Red Cell Distribution Width as a Prognostic Factor and Its Comparison with Lactate in Patients with Sepsis

Interpretable machine learning model for early prediction of 28-day mortality in ICU patients with sepsis-induced coagulopathy: development and validation

OBJECTIVE

Sepsis-induced coagulopathy (SIC) is extremely common in individuals with sepsis, significantly associated with poor outcomes. This study attempted to develop an interpretable and generalizable machine learning (ML) model for early predicting the risk of 28-day death in patients with SIC.

METHODS

In this retrospective cohort study, we extracted SIC patients from the Medical Information Mart for Intensive Care III (MIMIC-III), MIMIC-IV, and eICU-CRD database according to Toshiaki Iba's scale. And the overlapping in the MIMIC-IV was excluded for this study. Afterward, only the MIMIC-III cohort was randomly divided into the training set, and the internal validation set according to the ratio of 7:3, while the MIMIC-IV and eICU-CRD databases were considered the external validation sets. The predictive factors for 28-day mortality of SIC patients were determined using recursive feature elimination combined with tenfold cross-validation (RFECV). Then, we constructed models using ML algorithms. Multiple metrics were used for evaluation of performance of the models, including the area under the receiver operating characteristic curve (AUROC), area under the precision recall curve (AUPRC), accuracy, sensitivity, specificity, negative predictive value, positive predictive value, recall, and F1 score. Finally, Shapley Additive Explanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME) were employed to provide a reasonable interpretation for the prediction results.

RESULTS

A total of 3280, 2798, and 1668 SIC patients were screened from MIMIC-III, MIMIC-IV, and eICU-CRD databases, respectively. Seventeen features were selected to construct ML prediction models. XGBoost had the best performance in predicting the 28-day mortality of SIC patients, with AUC of 0.828, 0.913 and 0.923, the AUPRC of 0.807, 0.796 and 0.921, the accuracy of 0.785, 0.885 and 0.891, the F1 scores were 0.63, 0.69 and 0.70 in MIMIC-III (internal validation set), MIMIC-IV, and eICU-CRD databases. The importance ranking and SHAP analyses showed that initial SOFA score, red blood cell distribution width (RDW), and age were the top three critical features in the XGBoost model.

CONCLUSIONS

We developed an optimal and explainable ML model to predict the risk of 28-day death of SIC patients 28-day death risk. Compared with conventional scoring systems, the XGBoost model performed better. The model established will have the potential to improve the level of clinical practice for SIC patients.

Association between Red Cell Distribution Width and Outcomes of Nonagenarians Admitted to the Intensive Care Unit-A Retrospective Cohort Study

The red cell distribution width (RDW) measures the heterogeneity of the erythrocyte volume. Different clinical conditions are associated with increased RDW, and high levels (>14.5%) have been described as a predictive marker for unfavorable outcomes and mortality in critically ill patients. However, there is a lack of data on very elderly critically ill patients. Therefore, we aimed to investigate the association of RDW with outcomes in critically ill patients ≥ 90 years. A retrospective analysis was conducted for all consecutive critically ill patients ≥ 90 years who were admitted to the Department of Intensive Care Medicine of the Medical University Centre Hamburg-Eppendorf (Hamburg, Germany) with available RDW on admission. Clinical course and laboratory were analyzed for all patients with eligible RDW. High RDW was defined as (>14.5%). We clinically assessed factors associated with mortality. Univariable and multivariable Cox regression analysis was performed to determine the prognostic impact of RDW on 28-day mortality. During a 12-year period, we identified 863 critically ill patients ≥ 90 years old with valid RDW values and complete clinical data. In total, 32% (n = 275) died within 28 days, and 68% (n = 579) survived for 28 days. Median RDW levels on ICU admission were significantly higher in non-survivors compared with survivors (15.6% vs. 14.8%, p < 0.001). Overall, 38% (n = 327) had low, and 62% (n = 536) had high RDW. The proportion of high RDW (>14.5%) was significantly higher in non-survivors (73% vs. 57%, p < 0.001). Patients with low RDW presented with a lower Charlson Comorbidity Index (p = 0.014), and their severity of illness on admission was lower (SAPS II: 35 vs. 38 points, p < 0.001). In total, 32% (n = 104) in the low and 35% (n = 190) in the high RDW group were mechanically ventilated (p = 0.273). The use of vasopressors (35% vs. 49%, p < 0.001) and renal replacement therapy (1% vs. 5%, p = 0.007) was significantly higher in the high RDW group. Cox regression analysis demonstrated that high RDW was significantly associated with 28-day mortality [crude HR 1.768, 95% CI (1.355-2.305); p < 0.001]. This association remained significant after adjusting for multiple confounders [adjusted HR 1.372, 95% CI (1.045-1.802); p = 0.023]. High RDW was significantly associated with mortality in critically ill patients ≥ 90 years. RDW is a useful simple parameter for risk stratification and may aid guidance for the therapy in very elderly critically ill patients.

