Colistin Treatment Outcomes in Gram-Negative Bacterial Infections in the Northeast of Romania: A Decade of Change Through Pandemic Challenges

Abstract

Background: The rise in multidrug-resistant (MDR) Gram-negative bacteria (GNB) poses a critical threat to global health, exacerbated by the increased use of antibiotics during the SARS-CoV-2 pandemic. Colistin, often considered a last-resort antibiotic, has seen heightened usage, raising concerns about resistance and clinical outcomes. This study investigates the evolution of colistin resistance in GNB during the SARS-CoV-2 pandemic, with a focus on clinical outcomes in colistin-treated patients. Methods: Conducted in “Saint Spiridon” Hospital, Iasi, Romania, the study assesses antibiotic usage, trend in GNB resistance to colistin, and mortality rates across distinct pandemic phases (pre-pandemic, Delta, Omicron, and post-pandemic). A retrospective longitudinal analysis spanning 2011 to 2023 was performed. Statistical models were employed to analyze mortality risk and assess the pandemic’s impact on antimicrobial dynamics. Results: Colistin usage increased significantly during the pandemic, with the highest doses recorded during the Omicron phase. Resistance to colistin, particularly among A. baumannii and K. pneumoniae, was associated with increased mortality rates. The statistical model demonstrated robust predictive accuracy for mortality across the pandemic phases, with an AUC of 0.866 in the post-pandemic period. The findings underscore the critical role of A. baumannii as a driver of adverse outcomes, with co-infections significantly linked to systemic infections and higher mortality. Conclusions: The study highlights the evolving trends in colistin therapy and resistance, emphasizing the pandemic’s impact on antimicrobial stewardship and the persistent challenges posed by MDR pathogens. We emphasize the need for antimicrobial resistance surveillance, including the monitoring of colistin resistance, which is considered a last-resort antibiotic.

1. Introduction

The SARS-CoV-2 pandemic has significantly heightened the antimicrobial resistance (AMR) crisis, increasing the pressure on the healthcare and social care systems [1]. According to a 2022 CDC report, antimicrobial-resistant infections caused over 29,400 deaths in 2020 in United States (U.S.), with nearly 40% of these infections acquired in hospital settings. The pandemic’s impact on healthcare resources, marked by an increased use of medical devices, prolonged patient hospitalizations, staffing shortages, and compromised infection control measures, has driven a surge in healthcare-associated infections [2].

The World Health Organization (WHO) has identified AMR, particularly among Gram-negative species, as a growing global health threat [3].

The SARS-CoV-2 pandemic in 2020 created an opportunity to reduce antibiotic consumption, as the strict public health measures, including lockdowns, social distancing, and improved hygiene practices, led to a decrease in bacterial co-infections and overall healthcare utilization in many countries [4,5]. Additionally, the growing awareness of viral versus bacterial infections during the pandemic highlighted the need for more prudent antibiotic use, and many countries have reported reductions in antibiotic prescriptions, particularly for respiratory infections [6]. However, this potential was not fully realized in Romania, where antibiotic consumption remained high, likely due to empirical prescriptions in COVID-19 cases, self-medication, and gaps in antimicrobial stewardship programs [7]. Thus, an increased use of broad-spectrum antibiotics, such as penicillins with inhibitors, extended-spectrum cephalosporins, carbapenems, and reserve antibiotics like colistin was registered during the COVID-19 pandemic years in Romania [8]. A comparison of EARS-Net data from the pre-pandemic (2017) and post-pandemic (2021) periods revealed an increasing trend in AMR, particularly for Klebsiella pneumoniae and Acinetobacter baumannii isolates in Romania and other southern and eastern European regions [9]. This upward trend underscores the heightened risk of multidrug-resistant (MDR) pathogens that emerged during the COVID-19 pandemic.

Antimicrobial resistance in K. pneumoniae has emerged as a major concern in recent years [10,11]. Resistance to carbapenems is particularly concerning due to its association with cross-resistance to multiple other antibiotic classes, which greatly restricts treatment options and leads to increased rates of treatment failure [12]. In those cases, last-resort antibiotics such as colistin are frequently used in combination with other antibiotics [13,14]. However, the growing global prevalence of carbapenem-resistant and MDR K. pneumoniae has led to increased dependence on colistin in recent years, which has, in turn, contributed to the rise in colistin-resistant strains [10].

