Assessing SOFA score trajectories in sepsis using machine learning: A pragmatic approach to improve the accuracy of mortality prediction

INTRODUCTION

An increasing amount of longitudinal health data is available on critically ill septic patients in the age of digital medicine, including daily sequential organ failure assessment (SOFA) score measurements. Thus, the assessment in sepsis focuses increasingly on the evaluation of the individual disease's trajectory. Machine learning (ML) algorithms may provide a promising approach here to improve the evaluation of daily SOFA score dynamics. We tested whether ML algorithms can outperform the conventional ΔSOFA score regarding the accuracy of 30-day mortality prediction.

METHODS

We used the multicentric SepsisDataNet.NRW study cohort that prospectively enrolled 252 sepsis patients between 03/2018 and 09/2019 for training ML algorithms, i.e. support vector machine (SVM) with polynomial kernel and artificial neural network (aNN). We used the Amsterdam UMC database covering 1,790 sepsis patients for external and independent validation.

RESULTS

Both SVM (AUC 0.84; 95% CI: 0.71-0.96) and aNN (AUC 0.82; 95% CI: 0.69-0.95) assessing the SOFA scores of the first seven days led to a more accurate prognosis of 30-day mortality compared to the ΔSOFA score between day 1 and 7 (AUC 0.73; 95% CI: 0.65-0.80; p = 0.02 and p = 0.05, respectively). These differences were even more prominent the shorter the time interval considered. Using the SOFA scores of day 1 to 3 SVM (AUC 0.82; 95% CI: 0.68 0.95) and aNN (AUC 0.80; 95% CI: 0.660.93) led to a more accurate prognosis of 30-day mortality compared to the ΔSOFA score (AUC 0.66; 95% CI: 0.58-0.74; p < 0.01 and p < 0.01, respectively). Strikingly, all these findings could be confirmed in the independent external validation cohort.

CONCLUSIONS

The ML-based algorithms using daily SOFA scores markedly improved the accuracy of mortality compared to the conventional ΔSOFA score. Therefore, this approach could provide a promising and automated approach to assess the individual disease trajectory in sepsis. These findings reflect the potential of incorporating ML algorithms as robust and generalizable support tools on intensive care units.

Timely Cessation of Proton Pump Inhibitors in Critically Ill Patients Impacts Morbidity and Mortality: A Propensity Score-Matched Cohort Study

OBJECTIVE

Proton pump inhibitors (PPIs) are among the drugs most commonly used in critically ill patients. Although mainly applied temporarily for stress ulcer prophylaxis, their application is frequently not terminated. Potential adverse effects of PPI treatment could impact the outcome in case of unnecessary and, therefore, avoidable long-term continuation. We tested the hypotheses that nonindicated PPI therapy continued beyond hospital discharge is associated with increased morbidity, rehospitalization rate, and mortality.

DESIGN

Nationwide retrospective cohort study considering critically ill patients treated on German ICUs between January, 2017, and December, 2018 with a 2-year follow-up.

SETTING

A total of 591,207 patient datasets of a German healthcare insurer were screened.

PATIENTS

We identified 11,576 ICU patients who received PPI therapy for the first time during their index ICU stay without having an indication for its continuation.

INTERVENTIONS

The cohort was stratified into two groups: 1) patients without further PPI therapy and 2) patients with continuation of PPI therapy beyond 8 weeks after hospital discharge.

MEASUREMENTS AND MAIN RESULTS

Frequency of predescribed adverse events associated with PPI therapy, 1-year rehospitalization rate, and 2-year mortality were determined. The proportion of patients with continued PPI therapy without an objectifiable indication was 41.7% (4,825 of 11,576 patients). These patients had a 27% greater risk of pneumonia (odds ratio [OR] 1.27; 95% CI, 1.15-1.39; p < 0.001) and a 17% greater risk of cardiovascular events (OR 1.17; 95% CI, 1.08-1.26; p < 0.001). Continued PPI therapy was associated with a 34% greater risk of rehospitalization (OR 1.34; 95% CI, 1.23-1.47) and a nearly 20% greater 2-year mortality risk (hazard ratio 1.17; 95% CI, 1.08-1.27; p = 0.006).

CONCLUSIONS

These data demonstrate that an unnecessary continuation of PPI therapy after hospital discharge may significantly impact morbidity and mortality. To avoid potentially harmful overuse of a PPIs, intensivists should ensure timely cessation of a temporarily indicated PPI therapy.

