Prevalence and impact of malnutrition on readmission among hospitalized patients with heart failure in China

Prediction of 90 day readmission in heart failure with preserved ejection fraction by interpretable machine learning

AIMS

Certain critical risk factors of heart failure with preserved ejection fraction (HFpEF) patients were significantly different from those of heart failure with reduced ejection fraction (HFrEF) patients, resulting in the limitations of existing predictive models in real-world situations. This study aimed to develop a machine learning model for predicting 90 day readmission for HFpEF patients.

METHODS AND RESULTS

Data were extracted from electronic health records from 1 August 2020 to 1 August 2021 and follow-up records of patients with HFpEF within 3 months after discharge. Feature extraction was performed by univariate analysis combined with the least absolute shrinkage and selection operator (LASSO) algorithms. Machine learning models like eXtreme Gradient Boosting (XGBoost), random forest, neural network and logistic regression were adopted to construct models. The discrimination and calibration of each model were compared, and the Shapley Additive exPlanations (SHAP) method was used to explore the interpretability of the model. The cohort included 746 patients, of whom 103 (13.8%) were readmitted within 90 days. XGBoost owned the best performance [area under the curve (AUC) = 0.896, precision-recall area under the curve (PR-AUC) = 0.868, sensitivity = 0.817, specificity = 0.837, balanced accuracy = 0.827]. The Kolmogorov-Smirnov (KS) statistic was 0.694 at 0.468 in the XGBoost model. SHAP identified the top 12 risk features, including activities of daily living (ADL), left atrial dimension (LAD), left ventricular end-diastolic diameter (LVDD), shortness, nitrates, length of stay, nutritional risk, fall risk, accompanied by other symptoms, educational level, anticoagulants and edema.

CONCLUSIONS

Our model could help medical agencies achieve the early identification of 90 day readmission risk in HFpEF patients and reveal risk factors that provide valuable insights for treatments.

What is the additive value of nutritional deficiency to VA-FI in the risk assessment for heart failure patients?

OBJECTIVES

To assess the impact of adding the Prognostic Nutritional Index (PNI) to the U.S. Veterans Health Administration frailty index (VA-FI) for the prediction of time-to-death and other clinical outcomes in Veterans hospitalized with Heart Failure.

METHODS

A retrospective cohort study of veterans hospitalized for heart failure (HF) from October 2015 to October 2018. Veterans ≥50 years with albumin and lymphocyte counts, needed to calculate the PNI, in the year prior to hospitalization were included. We defined malnutrition as PNI ≤43.6, based on the Youden index. VA-FI was calculated from the year prior to the hospitalization and identified three groups: robust (≤0.1), prefrail (0.1-0.2), and frail (>0.2). Malnutrition was added to the VA-FI (VA-FI-Nutrition) as a 32nd deficit with the total number of deficits divided by 32. Frailty levels used the same cut-offs as the VA-FI. We compared categories based on VA-FI to those based on VA-FI-Nutrition and estimated the hazard ratio (HR) for post-discharge all-cause mortality over the study period as the primary outcome and other adverse events as secondary outcomes among patients with reduced or preserved ejection fraction in each VA-FI and VA-FI-Nutrition frailty groups.

RESULTS

We identified 37,601 Veterans hospitalized for HF (mean age: 73.4 ± 10.3 years, BMI: 31.3 ± 7.4 kg/m2). In general, VA-FI-Nutrition reclassified 1959 (18.6%) Veterans to a higher frailty level. The VA-FI identified 1,880 (5%) as robust, 8,644 (23%) as prefrail, and 27,077 (72%) as frail. The VA-FI-Nutrition reclassified 382 (20.3%) from robust to prefrail and 1577 (18.2%) from prefrail to frail creating the modified-prefrail and modified-frail categories based on the VA-FI-Nutrition. We observed shorter time-to-death among Veterans reclassified to a higher frailty status vs. those who remained in their original group (Median of 2.8 years (IQR:0.5,6.8) in modified-prefrail vs. 6.3 (IQR:1.8,6.8) years in robust, and 2.2 (IQR:0.7,5.7) years in modified-frail vs. 3.9 (IQR:1.4,6.8) years in prefrail). The adjusted HR in the reclassified groups was also significantly higher in the VA-FI-Nutrition frailty categories with a 38% increase in overall all-cause mortality among modified-prefrail and a 50% increase among modified-frails. Similar trends of increasing adverse events were also observed among reclassified groups for other clinical outcomes.

CONCLUSION

Adding PNI to VA-FI provides a more accurate and comprehensive assessment among Veterans hospitalized for HF. Clinicians should consider adding a specific nutrition algorithm to automated frailty tools to improve the validity of risk prediction in patients hospitalized with HF.

A nomogram predicting pneumonia after cardiac surgery: a retrospective modeling study

BACKGROUND

Postoperative pneumonia (POP) is the most prevalent of all nosocomial infections in patients who underwent cardiac surgery. The aim of this study was to identify independent risk factors for pneumonia after cardiac surgery, from which we constructed a nomogram for prediction.

