Interaction of lactate/albumin and geriatric nutritional risk index on the all‐cause mortality of elderly patients with critically ill heart failure: A cohort study

Abstract Background Whether there is a multiplicative interaction of lactate/albumin (L/A) ratio and geriatric nutritional risk index (GNRI) on the mortality of critically ill elderly patients with heart failure (HF) remains unclear. Hypothesis To assess the interaction of L/A ratio and GNRI on the all‐cause mortality in critically ill elderly patients with HF. Methods This was a retrospective cohort study and data were extracted from the Medical Information Mart for Intensive Care III (MIMIC‐III) database. The endpoints were 28‐day and 1‐year all‐cause mortality, and the independent variables were L/A ratio and GNRI. The multiplicative interaction of L/A ratio and GNRI on the mortality was examined using Cox proportional‐hazards model. Results A total of 5627 patients were finally included. Results showed that patients with higher L/A ratio or GNRI ≤ 58 had higher risk of 28‐day and 1‐year all‐cause mortality (all p < .01). We also found the significant multiplicative interaction effect between L/A ratio and GNRI score on the 28‐day and 1‐year all‐cause mortality (both p < .05). The increased L/A ratio was associated with higher risk of 28‐day and 1‐year all‐cause mortality in patients with GNRI ≤ 58 than those with GNRI > 58. Conclusions There was a multiplicative interaction effect between L/A ratio and GNRI score on the mortality, and low GNRI score was associated with the increased risk of all‐cause mortality with the increase of L/A ratio, suggesting the importance of nutrition‐oriented intervention in critically ill elderly HF patients with high L/A ratio.


| INTRODUCTION
Heart failure (HF), a syndrome caused by heart function disorder, is a main reason of cardiovascular morbidity and mortality. 1 The prevalence of HF ranges from approximately 1% in middle-aged people to more than 10% in elderly people. 2 About 25.5% of critically ill patients with HF died within 1 year after hospital discharge. 3 Evidence has showed that the mortality of HF is agedependent, and 1-year mortality in the elderly reaches up to 7.4% for 60-year-olds and 19.5% for 80-year-olds, which poses an overwhelming threat to human health and leads to a huge medical burden. 4 To decrease the mortality of critically ill elderly patients with HF, it is important to identify those at the high risk of mortality.
The elevated blood lactate is considered as a marker of tissue hypoperfusion in critical care medicine and used as a prognostic marker for helping make therapeutic decision. 5 However, lactate elevation may be caused by various conditions, including reduced lactate elimination due to renal or hepatic dysfunction and accelerated glycolysis; thus, predictive capacity solely based on the original lactate level might be low. 6 In addition to lactate, serum albumin, which acts as a negative acute-phase protein and reflects the severity of inflammation, has been reported as a factor to predict the mortality in HF patients. 7 However, albumin levels may be impacted by chronic inflammatory conditions and nutritional status. 8 Therefore, only using albumin level to predict the prognosis may have limitations. The lactate/albumin (L/A) ratio is believed to be a more effective prognostic marker than lactate or albumin alone for intensive care patients. 9 In critically ill adults with HF, L/A ratio has been confirmed as an important predictor of short-term and long-term mortality, while the diagnostic utility of L/A ratio in elderly patients has not been reported. 10 The nutritional status in critically ill patients is a common concern, and it is worth noting that the incidence of malnutrition among elderly patients is 71.24%, which is far higher than 28% among younger patients. 11 Previous studies have reported that malnutrition is a useful predictor of mortality in HF patients. [12][13][14] Geriatric nutritional risk index (GNRI), comprising three objective parameters (height, weight, and serum albumin), is a simple tool to assess the nutritional status in the elderly. 12 A clinical study has reported the superiority of GNRI to other scores in predicting shortterm mortality for elderly HF patients. 15 Nishi et al. reports that nutritional screening using the GNRI is contributive to predict the long-term mortality of elderly HF patients. 16 Currently, although L/A ratio and GNRI has been confirmed as independent predictors of mortality in HF, whether there is a multiplicative interaction effect of L/A ratio and GNRI on the mortality of critically ill elderly patients with HF remains unclear.
Therefore, we designed a cohort study to evaluate the association and multiplicative interaction of L/A ratio and GNRI on the short-term and long-term mortality in critically ill elderly patients with HF.

