Role of Clinical Characteristics and Biomarkers at Admission to Predict One-Year Mortality in Elderly Patients with Pneumonia

Epidemiological and clinical management aspects of pneumonias diagnosed in the emergency department in elderly patients in Spain: Results of the EDEN-29 study

OBJECTIVES

To estimate the incidence of pneumonia diagnosis in elderly patients in Spanish emergency departments (ED), need for hospitalization, adverse events and predictive capacity of biomarkers commonly used in the ED.

METHODS

Patients ≥65 years with pneumonia seen in 52 Spanish EDs were included. We recorded in-hospitaland 30-day mortality as adverse events, as well as intensive care unit (ICU) admission among hospitalizedpatients. Association of 10 predefined variables with adverse events was calculated and expressed as odds ratio (OR) with 95% confidence interval (CI), as well as predictive capacity of 5 commonly used biomarkers in the ED (leukocytes, hemoglobin, C-reactive protein, glucose, creatinine) was investigated using area under the receiver operating characteristic curve (AUC-ROC).

RESULTS

591 patients with pneumonia attended in the ED were included (annual incidence of 18,4 per 1000 inhabitants). A total of 78.0% were hospitalized. Overall, 30-day mortality was 14.2% and in-hospital mortality was 12.9%. Functional dependency was associated with both events (OR=4.453, 95%CI=2.361-8.400; and OR=3.497, 95%CI=1.578-7.750, respectively) as well as severe comorbidity (2.344, 1.363-4.030, and 2.463, 1.252-4.846, respectively). Admission to the ICU during hospitalization occurred in 3.5%, with no associated factors. The predictive capacity of biomarkers was only moderate for creatinine for ICU admission (AUC-ROC=0.702, 95% CI=0.536-0.869) and for leukocytes for post-discharge adverse event (0.669, 0.540-0.798).

CONCLUSIONS

Pneumonia is a frequent diagnosis in elderly patients consulting in the ED. Their functional dependence and comorbidity is the factor most associated with adverse events. The biomarkers analyzed do not have a good predictive capacity for adverse events.

Improving prognostication of pneumonia among elderly patients: usefulness of suPAR

PURPOSE

Elderly patients with suspected pneumonia represent a significant proportion of hospital admissions, which is a prognostic challenge for physicians. Our research aimed to assess the prognosis of patients with pneumonia using soluble urokinase plasminogen activator receptor (suPAR) combined with clinical data.

METHODS

In a prospective observational study including 164 patients > 65 years (mean age 84.2 (+/-7.64) years) who were hospitalized for a suspicion of pneumonia, suPAR was assessed for each patient, as was the prognosis score (PSI, CURB65) and inflammatory biomarkers (C-reactive protein, procalcitonin, white blood cells). The prognostic value of the suPAR for 30-day mortality was assessed using receiver operating characteristic (ROC) curve analyses. Optimal cut-offs with corresponding sensitivity (SE) and specificity (SP) were determined using the Youden index.

RESULTS

A suPAR > 5.1 ng/mL was predictive of 30-day mortality with a sensitivity of 100% and a specificity of 40.4%. A combination of the following parameters exhibited an SE of 100% (95% CI, 100-100) for an SP value of 64.9% (95% CI, 57.6-72.2) when at least two of them were above or below the following cut-off threshold values: suPAR > 9.8 ng/mL, BMI < 29.3 kg/m2 and PSI > 106.5.

CONCLUSION

The suPAR seems to be a promising biomarker that can be combined with the PSI and BMI to improve the prognosis of pneumonia among elderly patients. Prospective studies with larger populations are needed to confirm whether this new approach can improve patient outcomes.

TRIAL REGISTRATION

ClinicalTrials.gov (NCT02467192), 27th may 2015.

Decreased Haemoglobin Level Measured at Admission Predicts Long Term Mortality after the First Episode of Acute Pulmonary Embolism