A novel clinical prediction model of severity based on red cell distribution width, neutrophil-lymphocyte ratio and intra-abdominal pressure in acute pancreatitis in pregnancy

BACKGROUND

Acute pancreatitis in pregnancy (APIP) with a high risk of death is extremely harmful to mother and fetus. There are few models specifically designed to assess the severity of APIP. Our study aimed to establish a clinical model for early prediction of severity of APIP.

METHODS

A retrospective study in a total of 188 patients with APIP was enrolled. The hematological indicators, IAP (intra-abdominal pressure) and clinical data were obtained for statistical analysis and prediction model construction.

RESULTS

According to univariate and multivariate logistic regression analysis, we found that red cell distribution width (RDW), neutrophil-lymphocyte ratio (NLR) and Intra-abdominal pressure (IAP) are prediction indexes of the severity in APIP (p-value < 0.05). Our novel clinical prediction model was created by based on the above three risk factors and showed superior predictive power in primary cohort (AUC = 0.895) and validation cohort (AUC = 0.863). A nomogram for severe acute pancreatitis in pregnancy (SAPIP) was created based on the three indicators. The nomogram was well-calibrated.

CONCLUSION

RDW, NLR and IAP were the independent risk factors of APIP. Our clinical prediction model of severity in APIP based on RDW, NLR and IAP with predictive evaluation is accurate and effective.

Baseline red blood cell distribution width and perforin, dynamic levels of interleukin 6 and lactate are predictors of mortality in patients with sepsis

BACKGROUND

Sepsis is a critical illness often encountered in the intensive care unit. However, prognostic biomarkers for sepsis have limited sensitivity. This study aimed to identify more sensitive predictors of mortality through repeated monitoring of laboratory parameters.

METHODS

Patients with sepsis (Sepsis 3.0 criteria met) were recruited and divided into the survivor and nonsurvivor groups after 28 days. Data on blood biochemistry, lymphocyte subsets, and cytokines were obtained on the first and seventh hospitalization days. Univariate and multivariate Cox regression analyses were performed to explore the correlation between these variables and patient mortality.

RESULTS

Forty patients with sepsis were included. The mortality rate was 37.5%. Red blood cell distribution width-standard deviation (RDWSD) (hazard ratio [HR] = 1.107 [95% CI: 1.005-1.219], p = 0.040) and perforin level (HR = 1.001 [95% CI: 1-1.003], p = 0.035) on the first day, as well as lactate (HR = 112.064 [95% CI: 2.192-5729.629], p = 0.019) and interleukin 6 (IL-6) (HR = 1.005 [95% CI: 1.001-1.008], p = 0.014) levels on the seventh day, were independent risk factors of mortality. If the patients were divided into two groups based on RDWSD (normal: n = 31; increased: n = 9), the Kaplan-Meier curves showed that the group with increased RDWSD had a lower survival (p = 0.025).

CONCLUSION

Baseline RDWSD and perforin, along with dynamic IL-6 and lactate levels, were independent predictors of mortality in patients with sepsis.

Biomarkers of Sepsis

Sepsis is a highly complex disease caused by a deregulated host's response to infection [...].

Red blood cell distribution width at admission predicts outcome in critically ill patients with kidney failure: a retrospective cohort study based on the MIMIC-IV database

PURPOSE

We aimed to explore whether red blood cell distribution width (RDW) could serve as a biomarker to predict outcomes in critically ill patients with kidney failure in this study.

MATERIALS AND METHODS

This retrospective study was conducted with the Medical Information Mart for Intensive Care IV (MIMIC-IV).A total of 674 patients were divided into three groups based on tertiles of RDW. We used the generalized additive model, Kaplan-Meier curve, and Cox proportional hazards models to evaluate the association between RDW and clinical outcomes. We then performed subgroup analyses to investigate the stability of the associations between RDW and all-cause mortality.