| Study population
We included critically ill elderly patients with HF (age above 70 years) at first ICU admission and stayed more than 48 h in the hospital.
Patients who missed basic data on height, weight, lactate, and albumin, and had extreme height (height = 0 cm, height < 100 cm while weight > 55 kg, or height > 300 cm) were excluded from this study.

| Endpoints
Our study endpoints were 28-day and 1-year all-cause mortality after the date of ICU admission. From the Social Security Death Index records, survival information containing survival outcome and time of death was extracted. Patients were followed up until they were dead.

| Independent variables
The independent variables in our study were L/A ratio and GNRI. According to the knots of curve, patients were divided into two groups: GNRI ≤ 58 and GNRI > 58 (Supporting Information: Figure 1A,B).

| Covariates of interest
The covariates of interest included the following six aspects: (1)

| Statistical analysis
The continuous variables were presented as mean ± standard deviation (mean ± SD), and categorical variables were presented as number (n) and percentage (%). For groups divided by L/A ratio, differences of continuous variables were compared using analysis of variance or Kruskal-Wallis test and differences of categorical variables were compared using χ 2 test or Fisher's exact test. For groups stratified by GNRI score, differences of continuous variables were compared using Student's t test (normally distributed), and categorical variables were compared using χ 2 test or Fisher's exact test. Missing data were imputed using multiple imputations. The cumulative 28-day and 1-year all-cause mortality by L/A ratio quintiles and GNRI score group was shown through Kaplan-Meier (KM) method. Cut-off point of GNRI score was determined using the RCS method. Univariate and multivariate Cox proportional-hazards models were used to examine the association and multiplicative interaction of L/A ratio and GNRI score on the 28-day and 1-year all-cause mortality, and results were expressed as hazard ratio (HR) with 95% confidence interval (95% CI). A two-sided p < .05 was considered to be statistically significant. All statistical analyses were performed using R (v4.1.2, R Foundation for Statistical Computing).

| Association between L/A ratio and 28-day and 1-year all-cause mortality
KM curve showed a statistical significance between the five groups in 28day all-cause mortality (Supporting Information: Figure 2A) and 1-year allcause mortality (Supporting Information: Figure 2B) (both log-rank p < .001). Results of Cox proportional-hazards model were displayed in

| Association between GNRI score and 28-day and 1-year all-cause mortality
Through KM curve, we found the cumulative 28-day and 1-year allcause mortality between GNRI > 58 group and GNRI ≤ 58 group were F I G U R E 2 Forest plot for the HR of 28-day and 1-year all-cause mortality according to L/A ratio or GNRI. CI, confidence interval; GNRI, geriatric nutritional risk index; HR, hazard ratio; L/A, lactate/albumin. significantly different over time (log-rank p < .001 and log-rank p = .001, respectively) (Supporting Information: Figure 3A,B).
Supporting Information: Information: Table S2). The forest plot for the significance of the interaction of L/A ratio and GNRI on the 28-day and 1-year all-cause mortality was shown in Figure 3. Figure 4A shows the risk of 28-day all-cause mortality of critically ill elderly patients with HF was higher in GNRI ≤ 58 group than the score >58 group as the L/A ratio increased. The result was similar in 1-year all-cause mortality; with the increase of L/A ratio, GNRI ≤ 58 displayed the higher risk than the score >58 ( Figure 4B).