Background: Decreased hemoglobin concentration was reported to predict long term prognosis in patients various cardiovascular diseases including congestive heart failure and coronary artery disease. We hypothesized that hemoglobin levels may be useful for post discharge prognostication after the first episode of acute pulmonary embolism. Therefore, the aim of the current study was to evaluate a potential prognostic value of a decreased hemoglobin levels measured at admission due to the first episode of acute PE for post discharge all cause mortality during at least 2 years follow up. Methods: This was a prospective, single-center, follow-up, observational, cohort study of consecutive survivors of the first PE episode. Patients were managed according to ESC current guidelines. After the discharge, all PE survivors were followed for at least 24 months in our outpatient clinic. Results: During 2 years follow-up from the group of 402 consecutive PE survivors 29 (7.2%) patients died. Non-survivors were older than survivors 81 years (40−93) vs. 63 years (18−97) p < 0.001 presented higher sPESI 2 (0−4) vs. 1 (0−5), p < 0.001 driven by a higher frequency of neoplasms (37.9% vs. 16.6%, p < 0.001); and had lower hemoglobin (Hb) level at admission 11.7 g/dL (6−14.8) vs. 13.1 g/dL (3.1−19.3), p < 0.001. Multivariable analysis showed that only Hb and age significantly predicted all cause post-discharge mortality. ROC analysis for all cause mortality showed AUC for hemoglobin 0.688 (95% CI 0.782−0.594), p < 0.001; and for age 0.735 (95% CI 0.651−0.819) p < 0.001. A group of 59 subjects with hemoglobin < 10.5 g/dL showed mortality rate of 16.9% (OR for mortality 4.19 (95% CI 1.82−9.65), p-value < 0.00, while among 79 patients with Hb > 14.3 g/dL only one death was detected. Interestingly, patients in age > 64 years hemoglobin levels < 13.2 g/dL compared to patients in the same age but with >13.2 g/dL showed OR 3.6 with 95% CI 1.3−10.1 p = 0.012 for death after the discharge. Conclusions: Lower haemoglobin measured in the acute phase especially in patients in age above 64 years showed significant impact on the prognosis and clinical outcomes in PE survivors.

Performance of Machine Learning Algorithms for Predicting Adverse Outcomes in Community-Acquired Pneumonia

Background: The ability to assess adverse outcomes in patients with community-acquired pneumonia (CAP) could improve clinical decision-making to enhance clinical practice, but the studies remain insufficient, and similarly, few machine learning (ML) models have been developed.

Objective: We aimed to explore the effectiveness of predicting adverse outcomes in CAP through ML models.

Methods: A total of 2,302 adults with CAP who were prospectively recruited between January 2012 and March 2015 across three cities in South America were extracted from DryadData. After a 70:30 training set: test set split of the data, nine ML algorithms were executed and their diagnostic accuracy was measured mainly by the area under the curve (AUC). The nine ML algorithms included decision trees, random forests, extreme gradient boosting (XGBoost), support vector machines, Naïve Bayes, K-nearest neighbors, ridge regression, logistic regression without regularization, and neural networks. The adverse outcomes included hospital admission, mortality, ICU admission, and one-year post-enrollment status.

Results: The XGBoost algorithm had the best performance in predicting hospital admission. Its AUC reached 0.921, and accuracy, precision, recall, and F1-score were better than those of other models. In the prediction of ICU admission, a model trained with the XGBoost algorithm showed the best performance with AUC 0.801. XGBoost algorithm also did a good job at predicting one-year post-enrollment status. The results of AUC, accuracy, precision, recall, and F1-score indicated the algorithm had high accuracy and precision. In addition, the best performance was seen by the neural network algorithm when predicting death (AUC 0.831).

Conclusions: ML algorithms, particularly the XGBoost algorithm, were feasible and effective in predicting adverse outcomes of CAP patients. The ML models based on available common clinical features had great potential to guide individual treatment and subsequent clinical decisions.

Kinetics of inflammatory biomarkers to predict one-year mortality in older patients hospitalized for pneumonia: a multivariable analysis

OBJECTIVES

Long-term mortality is increased in older patients with pneumonia. We aimed to test whether residual inflammation is predictive of one-year mortality after pneumonia.

METHODS

Inflammation biomarkers (C-reactive protein [CRP], interleukin [IL]-6 and IL-8, tumor necrosis factor-α, serum amyloid A, neopterin, myeloperoxidase, anti-apolipoprotein A-1, and anti-phosphorylcholine IgM) were measured at admission and discharge in older patients hospitalized for pneumonia in a prospective study. Univariate and multivariate analyses were conducted using absolute level at discharge and relative and absolute differences between admission and discharge for all biomarkers, along with usual prognostic factors.

RESULTS

In the 133 included patients (median age, 83 years [interquartile range: 78-89]), one-year mortality was 26%. In univariate analysis, the relative difference of CRP levels had the highest area under the receiver operating characteristic curve (0.70; 95% confidence interval [CI] 0.60-0.80). A decrease of CRP levels of more than 67% between admission and discharge had 68% sensitivity and 68% specificity to predict survival. In multivariate analysis, lower body mass index (hazard ratio=0.87 [CI 95% 0.79-0.96], P-value=0.01), higher IL-8 (hazard ratio=1.02 [CI 95% 1.00-1.04], P-value=0.02), and higher CRP (1.01 [95% CI 1.00-1.02], P=0.01) at discharge were independently associated with mortality.

CONCLUSION

Higher IL-8 and CRP levels at discharge were independently associated with one-year mortality. The relative CRP difference during hospitalization was the best individual biomarker for predicting one-year mortality.