Introduction

Despite improvement in survival of immunocompromised patients admitted to the ICU [1, 2], this population still has a poor outcome, including high ICU-acquired infection (ICU-AI) incidence but also higher mortality than immunocompetent patients [1, 3]. Noteworthy, a previous study reported that ICU-AI in immunocompromised patients decreases survival, suggesting that ICU-AI prevention may improve outcomes [4]. Multiple-site decontamination (MSD) is a selective decontamination regimen that has been associated with decreased ICU-AI incidence in critical patients, but also with decreased mortality in some specific settings [5, 6]. Although the proportion of immunocompromised patients admitted to the ICU is increasing, data regarding specific management, including ICU-AI prophylaxis, in this setting are lacking and previous studies regarding selective decontamination were conducted in the previous century [7,8,9]. Interestingly, this population is at higher risk of nonbacterial infection, such as pulmonary aspergillosis or viral reactivation [4, 10], this type of micro-organisms being not directly targeted by MSD. We aim to investigate the effect of MSD in immunocompromised patients and hypothesized that MSD might improve outcomes through a decrease in ICU-AI incidence.

Patients and methods

Patients and setting

We conducted an ancillary analysis of a prospective pre/post observational study in 2 medico/surgical ICUs in Bretagne, western France. Anticipating a change in daily practice with MSD implementation in participating ICUs, an observational study with prospective collection of data has been conducted as reported elsewhere [11]. From 1 January 2020 until 31 December 2022, all adults who required mechanical ventilation for 24 h or more were eligible with the exception of those with liberty deprivation, pregnant women, and patients younger than 18 years old who were excluded from the study. Follow-up was pursued until ICU discharge or death, whichever occurs first.

Ethics

The study protocol received approval from the Rennes Hospital ethics committee (comité d’éthique du CHU de Rennes avis 19–52). Patients or closest relatives were informed of the anonymous prospective collection of the data and had the possibility not to participate in the study. In case of refusal, the data were not collected accordingly. This manuscript follows the STROBE statement for reporting cohort studies.

Intervention

As of May 5, 2021 (Saint-Brieuc), and June 1, 2021 (Vannes), MSD was implemented in participating ICUs in addition to standard care for the prevention of acquired infections in patients with expected intubation duration > 24 h. Patients admitted after the implementation date received MSD and constituted the post-implementation group (from 5 May 2021 to 31 December 2022, in Saint-Brieuc and from June 1, 2021, to December 31, 2022, in Vannes), whereas those admitted earlier, even when they were still hospitalized in the ICU after the implementation date, received standard-care and constituted standard-care group (from 1 January 2023 to 5 May 2021 in Saint-Brieuc and from 1 January 2023 to 1 June 2021 in Vannes). MSD is a variant of selective digestive decontamination, which consists of the administration of topical antibiotics including gentamicin (543 mg per day), colistin sulfate (400 mg per day), and amphotericin B (2 g per day), four times daily in the oropharynx and the gastric tube, 4% chlorhexidine body-wash once daily and a 5-day nasal mupirocin course without intravenous antibiotics [6, 11].MSD was applied in intubated patients with an expected intubation duration > 24 h from admission and during the full length of mechanical ventilation duration.

Strategies for ICU-AI prevention and diagnosis

Strategies for VAP and BSI prevention were left at each ICU’s discretion but they were no modifications of practices during the study period with the exception of MSD implementation in concerned ICUs. Standard care strategy for ICU-AI prevention was applied during both pre- and post-implementation periods and consists of a bundle of care that included semi-recumbent positioning (depending on its feasibility and tolerance), specific oral care with tooth brushing and mouth washing every 6 h and 4 times daily cuff pressure monitoring. There was no specific protocol for ulcer prophylaxis. Catheter dressings were performed with dry sterile compresses and were changed weekly or sooner in case of bleeding or spotting.

VAP diagnosis was systematically associated with a pulmonary sample that can be an endotracheal aspiration, a broncho-alveolar lavage, or a distally protected sample. During the study period, physicians were asked to complete a checklist for each VAP suspicion in order to collect data (clinical, biological, and radiological findings) for VAP classification by the dedicated team in each center [12].

Patients were classified as having possible, putative, probable, or proven aspergillosis according to the AspICU, Influenza-associated pulmonary aspergillosis (IAPA), and COVID-19-associated pulmonary aspergillosis (CAPA) criteria when indicated [13,14,15] (Supplementary Fig 1).