F I G U R E 3
Forest plot for the significance of interaction of L/A ratio and GNRI on the 28-day and 1-year all-cause mortality. CI, confidence interval; GNRI, geriatric nutritional risk index; HR, hazard ratio; L/A, lactate/albumin.

| DISCUSSION
Critically ill elderly patients with HF have a high mortality, which poses a high burden to the economy and medical resources. 3,4 Thus, predicting the risk of mortality of these patients can help clinicians make relevant clinical decisions earlier. In this study, we evaluated the association and multiplicative interaction of L/A ratio and GNRI score on the 28-day and 1-year all-cause mortality of critically ill elderly patients with HF. The results showed that higher L/A ratio was associated with higher risk of 28-day and 1-year all-cause mortality. Patients with low GNRI scores had an increased risk of 28-day and 1-year all-cause mortality compared to those with the high score. Also, we found the significant multiplicative interaction effect between L/A ratio and GNRI on the 28-day and 1-year all-cause mortality. With the increase of L/A ratio, the risk of 28-day and 1-year all-cause mortality were higher in low GNRI score group than the high score group.
L/A ratio has been reported as an effective prognostic marker than lactate or albumin alone for critically ill patients. 9 In our study, higher L/A ratio was associated with the higher risk of short-term and long-term all-cause mortality. The positive association between L/A ratio and the risk of all-cause mortality in critically ill patients with HF may be explained as follows. The level of lactate may be increased by insufficient tissue perfusion and anaerobic metabolism in critically ill HF patients resulting from poor cardiac function or long-term and large-scale use of diuretics. 19,20 Evidence has showed the positive association between the increased lactate level and the elevated risk of 1-year mortality in patients with HF. 5 In HF patients, lactate level is a marker of hypoperfusion and hypoxia, and elevated level is associated with hemodynamic instability. 5 Patients with severe HF F I G U R E 4 Association between L/A ratio and 28-day (A) and 1-year all-cause mortality (B) in different GNRI groups. GNRI, geriatric nutritional risk index; L/A, lactate/albumin. had chronic consumption or liver congestion, which may result in hypoalbuminemia. 21 Previous studies have showed that approximately 14% of HF patients had hypoalbuminemia, and hypoalbuminemia caused an increased risk of death. 21,22 Therefore, a high L/A ratio reflects an increase of lactate level and a decrease of albumin level, which elevates the risk of all-cause mortality in severe HF patients.
The nutritional guidelines have emphasized the nutrition screening for critical ill patients. 23,24 In elderly HF patients, cardiac dysfunction and venous congestion predispose to bowel edema and malabsorption, thereby resulting in malnutrition. 25 Malnutrition is highly prevalent in elderly HF patients and associated with the higher death rate. 26,27 GNRI is a valuable and objective indicator for risk of malnutritional status of elderly patients, and lower GNRI means the worse nutritional status. 25 In our study, we found that the low GNRI score was correlated with higher all-cause mortality, which was consistent with the results of previously studies. 16,25,28 The following mechanisms may explain the association between GNRI and mortality. GNRI is calculated by albumin, actual weight, and ideal weight. 18 Hypoalbuminemia is recognized as a factor contributing to developing HF. 29 Both hypoalbuminemia 30 and weight loss 31 displayed an effective and consistent outcome-predictability of death. Therefore, GNRI that combines albumin, actual weight, and ideal weight can show a synergistical effect on mortality.
In our study, a significant multiplicative interaction effect between L/A ratio and GNRI score was observed, implying that GNRI score affected the association between L/A ratio and mortality. Further, we found the risk of mortality caused by the increased L/A ratio was higher in the low GNRI score group than the high score group. This finding suggested that more attention on the nutritional status is needed in elderly patients with high L/A levels.
Our study explores the association and multiplicative interaction of L/A ratio and GNRI on the long-term and short-term all-cause mortality of critically ill elderly patients with HF. Some limitations are existed in our study. First, our data were extracted from the MIMIC-III database and data missing is inevitable, but the missing data have been imputed using multiple imputations. Second, the lactate level is collected at baseline; therefore, we cannot assess the impact of dynamic change of L/A ratio on the risk of mortality.

| CONCLUSION
In conclusion, our study found the association and multiplicative interaction of L/A ratio and GNRI score on the all-cause mortality of critically ill elderly HF patients, and increased L/A ratio was associated with the higher risk of all-cause mortality in the low GNRI score group. Our findings emphasized the importance of nutritionoriented management and intervention in reducing the risk of mortality among elderly HF patients with high L/A ratio in the ICU.

This study was funded by the Natural Science Foundation of Bengbu
Medical College (BYKY2019270ZD).