Before the emergence of the SARS-CoV-2 pandemic, colistin resistance was already recognized as a growing global health concern, primarily driven by the overuse of colistin in both human medicine and animal agriculture. The increasing reliance on colistin as a last-resort antibiotic for carbapenem-resistant Gram-negative bacterial (GNB) infections has led to the gradual selection of resistant strains, complicating treatment strategies for multidrug-resistant (MDR), extensively drug-resistant (XDR), and pan-drug-resistant (PDR) infections [15].

A global study on colistin resistance revealed that K. pneumonaie had the highest resistance rate among isolates in 2020, reaching 12.9% (4 out of 31 isolates) compared to approximately 2.9% in previous years [16]. A meta-analysis on Pseudomonas aeruginosa reported an increase in the overall colistin resistance rate from 2% in the period of 2006–2010 to 5% in the period of 2020–2023 [17]. A cross-sectional study performed in Egypt in the period of 2022–2023 found colistin resistance rates of 12.3% in K. pneumoniae, 7% in P. aeruginosa, and 73.7% in A. baumannii [18]. These findings underscore the escalating challenge of colistin resistance in critical healthcare settings, especially since 2020. Carbapenems are among the last remaining options for treating A. baumannii infections [19]. Importantly, the increasing prevalence of carbapenem-resistant strains led the World Health Organization (WHO) to designate carbapenem-resistant A. baumannii as a “Priority 1 critical pathogen” for antibiotic research in its 2017 report, issued a few years before the onset of the COVID-19 pandemic [20]. Colistin, a last-resort antibiotic, which was abandoned in the 1980s due to its nephrotoxic and neurotoxic side effects, has been reintroduced to combat the growing threat of carbapenem-resistant GNB [21]. However, the reintroduction of colistin led to a rise in the prevalence of colistin-resistant strains over the past decade [20]. The COVID-19 pandemic further exacerbated the antimicrobial resistance crisis, likely due to the widespread use of broad-spectrum antibiotics, including colistin, to treat secondary bacterial infections in critically ill COVID-19 patients. Many hospitals reported a surge in multidrug-resistant infections due to extended hospitalization days, invasive mechanical ventilation, and excessive antibiotic prescriptions during the pandemic [22]. A. baumannii, in particular, became a major concern in COVID-19 wards, with studies showing alarmingly high colistin resistance rates of 52% in an Iranian study [23] or 91.2% in Wuhan [24].

According to the EARS-Net surveillance study, colistin resistance in P. aeruginosa increased from 1% to 4% across Europe in the period of 2013–2016 [25]. In 2016, the majority of colistin-resistant isolates from Europe were reported in Italy and Greece. In China, the colistin resistance rates have been reported to range between 1 and 7% [26,27].

The COVID-19 pandemic significantly disrupted healthcare systems and practices, leading to the reallocation of antimicrobial stewardship resources to support the overwhelmed healthcare workforce. During the peak of the crisis, this shift reduced the capacity to implement antimicrobial stewardship initiatives and provide guidance on the rational use of antibiotics in COVID-19 patients [28]. The urgent focus on clinical care, combined with limited healthcare resources, contributed to an increased risk of inappropriate antimicrobial prescribing [5,29,30].

Colistin resistance is relatively uncommon in clinical isolates of Escherichia coli compared to A. baumannii and K. pneumoniae. Between 2010 and 2014, resistance rates were reported at 0.2% in clinical isolates and 0.9% in commercial meat samples. In Taiwan, E.coli resistance rates ranged from 1.1% to 8.7% between 2012 and 2015 [31].

The aim of the study was to investigate the evolution of colistin resistance in Gram-negative bacteria (GNB) during the SARS-CoV-2 pandemic, as well as in the pre-pandemic and post-pandemic periods, with a focus on the clinical outcomes of colistin-treated patients. Conducted in a hospital in the northeast of Romania, “Saint Spiridon” County Clinical Emergency Hospital, Iasi, the study seeks to assess the pandemic’s influence on colistin usage, resistance patterns, and mortality rates across distinct pandemic phases (pre-pandemic, Delta, Omicron, and post-pandemic) while also completing the overall picture of colistin resistance.

