Link between nutritional risk and depression in older adults

The official NHANES website (https://www.cdc.gov/nchs/nhanes/index.htm↗) provided the information used in this research. The National Health and Nutrition Examination Study (NHANES), a thorough study carried out by the Centers for Disease Control and Prevention (CDCP) and the National Center for Health Statistics (NCHS), seeks to provide nationally representative statistics on the civilian population of the United States. All participants provided informed permission before their interviews and tests, and the NCHS Ethics Review Board approved the data-collecting procedure for the NHANES. In this investigation, a dataset was created by extracting responses from the NHANES publicly accessible data files from 2011 to 2014. A total of 19,931 individuals participated in two consecutive NHANES survey cycles. This study excluded 16,299 individuals who were under the age of 60. Subsequently, individuals who did not have calculated GNRI-related data (n = 467) and those who were missing data on depressive symptoms (n = 219) were also excluded. The final study population consisted of 2,946 individuals. Figure 1 illustrates a flowchart that describes the criteria used for patient selection.

The GNRI was calculated based on several factors, including the subject's height (m), weight (kg), optimum weight (kg), and serum albumin (g/L) [7, 10]. The following is the GNRI calculating formula: GNRI = 1.489 × Albumin (g/L) + 41.7 × [body weight/ideal body weight], ideal body mass = 22 × Height (m) × Height (m). The weight/ideal weight ratio is set to 1 when the actual weight is greater than the desired weight. The patients included in this study were divided into two groups: High-GNRI group (GNRI ≥ 98) and Low-GNRI group (GNRI < 98) [9].

Depressive symptoms were assessed using the 9-item Patient Health Questionnaire (PHQ-9) Depression Scale, which consists of 9 questions based on the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) Depressive Symptoms [11]. Respondents were asked to rate the severity of their symptoms from 0 (not at all) to 3 (almost every day). The total score for the nine items ranged from 0 to 27 [12]. In this study, individuals with a total score greater than 9 on the PHQ-9 were considered to have clinically relevant depressive disorder (CRD), and a total score of 9 or less was considered non-clinical [13]. In the study conducted by Kurt Kroenke et al., it was found that using the mental health professional (MHP) reinterview as the criterion standard, a PHQ-9 score ≥ 10 demonstrated a sensitivity of 88% and a specificity of 88% for major depression [14].

Interviewers ascertained a participant's birthdate and other demographic data by self-report through the initial screening questionnaire. Age in years was a continuous variable. Gender was reported as male or female. Race was classified as Mexican American, other Hispanic, non-Hispanic white, non-Hispanic black, or other. Body mass index (BMI) was calculated by the following formula: BMI = weight (kg)/height2 (m2). The Systemic Immune-Inflammation Index (SII) level was determined by multiplying the platelet count by the neutrophil count/lymphocyte count [4]. Education attainment was assessed through the initial screening questionnaire and categorized as less than 9th grade, 9th-11th grade, high school, some college or associate degree, college or above. Marital status was categorized as married, widowed, divorced, separated, never married, or living with a partner. Household income is classified based on the poverty income ratio (PIR) into three categories: PIR < 1.5,1.5 ≤ PIR < 3.5, or PIR ≥ 3.5. Smoking status was established by responses to: "Do you now smoke cigarettes?" and "Do you now smoke cigarettes?". The classification of one's smoking status was based on the duration and number of cigarettes they have consumed throughout their lifetime. There are three distinct categories of individuals who engage in smoking: "never," "former," and "current." (never, smoked less than 100 cigarettes; former, smoked more than 100 cigarettes in life and smoke not at all now; now, smoked more than 100 cigarettes in life and smoke some days or every day) [15].

