Multidrug-resistant organisms refer to bacteria that present resistance to three or more classes of antimicrobial agents simultaneously; among them, carbapenem-resistant microorganisms (CRO), especially carbapenem-resistant gram-negative bacilli, are the most frequently implicated, including carbapenem-resistant Enterobacteriaceae (CRE), carbapenem-resistant Acinetobacter baumannii (CRAB), and carbapenem-resistant Pseudomonas aeruginosa (CRPA) (Magiorakos et al., 2012).

Since the past decade, the incidence of CRO infection has been increasing worldwide and has led to a poor prognosis (Xu et al., 2017; Touat et al., 2019; Merrick et al., 2021). CRO infection has also created challenging situations in China. The 2020 national antimicrobial surveillance report suggests that the carbapenem resistance rates of Klebsiella pneumoniae, Pseudomonas aeruginosa, and Acinetobacter baumannii are above 23%, 19%, and 72%, respectively (Hu et al., 2021). Considering the serious global situation, both the World Health Organization and the United States Centers for Disease Control and Prevention have established the CRO risk level as the top rank and have suggested the implementation of active and effective prevention and control measures for CRO infection to ensure patient safety (Guh et al., 2014; WHO Guidelines, 2017).

The emergence and spread of CRO resistance are associated with multiple factors such as inappropriate use of antimicrobials, poor practice of infection prevention and control, and so on (Bonomo et al., 2018). The principle of prevention and control is to take bundling measures to block the spread of CRO (Guh et al., 2014; Wilson et al., 2016; Friedman et al., 2017; Magiorakos et al., 2017; WHO Guidelines, 2017; Solter et al., 2018), the primary measures being strict management of the use of antimicrobials and active screening for infection (Wilson, 2016). The formulation of concrete prevention and control strategies relies on the outcomes of screening (Orsi et al., 2011; Richter and Marchaim, 2017). However, screening is time consuming and costly, and the ideal target population for screening and frequency of testing continue to remain debatable (Birgand et al., 2016; Al-Zahrani, 2018; Kois et al., 2020; Bedos et al., 2021; Merrick et al., 2021).

Recently, as a visualization tool, a nomogram based on multiple logistic regression was developed to predict the risk of CRE infection among liver transplantation recipients (Giannella et al., 2021) and patients in the intensive care unit (ICU) of a secondary referral hospital (Seo et al., 2021). However, there is no established risk prediction model for CRO infection among ICU patients. This study therefore aimed to develop a nomogram to determine the risk of acquiring CRO infection in patients with first ICU admission and to explore the length of ICU stay (ICU-LOS) and 28-day survival in patients with CRO infection.

This study showed that the risk factors for CRO infection were male sex, Hb-min, temperature-max, PICC use, dialysis treatment, and use of carbapenems. The nomogram was established to predict the risk of acquiring CRO infection based on these six variables determined by Logistic regression analysis. The nomogram was also internally validated, wherein a better performance was noted, and because of geographical, racial, and medication differences, external validation was not performed.

When patients are admitted to the ICU for the first time, we may tend to overlook their risk of acquiring CRO infection and may not perform monitoring. Public health workers can use this nomogram to easily assess the risk of acquiring CRO infection among ICU patients and to decide on the development of concrete prevention and control strategies.

In this study, approximately 4.04% of patients with first ICU admission had CRO infection, a proportion that was lower than that reported in other studies (Li et al., 2019; Giannella et al., 2021; Seo et al., 2021). Patients with CRO infection had significantly higher ICU-LOS, treatment costs, and mortality rates (Tansarli et al., 2013; Birgand et al., 2016; Ho et al., 2016; Avendano et al., 2020; Merrick et al., 2021). Early identification of patients with CRO infection is important for disease prevention and control. The prevalence of CRO infection in a region is influenced by multiple factors (Wilson, 2016) such as the overall medical care level, age, administration of antibiotics, disease severity, and so on. Moreover, CRO in this study was defined based on database data, which may have had many influences on the true infection rate.

CRO mainly include CRE, CRAB, and CRPA. Currently available surveillance and screening measures primarily target CRE (Gniadek et al., 2016), which is an essential component of CRO. In the US, it is reported that the common risk factors for CRE infection are the underlying disease severity, long-term hospital admissions, indwelling invasive devices, poor functional status, exposure to antimicrobials, and increasing colonization pressure (Guh et al., 2014). At a secondary referral hospital in Korea, a retrospective study suggested that the risk factors for carbapenem-resistant infection among ICU patients were indwelling nasogastric tube and central venous catheter (CVC), a high Acute Physiology and Chronic Health Evaluation (APACHE) II score, and administration of cephalosporin antibiotics (Seo et al., 2021). Additional risk factors were reported in a China study (Li et al., 2019). However, the risk of carbapenem resistance in glucose-nonfermenting gram-negative bacilli (CR-NF), such as Pseudomonas aeruginosa and Acinetobacter baumannii, is increasing significantly (Gniadek et al., 2016). However, the methods for the surveillance and screening of CR-NF are still controversial.

