Applying Machine Learning for Prescriptive Support: A Use Case with Unfractionated Heparin in Intensive Care Units.

Continuous unfractionated heparin is widely used in intensive care, yet its complex pharmacokinetic properties complicate the determination of appropriate doses. To address this challenge, we developed machine learning models to predict over- and under-dosing, based on anti-Xa results, using a monocentric retrospective dataset. The random forest model achieved a mean AUROC of 0.80 [0.77-0.83], while the XGB model reached a mean AUROC of 0.80 [0.76-0.83]. Feature importance was employed to enhance the interpretability of the model, a critical factor for clinician acceptance. After prospective validation, machine learning models such as those developed in this study could be implemented within a computerized physician order entry (CPOE) as a clinical decision support system (CDSS).

Tinzaparin, an alternative to subcutaneous unfractionated heparin, in patients with severe and end-stage renal impairment: a retrospective observational single-center study.

BACKGROUND: Tinzaparin could be easier to manage than unfractionated heparin in patients with severe renal impairment. However, clinical and pharmacologic data regarding its use in such patients are lacking. OBJECTIVES: The aims of this study were to determine, in patients with estimated glomerular filtration rate (eGFR) of <30 mL.min⁻1, tinzaparin pharmacokinetics (PK) parameters using a population PK approach and bleeding and thrombotic complications. METHODS: We performed a retrospective observational single-center study, including in-patients with eGFR of <30 mL.min⁻1 receiving prophylactic (4500 IU.d⁻1) or therapeutic (175 IU.kg⁻1.d⁻1) tinzaparin. Measured anti-Xa levels were analyzed using a nonlinear mixed-effects modeling approach. Individual predicted tinzaparin exposure markers at steady state were calculated for each patient and dosing regimen. The PK was also evaluated through Monte Carlo simulations based on the final covariate model parameter estimates. RESULTS: Over a 22-month period, 802 tinzaparin treatment periods in 623 patients were analyzed: two-thirds received a prophylactic dose, 66% had an eGFR of <20 mL.min⁻1, and 25% were on renal replacement therapy. In patients for whom anti-Xa measurements were performed (n = 199; 746 values), PK parameters, profiles, and maximum plasma concentrations were comparable with those in patients without renal impairment or in healthy volunteers. In the whole population, major bleeding occurred in 2.4% and 3.5% of patients receiving prophylactic and therapeutic doses over a median 9- and 7-day treatment period, respectively. No patients had thrombotic complications. CONCLUSION: Tinzaparin PK parameters and profiles were not affected by renal impairment. This suggests that tinzaparin, at therapeutic or prophylactic dose, could be an alternative to unfractionated heparin in hospitalized patients with severe renal impairment.

Impact of Intraoperative Allogeneic Platelet Transfusion on Healthcare-Associated Infections in Cardiac Surgery: Insights From a Large Single-Center Cohort Study.

OBJECTIVES: Despite significant improvement in patient blood management, cardiac surgery remains a high hemorrhagic risk procedure. Platelet transfusion is used commonly to treat thrombocytopenia-associated perioperative bleeding. Allogeneic platelet transfusion may induce transfusion-related immunomodulation. However, its association with postoperative healthcare-associated infections is still a matter of debate. The objective was to evaluate the impact of allogeneic platelet transfusion during cardiac surgery on postoperative healthcare-associated infection incidence. DESIGN: Retrospective cohort study. SETTING: Tertiary referral academic center. PARTICIPANTS: Patients undergoing cardiac surgery from 2012 to 2018. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Intraoperative platelet transfusion was defined as exposure in a causal model. The primary outcome was the incidence of healthcare-associated infections comprised of bloodstream infection, hospital-acquired pneumonia, and surgical-site infection. Among 7,662 included patients, 528 patients (6.8%) were exposed to intraoperative platelet transfusion, and 329 patients (4.3%) developed 454 postoperative infections. Bloodstream infection affected 106 patients (1.4%), hospital-acquired pneumonia affected 174 patients (2.3%), and surgical-site infection affected 148 patients (1.9%). Intraoperative platelet transfusion was associated with an increased risk of bloodstream infection after adjustment by multivariable logistic regression (odds ratio [OR] 2.85; 95% CI 1.40-5.8; p = 0.004; n = 7,662), propensity score matching (OR 3.95; 95% CI 1.57-12.0), p = 0.007; n = 766), and propensity score overlap weighting (OR 3.04; 95% CI 1.51-6.1, p = 0.002; n = 7,762). Surgical-site infection and hospital-acquired pneumonia were not significantly associated with platelet transfusion. CONCLUSIONS: These results suggested that intraoperative allogeneic platelet transfusion is a risk factor for bloodstream infection after cardiac surgery. These results supported the development of patient blood management strategies aimed at minimizing perioperative platelet transfusion in cardiac surgery.

