This retrospective cohort study was based on the National Health and Nutrition Examination Survey (NHANES) - a large national survey of the United States civilian population conducted by the Centers for Disease Control and Prevention's National Center for Health Statistics. The NHANES data were collected through a complex multi-stage probabilistic design that identifies strata based on geography and a certain percentage of the minority population (19). Furthermore, this study was supported by the National Center for Health Statistics Research Ethics Review Board, you can find it on this website: https://www.cdc.gov/nchs/nhanes/irba98.html. And the ethics approval numbers Protocol #98-12, Protocol #2005-06, Continuation of Protocol#2005-06, Protocol #2011-17and Continuation of Protocol#2011-2017. All participants provided written informed consent.

Data for the analysis were taken from NHANES 1999–2016 with a total of 91,964 participants. In the present study, we included people aged ≥60 years and excluded subjects without serum albumin, height, weight, follow-up data, and without hypertension disease. After applying the above criteria, 10,037 participants were finally included (Figure 1).

The Geriatric Nutrition Risk Index was calculated as GNRI = [1.489 × serum albumin (g/L)] + (41.7 × weight (kg)/ideal weight (kg)) The ideal weight was calculated by the following Lorenz equation: 22 × square of height. If the patient's weight exceeded the ideal weight, the ratio of weight to ideal weight was set to 1 (14). Classification of patients was based on the following thresholds: moderate to severe risk of malnutrition: <92; low risk: 92 to 98; no risk: >98 (17).

A history of hypertension and antihypertensive medication was collected by questionnaire. Blood pressure was measured by a trained physician using a mercury sphygmomanometer with an appropriately sized cuff. Blood pressure measurements were performed three times and the average of the three measurements was defined as systolic blood pressure (SBP) and diastolic blood pressure (DBP). Hypertension was defined as having a self-reported history of hypertension or use of antihypertensive medication or SBP ≥ 140 mmHg or DPB ≥ 90 mmHg.

We included patient demographic information, including age, gender, race, marital status, and education levels. Other covariates selected from previous literature reports or clinical experience that had an impact on the hypertensive population included body mass index (BMI), lymphocytes, neutrophils, serum creatinine, serum uric acid, triglycerides, fasting glucose, high density lipoprotein cholesterol (HDL-C), low density lipoprotein cholesterol (LDL-C), serum albumin, C-reactive protein (CRP), sodium and salt intake, smoking, alcohol consumption, physical activity, diabetes, cardiovascular disease, and chronic kidney disease (20). Race was categorized as Mexican American, non-Hispanic white, non-Hispanic black, or other. Education levels were classified as less than 9th Grade, and higher than 9th Grate or 9th Grate. Marital status was categorized as with partner (married or living with a partner), without partner (widowed, divorced, or separated), or never married. Body mass index (BMI) was calculated by the following formula: BMI = weight (kg)/height² (m²) (21). The sodium and salt intake were assessed from a 24-hour recall questionnaire (22). According to the median sodium and salt intake, there were “high sodium and salt intake” and “low sodium and salt intake” groups. Participants were categorized as “mild”, “moderate or heavy”, “former” and “never” based on the number of drinks per day he/she had drunk (23). Participants who were “mild” were considered to be drinking alcohol ≤1 drink in women and ≤2 drinks in men; Participants who were “heavy” were considered to be drinking alcohol ≤2 drinks in women and ≤3drinks in men or individuals had drunk ≥3 drinks of woman and ≥4drinks of man; Participants who used to drink alcohol but do not drink now are defined as “former” and never drink alcohol is defined as “never”. Smoking status was defined as the number and timeline of cigarettes in life (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). Physical activity was classified as ideal [≥8,000 metabolic equivalent task (MET) minutes/week], intermediate (600–7,999 MET minutes/week), and poor (<600 MET minutes/week) (24). Definition of cardiovascular disease: data were obtained from self-report data from the personal interview: “Have you ever been told by a doctor or other health professional that you have CHF (Congestive Heart Failure)/CHD (Coronary Heart Disease) /angina/heart attack/stroke?” If any of the above questions were answered in the affirmative, CVD was considered to be present. Glomerular filtration rate (eGFR) was estimated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI-2009) creatinine equation (25). Chronic kidney disease (CKD) was defined as estimated eGFR < 60 ml/min/1.73 m² (26). Diabetes mellitus was defined as meeting one of the following conditions: the doctor told you have diabetes, glycohemoglobin (HbA1c) (%) > 6.5, fasting glucose (mmol/L) ≥ 7.0, random blood glucose (mmol/L) ≥ 11.1, 2-hour oral glucose tolerance test (OGTT) blood glucose (mmol/L) ≥ 11.1 and use of diabetes medication or insulin. Furthermore, we categorized the use of antihypertensive medications according to the latest American College of Cardiology/American Heart Association (ACC/AHA) guidelines for hypertension (27). These included Beta blockers, Calcium channel blockers, Angiotensin-Converting Enzyme Inhibitors/Angiotensin II receptor antagonists (ACEI/ARB), and Diuretics.

