Predicting COVID-19 progression in hospitalized patients in Belgium from a multi-state model

Abstract

Objectives: To adopt a multi-state risk prediction model for critical disease/mortality outcomes among hospitalised COVID-19 patients using nationwide COVID-19 hospital surveillance data in Belgium.

Materials and methods: Information on 44,659 COVID-19 patients hospitalised between March 2020 and June 2021 with complete data on disease outcomes and candidate predictors was used to adopt a multi-state, multivariate Cox model to predict patients' probability of recovery, critical [transfer to intensive care units (ICU)] or fatal outcomes during hospital stay.

Results: Median length of hospital stay was 9 days (interquartile range: 5-14). After admission, approximately 82% of the COVID-19 patients were discharged alive, 15% of patients were admitted to ICU, and 15% died in the hospital. The main predictors of an increased probability for recovery were younger age, and to a lesser extent, a lower number of prevalent comorbidities. A patient's transition to ICU or in-hospital death had in common the following predictors: high levels of c-reactive protein (CRP) and lactate dehydrogenase (LDH), reporting lower respiratory complaints and male sex. Additionally predictors for a transfer to ICU included middle-age, obesity and reporting loss of appetite and staying at a university hospital, while advanced age and a higher number of prevalent comorbidities for in-hospital death. After ICU, younger age and low levels of CRP and LDH were the main predictors for recovery, while in-hospital death was predicted by advanced age and concurrent comorbidities.

Conclusion: As one of the very few, a multi-state model was adopted to identify key factors predicting COVID-19 progression to critical disease, and recovery or death.

Keywords: Belgium; COVID-19; hospital data; multistate modelling; risk prediction model.

FIGURE 1

Schematic representation of the multi-state model with four states and five transition events, including the event matrix for training and Testing set. ^aNo event represents censored or absorbing state. ^bTraining set was taken in 10-fold for building the prediction model on stacked multiple-imputed data. ^cTesting set for calculating prediction error includes complete cases only (excluding 57% of the Testing patients with missing data on any of the predicting variables).

FIGURE 2

Prediction error of the multi-state prediction model as compared with a null model, calculated using the Brier score, overall (A) and stratified by state (B).

FIGURE 3

Plots for the cumulative transition hazards (A) and state-occupation probabilities (B) in a multi-state model considering a high-risk patient (tick solid line) versus a reference patient (thin solid line). The reference patient represents an average patient characterised by being male, aged between 55 and 69 years old, admitted to a general hospital with a medium-high ICU occupancy at hospital admission, experiencing lower respiratory complaints, having zero co-morbidities, being a non-smoker, and presenting high levels of lymphocyte and medium-high levels (Q3) of lactate dehydrogenase (LDH) and C-reactive protein (CRP). The high-risk patient represents an existing patient with the worst risk profile, while having the same age, sex, hospital type and ICU occupancy at admission as the reference patient.

FIGURE 4

Nomograms to predict the transition probability for admission to recovery (A), ICU (B), and in-hospital death (C) after 2, 3, 7, and 14 days of hospitalisation in COVID-19 hospitalised COVID-19 patients. Patients’ values are located on the axis of each variable where 0 refers to “No” and 1 “Yes”; drawing an upward line at 90° angle to determine the number of points for that particular variable. The sum of these numbers of points is located on the total score axis; drawing a downward line at 90° angle to determine the probability of experiencing that particular transition at day 2, 3, 7, and 14.