The role of antimicrobial resistance on long-term mortality and quality of life in critically ill patients: a prospective longitudinal 2-year study

Intensive Care Unit (ICU) survivors may suffer tremendous changes in lifestyle post-discharge [1]. Critical illness may impair mental, physical and social health, limit patients’ mobility and independence, and damage self-confidence, sense of hope and positive outlook [2]. These devastating consequences of critical illness, although of limited “clinical relevance”, are now considered as “patient-important outcomes” (such as quality of life, functional/cognitive/neurological outcomes assessed after ICU discharge) and their role in clinical research is increasingly emphasized [3, 4]. In addition, the importance of assessing intermediate and long-term outcomes post-ICU discharge has also been encouraged [5].

Antibiotic resistant pathogens (ABRP) have been identified as one of the most important threats for the modern ICU, limiting treatment options and resulting in adverse clinical outcomes and excessive cost of care [6‐8]. In the Greek ICU, in particular, Gram-negative multi-drug resistant (MDR) and extensively-drug resistant (XDR) bacteria are considered endemic [9, 10]. The impact of antimicrobial resistance on ICU mortality has been widely studied so far [9‐20]; however, the putative influence of ABRP-induced ICU infections on post-ICU patient-important, intermediate and long-term outcomes remains largely undetermined so far.

The complex consequences of critical illness are usually referred to as “quality of life” indices, on the basis of several different instruments (questionnaires, phone or personal interviews, physical examinations, chart interviews, neuro-cognitive tests) which were developed in order to assess outcomes [21]. These markers reflect the individual aspect of ‘happy to live’ as a result of multidimensional perceptions of physical, psychological, and social variables [2]. An established outcome index is quality-adjusted life year (QALY) index, which receives a growing interest in ICU research, as it combines both quantity and quality of life gained as a result of ICU admission and implementation of advanced therapy [1, 22‐26].

The aim of our study was to examine whether antimicrobial resistance in ICU is associated with increased long-term mortality and problematic quality of life for a prolonged time period post-ICU admission. For this purpose, ICU patients were divided into three subgroups: patients having suffered ABRP-induced ICU infection(s) (ABRP-group), non-ABRP infection(s) (non-ABRP-group) and patients who manifested no infection at all while in ICU (“no-infection” group). Survival data and QALY measurements were prospectively recorded over a two-year follow-up period.

A total of 373 ICU patients were assessed for eligibility to participate in the study during the study period (January 1, 2014 to December 31, 2015). Twenty-two (22) of them were excluded for several reasons (eight patients did not provide written informed consent; fourteen patients refused to participate or were not found at follow-ups, so data regarding their longitudinal course post-discharge were considered as missing); thus, 351 patients were finally enrolled and analyzed.

Fifty-eight patients (16.52%) demonstrated infection(s) due to at least one (≥ 1) ABRP (ABRP group), 57 (16.24%) manifested infection(s) due to ≥ 1 non-ABRP (non-ABRP group), while 236 (67.24%) manifested no infection at all (“no-infection” group). Eleven patients of the ABRP group (11/58; 18.9%) demonstrated microbiological results of both ABRP and non-ABRP [11]. Patients’ microbiology in our ICU population is demonstrated in detail in Table 1.

Baseline characteristics and epidemiological background of our 351 patients divided into three subgroups according to having suffered intra-ICU infections (either associated with ABRP or with non-ABRP) or not are presented in Table 2. ICU infections due to ABRP were associated significantly with more severe initial clinical condition (increased APACHE II score upon admission) but not worse preadmission status (as the latter assessed by QALY values calculated for the previous year before admission), cardiac arrest as admission etiology, as well as acute respiratory distress syndrome (ARDS), continuous renal replacement therapy (CRRT) and MODS score; among them, only MODS score (OR: 0.676, 95%CI 0.584–0.782; P < 0.001) and CRRT (OR: 4.453, 95%CI 1.805–10.982; P = 0.001) manifested independent association with the occurrence of ABRP infections in ICU (Additional file 1: Table 1). The duration of pre-infection ventilation was 11.76 ± 10.2 days in our ABRP group; sixteen out of our 19 tracheostomized patients (84.2%) developed ABRP infection before tracheostomy was performed.

Intra-ICU and post-discharge outcome data in our study population overall (N = 351) as well as in the subset of ICU survivors (N = 253), divided into the three aforementioned subgroups, is presented in Table 3. ABRP infections were associated with adverse intra-ICU outcomes, including critical care myopathy and tracheostomy on discharge, prolonged mechanical ventilation and ICU stay, increased ICU costs and intra-ICU mortality (Table 3). Blood stream ABRP infections were found to constitute independent determinants of ICU cost of care (see also Additional file 1: Digital Content, Results section, as well as Additional file 1: Table 2).