METHODS

The clinical data of patients admitted to the Department of Cardiothoracic Surgery of Nanjing Drum Tower Hospital from October 2020 to September 2021 who underwent cardiac surgery were retrospectively analyzed, and the patients were divided into two groups according to whether they had POP: POP group (n=105) and non-POP group (n=1083). Preoperative, intraoperative, and postoperative indicators were collected and analyzed. Logistic regression was used to identify independent risk factors for POP in patients who underwent cardiac surgery. We constructed a nomogram based on these independent risk factors. Model discrimination was assessed via area under the receiver operating characteristic curve (AUC), and calibration was assessed via calibration plot.

RESULTS

A total of 105 events occurred in the 1188 cases. Age (>55 years) (OR: 1.83, P=0.0225), preoperative malnutrition (OR: 3.71, P<0.0001), diabetes mellitus(OR: 2.33, P=0.0036), CPB time (Cardiopulmonary Bypass Time) > 135 min (OR: 2.80, P<0.0001), moderate to severe ARDS (Acute Respiratory Distress Syndrome )(OR: 1.79, P=0.0148), use of ECMO or IABP or CRRT (ECMO: Extra Corporeal Membrane Oxygenation; IABP: Intra-Aortic Balloon Pump; CRRT: Continuous Renal Replacement Therapy )(OR: 2.60, P=0.0057) and MV( Mechanical Ventilation )> 20 hours (OR: 3.11, P<0.0001) were independent risk factors for POP. Based on those independent risk factors, we constructed a simple nomogram with an AUC of 0.82. Calibration plots showed good agreement between predicted probabilities and actual probabilities.

CONCLUSION

We constructed a facile nomogram for predicting pneumonia after cardiac surgery with good discrimination and calibration. The model has excellent clinical applicability and can be used to identify and adjust modifiable risk factors to reduce the incidence of POP as well as patient mortality.

Risk factors of readmission and the impact of outpatient management in heart failure patients: A national study in Japan

AIMS

Heart failure is a significant disease, and its high readmission rate is a big concern. We must identify readmission risk factors and optimize outpatient management to prevent them. This study aims to investigate the readmission risk factors, including outpatient management represented by the number of outpatient visits, and to identify the factors related to frequent outpatient visits.

METHODS AND RESULTS

We used the diagnosis-procedure-combination database between April 2016 and March 2022. Based on the number of outpatient visits within 60 days after discharge, we categorized patients into <1 visits/month, (1<, ≦2) visits/month, and <2 visits/month and observed the occurrence of 60 days readmission. We performed multiple logistic regression analyses to reveal the readmission risk factors and the association between the number of outpatient visits and readmission. As a subgroup analysis, we conducted the same research in the low- and high-readmission risk groups. We compared medical contents between (1<, ≦2) visits/month and <2 visits/month. We analysed 101 239 patients and identified the following factors as a risk of readmission: older age (P < 0.001), female (P = 0.009), longer length-of-hospital-stay (P < 0.001), artificial ventilator (P < 0.001), tolvaptan (P < 0.001), top 50% dosage of loop diuretics (P = 0.036), bottom 50% dosage of class III antiarrhythmic agents (P < 0.001), hypertension (P = 0.005), atrial fibrillation (P < 0.001), dilated cardiomyopathy (P < 0.001), valvular disease (P = 0.021), myocardial infarction (P < 0.001), diabetes (P < 0.001), and renal disease (P < 0.001). We revealed that the risk of readmission increases in <2 visits/month compared to (1<, ≦2) visits/month (P < 0.001), whereas the risk of readmission decreases in ≦1 visits/month compared with (1<, ≦2) visits/month (P < 0.001). In the subgroup analysis, we found the possibility that some risk factors are specific to the subgroup. We identified that the following factors were related to frequent outpatient visits: older age (P < 0.001), home medical care (P = 0.007), tolvaptan (P < 0.001), top 50% dosage of loop diuretics (P < 0.001), diabetes (P < 0.001), renal disease (P = 0.009), 0-2 weeks follow-up (P < 0.001), 2-4 weeks follow-up (P < 0.001), cardiac rehabilitation (P < 0.001), and echocardiography (P < 0.001).

CONCLUSIONS

This study comprehensively identified risk factors for readmission and found outpatient visit is personalized by readmission risk. There is still room to optimize outpatient management. We suggest optimizing outpatient management according to our identified characteristics.