Healthcare-associated infections in patients with severe COVID-19 supported with extracorporeal membrane oxygenation: a nationwide cohort study.

BACKGROUND: Both critically ill patients with coronavirus disease 2019 (COVID-19) and patients receiving extracorporeal membrane oxygenation (ECMO) support exhibit a high incidence of healthcare-associated infections (HAI). However, data on incidence, microbiology, resistance patterns, and the impact of HAI on outcomes in patients receiving ECMO for severe COVID-19 remain limited. We aimed to report HAI incidence and microbiology in patients receiving ECMO for severe COVID-19 and to evaluate the impact of ECMO-associated infections (ECMO-AI) on in-hospital mortality. METHODS: For this study, we analyzed data from 701 patients included in the ECMOSARS registry which included COVID-19 patients supported by ECMO in France. RESULTS: Among 602 analyzed patients for whom HAI and hospital mortality data were available, 214 (36%) had ECMO-AI, resulting in an incidence rate of 27 ECMO-AI per 1000 ECMO days at risk. Of these, 154 patients had bloodstream infection (BSI) and 117 patients had ventilator-associated pneumonia (VAP). The responsible microorganisms were Enterobacteriaceae (34% for BSI and 48% for VAP), Enterococcus species (25% and 6%, respectively) and non-fermenting Gram-negative bacilli (13% and 20%, respectively). Fungal infections were also observed (10% for BSI and 3% for VAP), as were multidrug-resistant organisms (21% and 15%, respectively). Using a Cox multistate model, ECMO-AI were not found associated with hospital death (HR = 1.00 95% CI [0.79-1.26], p = 0.986). CONCLUSIONS: In a nationwide cohort of COVID-19 patients receiving ECMO support, we observed a high incidence of ECMO-AI. ECMO-AI were not found associated with hospital death. Trial registration number NCT04397588 (May 21, 2020).

Interventions to improve appropriateness of laboratory testing in the intensive care unit: a narrative review.

Healthcare expenses are increasing, as is the utilization of laboratory resources. Despite this, between 20% and 40% of requested tests are deemed inappropriate. Improper use of laboratory resources leads to unwanted consequences such as hospital-acquired anemia, infections, increased costs, staff workload and patient stress and discomfort. The most unfavorable consequences result from unnecessary follow-up tests and treatments (overuse) and missed or delayed diagnoses (underuse). In this context, several interventions have been carried out to improve the appropriateness of laboratory testing. To date, there have been few published assessments of interventions specific to the intensive care unit. We reviewed the literature for interventions implemented in the ICU to improve the appropriateness of laboratory testing. We searched literature from 2008 to 2023 in PubMed, Embase, Scopus, and Google Scholar databases between April and June 2023. Five intervention categories were identified: education and guidance (E&G), audit and feedback, gatekeeping, computerized physician order entry (including reshaping of ordering panels), and multifaceted interventions (MFI). We included a sixth category exploring the potential role of artificial intelligence and machine learning (AI/ML)-based assisting tools in such interventions. E&G-based interventions and MFI are the most frequently used approaches. MFI is the most effective type of intervention, and shows the strongest persistence of effect over time. AI/ML-based tools may offer valuable assistance to the improvement of appropriate laboratory testing in the near future. Patient safety outcomes are not impaired by interventions to reduce inappropriate testing. The literature focuses mainly on reducing overuse of laboratory tests, with only one intervention mentioning underuse. We highlight an overall poor quality of methodological design and reporting and argue for standardization of intervention methods. Collaboration between clinicians and laboratory staff is key to improve appropriate laboratory utilization. This article offers practical guidance for optimizing the effectiveness of an intervention protocol designed to limit inappropriate use of laboratory resources.