The ABRP group was further divided into MDR subgroup (N = 34) and XDR subgroup (N = 24; 8 of them manifested both XDR/MDR isolates). XDR subgroup was associated with more days on mechanical ventilation (31.9 ± 21 vs. 17.3 ± 11.5, P = 0.001), prolonged ICU length of stay (39.5 ± 24.5 vs. 22.6 ± 15.1, P = 0.002), higher intra-ICU mortality [62.5% (15/24) vs. 26.5% (9/34), P = 0.006] and higher MODS score (5.7 ± 3.9 vs. 3.3 ± 3.3, P = 0.016), yet similar APACHE II on admission (22.1 ± 4.9 vs. 21.2 ± 4.1, P = 0.431) compared to MDR subset.

Interestingly, patients who developed ICU infections due to XDR pathogens were of similar age (63.3 ± 13.8 vs. 66.2 ± 14.7, P = 0.355), predominantly males [22/24 (91.7%) vs. 209/327 (63.9%), P = 0.006], and demonstrated higher APACHE II on admission (22.1 ± 4.9 vs. 19.2 ± 7.2, P = 0.05) and MODS score (5.7 ± 3.9 vs. 1.6 ± 2.2, P < 0.001), compared to the rest ICU population.

The most important finding of this study is that ICU infections due to antimicrobial resistant pathogens may exert a prolonged adverse impact on critically ill patients’ longevity and well-being post-ICU discharge. This information is important, since it may indicate that despite the implementation of intensive care treatment, patients who develop infections due to ABRPs while in ICU are expected to have limited benefits in patient-important quality-of-life outcomes and long-term survival. Interestingly, our findings may also suggest that XDR infections represent an independent determinant of long-term survival rather than a mediating factor in declining health status of patients with more severe critical illness; however, this remains to be elucidated in larger-scale future studies.

The present study provides evidence that antimicrobial resistance in ICU is associated with adverse intra-ICU outcomes, including prolonged mechanical ventilation and ICU length of stay, and increased ICU costs [32]; in addition, our results may suggest that ABRP infections possibly impact negatively ICU mortality (Table 3). These data is in line with previous reports [10‐14], while no significant impact on ICU mortality was found in other studies [9, 15‐19]. Remarkably, a number of these studies were Greek as well [9, 10, 14, 18, 19], and mainly included gram-negative ABRPs [9‐19]. However, there is still limited data in the literature regarding the relationship between ABRP infections and mortality rates post-discharge [33].

In our series, intra-ICU mortality was 27.9%, and rose abruptly to 52.4% in only 90 days post-ICU admission. Beyond 90-days, long term-mortality reached a “plateau” phase, rising at a value of 61.5% at 2-year follow-up. Our results are in line with previous reports, also suggesting unacceptably high mortality rates post-ICU discharge [34]. This has been attributed to multiple reasons, such as higher pathogen virulence and/or weaker hosts [35, 36], specific type of infections (such as pneumonia or peritonitis) [37], inappropriate empirical antimicrobial therapy and/or delays in administering effective antimicrobial treatment [19, 38, 39]. Moreover, inability of the staff in lower levels of care to keep up with the needs of critically ill patients [40], absence of Intermediate Care Unit facilities [34] and premature discharge from ICU due to the constant need for beds [41] may be related to enhanced mortality rates post-discharge. However, the longitudinal impact of antimicrobial resistance on both intermediate and long term survival has not been examined so far. Our series indicates that ABRP infections in ICU (especially due to XDRs) may be associated with increased 90-day and 2-year mortality (Table 3 and Additional file 3: Fig. 2). In addition, ABRP group (mainly XDR subgroup) demonstrated worse survival curves not only in ICU patients overall, but also in the subset of ICU survivors; in the latter, it is of note that antimicrobial (possibly XDR) resistance was associated with increased short-term, intermediate as well as long-term mortality post-discharge (Fig. 1 and Additional file 4: Fig. 3). Accordingly, one could argue that ICU survivors with a history remarkable for antimicrobial resistance while in ICU may continuously be at high risk for restricted life expectancy post-discharge; however, this hypothesis, as well as the possible mechanisms mediating this relationship, warrant further investigation in future studies.