The study categorized participants based on their average daily alcohol consumption into three groups: 'mild,' 'moderate,' and 'heavy.' Those in the 'moderate' category consumed 2 drinks per day for women and 3 drinks per day for men, while individuals in the 'heavy' category consumed at least 3 drinks per day for women and at least 4 drinks per day for men. Participants who did not fall into either of these categories were considered mild drinkers. The study aimed to determine the occurrence of cardiovascular disease (CVD) outcomes by self-reported diagnoses of five major CVD events, which involved congestive heart failure (CHF), coronary heart disease (CHD), angina pectoris, heart attack, and stroke. Participants who answered yes to the question,' Have you ever been told by a physician that you had CHD/CHF/angina/a heart attack or a stroke?' were identified as having CVD. It is noteworthy that relying on self-reported measures can introduce recall bias, which can affect data interpretation [16]. Hypertension is diagnosed based on four criteria, and individuals who meet any one of these criteria are categorized as hypertensive patients. The diagnostic criteria consist of self-reported hypertension history, the utilization of antihypertensive medication, a systolic blood pressure (SBP) ≥ 140mmHg, or a diastolic blood pressure (DBP) ≥ 90mmHg [17]. The diagnosis of hyperlipidemia refers to 6 criteria, and participants who meet any one of these criteria are defined as hyperlipidemia patients. The criteria for diagnosing hyperlipidemia are as follows: (1) total cholesterol ≥ 200 mg/dL, (2) triglycerides ≥ 150 mg/dL, (3) HDL cholesterol < 40 mg/dL for males and < 50 mg/dL for females, (4) LDL ≥ 130 mg/dL, (5) participants using lipid-lowering medications, or (6) patients self-reporting a diagnosis of hyperlipidemia [18]. Participants who meet the following two criteria are defined as diabetes patients. Diabetes diagnosis criteria were (1) participants using antidiabetic medications; or (2) patients self-reporting a diabetes diagnosis.

The study employed the R software (version 4.2.3) developed by the R project (http://www.r-project.org↗) for data analysis. The specific R packages used in this study are as follows: "doBy" package, which provides functionality for grouping data frames by factor variables; "plotrix" package, used to create various charts; "stringi" and "stringr" packages, both used for manipulating and operating on strings; "rms" package, used for fitting predictive models; "car" package, used for testing various regression models; "kableExtra" package, used for creating and customizing aesthetically appealing and readable tables, and "readxl package, used for reading Microsoft Excel files in R (.xls and.xlsx formats). To address the complexities inherent in survey designs, survey non-response, and post-stratification adjustments, NHANES created weights using the Mobile Examination Center (MEC) sample weights and the relevant home-examined sample design variables (strata, primary sampling unit). These weights were utilized to ensure representative results, aligning with the demographic characteristics of the entire United States population. Consequently, all analyses presented in this study were conducted by appropriately weighting the samples in accordance with NHANES analytical guidelines [19].

Initially, the sample was divided into two groups based on the depression threshold: depressed and non-depressed. Continuous variables are expressed as weighted means ± standard deviations. P value was calculated by weighted linear regression model. Categorical variables were expressed as numbers and percentages and compared using the chi-square test. Multivariable logistic regression analysis was used to study the association between GNRI and depression. Furthermore, subgroup analysis and interaction tests were performed according to age, gender, race, education, marital status, BMI, smoking, hypertension, cardiovascular disease, hyperlipidemia, and diabetes. Ultimately, through the construction of a restricted cubic spline plot, we aim to ascertain whether there exists a non-linear correlation between the GNRI and depression prevalence.

Table 1 presents the weighted characteristics of participants divided into two groups based on their depression status. A total of 2946 participants took part in the study, out of which 280 individuals were identified as adults with suggested depression. In the primary analysis, age (69.453 ± 6.72 vs. 67.945 ± 6.59, P = 0.00136) emerged as a significant factor influencing depression. The average age of participants with depression was lower compared to those without depression, suggesting that individuals with depression were younger. Body mass index (BMI), a measure of obesity, was found to be significantly different between the two groups (28.834 ± 6.11 vs. 31.241 ± 7.69, P < 0.00001). The data showed that the average BMI of the depression group was higher than that of the non-depression group, indicating a tendency for higher obesity levels among depressed patients. Moreover, the depressed group displayed a higher proportion of individuals with low education levels, while the non-depressed group was primarily composed of individuals with a high school education or higher. This suggests that education level may be associated with the risk of depression (P < 0.00001). Additionally, the depressed group had a higher proportion of individuals who were divorced or widowed, while the non-depressed group had a higher proportion of individuals who were married (P < 0.00001). Changes in marital status may consequently influence mental health.