RESULTS

Nonlinear and J-shaped curves were observed in the generalized additive model. Kaplan-Meier analysis showed that patients with elevated RDW had a lower survival rate. The Cox regression model indicated that high levels of RDW were most closely associated with ICU mortality and 30-day mortality (HR = 4.71, 95% CI: 1.69-11.64 and HR = 6.62, 95% CI: 2.84-15.41). Subgroup analyses indicated that the associations between RDW and all-cause mortality were stable.

CONCLUSIONS

Elevated levels of RDW were associated with an increased risk of all-cause mortality, and RDW could be an independent prognostic factor for kidney failure.

Diagnostic value of RDW for the prediction of mortality in adult sepsis patients: A systematic review and meta-analysis

Background

Red blood cell distribution width (RDW) is a common biomarker of bacterial infections, and it can be easily obtained from a routine blood test. We investigate the diagnostic value of RDW for the prediction of mortality in adult sepsis patients through a review and meta-analysis. We registered this review in PROSPERO (Registration Number: CRD42022357712), and the details of the registration are included in Appendix 1.

Methods

We searched PubMed, Cochrane Library, Springer, and Embase between Jan. 1, 2000, and May 30, 2022, for primary studies about this research. We collected articles that investigated RDW for varying degrees of sepsis patients—those who suffered from sepsis, severe sepsis, or sepsis shock. Studies of healthy people and sepsis of children and neonates were excluded from our research. The definition of study characteristics and data extraction were finished by two independent researchers and discrepancies resolved by consensus. The combined sensitivities and specificities were calculated by meta-analysis using STATA14.0. The sensitivity of the included studies was analyzed by excluding studies that had potential heterogeneity. A summary operating characteristic curve was made to evaluate the diagnostic value for the prediction of mortality in adult sepsis patients. The Fagan test was used to explore likelihood ratios and posttest probabilities. Finally, we investigated the source of heterogeneity using meta-regression.

Results

Twenty-four studies, including 40,763 cases altogether, were included in this analysis. Bivariate analysis indicated a combined sensitivity of 0.81 (95% CI 0.73–0.86) and specificity of 0.65 (95% CI 0.54–0.75). The area under the summary receiver operating characteristic curve was 0.81 (95% CI 0.77–0.84). Substantial heterogeneity resided in the studies (I^{2 =}96.68, 95% CI 95.95–97.4). Meta-regression showed that the reference description, prospective design, and blinded interpretation of the included studies could be responsible for the heterogeneity.

Conclusions

RWD is an available and valuable biomarker for prediction of mortality in adult sepsis patients.

Systematic review registration

https://www.crd.york.ac.uk/PROSPERO/, identifier CRD42022357712.

NLR, MLR, PLR and RDW to predict outcome and differentiate between viral and bacterial pneumonia in the intensive care unit

The neutrophil-to-lymphocyte ratio (NLR), monocyte-to-lymphocyte ratio (MLR), platelet-to-lymphocyte ratio (PLR), and red cell distribution width (RDW) are emerging biomarkers to predict outcomes in general ward patients. However, their role in the prognostication of critically ill patients with pneumonia is unclear. A total of 216 adult patients were enrolled over 2 years. They were classified into viral and bacterial pneumonia groups, as represented by influenza A virus and Streptococcus pneumoniae, respectively. Demographics, outcomes, and laboratory parameters were analysed. The prognostic power of blood parameters was determined by the respective area under the receiver operating characteristic curve (AUROC). Performance was compared using the APACHE IV score. Discriminant ability in differentiating viral and bacterial aetiologies was examined. Viral and bacterial pneumonia were identified in 111 and 105 patients, respectively. In predicting hospital mortality, the APACHE IV score was the best prognostic score compared with all blood parameters studied (AUC 0.769, 95% CI 0.705–0.833). In classification tree analysis, the most significant predictor of hospital mortality was the APACHE IV score (adjusted P = 0.000, χ2 = 35.591). Mechanical ventilation was associated with higher hospital mortality in patients with low APACHE IV scores ≤ 70 (adjusted P = 0.014, χ2 = 5.999). In patients with high APACHE IV scores > 90, age > 78 (adjusted P = 0.007, χ2 = 11.221) and thrombocytopaenia (platelet count ≤ 128, adjusted P = 0.004, χ2 = 12.316) were predictive of higher hospital mortality. The APACHE IV score is superior to all blood parameters studied in predicting hospital mortality. The single inflammatory marker with comparable prognostic performance to the APACHE IV score is platelet count at 48 h. However, there is no ideal biomarker for differentiating between viral and bacterial pneumonia.