In this study, we sought to investigate the acquisition risk of CRO infection. Interestingly, we found that male sex was associated with the risk of acquiring CRO infection. This may be related to the constitutive ratio of CRO-infected individuals and differences in the immunity between sexes (Klein et al., 2015). In a tertiary teaching hospital in China, it is reported that 80% patients with CRE infection were male, and male sex is an independent risk factor for CRE infection (Liu, 2016). And in ICU, compared with female patients, male patients were more likely to undergo invasive treatments, such as deep venous puncture, mechanical ventilation and dialysis (Xu et al., 2019), which may increase the risk of CRO infection. Further studies are needed to investigate the relationship between sex and the risk of developing CRO infection. Anemia and fever are important characteristic features of CRO infection. Anemia is frequently seen in different forms of the infection (Kwaan, 2011), which may be related to abnormalities in iron metabolism (Jonker and Boele van Hensbroek, 2014). PICC refers to catheters placed into the central vein by peripheral venipuncture. A meta-analysis showed that PICC-related central line-associated bloodstream infection (CLABSI) occurred as frequently as CVC-related CLABSI (Chopra et al., 2013). As per our understanding, indwelling PICC is often used for specific categories of patients or for patients requiring prolonged use, which will lead to an increased risk of CRO infection. Dialysis is one of the important supportive measures for ICU patients. The latest matched case-control study suggested that ICU patients undergoing dialysis may be at increased risk for CRE infection, a finding that is consistent with those reported in a study (Aleidan et al., 2021). The same may be true for the risk of CRO infection. Inappropriate use of antimicrobials is an important reason for the development of bacterial resistance. Studies have suggested a positive association between exposure to broad-spectrum antimicrobials and carbapenem resistance (Sampaio and Gales, 2016; Moawad et al., 2018; Yang et al., 2018; Alzomor et al., 2019). Strict management of antimicrobial use remains the cornerstone to avoid bacterial resistance.

Currently, multiple risk factors are strongly associated with CRO infection. However, there is established no scoring system available to predict the risk of CRO infection. Partly, the reasons may be the differences in medical conditions, economic development levels, and antimicrobial usage in different regions (Liu et al., 2015). In this study, we developed a nomogram to predict CRO acquisition risk in patients with first ICU admission, which showed better performance. Clinicians can dynamically assess each patient’s CRO acquisition risk using the developed nomogram. If the score increases significantly, then we should make more efforts to control the risk factors and conduct targeted surveillance. For example, if a male patient in an ICU presents with fever and anemia, has the use of carbapenems and an implanted catheter, and undergoes dialysis, then we should actively manage the fever and anemia; reassess the need for catheters, dialysis, and carbapenems; and complete active screening for infection. By controlling for the emerging risk factors, the occurrence of CRO infection could be decreased. Meanwhile, we can identify patients with CRO infection at an early stage and reduce nosocomial transmission through targeted active surveillance. Accordingly, costs associated with the prevention and control of the infection will also decrease.

This study had several limitations. First, the incidence of CRO infection was calculated based on culture-positive patients in the MIMIC-IV database. It is possible that a few patients with CRO infection remained undiagnosed. Second, because of geographical variations, risk factors for CRO infection may not have been entirely consistent. Finally, the newly established nomogram was not validated based on external data. And the sensitivity of the nomogram was medium (0.55), the specificity was good (0.85), the negative predictive value was high (0.98), but the positive predictive value was not high (0.14) ( Table S2 ). It may be related to the sample size and the prevalence of CRO infection. As a complementary measure, we plotted a dynamic nomogram for more extensive validation.

In conclusion, a nomogram that included six variables was developed to predict CRO acquisition risk in ICU patients based on a LASSO regression and Logistic analysis, which showed a good performance through internal validation. Based on scoring using the nomogram, we can aim to control the risk factors and conduct targeted active surveillance in a timely manner for the purpose of reducing the occurrence of CRO infection, identifying patients with CRO infection at an early stage, reducing nosocomial transmission, and decreasing costs related to prevention and control. Further external validation is needed to optimize and improve the nomogram.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material .

Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

MY and JZ: concept. JZ and WL: methodology and writing of the manuscript. WS and XC: data processing. WX and YL: software. MY and TH: review and editing. All authors contributed to the article and approved the submitted version.

This study was supported by a research grant from the National Natural Science Foundation of China (No.82072134) and the Clinical Research cultivation Program of the Second Affiliated Hospital of Anhui Medical University (No.2020LCYB03).

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.