Emerging SARS-CoV-2 BA.2.2 threatens patients’ lives all over the world.

The COVID-19 pandemic has continued to overwhelm healthcare systems worldwide. The outbreak of Omicron sub-variant BA.2 in Shanghai in March 2022 has resulted in heavy medical burden to the healthcare system and economy recession (1). Due to the effective vaccination and variation of the virus, clinical characteristics of Omicron inflicted patients are totally different from those who were infected by COVID 19 when it broke out in Wuhan in 2020 (2). Effective triage of patients presenting to the hospital for risk of progressive deterioration is crucial to inform clinical decision making and to facilitate effective resource allocation, including hospital beds, critical care resources, and targeted drug therapies (3). Since the majority of Omicron sub-variant BA.2 patients show mild symptoms and a low mortality rate, early identification of subgroups of patients at a high risk of death or deterioration and requiring ventilation enables precise delivery of treatments.

Many multivariable clinical prognostic models for patients with COVID-19 have been developed to predict adverse outcomes, such as mortality or clinical deterioration (4–7). The deterioration model developed by the International Severe Acute Respiratory and emerging Infections Consortium Coronavirus Clinical Characterization Consortium (ISARIC4C) study (4) has combined 11 predictors to predict clinical deterioration and achieved a good clinical utility (c-statistic = 0.77), but the population included both confirmed and suspected COVID-19 patients and the missing data in the study led to the difficulty in integrating important variables into the model, such as D-dimer, which may cause the negligence of the clinical characteristics. Fang et al. (5) established a prognostic model based on 10 variables, which achieved a good prognostic value (AUC = 0.89), but it did not include patients aged ≥89, and it had a decent percentage of missing data. COVID-GRAM (6) established a deterioration model which included 1,590 Chinese patients and combined 10 indicators to build an online free risk calculator, which had a good prognostic value (AUC = 0.88), but the mean age of the included population was 48.9 years, and it was based on the COVID-19 outbreak in 2020 in Wuhan, China. The end-stage liver disease (MELD)(7) score model included 4,213 COVID-19 confirmed patients, but the model was also based on data collected during the outbreak of COVID-19 in 2020. Due to the difference in multiple respects between the two outbreaks (8), there is a surging demand to build a new model to estimate the risk of deterioration for patients inflicted by Omicron variants.

In the present study, a large cohort of COVID-19 patients with confirmed Omicron variant BA2.2 was utilized to develop a prognostic model for in-hospital clinical deterioration (requirement for ventilatory support or death) and its efficacy was validated in a separate cohort.

The selection process of the study population was illustrated in Figure 1. A total of 1,830 patients with COVID-19 were enrolled and 575 excluded due to their age younger than 65. Baseline characteristics between the training cohort and the validation cohort were shown in Tables 1, 2. The average time from PCR confirmation to admission was 2 days. Nearly half of them had hypertension (49.5%). The majority of these elderly patients showed mild to moderate symptoms, with a small proportion of them having severe and critical symptoms. Most patients received the Paxlovid (84.5%) and anticoagulation (71.0%) treatments. The proportion of deterioration rate was 11.6%. No difference was observed in above variables between the training cohort and the validation cohort.

In the present study, a prediction model on the outbreak of Omicron sub-variant BA.2 in Shanghai was built using clinical data from our hospital. It was found that high levels of CRP, NLR, INR, BUN, and decreased A/G at admission were significantly correlated with the probability of clinical deterioration. Using these 5 factors, an effective prognostic nomogram was constructed, which had a significantly high sensitivity and specificity to identify individuals with a high risk of deterioration. DCA and CICA further demonstrated the superiority of our nomogram evidenced by the net clinical benefit, which is invaluable for individualized assessment of in-hospital deterioration. To our best knowledge, this study is the first to build such a prediction model targeting the Omicron sub-variant BA.2. Other available models are based on clinical information of alpha or delta variant breaking out in 2020 or 2021.