Antithrombin Levels and Heparin Responsiveness during Venoarterial Extracorporeal Membrane Oxygenation: A Prospective Single-center Cohort Study.

BACKGROUND: Unfractionated heparin, administered during venoarterial extracorporeal membrane oxygenation to prevent thromboembolic events, largely depends on plasma antithrombin for its antithrombotic effects. Decreased heparin responsiveness seems frequent on extracorporeal membrane oxygenation; however, its association with acquired antithrombin deficiency is poorly understood. The objective of this study was to describe longitudinal changes in plasma antithrombin levels during extracorporeal membrane oxygenation support and evaluate the association between antithrombin levels and heparin responsiveness. The hypothesis was that extracorporeal membrane oxygenation support would be associated with acquired antithrombin deficiency and related decreased heparin responsiveness. METHODS: Adults receiving venoarterial extracorporeal membrane oxygenation were prospectively included. All patients received continuous intravenous unfractionated heparin using a standardized protocol (target anti-Xa 0.3 to 0.5 IU/ml). For each patient, arterial blood was withdrawn into citrate-containing tubes at 11 time points (from hour 0 up to day 7). Anti-Xa (without dextran or antithrombin added) and antithrombin levels were measured. The primary outcome was the antithrombin plasma level. In the absence of consensus, antithrombin deficiency was defined as a time-weighted average of antithrombin less than or equal to 70%. Data regarding clinical management and heparin dosage were collected. RESULTS: Fifty patients, including 42% postcardiotomy, were included between April 2020 and May 2021, with a total of 447 samples. Median extracorporeal membrane oxygenation duration was 7 (interquartile range, 4 to 12) days. Median antithrombin level was 48% (37 to 60%) at baseline. Antithrombin levels significantly increased throughout the follow-up. Time-weighted average of antithrombin levels was 63% (57 to 73%) and was less than or equal to 70% in 32 (64%) of patients. Overall, 45 (90%) patients had at least one antithrombin value less than 70%, and 35 (70%) had at least one antithrombin value less than 50%. Antithrombin levels were not significantly associated with heparin responsiveness evaluated by anti-Xa assay or heparin dosage. CONCLUSIONS: Venoarterial extracorporeal membrane oxygenation support was associated with a moderate acquired antithrombin deficiency, mainly during the first 72 h, that did not correlate with heparin responsiveness.

Combined Platelet and Red Blood Cell Recovery during On-pump Cardiac Surgery Using same™ by i-SEP Autotransfusion Device: A First-in-human Noncomparative Study (i-TRANSEP Study).

BACKGROUND: Centrifugation-based autotransfusion devices only salvage red blood cells while platelets are removed. The same™ device (Smart Autotransfusion for ME; i-SEP, France) is an innovative filtration-based autotransfusion device able to salvage both red blood cells and platelets. The authors tested the hypothesis that this new device could allow a red blood cell recovery exceeding 80% with a posttreatment hematocrit exceeding 40%, and would remove more than 90% of heparin and 75% of free hemoglobin. METHODS: Adults undergoing on-pump elective cardiac surgery were included in a noncomparative multicenter trial. The device was used intraoperatively to treat shed and residual cardiopulmonary bypass blood. The primary outcome was a composite of cell recovery performance, assessed in the device by red blood cell recovery and posttreatment hematocrit, and of biologic safety assessed in the device by the washout of heparin and free hemoglobin expressed as removal ratios. Secondary outcomes included platelet recovery and function and adverse events (clinical and device-related adverse events) up to 30 days after surgery. RESULTS: The study included 50 patients, of whom 18 (35%) underwent isolated coronary artery bypass graft, 26 (52%) valve surgery, and 6 (12%) aortic root surgery. The median red blood cell recovery per cycle was 86.1% (25th percentile to 75th percentile interquartile range, 80.8 to 91.6) with posttreatment hematocrit of 41.8% (39.7 to 44.2). Removal ratios for heparin and free hemoglobin were 98.9% (98.2 to 99.7) and 94.6% (92.7 to 96.6), respectively. No adverse device effect was reported. Median platelet recovery was 52.4% (44.2 to 60.1), with a posttreatment concentration of 116 (93 to 146) · 109/l. Platelet activation state and function, evaluated by flow cytometry, were found to be unaltered by the device. CONCLUSIONS: In this first-in-human study, the same™ device was able to simultaneously recover and wash both platelets and red blood cells. Compared with preclinical evaluations, the device achieved a higher platelet recovery of 52% with minimal platelet activation while maintaining platelet ability to be activated in vitro.

