The nutritional assessment in critically ill COVID-19 patients is an important element for outcome prediction. In this study patients with high nutritional risk (mNUTRIC ≥ 5 points) at ICU admission showed an increased probability of 28-day mortality than those with low nutritional risk (mNUTRIC < 5 points). Moreover, the mNUTRIC score had the highest sensitivity and specificity for outcome prediction compared to SOFA and APACHEII.
A high prevalence of malnutrition (37.5%) was recently reported in a general cohort of COVID-19 inpatients and 26% of them showed severe malnutrition17. The same findings were observed in elderly patients with COVID-19 in a cross sectional study in Wuhan, China4. Many possible features can lead to malnutrition in COVID-19 patients. The presence of dyspnea, dysgueusia, anosmia, anorexia, dysphagia, nausea, vomiting and diarrhea can decrease food intake. Simultaneously, caloric and protein intake may be inadequate during the pre-intubation period5.
A state of hyper-metabolism and increased energy expenditure measured by indirect calorimetry were recently observed in a small cohort of critically ill COVID-19 patients18. Therefore, acute underfeeding is a possible consequence and it can lead to immunosuppression and to inflammatory response impairment19. In fact, almost all our patients showed lymphopenia and even if this finding is common in critically ill COVID-19 patients20, malnutrition can itself increase lymphopenia and the risk of infectious complications21.
In ICU the prevalence of malnutrition in COVID-19 patients is even higher (66.7%)22 and the hypermetabolic phase is prolonged up to 21 days since intubation23.
The high prevalence of malnutrition in COVID-19 patients and the potential role of individualised nutritional support to improve clinical outcomes and survivals24 suggest the need to screen the inpatients nutritional risk and to implement individualised nutritional support in patients at risk25.
The NUTRIC score was developed as a scoring method for quantifying nutrition risk in ICU patients26 and the American Society for Parenteral and Enteral Nutrition (ASPEN) recommended this score together with the Nutrition Risk Screening 2002 (NRS 2002) for nutritional risk screening in critically ill patients7. Conversely, the European Society for Parenteral and Enteral Nutrition (ESPEN) no longer recommended the use of NUTRIC score since is does not include nutritional parameters and it is heavily influenced by APACHE II and SOFA6. However, no standardized method currently exists to assess the patient nutritional risk in ICU. The mNUTRIC was the evolution of NUTRIC since IL-6 value was removed from the score because the contribution of this variable was not statistically and clinically useful26. Moreover, as it was recently noted, the mNUTRIC and NRS-2002 scores have similar performance in predicting hospital mortality but the mNUTRIC has a better discriminatory ability for mortality prediction in critically ill patients27. In our study a high nutritional risk (mNUTRIC ≥ 5 points) was found in 41.8% of the cohort and the probability for 28-day mortality had more than doubled compared to low nutritional risk (mNUTRIC < 5 points) patients. As is was recently reported by Zhang et al., a mNUTRIC score ≥ 5 was observed in 61% of critically ill COVID-19 patients and the mortality of this group (87%) was significantly higher than in the low nutritional risk group (49%)12.
Our results are in line with this cited paper and the optimal mNUTRIC cut-off was confirmed as 5 points score for high nutritional risk assessment. On the contrary, almost all our patients required mechanical ventilation and the principal reason for ICU admission was related to the severity of respiratory failure. Since the median time from respiratory symptoms onset to ICU admission was 12 days (9–16 IQR), it is possible to argue that the increased catabolism and the poor nutritional intake worsened the respiratory failure. A global mortality of 46% was found in our critically ill Covid-19 patients and this data is in line with previous findings in our country28.
Along with the nutritional risk assessment, a recently published meta-analysis showed that higher levels of inflammatory markers such as C-reactive protein (CRP) have been associated with COVID-19 severity and could be considered as prognostic factor29. Our predictive model showed that CRP plasmatic levels were independently associated with increased risk of 28-day mortality. In particular, for every increase in one point in CRP levels, the 28-day mortality hazard increased by 2%. It is noteworthy that our results are in keeping with previous studies reporting the role of increased CRP levels in predicting disease severity30 and the need of mechanical ventilation31 in COVID-19 patients. Moreover, a positive correlation between CRP concentrations and the extension of lung lesions was also found in COVID-19 patients32.
Obesity is a frequently reported risk factor for ICU admission and the need of invasive mechanical ventilation33. On the contrary, in our multivariable analysis obesity was not a risk factor for mortality in ICU patients. It is important to note that only 9 patients in our cohort showed a BMI ≥ 30 kg/m2 in the high and in the low nutritional risk group. Consequently, this variable could have been underpowered. On the other hand, the mortality risk associated with the underweight is frequently reported with an excessively wide confidence interval34. However, as we previously found in a large cohort of critically ill COVID-19 patients, only morbid obesity (BMI > 40 kg/m2) was a risk factors for death35. Probably, due to the inability of BMI to differentiate between fat and lean body mass and to account for edema and sodium retention, this variable is less useful for outcome prediction. Moreover, no significant associations was found between body composition and disease severity after bioelectric impedance measurements36.
The validation of a prognostic model is a fundamental step to implement its use in clinical practice. As it was recently outlined, a possible risk of bias can occur in many prognostication model for patients with COVID-19 due to a lack of validation37. Therefore, we decided to use a bootstrap resampling technique for internal validation of our model. The bootstrap resampling draws random samples with replacement from the derivation cohort. Consequently, the prognostic model is evaluated both in the derivation cohort and in the bootstrap sample in order to assess its performance such as discrimination and calibration and to reduce the risk of potential false positive prediction estimates.
To our knowledge, this is the first internal validated model with the use of mNUTRIC score to predict 28-day mortality in critically ill COVID-19 patients. Even if our model showed a good discriminatory accuracy for 28-day mortality prediction in critically ill COVID-19 patients, more prospective studies are needed to evaluate the role of mNUTRIC and how to reduce malnutrition onset in these patients.
This study has several limitations. First, it is a single centre retrospective study during a pandemic that heavily affect medical resources. Therefore, some data might present inaccuracies and bias.
A limited number of subjects were enrolled, however this issue is quite counteracted by the internal validation process. Nutritional risk assessment was performed at ICU admission and no dynamic evaluations were provided. Consequently, we cannot infer the role of nutritional support in terms of outcome and if the nutritional risk changed over time. Another major problem is the lack of albumin and pre-albumin levels or others rapid-turnover visceral proteins or muscle mass assessment in our patients as classical laboratory indexes for nutritional assessment. These data were available in only a small amount of patients without reaching the required minimum sample size to be analysed. Although a low concentration of albumin and pre-albumin was found both in high and low nutritional risk patients12, the principal endpoint of our study was simply the evaluation of 28-day mortality and at the time of writing this paper all the patients concluded the follow-up period.
Further analyses are certainly needed to confirm these results and to evaluate the role of mNUTRIC to predict complications such as secondary infection, organ failure or thrombosis during the ICU length of stay of critically ill COVID-19 patients.