2. Results

The patient population shows fluctuations over time, peaking in 2014 (502 patients) and declining sharply in 2020 (247 patients), likely reflecting the COVID-19 pandemic’s impact on hospital admissions (

Table 1. General characteristics of patients.

	Total, N %	F, n % M, n %	Age Median [CI]
2011	393.00 (7.73)	177 (45.04) 216 (54.96)	65 [55–74]
2012	375.00 (7.38)	148 (39.47) 227 (60.53)	63 [54–74]
2013	437.00 (8.60)	151 (34.55) 286 (65.45)	63 [52.5–75]
2014	502.00 (9.88)	185 (36.85) 317 (63.15)	65 [54–75]
2015	423.00 (8.33)	158 (37.35) 265 (62.65)	65 [57–74]
2016	420.00 (8.27)	144 (34.29) 276 (65.71)	65 [55–74]
2017	370.00 (7.28)	140 (37.84) 230 (62.16)	64 [53–75]
2018	417.00 (8.21)	149 (35.73) 268 (64.27)	64 [53–73]
2019	358.00 (7.05)	144 (40.22) 214 (59.78)	64 [52–72]
2020	247.00 (4.86)	98 (39.68) 149 (60.32)	63 [52–73]
2021	304.00 (5.98)	110 (36.18) 194 (63.82)	63 [53–74]
2022	431.00 (8.48)	148 (34.34) 283 (65.66)	65 [54–73]
2023	404.00 (7.95)	157 (38.86) 247 (61.14)	66 [54–73]

Abbreviations: N = total number; F = female; M = male; n = number; CI = confidence interval. The smallest N values are highlighted in red.

).

Males consistently form a higher proportion of the patients (around 60% annually), while females account for approximately 40%.

The median age remains stable, ranging between 63 and 66 years, indicating that these infections predominantly affect older adults.

The distribution of bacterial isolates (A. baumannii, P. aeruginosa, K. pneumoniae, E. coli, and others) varied over the study period from 2011 to 2023. Figure 1 captures the trends in infection patterns and highlights significant shifts in pathogen prevalence over time, evaluated using chi-squared statistical tests (Figure 1). While the total number of A. baumannii isolates did not change before and after the pandemic, there was a relative decrease in P. aeruginosa isolates in 2020.

The percentage distribution of GNB in colistin-treated patients over the period of 2011 to 2023 is illustrated, highlighting temporal trends in Figure 2. Each sub-panel (A–D) focuses on a specific GNB species (A. baumannii, P. aeruginosa, K. pneumoniae, and E. coli), with statistical significance (p < 0.0001) derived from chi-squared tests, indicating important variations and potential shifts in prevalence or treatment efficacy over time (Figure 2). The highest percentage of A. baumannii isolates was registered in 2020 (Figure 2A), while P. aeruginosa prevalence showed a generally increasing trend from 29% in 2011 to 47% in 2023 (Figure 2B). Interestingly, the proportion of K. pneumoniae isolates showed a decreasing trend from 2012 (30%) to 2019 (16%), which increased again starting with 2020, reaching a value over 28% (Figure 2C). The increase in the other isolates was accompanied by a reduction in the percentage of E. coli strains over the study period from 25% in 2011 to 11% in 2023, with an extra peak of 21% in 2020 (Figure 2D).

The distribution of sample types, from which various GNB were isolated between 2011 and 2023, are depicted in Figure 3. Each subpanel (A–D) focuses on a specific pathogen (A. baumannii, P. aeruginosa, K. pneumoniae, and E. coli), showing trends across different sample sources, including wound cultures, endotracheal aspirates, urine cultures, blood cultures, fluid cultures, abscess cultures, and others (Figure 3).

The majority of A. baumannii isolates were obtained from wound cultures and endotracheal aspirates. The increasing trend in wound culture isolates emphasizes their role in surgical site infections and trauma cases (Figure 3A).

P. aeruginosa isolates showed a balanced distribution across the years, with a high predominance in wound cultures and endotracheal aspirates (Figure 3B). This stability trend might reflect P. aeruginosa’s adaptability to multiple anatomical sites, including wounds, the respiratory tract, and urinary systems. Its role in chronic infections, such as in patients with cystic fibrosis or chronic wounds, may explain the lack of significant fluctuations. A steady dominance of wound cultures was followed by blood and urine cultures.

