Carbapenem-resistant Gram-negative bacteria are a prevalent nosocomial infection with poor prognosis and high mortality, especially in intensive care^2.3. More data on CRGNB infections in the ICU are needed to explore prognostic factors and outcomes to guide treatment decisions. In this multicenter CRGNB nosocomial pneumonia intensive care study, significant predictors of in-hospital mortality included length of hospital stay before pneumonia greater than 9 days, SOFA score greater than 7, supportive care with vasopressor therapy and increased antimicrobial therapy. Subgroup analysis showed that patients with multi-organ involvement or a SOFA score greater than 7 had a better survival outcome when receiving colistin-based therapy.

Through nationwide surveillance in Taiwan since 2011, increased resistance to carbapenems has been demonstrated in A.baumannii, K. pneumoniae, and P. aeruginosa 69.8 to 75.3%, 10.5 to 44.4%, and 17.7 to 22.6%, respectively²³. Previous studies have demonstrated that risk factors, such as carbapenem consumption, LOS, diabetes mellitus, multiple lobar involvement, and prior exposure to antimicrobial therapy, are associated with acquisition of infection. at the NBGRC^24,25,26,27. In addition, the mortality rate of nosocomial pneumonia caused by CRGNB is also high. When pooling data from previous clinical studies, the mortality rate was greater than 40%, even in the subgroup of patients receiving combination therapy with colistin^9.10. As the only prognostic factor amenable to intervention, the appropriateness of empiric antimicrobial therapy has been widely evaluated. Despite the promising results of the first observational studies^7,8,28a more recent randomized controlled trial failed to demonstrate benefit of empiric treatment with colistin to improve its suitability for ventilator-associated pneumonia²⁹. There are concerns about the potential for development of nephrotoxicity and emergence of antimicrobial resistance following treatment with colistin^30.31. Given the poor prognosis and controversial role of colistin-based therapy for nosocomial pneumonia caused by CRGNB, our study aimed to investigate prognostic factors to improve treatment decisions.

In this study, we present a multicenter retrospective study of 690 critically ill patients with nosocomial pneumonia caused by CRGNB. Consistent with previous studies^9,10,22, approximately 80% of patients had late nosocomial pneumonia (defined as pneumonia after five or more days of hospitalization) with a median length of stay before onset of pneumonia of 13 days. About 46% of patients ultimately died in hospital. In addition, we noted that the length of stay in hospital following the diagnosis of pneumonia was prolonged, with a median of 28 days. These findings reinforce the importance that while treating patients with nosocomial pneumonia caused by CRGNB, clinicians should be aware of the high mortality and economic burden. Every effort should be made to prevent nosocomial cross-transmission between patients.

Using multivariate logistic regression analysis, we identified that a length of hospital stay greater than 9 days before pneumonia onset, a SOFA score greater than 7, and supportive care with vasopressor treatment negatively affected the prognosis. The escalation of antimicrobial treatment positively influenced patient survival. Rapid escalation of treatment is an important factor since it is currently the only intermediate factor that has been shown to improve clinical outcomes.

Previous studies have demonstrated that a prolonged length of stay before the onset of pneumonia, with a threshold ranging from 4 to 7 days, was associated with the risk of acquiring potentially resistant pathogens, leading to an increase in mortality.^1,32,33. In our study, which involved patients with nosocomial pneumonia caused by CRGNB, we found that prolonged hospitalization for more than 9 days was an independent prognostic factor, independent of comorbidity and disease severity, with a sensitivity, specificity and area under the receiver operating characteristic curve (AUROC) of 71.4%, 48.9% and 0.63, respectively. This phenomenon could be explained by the fact that patients with a prolonged hospital stay tend to be more frail and have a greater complexity of the disease, which leads to high mortality.³⁴.

Septic shock and multi-organ failure are considered common and life-threatening complications of pneumonia³⁵especially in cases caused by NBRMC⁸. Consistent with previous studies, we noted that a high SOFA score and vasopressor requirement were important risk factors for mortality. The present study demonstrated that a SOFA score greater than 7 was associated with increased mortality with sensitivity, specificity, and AUROC of 65.4%, 65.1%, and 0.70, respectively. To identify the organ damage pattern, we applied a k-means algorithm using SOFA subscore data. K-means is one of the most widely used clustering techniques in unsupervised learning³⁶. Regarding the complex and heterogeneous nature of critically ill patients, phenotype identification has been well introduced in previous studies, especially in the area of ​​sepsis.^37.38 and acute respiratory distress syndrome³⁹. In our study, using a simple algorithm and six parameters, with no other clinical and biological features, we identified two distinct subtypes with different trajectories and outcomes. This finding suggests that the assessment of individual SOFA subscores rather than the total score alone might be useful in clinical practice for severity stratification.

In the present study, treatment escalation was recognized as the only modifiable factor to decrease in-hospital mortality. Due to the high rate of prescriptions for escalation therapy, we further assessed the influence of colistin therapy on patient outcomes. Although colistin is often considered a reliable option for the treatment of CRGNB infection, previous meta-analyses have shown significant effects only for clinical cure and microbiological eradication, but not for mortality.^10.20, which was similar to our results. Since different severity stratification criteria can be a factor influencing treatment response and clinical outcome, we further analyzed and demonstrated that colistin-based treatment was associated with better survival outcomes in sub -groups of patients with a SOFA score greater than 7 and involvement of several organs. These results reinforce that while treating critically ill patients with nosocomial pneumonia in the setting of high CRGNB prevalence, high SOFA scores and the presence of multi-organ involvement should not be overlooked by clinicians. clinicians. In this condition, rapid escalation of colistin therapy may improve patient outcomes.

Our multicenter design overcomes the potential for single-center bias and enhances the external validity of our results. Additionally, we studied a large critically ill population in five tertiary teaching hospitals covering different regions in Taiwan, confirming its high representative reliability. However, our study has several limitations. First, given the nature of the retrospective analysis, missing data were unavoidable and the pattern of antimicrobial therapy prescription was heterogeneous and complex. Therefore, it is not possible to directly compare various treatment strategies and determine the most effective therapy for nosocomial pneumonia caused by CRGNB. Second, in vitro colistin susceptibility was not tested at all participating hospitals. In addition, the reporting of upper limit MIC values ​​for meropenem/doripenem was not at all consistent. Therefore, it is impractical to assign patients based on treatment inadequacy and assess its influence on patient outcomes. Third, as an important complication of pneumonia, acute respiratory distress syndrome was not recorded and could be a confounder in this study. Fourth, due to lack of ground truth reference, k-means clustering performance was only evaluated by internal validation. Further prospective studies should be performed to assess external validity and confirmation of robustness.