Validation of Prognostic Clinical Risk Scores in Predicting Outcome for Patients With COVID-19 at Initial Triage

Dr Adnan Agha

Status: Enrolling

Conditions: COVID-19

Treatments: Other: logistic regression of known prognostic markers of severity of COVID19

Study type: Observational

Funder types: Other

Identifiers

NCT05582382

CMHS_IntMed_DOH/CVDC/2020/1251

Details and patient eligibility

About

Background Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) causing Covid-19 pandemic continues to be a global health threat with a massive burden on health care systems resulting in more than six million deaths in 188 countries. Because of wide clinical spectrum of disease severity, having clinically applicable prognostic tools for early identification of patients at high risk of progression to severe / critical illness is essential to guide clinical decision making and resource allocation efforts. So far, clinical prognostic tools have focused on host factors, but more recent data indicated a significant association between SARS-CoV-2 variants and the development of complications such as long COVID.

Objectives

1. Validation of the ALA & ALKA prediction tools for initial evaluation of patients diagnosed with COVID-19 infection.
2. Comparison of performance of the ALA & ALKA prediction tools with the currently clinical risk assessment scoring system used during initial evaluation of patients diagnosed with COVID-19 infection.
3. Evaluation of the clinical risk assessment scoring based on number of comorbidities in prediction of COVID-19 related complications
4. Assessment of the association between SARS-CoV-2 variants and the risk of COVID-19 severity
5. Assessment of the impact of SARS-CoV-2 variants on the performance of ALA & ALKA prediction tools

Methods Data will be abstracted from electronic medical records including demographics, clinical manifestation, comorbidities, and initial laboratory data in patients with Covid 19 infection of around 2000 patients presented initially to COVID assessment centre, including SARS CoV-2 sequencing data. Furthermore, population level SARS-CoV-2 RNA sequence data will also be examined and correlated with COVID-19 severity and the performance of prediction tools.

Full description

Background:

Since December 2019, when severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) causing COVID -19 disease emerged in Wuhan city and on 11 March 2020 rapidly spread into the rest of the world including UAE as a pandemic. COVID-19 continues to be a global health threat with a massive burden on health care systems resulting in more than six million deaths in 188 countries (1).

COVID-19 infection is characterized by a wide clinical spectrum of disease severity ranging from asymptomatic illness to severe disease that may progress to life-threatening complications such as shock and acute respiratory distress syndrome (2). Thus, having clinically applicable prognostic tools for early identification of symptomatic patients at high risk of progression to severe / critical illness is essential to guide allocating limited healthcare resources (3). So far, clinical prognostic tools have focused on host factors, but more recent data indicated a significant association between SARS-CoV-2 variants and the development of complications such as long COVID (4).

Currently, the clinical assessment for patients with COVID-19 infection is based on patient's age, number of comorbidities, subjective symptoms, and extent of pulmonary infiltrate on radiological examination which makes early prediction of severe / critical illness rather difficult (5-7). A recently published prognostic prediction tools (ALA & ALKA) were proposed to aid triaging patients with COVID-19 infection on initial diagnosis (8). These prediction tools are based on simple readily available laboratory tests and therefore may offer a clear advantage over other tools to guide discharge and admission decisions in triage assessment centers Nevertheless, external validation of these simple tools using another cohort of patients would provide a stronger evidence to support their utility in triaging patients on initial diagnosis. In addition, it will also allow further optimization of these tools to improve their utility as clinical decision support tools to triage patients on initial diagnosis. Patients deemed to be high risk based on these predictive tools could be triaged to hospital admission where intensive care unit (ICU) is available in anticipation of worse outcome. Therefore, these patients may benefit from earlier initiation of the required level of care and support including specific therapy.

The aim of this study is to validate and compare the ALA & ALKA prediction tools with the currently clinical risk assessment scoring system proposed for initial evaluation of patients with COVID-19 infection.

Methodology:

An observational longitudinal follow up of all consecutive patients with positive SARS-CoV-2 testing on nasopharyngeal swabs per WHO definitions presenting to the emergency department . Furthermore, population level SARS-CoV-2 RNA sequence data will also be examined and correlated with COVID-19 severity and the performance of prediction tools.

Data will be abstracted from electronic medical records using a data collection tool. This includes demographics, clinical manifestation, number of comorbidities, initial laboratory and radiological examination results and their final outcomes as detailed below.

The risk assessment score at initial presentation will be calculated for each patient using clinical assessment scoring of ALA & ALKA and compared with the currently proposed clinical risk assessment scoring system

The utility of the risk score in triaging patients on their initial visits to emergency department (ED) will be validated against the following measured outcomes:

1. Hospital admission on the first encounter to ED
2. Admission to ICU for the duration of the COVID-19 hospitalization
3. In hospital and out of hospital mortality
4. Return to ED following initial discharge (within the current covid illness period, Maximum 30 days from the initial diagnosis)

Sample Collection Process:

Data will be abstracted from electronic medical records using a data collection tool. The data would include demographics, clinical manifestation, comorbidities, laboratory and radiological results, and final outcomes.

The assessment risk score at initial presentation will be calculated using a free web-based online calculator.

Data Handling & Analysis:

Descriptive statistics will be generated for all variables. Multivariate logistic regression models to fit for outcomes. Variables incorporated in the COVID-19 risk of score will be included in the regression analysis to predict the outcomes. Multivariate logistic regression results will be presented in terms of adjusted Odds Ratios with corresponding 95% confidence intervals and p-values.

Discrimination will be evaluated using C-Statistic, along with its corresponding 95% Confidence Intervals and Receiver Operating Characteristic (ROC) curve. C-Statistics ≥ 0.7 will be considered good and ≥ 0.8 will be considered excellent (9). Calibration will be assessed based on the predicted probability for the outcome as predicted from the regressions. Calibration curves will be generated. P-values <0.05 is considered statistically significant. All analysis will be performed using SPSS software (version 28, IBM Corp, NY, USA).

Enrollment: 2,000 estimated patients

Sex: All

Ages: 16 to 99 years old

Volunteers: No Healthy Volunteers

Inclusion criteria

• All consecutive patients with positive SARS-CoV-2 testing on nasopharyngeal swabs per WHO definitions presenting to the emergency department
• All patients admitted to the hospital for isolation purposes only

Exclusion criteria

• Inconclusive PCR results on initial or repeat results with 24 hours