ICU-acquired infections diagnoses were prospectively recorded by an external committee that was not blinded to the pre- or post-intervention study period. Herein, ICU-AI diagnosis was suspected by the treating physician but the final diagnosis was confirmed by a dedicated member of the nosocomial infection committee (CLIN) in each hospital. The CLIN was composed of a microbiologist, infectiologist, and physician including a member of the intensive care unit in each hospital.

In all participating ICUs, patients were screened for MDRO rectal carriage at admission, weekly afterward, and at discharge on rectal swabs. As described elsewhere, patients with no prior colonization (no colonization at admission) who tested positive for MDRO on either rectal screening or on a blood or respiratory sample were considered as having MDRO acquisition [16].

Definition

Immunodepression was considered in patients with ongoing solid organ neoplasia (active or in remission for less than 5 years), haematological malignancy, severe neutropenia (< 0.5 G/L) acquired immunodeficiency syndrome, organ transplant or taking immunosuppressive drugs including long term steroids > 10 mg prednisolone per day during > 28 days) [4, 17].

Each center had a CLIN for the prevention and prospective census of ICU-AI and applied the recommendations of the French Society for Hospital Hygiene for the prevention and treatment of infection (available at https://sf2h.net/publications/actualisation-precautions-standard-2017). Infection was considered acquired in the ICU when it was diagnosed 48 h after admission and was not incubating on admission. ICU-AI was diagnosed by the treating physician. BSI was defined as a positive blood culture occurring 48 h or more after admission. Regarding common skin contaminants, 2 positive blood cultures drawn on separate occasions were required [18]. The diagnosis of VAP was considered in patients ventilated for 48 h or more and until 48 h after extubation and was based on clinical signs (fever, purulent sputum, hypoxia), radiological findings (new infiltrate on chest-X-ray or CT scan), and leukocytosis [19]. All VAPs were bacteriologically confirmed. Respiratory samples for VAP diagnosis were performed either using fiberoptic broncho-alveolar lavage or endotracheal aspiration, according to local practices. The threshold for lung samples positivity was 104 cfu/mL for BAL and 105 cfu/mL for tracheal aspirate. Microorganisms responsible for infection were considered as MDRO according to the European Society of Clinical Microbiology and Infectious Disease definition [20].

Primary and secondary endpoints

The primary endpoint was ICU-AI incidence and the secondary endpoints were specific VAP and BSI incidences as well as ICU mortality. Finally, we aim to describe the microbiology of ICU-AI episodes in both groups.

Statistical analysis

Statistical analysis was performed with the statistical software R 4.1.1. Categorical variables were expressed as percentages and continuous variables as median and interquartile range (IQR). The chi-square test and Fisher exact test were used to compare categorical variables and the Man-Whitney U test or the Wilcoxon for continuous variables.

Predictors of acquired infections were estimated using a uni- and multivariable Poisson regression model while predictors of ICU death were analyzed using a uni- and multivariable Cox proportional hazard model and Kaplan–Meier survival curves with log-rank test.

In order to account for competing events such as discharge from the ICU alive, a second analysis using uni (Model 1) and multivariable (Model 2) competitive risk analysis was used to estimate the probability of developing ICU-AI. A third Fine and Gray model (Model 3) was finally performed to analyze the association between exposure and VAP with extubation being a competing event with VAP onset. Using the “cmprsk” package, we performed a fine and gray model to estimate the sub-distribution hazard ratio (sdHR). A multivariable Cox proportional hazard model was used for survival analysis. Variables associated with events (either BSI or death) with a p value < 0.2 in univariate analysis were included in a multivariable model. Of note, for outcome comparison, only the first BSI was taken into account.

Since SAPS II includes age, collinearity is present between this variable and the variable age. To go through, the variable included in the multivariable analysis consisted of SAPS II with the exception of the age component. Multivariable analyses were performed with the inclusion of non-redundant variables associated with the event (ICU-AI or death) with a p value < 0.2 in the univariate analysis. There were no missing data in the dataset. All tests were two-sided, and p < 0.05 was considered statistically significant.