Statistical analyses of data from this study were performed using R software (The R Foundation; http://www.r-project.org; version 4.2.2). NHANES created weights to account for the complex survey design, survey nonresponse, and post-stratification adjustments to match the total population residing in the United States. Therefore, in all the following analyses, we weighted the samples according to NHANES analysis guidelines (28).

Initially, the baseline characteristics of patients were compared by dividing the sample into three groups: no-risk, low-risk, and moderate-to-severe risk of adverse events according to GNRI thresholds. Continuous variables were expressed as weighted means ± standard deviations. Differences between the 3 study groups were analyzed by One-way analysis of variance (ANOVA). Categorical variables were expressed as numbers and percentages and compared using the Rao-Scott chi-squared test. NHANES public-use linked mortality file data as of December 31, 2019 were used to determine the mortality status of the follow-up population (29).Survival rates for different nutritional risk groups were then analyzed using the Kaplan-Meier survival analysis. Furthermore, Cox proportional hazard regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for cardiovascular mortality and all-cause mortality. The nonlinear relationship between GNRI and the presence of all-cause mortality and cardiovascular mortality was examined using a multivariate-adjusted Cox restricted cubic spline regression model. Finally, subgroup analyses were performed stratified by sodium and salt intake, physical activity, cardiovascular disease, diabetes mellitus, and chronic kidney disease. In all of these analyses, a two-sided P < 0.05 was considered statistically significant.

A total of 10,037 elderly patients with hypertension were included in this study. Based on the calculation of the weights for each sample, our sample represents approximately 3,322,957 elderly hypertensive patients in the United States. The median follow-up time was 106 months. Baseline characteristics of the total population and groups according to different GNRI levels are listed in the table (Table 1). In the total population, the mean age of patients was 70.7 years and the mean GNRI was 104.1. Among the 3 groups compared by GNRI classification, age, lymphocyte count, neutrophil count, serum creatinine, triglycerides, LDL cholesterol, C-reactive protein, diastolic blood pressure, gender, race, marital status, cardiovascular disease, diabetes mellitus, chronic kidney disease, smoking, and alcohol consumption were statistically different (P < 0.05). Participants in the moderate to severe malnutrition risk group of GNRI (GNRI < 92) had higher lymphocytes, C-reactive protein levels, higher prevalence of diabetes, and chronic kidney disease. In addition, BMI, serum uric acid, HDL cholesterol, and systolic blood pressure were not statistically different (P > 0.05).

During a median follow-up period of 106 months, 4255 (42.3%) all-cause mortality and 1207 (17.2%) cardiovascular disease mortality occurred in patients with hypertension. We used Kaplan-Meier to analyze survival between different groups at moderate to severe risk of malnutrition, low risk, and no risk of GNRI. The results showed that different levels of GNRI differed between all-cause and cardiovascular mortality (log-rank of trend p < 0.0001). The reduction in survival was more significant for moderate to severe malnutrition risk of GNRI for both all-cause and cardiovascular mortality (Figure 2).

In the COX proportional hazard regression model, GNRI is associated with the risk of all-cause mortality and cardiovascular mortality. In model 1, no variables were adjusted for. Elevated GNRI was associated with reduced risk of all-cause mortality and cardiovascular mortality [(HR, 0.941, 95% CI 0.933, 0.949) and [HR = 0.935, 95% CI (0.922, 0.949), respectively]. Similar findings were found for the adjusted Models 2, 3, and 4. And in the fully adjusted model 4, GNRI had hazard ratios of 0.958 (95% CI, 0.949, 0.967), and 0.956 (95% CI, 0.941, 0.972) for all-cause mortality and cardiovascular mortality, respectively. When GNRI was used as a categorical variable, the moderate to severe malnutrition risk of GNRI was shown to be associated with the risk of all-cause mortality and cardiovascular mortality in the unadjusted model with hazard ratios of 2.730 (95% CI, 2.006,3.714) and 2.891 (95% CI, 1.639,5.101), respectively. In the fully adjusted model (model 4), the hazard ratios for the risk of moderate to severe malnutrition in GNRI were 2.112 (95% CI, 1.377, 3.240) and 2.604 (95% CI, 1.603,4.229) for all-cause mortality and cardiovascular mortality, respectively (Table 2).

In a restricted cubic spline regression model (RCS) fully adjusted for confounders, the increase in GNRI was associated with a reduced risk of all-cause mortality and cardiovascular mortality in the elderly hypertensive population, but they both showed a non-linear trend (non-linear p < 0.001 and non-linear p = 0.015, respectively) (Figure 3).

We examined the association between GNRI and all-cause mortality and cardiovascular mortality in the subgroup analysis (Table 3). The results showed that GNRI interacted with sodium and salt intake and physical activity in all-cause mortality. For cardiovascular mortality, there was no interaction between GNRI and sodium and salt intake, physical activity, cardiovascular disease, diabetes mellitus, and chronic kidney disease. In addition, GNRI was not associated with all-cause mortality and cardiovascular mortality in an older hypertensive population in which physical activity was ideal (P > 0.05).