Factors Influencing Anti-Xa Assays: A Multicenter Prospective Study in Critically Ill and Noncritically Ill Patients Receiving Unfractionated Heparin.

BACKGROUND: The presence of dextran sulfate (DS) in reagents and the type of blood collection tube (citrate/citrated-theophylline-adenosine-dipyridamole [CTAD]) can lead to discrepancies between unfractionated heparin (UFH) anti-Xa levels. OBJECTIVES: To evaluate the extent of the effect (1) of different reagents containing or not containing DS and (2) of the blood collection tubes, on UFH anti-Xa levels, in various clinical situations (NCT04700670). METHODS: We prospectively included patients from eight centers: group (G)1, cardiopulmonary bypass (CPB) after heparin neutralization (n = 39); G2, cardiothoracic intensive care unit (ICU) after CPB (n = 35); G3, medical ICU (n = 53); G4, other medical inpatients (n = 38). Blood was collected into citrated and CTAD tubes. Chromogenic anti-Xa assays were centrally performed, using seven reagent/analyzer combinations including two without DS. The association between anti-Xa levels and covariates was tested using a linear mixed-effects model. RESULTS: We analyzed 4,546 anti-Xa values from 165 patients. Median anti-Xa levels were systematically higher with reagents containing DS, whatever the patient group, with the greatest effect observed in G1 (0.32 vs. 0.05 IU/mL). Anti-Xa levels were slightly higher in CTAD than in citrate samples, irrespective of the assay. The model showed: (1) a significant dextran-patient group interaction (p < 0.0001), the effect of DS on anti-Xa levels varying from 30.9% in G4 to 296% in G1, and (2) a significant effect of CTAD, varying between patient groups (p = 0.0302). CONCLUSION: The variability of anti-Xa levels with a great overestimation of the values, using a reagent containing DS, can lead to different treatment decisions, especially after heparin neutralization by protamine. Clinical consequences of these differences remain to be demonstrated.

Veno-arterial extracorporeal membrane oxygenation for circulatory failure in COVID-19 patients: insights from the ECMOSARS registry.

OBJECTIVES: The clinical profile and outcomes of patients with Coronavirus Disease 2019 (COVID-19) who require veno-arterial extracorporeal membrane oxygenation (VA-ECMO) or veno-arterial-venous extracorporeal membrane oxygenation (VAV-ECMO) are poorly understood. We aimed to describe the characteristics and outcomes of these patients and to identify predictors of both favourable and unfavourable outcomes. METHODS: ECMOSARS is a multicentre, prospective, nationwide French registry enrolling patients who require veno-venous extracorporeal membrane oxygenation (ECMO)/VA-ECMO in the context of COVID-19 infection (652 patients at 41 centres). We focused on 47 patients supported with VA- or VAV-ECMO for refractory cardiogenic shock. RESULTS: The median age was 49. Fourteen percent of patients had a prior diagnosis of heart failure. The most common aetiologies of cardiogenic shock were acute pulmonary embolism (30%), myocarditis (28%) and acute coronary syndrome (4%). Extracorporeal cardiopulmonary resuscitation (E-CPR) occurred in 38%. In-hospital survival was 28% in the whole cohort, and 43% when E-CPR patients were excluded. ECMO cannulation was associated with significant improvements in pH and FiO2 on day 1, but non-survivors showed significantly more severe acidosis and higher FiO2 than survivors at this point (P = 0.030 and P = 0.006). Other factors associated with death were greater age (P = 0.02), higher body mass index (P = 0.03), E-CPR (P = 0.001), non-myocarditis aetiology (P = 0.02), higher serum lactates (P = 0.004), epinephrine (but not noradrenaline) use before initiation of ECMO (P = 0.003), haemorrhagic complications (P = 0.001), greater transfusion requirements (P = 0.001) and more severe Survival after Veno-Arterial ECMO (SAVE) and Sonographic Assessment of Intravascular Fluid Estimate (SAFE) scores (P = 0.01 and P = 0.03). CONCLUSIONS: We report the largest focused analysis of VA- and VAV-ECMO recipients in COVID-19. Although relatively rare, the need for temporary mechanical circulatory support in these patients is associated with poor prognosis. However, VA-ECMO remains a viable solution to rescue carefully selected patients. We identified factors associated with poor prognosis and suggest that E-CPR is not a reasonable indication for VA-ECMO in this population.