Results

Population

Overall, 1654 patients were admitted to participating ICUs during the study period, of whom 758 were not intubated or remained < 24 h in the ICU. Among the 896 remaining patients, 601 were considered immunocompetent, giving 295 immunocompromised patients available for analysis (143 in the pre-implementation group vs 151 in the post-implementation group) (Fig. 1). Age was 68 years [60–73], 61% were male, the main reason for admission was medical (88%) and SAPS II was 48 [34–62]. Solid organ cancer was the main reason for immunosuppression (53%), with a majority of lung cancer (75/156), urological cancer (52/156), hepatobiliary cancer (36/156), and breast cancer (35/156). Immunomodulatory treatments were noticed in 135 patients (46%) including 64/135 with recent chemotherapy and 61/135 with immunotherapy for auto-immune or auto-inflammatory disease. Fifty-nine patients (20%) were admitted with neutropenia < 0.5 G/L, 51 (17%) patients had hematological malignancy and 23 (8%) had solid organ transplantation. Seven (2%) patients were colonized with a MDRO at admission (6 ESBL-PE and 1 vancomycin-resistant Enterococcus). Length of stay was 7 days [4,5,6,7,8,9,10,11,12,13] and 57 patients (19%) died in the ICU. Baseline characteristics at ICU admission did not differ between patients admitted before MSD implementation (pre-implementation group) and after implementation (MSD/ post-implementation group) (Table 1).

Fig. 1
figure 1

Flow chart

Full size image
Table 1 Baseline characteristics and outcomes of study patients
Full size table

Acquired infections

They were 35 ICU-AI (25 VAP and 10 BSI) in 29 patients (20%) (corresponding to 1 671 patient days) in the standard-care group as compared with 10 ICU-AI (4 VAP and 6 BSI) in 9 patients (6%) (corresponding to 1440 patient days) in the post-implementation group (p < 0.001 for ICU-AI, p < 0.001 for VAP and p = 0.377 for BSI) (Table 1) (Supplementary Figure 1).

In a multivariable Poisson regression model, MSD was independently associated with a decreased incidence of ICU-AI (incidence rate ratio [IRR] = 0.39; 95%CI [0.20–0.87] p = 0.008). Conversely, COVID-19 (IRR = 1.85; 95%CI [1.04–1.30] p = 0.036), vascular catheter (IRR = 4.94; 95%CI [1.19–20.39] p = 0.027) and immunomodulatory treatment (IRR = 1.91; 95%CI [1.08–3.68] p = 0.027) were associated with an increased risk (Table 2). In a second model taking into account discharge from ICU as a competing risk using a Fine and Gray model, results were similar with a decreased risk of ICU-AI in patients receiving MSD (sdHR = 0.40 95%CI [0.19–0.84] p = 0.015). Similarly, a decreased risk of VAP was observed in model 3 (competing risk analysis with extubation being the competing event of VAP) (Supplementary Table 1).

Table 2 Risk factors for acquired infection (Poisson regression test)
Full size table

There were no differences in the distribution of microorganisms responsible for ICU-AI between groups (Table 3). Interestingly, 9/38 patients with ICU-AI had pulmonary Aspergillosis, making this micro-organism as frequent as non-fermenting Gram-negative Bacilli. Enterobacteriaceae were present in 8/39 patients, coagulase-negative Staphylococci in 5/39.

Table 3 Data regarding ICU-AI
Full size table

Finally, 7 patients in the pre-implementation period vs 2 patients in the post-implementation period acquired a MDRO colonization while in the ICU (p = 0.096).

Survival

There were 35/143 deaths in the standard-care group as compared with 22/151 in the post-implementation group (p = 0.046) (Table 1 and Fig. 2). This difference remained in a multivariable Cox proportional hazard model (HR = 0.58; 95CI [0.34–0.95] p = 0.048) (Table 3). Conversely, higher SAPS II (without age component) (HR = 1.02; 95%CI [1.00–1.03] p = 0.008) and higher age (HR = 1.03 [1.00–1.07] p = 0.020) were associated with poor outcome (Table 4).

Fig. 2
figure 2

Cumulative incidence of ICU-AI in both groups

Full size image
Table 4 Risk factors for death in the ICU (Cox proportional hazard model)
Full size table

In a second model assessing the association between ICU-AI and death, patients with ICU-AI had a higher risk of death with time (HR = 1.89; [1.00–3.51] p = 0.049) (Supplementary Table 2).

Discussion

In the present study involving critically ill immunocompromised patients, we observed a decreased incidence of ICU-AI, especially VAP, associated with a higher survival rate in patients treated with MSD.

