Evolution of Clinical Phenotypes of COVID-19 Patients During Intensive Care Treatment: An Unsupervised Machine Learning Analysis (preprint)

Background: Identification of distinct clinical phenotypes in critically ill COVID-19 patients could improve understanding of the disease heterogeneity and enable prognostic and predictive enrichment facilitating more personalized treatment. However, previous attempts did not take into account temporal dynamics of the disease. By including the dimension of time, we aim to gain further insights into the heterogeneity of COVID-19.Methods: We used highly granular data from 3202 adult critically ill COVID patients in the multicenter Dutch Data Warehouse that were admitted to one of 25 Dutch ICUs between February 2020 and March 2021. Parameters including demographics, clinical observations, medications, laboratory values, vital signs, and data from life support devices were selected based on relevance and availability. Twenty-one consecutive datasets were created that each covered 24 hours of ICU data for each day of ICU treatment up until day 21. After aggregation and multiple imputation of the temporal data, clinical phenotypes in each dataset were identified by performing multiple cluster analyses. Clinical phenotypes were identified by aggregating values from all patients per cluster. Both evolution of the clinical phenotypes over time and patient allocation to these clusters over time were tracked.Results: The final patient cohort consisted of 2438 critically ill COVID-19 patients with a registered ICU mortality outcome. Forty-one parameters were chosen for the cluster analysis. On admission, both a mild and a more severe clinical phenotype were found. After day 4, the severe phenotype split into an intermediate and a severe phenotype for 11 consecutive days. Heterogeneity between phenotypes appears to be strongly driven by inflammation and dead space ventilation. During the 21-day period only 8.2% and 4.6% of patients in the initial mild and severe clusters remained assigned to the same phenotype respectively. The clinical phenotype half-life was between 5 and 6 days for the mild and severe phenotypes, and about 3 days for the medium severe phenotype.Conclusions: Patients typically do not remain in the same cluster throughout intensive care treatment. This may have important implications for prognostic or predictive enrichment. Prominent dissimilarities between clinical phenotypes are predominantly driven by inflammation and dead space ventilation.

A pragmatic approach to estimating average treatment effects from EHR data: the effect of prone positioning on mechanically ventilated COVID-19 patients (preprint)

Despite the recent progress in the field of causal inference, to date there is no agreed upon methodology to glean treatment effect estimation from observational data. The consequence on clinical practice is that, when lacking results from a randomized trial, medical personnel is left without guidance on what seems to be effective in a real-world scenario. This article proposes a pragmatic methodology to obtain preliminary but robust estimation of treatment effect from observational studies, to provide front-line clinicians with a degree of confidence in their treatment strategy. Our study design is applied to an open problem, the estimation of treatment effect of the proning maneuver on COVID-19 Intensive Care patients.

COVID Outcome Prediction in the Emergency Department (COPE) (preprint)

Background and aimThe COVID-19 pandemic is putting extraordinary pressure on emergency departments (EDs). To support decision making about hospital admission, we aimed to develop a simple and valid model for predicting mortality and need for admission to an intensive care unit (ICU) in suspected-COVID-19 patients presenting at the ED. MethodsFor model development, we included patients that presented at the ED and were admitted to 4 large Dutch hospitals with suspected COVID-19 between March and August 2020, the first wave of the pandemic in the Netherlands. Based on prior literature we included patient characteristics, vital parameters and blood test values, all measured at ED admission, as potential predictors. Logistic regression analyses with post-hoc uniform shrinkage was used to obtain predicted probabilities of in-hospital death and of being admitted to the ICU, both within 28 days after admission. Model performance (AUC; calibration plots, intercepts and slopes) was assessed with temporal validation in patients who presented between September and December 2020 (second wave). We used multiple imputation to account for missing predictor values. ResultsThe development data included 5,831 patients who presented at the ED and were hospitalized, of whom 629 (10.8%) died and 5,070 (86.9%) were discharged within 28 days after admission. A simple model - named COVID Outcome Prediction in the Emergency Department (COPE) - with linear age and logarithmic transforms of respiratory rate, CRP, LDH, albumin and urea captured most of the ability to predict death within 28 days. Patients who were admitted in the first month of the pandemic had substantially increased risk of death (odds ratio 1.99; 95% CI 1.61-2.47). COPE was well-calibrated and showed good discrimination for predicting death in 3,252 patients of the second wave (AUC in 4 hospitals: 0.82; 0.82; 0.79; 0.83). COPE was also able to identify patients at high risk of needing IC in second wave patients below the age of 70 (AUC 0.84; 0.81), but overestimated ICU admission for low-risk patients. The models are implemented as a web-based application. ConclusionCOPE is a simple tool that is well able to predict mortality and ICU admission for patients who present to the ED with suspected COVID-19 and may help to inform patients and doctors when deciding on hospital admission.

Cardiovascular risk factors are independently associated with COVID-19 mortality: a prospective cohort study (preprint)

Objectives: Recent reports suggest a high prevalence of hypertension and diabetes in COVID-19 patients, but the role of cardiovascular disease (CVD) risk factors in the clinical course of COVID-19 is unknown. We evaluated the time-to-event relationship between hypertension, dyslipidemia, diabetes, and COVID-19 outcomes. Design: We analyzed data from the prospective Dutch COVID-PREDICT cohort, an ongoing prospective study of patients admitted for COVID-19 infection. Setting: Patients from 8 participating hospitals, including two university hospitals from the COVID-PREDICT cohort were included. Participants: Admitted, adult patients with a positive COVID-19 polymerase chain reaction (PCR) or high suspicion based on CT-imaging of the thorax. Patients were followed for major outcomes during hospitalization. CVD risk factors were established via home medication lists and divided in antihypertensives, lipid lowering therapy, and antidiabetics. Primary and secondary outcomes measures: The primary outcome was mortality during the first 21 days following admission, secondary outcomes consisted of ICU-admission and ICU-mortality. Kaplan-Meier and Cox-regression analyses were used to determine the association with CVD risk factors. Results: We included 1604 patients with a mean age of 66+-15 of whom 60.5% were men. Antihypertensives, lipid lowering therapy, and antidiabetics were used by 45%, 34.7%, and 22.1% of patients. After adjustment for age and sex, the presence of [≥]2 risk factors was associated with increased mortality risk (HR 1.52, 95%CI 1.15-2.02), but not with ICU-admission. Moreover, the use of [≥]2 antidiabetics and [≥]2 antihypertensives was associated with mortality independent of age and sex with HRs of respectively 2.09 (95%CI 1.55-2.80) and 1.46 (95%CI 1.11-1.91). Conclusions: The accumulation of hypertension, dyslipidemia and diabetes leads to a stepwise increased risk for short-term mortality in hospitalized COVID-19 patients independent of age and sex. Further studies investigating how these risk factors disproportionately affect COVID-19 patients are warranted.