicts of Interest:
The authors declare no conflict of interest.