Association of Arterial Blood pH at Cannulation With 1 Year Survival Among Veno-Arterial Extracorporeal Membrane Oxygenation Recipients: The Three Seven Rule.

We aimed to describe the clinical outcomes of veno-arterial extracorporeal membrane oxygenation (VA-ECMO) therapy in our institution considering clinical context and pH at cannulation. All patients treated by VA-ECMO during the 2005-2020 period with 1 year complete follow-up were included. Our cohort was divided in three groups according to the pH level at cannulation: pH <7 (group 1), pH 7-7.2 (group 2), and pH>7.2 (group 3). Survival was analyzed using Kaplan-Meier method. Association between pH group and survival was estimated using a Cox model. A total of 572 patients were included: 60 patients in group 1, 115 in group 2, and 397 in group 3. Refractory cardiogenic shock (36%) was the main indication. One year survival rates were 13%, 36%, and 43% in groups 1, 2, and 3, respectively ( p < 0.001). Death mainly occurred within the first month. The strong correlation between pH and lactates led to propose a simple "three seven rule": pH <7 and lactate >7 was associated with <7% survival. Veno-arterial extracorporeal membrane oxygenation should be considered with caution in patients with pH <7. Lactates and pH might be important parameters to elaborate a new score to predict survival in this population. The "three seven rule" can be very relevant when facing emergency situations.

Multiple-site decontamination to prevent acquired infection in patients with veno-venous ECMO support.

BACKGROUND: Acute distress respiratory syndrome (ARDS) patients with veno-venous extra corporeal membrane oxygenation (ECMO) support are particularly exposed to ECMO-associated infection (ECMO-AI). Unfortunately, data regarding AI prophylaxis in this setting are lacking. Selective decontamination regimens decrease AI incidence, including ventilator-associated pneumonia (VAP) and bloodstream infection (BSI) in critically ill patients. We hypothesized that a multiple-site decontamination (MSD) regimen is associated with a reduction in the incidence of AI among VV-ECMO patients. METHODS: We conducted a retrospective observational study in three French ECMO referral centers from January 2010 to December 2021. All adult patients (> 18 years old) who received VV-ECMO support for ARDS were eligible. In addition to standard care (SC), 2 ICUs used MSD, which consists of the administration of topical antibiotics four times daily in the oropharynx and the gastric tube, once daily chlorhexidine body-wash and a 5-day nasal mupirocin course. AIs were compared between the 2 ICUs using MSD (MSD group) and the last ICU using SC. RESULTS: They were 241 patients available for the study. Sixty-nine were admitted in an ICU that applied MSD while the 172 others received standard care and constituted the SC group. There were 19 ECMO-AIs (12 VAP, 7 BSI) in the MSD group (1162 ECMO-days) compared to 143 AIs (104 VAP, 39 BSI) in the SC group (2376 ECMO-days), (p < 0.05 for all infection site). In a Poisson regression model, MSD was independently associated with a lower incidence of ECMO-AI (IRR = 0.42, 95% CI [0.23-0.60] p < 0.001). There were 30 multidrug resistant microorganisms (MDRO) acquisition in the SC group as compared with two in the MSD group (IRR = 0.13, 95% CI [0.03-0.56] p = 0.001). Mortality in ICU was similar in both groups (43% in the SC group vs 45% in the MSD group p = 0.90). Results were similar after propensity-score matching. CONCLUSION: In this cohort of patients from different hospitals, MSD appeared to be safe in ECMO patients and may be associated with improved outcomes including lower ECMO-AI and MDRO acquisition incidences. Since residual confounders may persist, these promising results deserve confirmation by randomized controlled trials.