Prognostic value of nutritional screening tools in hospitalized patients with decompensated heart failure: A systematic review and meta-analysis

Nutritional risk screening is fundamental to prevent undesirable outcomes in heart failure (HF). Current reviews of nutritional screening tools encompass both hospitalized and outpatient settings, which may not be suitable because of different clinical manifestations. We hypothesize that multidimensional tools would better identify prognosis of decompensated patients because the tools assess more than isolated aspects. This systematic review aims to explore the association of multidimensional nutritional risk screening tools and prognosis in patients hospitalized with decompensated HF. Five databases were searched for studies that assessed nutritional risk through multidimensional screening tools and its association with prognosis in adults hospitalized with decompensated HF. The 95% confidence interval and relative risk were computed using a random-effects model. Inverse variance method was used. Thirty-eight studies were included. Most studies demonstrated higher nutritional risk was significantly associated with worse prognosis. Quantitative analysis identified higher nutritional risk by using the Mini Nutritional Assessment Short Form (MNA-SF), Controlling Nutritional Status, Geriatric Nutritional Risk Index, and Prognostic Nutritional Index to be associated with all-cause mortality. The MNA-SF demonstrated greater magnitude of association with all-cause mortality in older subjects (relative risk, 4.85; 95% confidence interval, 2.0-11.75). Higher nutritional risk was associated with poor prognosis and higher mortality in patients hospitalized with decompensated HF, especially when screened by MNA-SF. Tools were not directly compared. That might reinforce the importance of evaluating multiple aspects when screening hospitalized HF patients once symptoms associated with decompensation frequently mask the underlying nutritional status and risk. PROSPERO registration number (CRD42021256271).

The clinical value of the Controlling Nutritional Status score for predicting prognosis in systolic heart failure cases in the vulnerable phase

Background

Malnutrition, a commonly encountered complication of heart failure, has an association with poor prognosis. The vulnerable phase of heart failure constitutes the most vulnerable stage of heart failure cases after discharge (usually within 3 months). At present, the prognostic value of Controlling Nutritional Status (CONUT) score in the vulnerable phase of systolic heart failure is unclear.

Methods

Totally 187 systolic heart failure cases were retrospectively assessed at the Second Affiliated Hospital of Dalian Medical University. Based on CONUT score at admission, cases were assigned to 3 groups, including the normal nutrition, and mild and moderate or severe malnutrition groups. The primary endpoint was all-cause death in the 90 days following discharge. The secondary, composite outcome encompassed all-cause death and rehospitalization due to heart failure. Kaplan-Meier method and log-rank test were performed to compare outcome event rates between groups. The independent risk factors for outcome events were obtained by multivariate COX regression analysis. The receiver operating characteristic (ROC) curve analysis and the Delong test were used to compare the prediction performance of the CONUT score and other independent risk factors for all-cause death.

Results

During the 90 days of follow-up, 8.6% of HF patients had the primary endpoint and 23.5% had the secondary outcome. All-cause mortality was markedly elevated in the moderate or severe malnutrition group (Logrank: p < 0.001). Compared with the normal nutrition group, composite endpoint events had starkly increased incidence rates in both malnutrition groups, and the incidence increased with the severity of malnutrition (Logrank: p < 0.05). Multivariate COX risk analysis revealed higher CONUT score [hazard ratio (HR) = 1.791, 95% confidence interval (CI) 1.379-2.327], age (HR = 1.08, 95% CI 1.028-1.134), B-type natriuretic peptide (BNP) (HR = 1.001, 95% CI 1.000-1.001), and aspartate aminotransferase (AST) (HR = 1.008, 95% CI 1.001-1.015) at admission as independent predictive factors of all-cause mortality. And higher CONUT score (HR = 1.162, 95% CI 1.024-1.318) and lower estimated glomerular filtration rate (eGFR) (HR = 0.98, 95% CI 0.966-0.993) for the secondary endpoint. The addition of the CONUT score significantly increased the predictive performance of age, BNP and AST, as well as their combination for all-cause death (Delong test: all p < 0.05).

Conclusion

The CONUT score at admission independently predicts poor prognosis during the vulnerable phase in patients with systolic heart failure and may be combined with conventional risk factors to further improve the predictive efficacy.

Malnutrition and infection lead to poor prognosis and heavy financial burden of patients with chronic heart failure

Background

Chronic heart failure (CHF) is a major public health concern, as it is associated with poor prognosis and heavy financial burden. In recent years, there has been increasing interest in medications for CHF in China, but few studies pay attention to the effects of nutrition and infection.

Methods and results

This was a retrospective study collected patients with CHF admitted to the Department of Cardiology of Qilu Hospital of Shandong University from January 2017 to May 2018. Patients were classified according to the prognosis and the financial burden. Through comparison and regression analysis, we found that the factor associated with worse prognosis were decreased heart rate, albumin and prealbumin; β-blockers and mineralocorticoid receptor antagonism (MRA) were the factor improved the prognosis of patients with CHF; the factor overburdening financial condition were infection, decreased prealbumin, high Alanine aminotransferase (ALT), usage of recombinant human brain natriuretic peptide (rhBNP) and Levosimendan; aspirin and Sacubitril/Valsartan were the factor releasing financial burden of patients with CHF. Then, we grouped by Controlling Nutritional Status (CONUT) score, which enabled evaluation of the patient's protein reserve and immune defenses. Patients in the malnutrition group had higher infection ratios, longer hospital stays, and greater hospital expenses than the normal group. The improvement ratios of therapeutic outcomes in the moderate or severe malnutrition group were lower than in the normal and mild malnutrition group.

Conclusion

Malnutrition and infection caused poor prognosis and increased financial burden of patients with CHF. The high CONUT score indicated the CHF patient's unfavorable prognosis and heavy financial burden.