The Aquaporin 3 Polymorphism (rs17553719) Is Associated with Sepsis Survival and Correlated with IL-33 Secretion

Sepsis is a common life-threatening disease caused by dysregulated immune response and metabolic acidosis which lead to organ failure. An abnormal expression of aquaporins plays an important role in organ failure. Additionally, genetic variants in aquaporins impact on the outcome in sepsis. Thus, we investigated the polymorphism (rs17553719) and expression of aquaporin-3 (AQP3) and correlated these measurements with the survival of sepsis patients. Accordingly, we collected blood samples on several days (plus clinical data) from 265 sepsis patients who stayed in different ICUs in Germany. Serum plasma, DNA, and RNA were then separated to detect the promotor genotypes of AQP3 mRNA expression of AQP3 and several cytokines. The results showed that the homozygote CC genotype exhibited a significant decrease in 30-day survival (38.9%) compared to the CT (66.15%) and TT genotypes (76.3%) (p = 0.003). Moreover, AQP3 mRNA expression was significantly higher and nearly doubled in the CC compared to the CT (p = 0.0044) and TT genotypes (p = 0.018) on the day of study inclusion. This was accompanied by an increased IL-33 concentration in the CC genotype (day 0: p = 0.0026 and day 3: p = 0.008). In summary, the C allele of the AQP3 polymorphism (rs17553719) shows an association with increased AQP3 expression and IL-33 concentration accompanied by decreased survival in patients with sepsis.

AQP5-1364A/C Polymorphism Affects AQP5 Promoter Methylation

The quantity of aquaporin 5 protein in neutrophil granulocytes is associated with human sepsis-survival. The C-allele of the aquaporin (AQP5)-1364A/C polymorphism was shown to be associated with decreased AQP5 expression, which was shown to be relevant in this context leading towards improved outcomes in sepsis. To date, the underlying mechanism of the C-allele—leading to lower AQP5 expression—has been unknown. Knowing the detailed mechanism depicts a crucial step with a target to further interventions. Genotype-dependent regulation of AQP5 expression might be mediated by the epigenetic mechanism of promoter methylation and treatment with epigenetic-drugs could maybe provide benefit. Hence, we tested the hypothesis that AQP5 promoter methylation differs between genotypes in specific types of immune cells.: AQP5 promoter methylation was quantified in cells of septic patients and controls by methylation-specific polymerase chain reaction and quantified by a standard curve. In cell-line models, AQP5 expression was analyzed after demethylation to determine the impact of promoter methylation on AQP5 expression. C-allele of AQP5-1364 A/C promoter polymorphism is associated with a five-fold increased promoter methylation in neutrophils (p = 0.0055) and a four-fold increase in monocytes (p = 0.0005) and lymphocytes (p = 0.0184) in septic patients and healthy controls as well. In addition, a decreased AQP5 promoter methylation was accompanied by an increased AQP5 expression in HL-60 (p = 0.0102) and REH cells (p = 0.0102). The C-allele which is associated with lower gene expression in sepsis is accompanied by a higher methylation level of the AQP5 promoter. Hence, AQP5 promoter methylation could depict a key mechanism in genotype-dependent expression.

The impact of the COVID-19 pandemic on non-COVID induced sepsis survival

BACKGROUND

The COVID-19 pandemic has taken a toll on health care systems worldwide, which has led to increased mortality of different diseases like myocardial infarction. This is most likely due to three factors. First, an increased workload per nurse ratio, a factor associated with mortality. Second, patients presenting with COVID-19-like symptoms are isolated, which also decreases survival in cases of emergency. And third, patients hesitate to see a doctor or present themselves at a hospital. To assess if this is also true for sepsis patients, we asked whether non-COVID-19 sepsis patients had an increased 30-day mortality during the COVID-19 pandemic.

METHODS

This is a post hoc analysis of the SepsisDataNet.NRW study, a multicentric, prospective study that includes septic patients fulfilling the SEPSIS-3 criteria. Within this study, we compared the 30-day mortality and disease severity of patients recruited pre-pandemic (recruited from March 2018 until February 2020) with non-COVID-19 septic patients recruited during the pandemic (recruited from March 2020 till December 2020).

RESULTS

Comparing septic patients recruited before the pandemic to those recruited during the pandemic, we found an increased raw 30-day mortality in sepsis-patients recruited during the pandemic (33% vs. 52%, p = 0.004). We also found a significant difference in the severity of disease at recruitment (SOFA score pre-pandemic: 8 (5 - 11) vs. pandemic: 10 (8 - 13); p < 0.001). When adjusted for this, the 30-day mortality rates were not significantly different between the two groups (52% vs. 52% pre-pandemic and pandemic, p = 0.798).

CONCLUSIONS

This led us to believe that the higher mortality of non-COVID19 sepsis patients during the pandemic might be attributed to a more severe septic disease at the time of recruitment. We note that patients may experience a delayed admission, as indicated by elevated SOFA scores. This could explain the higher mortality during the pandemic and we found no evidence for a diminished quality of care for critically ill sepsis patients in German intensive care units.