In this retrospective, single-center study the ability of different prognostic scores to predict intra-hospital mortality and ICU admission in patients with SARS-CoV-2 infection was evaluated. We decided to include most of the scores commonly used at the Emergency Room for the management of CAP (i.e. PSI, CURB-65 and extended CURB-65) [10-12] and to include also COVID-19 specific items, such as the recently proposed CALL score and the CRP value [16-17].
As for mortality, the PSI score performed best, with a median value belonging to class of risk IV, meaning an overall mortality of 9.3%, very similar to what observed in our cohort of patients, whereas median CURB-65 and extended CURB-65 accounted for low-to-moderate risk groups (up to 6.8% of estimated mortality) [10-12]. This finding was then reflected by the higher AUC of PSI score than of CURB-65 and extended CURB-65 scores, respectively, and suggests that CURB-65 or extended CURB-65 alone may not be fully suitable for COVID-19. Overall, the tested scores performed worse in predicting the need of ICU transfer.
Interestingly, both the CALL score and the CRP value, specifically designed for subjects with SARS-CoV-2 infection, were not able to predict the outcomes, rendering them not useful tools for discriminating patients at risk of death or ICU transfer, at least in our setting. In fact, the median value of CALL score was 10 (8-15), which corresponded to high risk of disease progression, significantly higher than that observed in our study .
The most interesting result of the present study was that expanding the CURB-65 score through the adjunt of hypoalbuminemia at hospital admission was able to increase the efficiency of predicting intra-hospital mortality in patients with SARS-CoV-2 pneumonia.
Several studies have identified a close correlation between hypoalbuminemia and mortality in CAP patients, whereas little is known as for SARS-CoV-2 pneumonia [20-26]. In fact, it has been recently demonstrated that an albumin level <3.5 g/dL was associated with worse outcome and coagulopathy [25-26]. In order to understand the pivotal role of albumin in SARS-CoV-2 infection, we should be aware that serum albumin possesses anti-oxidant and anti-inflammatory properties and behaves as an inverse acute phase reactant during acute systemic inflammation [27-29]. Furthermore, stressed and inflamed cells increase the uptake of albumin from the circulation, thus highlighting a complex relationship between inflammation/infection and albumin level in the extracellular matrix [30-31]. Last but not least, albumin down-regulates the expression in the cells of the ACE2 receptors, which are known to be crucial in mediating and expanding SARS-CoV-2 infection [32-33]. Therefore, patients with reduced levels of albumin are predisposed to poor survival and, according to our report, the assessment of baseline albumin serum level at the Emergency Room in addition to the CURB-65 score may represent a crucial and easy-to-perform tool for the early identification of patients at higher mortality risk.
While several scores have been investigated and further validated for optimal for the management of CAP, for COVID-19 patients drawing a definite conclusion among the many proposed prognostic biomarkers and scores appears to be very difficult [13-17]. The CALL score and a risk score including several parameters calculated at hospital admission (age, sex, ethnicity, oxygen saturations, radiology, neutrophil count, CRP, albumin, creatinine, comorbidities) were recently developed to predict 28-day mortality and ICU admission [9,16]. Furthermore, a predictive model (named the HNC-LL score and based on neutrophil count, lymphocyte count, LDH level, CRP level and hypertension) was also developed and validated to enable early and accurate identification of patients with SARS-CoV-2 infection at a high risk of severe disease . Finally, admission serum CRP at a cut-off value of 41.4 mg/L was shown to correlate well with disease severity and tended to be a good predictor of adverse outcome (i.e. mortality) in patients with SARS-CoV-2 pneumonia .
Taking into consideration the abovementioned scores, we decided to exclude from our analysis the score proposed by Galloway et al since disease severity was radiologically assessed with adoption of on chest radiograph examination that was replaced by chest CT in our patients population . The HNC-LL score was highly predictive of severe disease, with the definition of severity including both subjects requiring intensive care and those with respiratory distress but not requiring ICU admission . Even in this case, we did not include the HNC-LL score because we were mostly interested on mortality and ICU admission itself.
The present report has some clinical implications which should be highlighted. In fact, relying on a rapid and easy-to-perform instrument able to predict worse outcomes at hospital admission might represent a crucial step to early identify subjects with SARS-CoV-2 pneumonia at higher risk of intra-hospital mortality. Furthermore, the growing evidence that hypoalbuminemia plays a role in the survival rate of patients with COVID-19 might suggest its potential therapeutic use, especially taking into account the lack of a definite and specific (at least at the moment of writing) therapy.
The present study has several limitations. First, the retrospective and single-center nature of the study and the small number of included patients may have narrowed the interpretation of our results to a specific setting of population. Second, not all the published predictive scores in patients with SARS-CoV-2 pneumonia were included in our analysis. Finally, we did not evaluate whether subjects at hospital admission presented with de-novo hypoalbuminemia or with pre-existing conditions associated with low-level of albumin. Such effect, however, would better reflect the features of patients with SARS-CoV-2 pneumonia at higher risk of worse outcomes.
In conclusion, our results suggested that the addition of the observation of a low albumin level to the easy-to-calculate and well-known CURB-65 score at hospital admission is able to improve the quality of prediction of intra-hospital mortality in patients with SARS-CoV-2 infection. Further studies are needed to confirm this observation.