K.pneumoniae isolates showed a visible rise in wound cultures in 2020 (of 76.05%) when compared to the previous 3 years (in average 49.88%) or the following 3 years (54.34%), Figure 3C. Interestingly, there was a noticeable increase in blood cultures from less than 6% before the COVID-19 pandemic to over 10% in 2022 with a maximum of 12.96% in 2023. The rise in blood culture isolates reflected the growing burden of bloodstream infections caused by K. pneumoniae, especially multidrug-resistant strains. Its presence in diverse sources highlights its versatility as both a community-acquired and nosocomial pathogen.

Besides being isolated from wound cultures, E. coli showed the highest prevalence in urine cultures compared to the other GNB, consistent with E. coli’s role as a leading cause of urinary tract infections (UTIs). However, a gradual decline in isolates from urine cultures is observed over the years, coupled with slight increases in other sources, such as blood and wound cultures, possibly due to a rising in cases of invasive E. coli infections or improved detection methods in hospital settings (Figure 3D).

Trends in the frequency of multiple GNB infections between 2011 and 2023 are illustrated in Figure 4. The data are divided into four panels (A–D), each focusing on different co-infection types or combinations. There is a notable increase in the frequency of multiple infections during certain years, peaking significantly in 2020–2021 (p < 0.0001). The spike in co-infections during these years (over 42%) may reflect changes in clinical practices, bacterial resistance patterns, or increased antibiotic use, potentially linked to the pandemic-related stress on healthcare systems (Figure 4A). In only a few cases, counting for less than 8%, more than two GNB were isolated (Figure 4B). In 2020, the majority of multiple infection cases (85%) were positive for A. baumannii (Figure 4C), with the rest being positive for either P. aeruginosa, K. pneumoniae, or E. coli (Figure 4D).

The mortality outcomes for patients with GNB infections from 2011 to 2023 are presented in Figure 5. The data are broken down into overall mortality outcomes (Panel A) and specific outcomes for individual pathogens, including A. baumannii (Panel B), P. aeruginosa (Panel C), K. pneumoniae (Panel D), and E. coli (Panel E). Each bar represents the percentage distribution of cases by outcome, namely deceased, no change, ameliorated, or cured. Statistical significance for changes in mortality rates over time is calculated using chi-squared tests.

Mortality rates (deceased cases) remained stable over time, with minor fluctuations. The proportion of cases with “cured” or “ameliorated” outcomes also remains largely consistent (Figure 5). The lack of significant variation suggests stable treatment outcomes for GNB infections overall, possibly due to a consistent use of last-resort antibiotics like colistin.

For A. baumannii, mortality rates show a statistically significant increase over time (p = 0.0025), particularly during and after 2017. This trend aligns with the pandemic years (2019–2021), where hospitalization days and ventilator use likely increased A. baumannii-related infections. For P. aeruginosa, mortality rates appear stable, with minor increases after 2015, but overall changes are not statistically significant. P. aeruginosa remains a persistent pathogen but does not show the same worsening outcomes as A. baumannii, likely due to its slightly better susceptibility to treatment options.

For K. pneumoniae, mortality rates exhibit a sharp increase during the pandemic years, with significant shifts in outcomes observed. The percentage of “cured” cases decreases, reflecting a worsening overall prognosis for K. pneumoniae infections. The rise in mortality rates aligns with the spread of carbapenem-resistant K. pneumoniae (CRKP), necessitating colistin use.

For E. coli, mortality rates remain low compared to other pathogens, though significant differences are observed (p < 0.0001). A gradual decline in cured cases suggests a slight worsening of outcomes over time. In terms of statistical significance (p < 0.0001), despite lower mortality, the significant differences reflect subtle changes in treatment efficacy or patient populations.

Since the COVID-19 pandemic years of 2020–2021 provoked the above-mentioned shifts in GNB distribution, we next analyzed the frequency of GNB infections treated with colistin during four distinct periods—pre-pandemic, Delta pandemic, Omicron pandemic, and post-pandemic (2017–2023)—which is presented over the analyzed timeline. Each sub-panel highlights trends for specific pathogens, including A. baumannii (Panel A), P. aeruginosa (Panel B), K. pneumoniae (Panel C), and E. coli (Panel D). The statistical significance of differences across these periods is assessed using chi-squared tests.