From the baseline data, it can be seen that even the elderly in Shanghai showed mild to moderate symptoms although Omicron sub-variant BA.2 is so contagious. The deterioration rate is quite low, which is different from the outbreak of COVID-19 in Wuhan in 2020 (16). This might be attributed to the free vaccination among the public and strict quarantine policies in China (17, 18).

In the present study, continuous variables were used to build the prediction model, rather than categorical variables which are not always reliable and highly dependent on self-awareness. In contrast, laboratory tests are relatively accurate and readily available. High levels of CRP, NLR, INR, BUN, and low level of A/G are positively correlated with the deterioration rate. Compared with previous prediction models where age is always considered a major factor that influences the progression (4, 19, 20), our model did not take age into account. This is due to the observed characteristics in our study and a previous one (21), all patients were the elderly aged 65 or older. Initially, 1,830 patients were admitted to our hospital, but nearly all of the endpoint events occurred in elderly patients (age ≥ 65 years), only 3 patients had clinical deterioration among 575 patients younger than 65. In order to minimize the bias, we decided to focus on the elderly patients. Besides, age ≥ 65 is considered to be at highest risk for severe COVID-19 associated illness (22). In this elderly population, age is not an important factor that contributes to the endpoint events.

According to the previous research, the mechanism of COVID-19 infection is associated with inflammatory cytokine storm, oxidative stress, disseminated intravascular coagulation and distribution of ACE2 along the vascular endothelium, which lead to the multisystem dysfunction (23). Correlations between high levels of CRP and NLR and disease progression have been proven in many trials (24, 25), which have shown that Omicron BA.2 induces an inflammatory process evidenced by the activation of neutrophils and lymphocytes as well as the release of IL-6, CRP, and other inflammatory factors. INR is an index that reflects the functional status of the coagulation system. Increased INR values were significantly associated with COVID-19 severity and mortality (26). BUN reflects the function of the kidney, which has been validated as a significant risk factor of the disease severity (27). The reduced albumin-globulin ratio has been found to be a risk factor of COVID-19 severity in patients with cancer (28), and of great diagnostic significance in predicting the progression to severe disease states (29), which is consistent with our results.

Our study has a number of strengths: first, a large sample size and sufficient patients’ information guarantees the credibility of our conclusion. All of the included variables had no missing values and all the endpoints were collected during the study period. After the last confirmed COVID-19 patient was discharged, the hospital was closed for full disinfection. Second, our model is a practical quantitative prediction tool based on 5 features which are commonly used and easily obtained from routine blood tests. The performance of our nomogram is efficient for clinical practice. Third, the first-hand information of the Omicron outbreak in Shanghai in 2022 was of great practical use for medical practitioners in other countries where Omicron sub-variant BA.2 prevails.

The study also has a number of limitations. As a retrospective cohort study, we did not collect all the predictors that had been reported to be related to the progression of the disease, such as BMI, pro-BNP. Different clinical approaches were adopted at the early and late stages of this outbreak, full conformity to the recommended treatments is not guaranteed at different stages. Second, this is a single-centre study, and validation was performed in the same setting as the training cohort, an external validation would be much better to prove the effectiveness of the model. Furthermore, the construction of the model was based on hospitalized patients, our conclusion can not be extrapolated to the general population, but needs to be adjusted based on the different population characteristics.

In summary, our data suggest that our nomogram can predict the risk of deterioration (requirement for ventilatory support or death) in elderly patients and it is clinically practical, which will facilitate medical decision making and allocating medical resources to those with critical conditions in order to reduce mortality.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding authors.

The studies involving human participants were reviewed and approved by the associated Ethics Committee of Shanghai Fourth People’s Hospital approved the study and waived the need for informed consent. The patients/participants provided their written informed consent to participate in this study.

YB, YW, and ZY designed the initial concept. YB, YW, BX, JX, YH, HL, and ZY participated in acquisition and interpretation of data. YW and ZY performed the statistical analysis. YW, HL, and YB wrote the manuscript. All authors reviewed the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.