Prediction models identify patients at low risk for COVID-19 infection

Singapore researchers have developed prediction models that distinguish patients at low risk for the novel coronavirus disease (COVID-19) infection, which then leads to better rationalization of use of healthcare resources, reports a study.

“Validation of the models in external cohorts is recommended before adopting their use in clinical practice,” the researchers said.

This single-centre retrospective observational study analysed 1,228 patients admitted to Changi General Hospital’s respiratory surveillance wards from 10 February to 30 April 2020. The researchers derived prediction models for COVID-19 from a training cohort using variables based on demographics, clinical symptoms, exposure risks, and blood investigations fitted into logistic regression models. They then validated these models on a test cohort.

Fifty-two patients (4.2 percent) were diagnosed with COVID-19. Two predictions models were derived: the first based on age, presence of sore throat, dormitory residence, blood haemoglobin level (Hb), and total white blood cell counts (TW); the second on presence of headache, contact with infective patients, Hb, and TW. [Singapore Med J 2021;doi:10.11622/smedj.2021019]

The diagnostic performance of both models was good, with areas under the receiver operating characteristic curve of 0.934 and 0.866, respectively. The risk score cutoffs of 0.6 for model 1 and 0.2 for model 2 demonstrated 100-percent sensitivity, which allowed the identification of patients with low risk for COVID-19. By limiting the screening for COVID-19 to only those with elevated risk, the number of isolation days for surveillance patients had been reduced by up to 41.7 percent and of COVID-19 swab testing by up to 41.0 percent.

“Enhanced surveillance through systematic screening of patients with acute respiratory infection (ARI) is an effective but highly resource-intensive exercise,” the researchers said. “Up to 10 percent of acute hospital beds, including isolation rooms, may be used for such surveillance programmes alone.” [Infect Control Hosp Epidemiol 2020;41:820-825]

Expected trade-offs of this strategy were missed cases of COVID-19. However, basic infection control measures such as surgical masks and standard hand hygiene practices have been effective in reducing the risks of nosocomial spread to healthcare workers and patients with inadvertent exposure to COVID-19 cases. [Ann Intern Med 2020;172:766-767; J Hosp Infect 2020;105:119-127]

In a previous systematic review of prediction models for COVID-19 diagnosis, existing, nonpeer-reviewed studies showed substantial selection bias, and several models included variables (eg, interleukin-6 and computed tomography of the chest) that are not routinely performed in clinical practice. [medRxvi 2020;doi:10.1101/2020.03.19.20039099v1; medRxiv 2020;doi:10.1101/2020.03.05.20031906]

“Our prediction models were based on an unbiased cohort, as all patients who were admitted for pneumonia and symptoms of ARI underwent SARS-CoV-2 testing,” the researchers said. “Consequently, our results are likely to be more useful for clinical triaging, as the main challenge lies in risk-stratifying patients with no travel history to high-risk regions or contact history with COVID19 cases.”

In addition, both predictions models use variables that are readily available from routine clinical history and blood investigations, allowing practical implementation.