Regarding financial factors, the depressed group exhibited a higher proportion of individuals with a lower household income ratio (PIR), while the no-depression group displayed the opposite pattern. Financial difficulties may contribute to psychological stress and the onset of depression (P < 0.00001). Analysis of disease factors and unhealthy habits indicated that the number of smokers among individuals with depression was higher compared to those without depression (16.27% vs. 10.84%, P = 0.03358). Similarly, the prevalence of cardiovascular disease (CVD) among the depressed group was higher than the non-depressed group (35.5% vs. 20.78%, P < 0.00001). Additionally, the prevalence of hypertension was higher in the depressed group (72.96% vs. 66.17%, P = 0.04001), as well as the presence of diabetes (30.03% vs. 19.94%, P = 0.00038). These findings demonstrate significant differences in baseline characteristics among all variables except alcohol consumption and hyperlipidemia status (P < 0.05).

The GNRI in non-depressed individuals exhibited a higher value compared to that in depressed individuals. This disparity was found to be statistically significant (P = 3 × 10^ (-7)), and the two groups showed an overall tendency towards tighter clustering with minimal dispersion (Fig. 2).

Multivariable logistic regression analysis (Table 2) was used to investigate the association between GNRI and depression scores (DPs). In the unadjusted model, the results showed a significant negative association between GNRI levels and DPs in the individual without depression (OR -0.062, P < 0.00001). However, no significant negative correlation between GNRI levels and DPs was observed in individuals with depression (OR = -0.021, P = 0.58193). There existed a significant negative correlation between GNRI levels and DPs in the overall sample (OR -0.056, P < 0.00001). In the adjusted model I, age, gender, and race were considered as adjustment variables. Adjusted model II involved additional adjustment for BMI, education, marital status, PIR, smoking, CVD, diabetes, and hypertension, in addition to age, sex, and race. The results demonstrated consistency with the unadjusted model. Consequently, a preliminary inference can be drawn that there exists an inverse association between GNRI level and depression scores, indicating that individuals with lower GNRI levels are more susceptible to depression and GNRI can act as predictive factors for depression. However, in depressed patients, GNRI cannot be utilized to predict the severity of depression due to the absence of a significant association between depression scores and GNRI.

Table 3 presents the results of a logistic regression analysis examining the relationship between the high nutritional risk index (High-GNRI) and the prevalence of depression, which is measured as a dichotomous variable. The variable adjustments of the three models given in the table are consistent with Table 2. The analysis of the data revealed varying degrees of negative correlations between high GNRI levels and depression in the three models, using the Low Nutritional Risk Index (Low GNRI) as a reference. The OR values and P values indicated that when GNRI increases by one unit, the risk of depression will be reduced by 0.447 times (P < 0.00001), 0.436 times (P < 0.00001), and 0.564 times (P = 0.00693) in the three models. In summary, a high nutritional risk index may serve as a protective factor against the development of depression.

Through stratified analysis, each sample in the study was categorized and analyzed independently to ascertain the influence of confounding factors and specific population. The results presented in Table 4 (with Fig. 3 depicting the forest plot post data visualization) reveal that age has been segregated into two groups, ≤ 70 and > 70, ≤80, with corresponding OR and P values of (0.466 and 0.003) and (0.359 and 0.0002), respectively. Further, the data suggest that certain demographic groups, including women (P < 0.0001), non-Hispanic blacks (P = 0.0003), those with middle BMI (P = 0.0005), some college or AA degree (P = 0.0003), married (P = 0.0001), PIR between 1.50 and 3.49 (P = 0.0002), former smokers (P = 0.0002), those without CVD (P < 0.0001), those with hypertension (P < 0.0001) and diabetes (P = 0.0027), could be a significantly relevant population that is particularly vulnerable to the impact of GNRI, leading to a decreased occurrence of depression.

The restricted spline regression model has revealed a nonlinear, negative correlation between the GNRI and the prevalence of depression in the elderly population (nonlinear P = 0.013). Moreover, after controlling for confounding factors (age, sex, and race), GNRI was found to have an inverse association with the risk of depression (nonlinear P = 0.007; Fig. 4). As GNRI increases, the prevalence of depression gradually decreases, provided that GNRI is less than 104.17814.