In the recent years, patient-important outcomes are gaining wide acceptance in most fields of clinical research. Apart from decreasing mortality, clinical decision-making also aims at improving quality of life and functional status in ICU survivors [4, 23]. Up to the present, several aspects of critical illness have been investigated as possible determinants of long term outcomes (including QALY assessment) post-ICU discharge, such as the duration of mechanical ventilation/ICU stay [42, 43], septic shock [44, 45], ARDS [46], COPD [47], cardiac arrest on admission [48] etc. To the best of our knowledge, the possible role of antimicrobial resistance on quality-of-life, physical and psychological well-being for a prolonged period post-ICU discharge is examined for the first time in the present investigation. Interestingly, our study may suggest that antimicrobial resistance is likely to be associated with significant impairment in health-related quality-of-life for months, even years after ICU admission. Remarkably, ABRP infections in our series were associated with reduced 6-month to 2-year QALY kinetics (Fig. 2, Additional files 5 and 6: Figs. 4 and 5); similarly to long-term mortality, this relationship mainly concerned patients who managed to survive ICU (Table 3, Additional file 5: Fig. 4); in this respect, our findings are important, since it is suggested that history of ABRP infections in ICU, apart from increased mortality, may also predict limited anticipated benefits in non-neurocritical (as was the case in our study) critically ill patients post-discharge.

Antimicrobial resistance was associated with several aspects of critical illness (Table 2) and prolonged mechanical ventilation/ICU stay in our series (Table 3). Certainly, ABRP infections occurred predominantly in the early phase of critical illness rather than in the late stage of ICU stay (as indicated by relatively short duration of pre-infection ventilation and higher incidence of antimicrobial resistance before tracheostomy). Undoubtedly, the more severe critical illness may have predisposed to both ABRP infections and adverse long-term outcomes [33]; on the other hand, ABRP infections may have resulted in a more complicated and prolonged clinical course, which in turn may have led to severe persistent impairments and functional limitations. Interestingly, our study provides evidence that even in “sicker” ICU patients, infections due to either ABRPs or XDR pathogens may bear additional risk for increased long term (2-year) mortality (Table 4), but not affect 2-year QALY scores independently (Additional file 1: Table 4). Notably, the harmful effect of antimicrobial resistance (after adjusting for possible covariates of severe underlying illness) on long-term mortality is likely to be restricted to the XDR organisms rather than the MDR subset (Table 4, Additional file 1: Table 3, and Additional file 4: Fig. 3). Furthermore, specific therapeutic approaches aiming at treating resistant pathogens, such as colistin administration (Additional file 8: Fig. 7), may have also influenced long-term outcomes. Of course, colistin was the main regimen used for ABRP infections, hence its possible long-term consequences possibly reflect the ones of coinciding antimicrobial resistance (especially of XDR organisms, Additional file 9: Fig. 8). Finally, post-ICU care facilities and rehabilitation centers may contain numerous niches of MDR/XDR bacteria (i.e. infected patients, biofilms, toilets, surfaces, keyboards), which in turn may constitute “virulence-like” community-resistance reservoirs for secondary infection outbreaks [8, 44]. Nevertheless, our limited data may indicate a negative effect of antimicrobial resistance in declining health status, especially in the subset of the more severe critically ill. In this respect, prevention against ABRP infections might be of fundamental importance in modifying the clinical course of patients with an “a priori” dismal prognosis.

A few limitations to the study deserve mention. First, we used consecutive two-year sampling in order to have valid and meaningful results on the role of antimicrobial resistance in affecting long-term outcomes; however, we certainly admit that our findings (especially on the role of XDRs on critically ill patients’ longevity post-discharge) should be validated in larger-scale studies in the future. Second, most of our infections were non-catheter-related blood stream infections, while respiratory infections were far less common in our cohort. One could argue that the synthesis of our ICU infections and its specific microbiology (mainly gram negative bacteria) may have interfered with our findings. Certainly, genetic and geographic differences may exert diverse influence on the clinical phenotypes of special diseases in different populations. Furthermore, it has been shown [49] that some organ failures (i.e. cardiovascular system) may adversely affect outcomes of ICU patients more than others; however, the MODS score we used was inadequate to examine potential longitudinal differences among specific organ failures. Moreover, a drawback of the study is the likelihood of misclassifying a culture-negative infection as no infection at all. However, this possibility was small, by carefully culturing in concert with the routine microbiological sampling of the department. Finally, meningitis/ventriculitis due to ABRPs, which has shown significant diagnostic performance in post-neurosurgical patients [50], was not examined in our series.

The present prospective study demonstrated a significant association between ABRP (especially XDR) infections in ICU and increased mortality/inability rates for a prolonged period post-discharge. This relationship is likely to mainly concern patients with more severe critical illness, who manage to survive ICU though; in this respect, a history remarkable of antimicrobial resistance while in ICU should raise awareness for increased risk post-discharge and reconsider the anticipated benefits in long-term longevity and well-being, especially in patients with more severe critical illness and/or complex course of hospitalization.