The frequency of A. baumannii infections increased significantly during the Delta pandemic phase but declined during the Omicron and post-pandemic phases. The peak during Delta reflects the surge in A. baumannii-related infections, often associated with ventilator-associated pneumonia (VAP) and prolonged hospitalization days. The rise during the Delta period highlights the strain on healthcare systems, with increased hospitalization days and invasive procedures contributing to a spike in A. baumannii infections. The subsequent decline might suggest an improvement in the infection control and patient management in later phases.

The frequency of P. aeruginosa infections remained relatively stable across all periods, with no significant fluctuations. The stability reflects P. aeruginosa’s role as a persistent nosocomial pathogen that consistently impacts critically ill patients, regardless of external factors like the pandemic.

The frequency of K. pneumoniae infections increased significantly during the Delta and Omicron pandemic phases, with a subsequent decline post-pandemic. Peaks during Delta and Omicron reflect the pathogen’s growing resistance, especially in patients with bloodstream and respiratory infections.

The frequency of E. coli infections showed a minor increase during Delta but decreased in the Omicron and post-pandemic phases. E. coli plays a smaller role in colistin-treated infections, as it is more commonly associated with community-acquired rather than nosocomial infections. Statistical significance (p = 0.1952, not significant) indicates no significant differences in the frequency of E. coli infections across the study periods. The minor increase during Delta may reflect healthcare disruptions or the inclusion of more severely ill patients during the pandemic. The general stability aligns with E. coli’s reduced reliance on colistin compared to other pathogens (Figure 6).

Overall, Figure 6 highlights the dynamic shifts in GNB infections treated with colistin across pandemic phases. The data emphasize the significant burden of A. baumannii and K. pneumoniae during pandemic peaks, correlating with increased hospitalization days and invasive procedures in our hospital over the course of the analyzed period.

The persistent mortality burden of GNB infections treated with colistin during the pandemic phases is highlighted in Figure 7. Despite changes in the A. baumannii frequency over the pandemic phases, no statistically significant differences in mortality rates were noticed (Figure 7A), with similar data for P. aeruginosa (Figure 7B). Those stable mortality rates for A. baumannii and P. aeruginosa might reflect their persistent lethality and resistance patterns regardless of the pandemic phase. However, a notable increase in the mortality rates during the Omicron phase was observed for K. pneumoniae (Figure 7C) and E. coli (Figure 7D), possibly due to the healthcare system strain and delays in treatment during the pandemic’s peak. These results might highlight the continuous need for enhanced infection control measures, novel antibiotics, and sustained surveillance to address these resistant pathogens.

The lack of significant reductions in mortality for most pathogens suggests that challenges in enhancing colistin efficacy or resistance control persisted over time. The consistently high mortality rates, exceeding 45% for A. baumannii and K. pneumoniae, underline the importance of exploring and implementing alternative treatment options to improve patient outcomes.

The evolving trends in colistin therapy demonstrate increased use during the COVID-19 pandemic and underscore the ongoing challenge of MDR infections in the post-pandemic period. Colistin doses, expressed in million IU administered over the hospitalization period, steadily increased over the years, peaking sharply in 2020 during the COVID-19 pandemic. Following 2020, colistin doses stabilized but remained higher than pre-2020 levels (Figure 8).

Colistin played a critical role in managing MDR infections during the pandemic, while pathogens such as A. baumannii and K. pneumoniae continue to pose significant challenges. However, the weak link between colistin dose and survival emphasizes the importance of developing new antibiotics and optimizing treatment protocols.

The Delta pandemic phase was marked by higher colistin doses (Figure 9A) and extended treatment durations (Figure 9B), reflecting the increased burden of severe MDR infections in ICU settings. A. baumannii and K. pneumoniae stand out as pathogens requiring the most aggressive colistin use, underscoring their resistance profiles and association with poor outcomes. Reductions in treatment durations and doses post-pandemic suggest healthcare systems are regaining stability and managing infections more effectively. The weak correlation between colistin dose and survival rate underscores the complexity of treating MDR infections and highlights the need for additional therapeutic strategies to enhance patient outcomes (Figure 9C).