Impact of Prone Position in COVID-19 Patients on Extracorporeal Membrane Oxygenation.

OBJECTIVES: Prone positioning and venovenous extracorporeal membrane oxygenation (ECMO) are both useful interventions in acute respiratory distress syndrome (ARDS). Combining the two therapies is feasible and safe, but the effectiveness is not known. Our objective was to evaluate the potential survival benefit of prone positioning in venovenous ECMO patients cannulated for COVID-19-related ARDS. DESIGN: Retrospective analysis of a multicenter cohort. PATIENTS: Patients on venovenous ECMO who tested positive for severe acute respiratory syndrome coronavirus 2 by reverse transcriptase polymerase chain reaction or with a diagnosis on chest CT were eligible. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: All patients on venovenous ECMO for respiratory failure in whom prone position status while on ECMO and in-hospital mortality were known were included. Of 647 patients in 41 centers, 517 were included. Median age was 55 (47-61), 78% were male and 95% were proned before cannulation. After cannulation, 364 patients (70%) were proned and 153 (30%) remained in the supine position for the whole ECMO run. There were 194 (53%) and 92 (60%) deaths in the prone and the supine groups, respectively. Prone position on ECMO was independently associated with lower in-hospital mortality (odds ratio = 0.49 [0.29-0.84]; p = 0.010). In 153 propensity score-matched pairs, mortality rate was 49.7% in the prone position group versus 60.1% in the supine position group (p = 0.085). Considering only patients alive at decannulation, propensity-matched proned patients had a significantly lower mortality rate (22.4% vs 37.8%; p = 0.029) than nonproned patients. CONCLUSIONS: Prone position may be beneficial in patients supported by venovenous ECMO for COVID-19-related ARDS but more data are needed to draw definitive conclusions.

Bleeding and thrombotic events in patients with severe COVID-19 supported with extracorporeal membrane oxygenation: a nationwide cohort study.

PURPOSE: To describe bleeding and thrombotic events and their risk factors in patients receiving extracorporeal membrane oxygenation (ECMO) for severe coronavirus disease 2019 (COVID-19) and to evaluate their impact on in-hospital mortality. METHODS: The ECMOSARS registry included COVID-19 patients supported by ECMO in France. We analyzed all patients included up to March 31, 2022 without missing data regarding bleeding and thrombotic events. The association of bleeding and thrombotic events with in-hospital mortality and pre-ECMO variables was assessed using multivariable logistic regression models. RESULTS: Among 620 patients supported by ECMO, 29% had only bleeding events, 16% only thrombotic events and 20% both bleeding and thrombosis. Cannulation site (18% of patients), ear nose and throat (12%), pulmonary bleeding (9%) and intracranial hemorrhage (8%) were the most frequent bleeding types. Device-related thrombosis and pulmonary embolism/thrombosis accounted for most of thrombotic events. In-hospital mortality was 55.7%. Bleeding events were associated with in-hospital mortality (adjusted odds ratio (adjOR) = 2.91[1.94-4.4]) but not thrombotic events (adjOR = 1.02[0.68-1.53]). Intracranial hemorrhage was strongly associated with in-hospital mortality (adjOR = 13.5[4.4-41.5]). Ventilation duration before ECMO ≥ 7 days and length of ECMO support were associated with bleeding. Thrombosis-associated factors were fibrinogen ≥ 6 g/L and length of ECMO support. CONCLUSIONS: In a nationwide cohort of COVID-19 patients supported by ECMO, bleeding incidence was high and associated with mortality. Intracranial hemorrhage incidence was higher than reported for non-COVID patients and carried the highest risk of death. Thrombotic events were less frequent and not associated with mortality. Length of ECMO support was associated with a higher risk of both bleeding and thrombosis, supporting the development of strategies to minimize ECMO duration.