In recent decades, the management of immunosuppression has become a daily issue for physicians in general but particularly in the ICU [21]. The rise of patients with deficient immune systems had led to an increase in the need for ICU admission for these specific populations. The landscape of immunosuppression has evolved with a lower proportion of uncontrolled HIV patients, while new immunosuppressive treatments have emerged and are increasingly used [22, 23]. In addition, the increasing number of solid organ transplantation and recent advances in the field of hematological malignancies have contributed to improved survival for these patients but also increase the risk of life-threatening events making this population particularly at risk for needing ICU admission [2, 22, 24]. Beyond these epidemiological shifts, these patients are characterized by their severity, which is reflected in higher mortality rates and longer ICU lengths of stay, increasing exposure to nosocomial infections [3, 4].

Although a recent study reported similar ICU-AI incidence in immunocompromised patients than in non-immunocompromised patients, the impacts of those infections on the fate of immunocompromised patients deserve to be highlighted [4, 18]. The compromised host immune response as well as pathogen-involved specificities may contribute to this higher mortality in these patients. Impaired host response may result in an atypical clinical presentation (absence of fever, torpid course), while pre-existing disease may result in radiological abnormalities (pulmonary infiltrates), making the rapid diagnosis of ICU-acquired infections a challenge. As a result, diagnosis and treatment of these infections are often delayed [24]. In addition, the higher proportion of multidrug-resistant bacteria and the broader spectrum of pathogens involved may lead to inappropriate empirical treatments [4, 25]. Therefore, strategies to prevent nosocomial infections should be particularly considered in these patients.

Selective digestive decontamination was initially investigated in patients with hematological diseases and in liver recipients because these patients were at-risk of nosocomial infections [7,8,9]. Moreover, among the unrestricted ICU population, the assessment of selective decontamination regimen has evidenced its effectiveness in preventing ICU-acquired infections [26]. However, due to concerns about rebound infection on cessation of SDD, such a strategy is no longer used in those patients. Although recommended in recent guidelines as a validated strategy to prevent VAP, implementation of selective decontamination in intensive care units remains low [27]. Among factors contributing to the low widespread of this strategy the fear of antimicrobial resistance, may participate in such a poor compliance with current guidelines. However, a study assessing this issue evidenced the absence of the effect of selective decontamination regimens on multidrug-resistant bacteria colonization and acquired infections [13, 28, 29]. Noteworthy, a previous study conducted in participating ICUs reported lower consumption of high-risk promoting resistance antibiotics when MSD was implemented [11]. This may explain the favorable impact of MDRO acquisition with MSD implementation [13, 28].

The high proportion of fungi (reaching nearly 24% of the pathogens involved) in patients with VAP is remarkable. Invasive fungal infections are a common cause of ARDS in immunocompromised populations [30] and immune disorders have been reported as a risk factor for ICU-acquired pulmonary aspergillosis [10, 31], prompting broad screening for these pathogens in immunocompromised populations.

To our knowledge, our study is the first to evaluate decontamination strategies in critically ill immunocompromised populations. However, some limitations must be acknowledged. First, as our study was conducted in adult intensive care units located in western France, where the prevalence of multidrug-resistant bacteria is low, our results may not be generalizable to other settings. Moreover, the long-term impacts of MSD on both environmental and individual ecology remain a crucial issue. Herein, although we did not observe any impact on MDRO acquisition, the sample size precludes a strong conclusion. Second, the heterogeneous entity of immunosuppression may warrant a granular analysis of the effect of preventive strategies in these different settings. Furthermore, the definition of immunocompromised is challenging, in the present study we used a definition that was previously used in ICU patients. However, as immunosuppressive drugs are evolving, the definition of immunocompromised patients must also evolve and could be debated. Third, although the frequency of colonization with MDR-resistant bacteria in immunosuppressed patients has recently been shown to be lower than in other populations [17], the effects of MSD on colonization with MDR-resistant bacteria have not been evaluated. Finally, the study design (pre-post design without randomization, unblinded assessment of ICU-AI cases) precludes any conclusion to be drawn. The follow-up of included patients was limited to ICU stay, accordingly, the long-term impact of SDD on patients’ outcomes could not be assessed (particularly potential invasive fungal infection rebound on withdrawal of SDD). Randomized controlled trials are needed to properly study the effects of MSD in these specific populations.

In conclusion, in ICU immunocompromised patients, MSD appeared to be associated with improved outcomes including a decreased incidence of ICU-AI, especially VAP.

The burden of ICU-acquired infections on the fate of critically-ill immunocompromised patients emphasizes the need for preventive strategies.