The trend in colistin doses administered every year over the six-year period from 2019 to 2023, providing insights into the fluctuating usage of this critical antibiotic, is illustrated in Figure 10. These variations also reflect changes in clinical practices, healthcare system pressures, and possibly the prevalence of MDR infections during and after the COVID-19 pandemic. By analyzing this trend, we can better understand how external factors, such as pandemics and antimicrobial resistance patterns, influence the reliance on colistin as a treatment option (Figure 10).

Next, we aimed to identify which clinical or paraclinical factors have the greatest influence on mortality and identify mathematical models that can predict mortality based on our observations. Starting from the data achieved from the AUC analysis of each individual parameter for each pandemic phase, we identified three distinct potential mathematical models (Supplementary Tables S1–S4). Among all described models, Model_1 demonstrated strong predictive accuracy for mortality across different pandemic phases as confirmed by ROC analysis (Figure 11 and

Table 2. ROC evaluation of distinct mathematical models based on paraclinical and clinical parameters in association with mortality for each pandemic phase.

Model	AUC	S.E.	p-Value	95% CI
Pre-pandemic
Model_1	0.811	0.014	<0.0001	0.785–0.838
Model_2	0.756	0.015	<0.0001	0.726–0.786
Model_3	0.793	0.014	<0.0001	0.765–0.821
Delta
Model_1	0.817	0.023	<0.0001	0.773–0.862
Model_2	0.785	0.024	<0.0001	0.738–0.832
Model_3	0.800	0.025	<0.0001	0.751–0.849
Omicron
Model_1	0.796	0.019	<0.001	0.758–0.833
Model_2	0.772	0.020	<0.001	0.732–0.811
Model_3	0.772	0.020	<0.001	0.733–0.811
Post-pandemic
Model_1	0.866	0.024	<0.0001	0.819–0.912
Model_2	0.836	0.026	<0.0001	0.785–0.887
Model_3	0.847	0.024	<0.0001	0.800–0.895

Abbreviations: AUC = area under curve; p = statistical significance coefficient; M = male; CI = confidence interval; Model_1 = gender (M), age, hospitalization days, colistin days, A. baumannii, P. aeruginosa, colistin-resistant GNB, wound culture; Model_2 = gender (M), age, hospitalization days, colistin days, colistin-resistant GNB, wound culture, abscess culture, SARS-CoV-2 infection; Model_3 = gender (M), hospitalization days, colistin dose, colistin days, A. baumannii, P. aeruginosa, colistin-resistant GNB, wound culture, SARS-CoV-2 infection.

). The consistently high AUC values highlight its clinical utility, with improvements in the post-pandemic period pointing to opportunities for further optimization. The AUC values across all phases (ranging from 0.796 to 0.866) indicate that the model consistently performs well in predicting mortality, regardless of external disruptions. The Delta phase shows slightly better predictive accuracy compared to the pre-pandemic due to heightened mortality risks and reliance on key predictors like hospitalization days and colistin duration. Post-pandemic accuracy improves significantly, reflecting improved healthcare outcomes and management practices. The inclusion of factors like gender, age, hospitalization days, colistin usage, and specific pathogens (A. baumannii, P. aeruginosa, and resistant GNB) ensures robust model performance (Figure 11).

To further determine the independent predictors for mortality for each pandemic phase, we performed both univariate and multivariate analysis. In the pre-pandemic phase, mortality predictors included (i) age and prolonged colistin treatment (HR > 1, p < 0.0001); (ii) pathogen-specific impact, where A. baumannii showed the highest hazard ratio (HR 6.786) for mortality, followed by K. pneumoniae and P. aeruginosa, underscoring the severity of infections caused by these pathogens; and (iii) culture types, where endotracheal aspirates (HR 3.152) were significantly associated with mortality, highlighting the severity of respiratory infections in these patients (

Table 3. Univariate and multivariate mortality regression analysis of paraclinical and clinical variables in colistin-treated patients during the pre-pandemic period.

Variable (Pre)	Univariate Analysis			Multivariate Analysis
	HR	95% CI	p-Value	HR	95% CI	p-Value
Gender (M)	0.899	0.655–1.232	0.509
Age	1.036	1.024–1.047	<0.0001	1.037	1.025–1.048	<0.0001
Hospitalization days	0.960	0.948–0.972	<0.0001	0.960	0.947–0.972	<0.0001
Colistin dose	0.996	0.989–1.002	0.185
Colistin days	1.117	1.045–1.193	0.001	1.077	1.044–1.110	<0.0001
A. baumannii	7.083	4.637–10.81	<0.0001	6.786	4.495–10.245	<0.0001
P. aeruginosa	2.367	1.585–3.534	<0.0001	2.240	1.516–3.310	<0.0001
K. pneumoniae	3.045	2.046–4.530	<0.0001	3.023	2.037–4.483	<0.0001
E. coli	1.890	1.166–3.060	0.010	1.889	1.171–3.045	0.009
Co-infection with colistin-resistant GNB	1.690	1.050–2.719	0.031	1.626	1.016–2.600	0.043
Wound culture	0.439	0.228–0.842	0.013	0.315	0.214–0.463	<0.0001
Endotracheal aspirate	4.552	2.290–9.047	<0.0001	3.152	2.077–4.782	<0.0001
Urine culture	0.861	0.352–2.105	0.743
Blood culture	2.418	0.957–6.108	0.062
Fluid culture	2.300	0.983–5.382	0.055
Abscess culture	1.195	0.430–3.312	0.732

Abbreviations: M = male; GNB = Gram-negative bacteria; HR = hazard ratio; CI = confidence interval; p = statistical significance coefficient; significant p-values are highlighted in bold.

).

In the Delta period, A. baumannii remained a major predictor of mortality (HR 5.548); SARS-CoV-2 co-infection emerged as a significant risk factor (HR 1.827, p < 0.05), reflecting the compounded risk of COVID-19 in these patients; and colistin dose and treatment duration had less consistent associations with outcomes, likely due to varying treatment protocols during the pandemic (

Table 4. Univariate and multivariate mortality regression analysis of paraclinical and clinical variables in colistin-treated patients during the Delta pandemic period.

Variable (Delta)	Univariate Analysis			Multivariate Analysis
	HR	95% CI	p-Value	HR	95% CI	p-Value
Gender (M)	0.938	0.555–1.584	0.811
Age	1.037	1.018–1.056	<0.0001	1.027	1.010–1.043	0.001
Hospitalization days	0.953	0.933–0.971	<0.0001	0.975	0.961–0.988	<0.0001
Colistin dose	1.001	0.992–1.009	0.811
Colistin days	1.091	0.989–1.203	0.081
A. baumannii	6.620	3.344–13.102	<0.0001	5.548	3.070–10.024	<0.0001
P. aeruginosa	2.046	1.086–3.850	0.027	1.709	0.997–2.926	0.051
K. pneumoniae	1.444	0.807–2.579	0.215
E. coli	1.357	0.610–3.017	0.454
Co-infection with colistin-resistant GNB	1.477	0.656–3.324	0.346
Wound culture	0.534	0.129–2.206	0.386
Endotracheal aspirate	4.369	1.029–18.533	0.046	6.159	3.499–10.838	<0.0001
Urine culture	0.677	0.112–4.087	0.671
Blood culture	3.683	0.689–19.662	0.127
Fluid culture	3.823	0.692–21.109	0.124
Abscess culture	0.198	0.016–2.378	0.202
SARS-CoV-2 infection confirmed	2.011	1.046–3.864	0.036	1.827	1.002–3.346	0.049

Abbreviations: M = male; GNB = Gram-negative bacteria; HR = hazard ratio; CI = confidence interval; p = statistical significance coefficient; significant p-values are highlighted in bold.

).

During the Omicron period, we observed that K. pneumoniae and E. coli showed increased hazard ratios compared to the Delta period, indicating a shift in the dominant pathogens or resistance patterns. Additionally, the endotracheal aspirate culture type remained a strong predictor of poor outcomes (HR 5.949, p < 0.0001) (

Table 5. Univariate and multivariate mortality regression analysis of paraclinical and clinical variables in colistin-treated patients during the Omicron pandemic period.

Variable (Omicron)	Univariate Analysis			Multivariate Analysis
	HR	95% CI	p-Value	HR	95% CI	p-Value
Gender (M)	0.854	0.537–1.356	0.504
Age	1.044	1.027–1.060	<0.0001	1.031	1.014–1.048	<0.0001
Hospitalization days	0.952	0.935–0.968	<0.0001	0.954	0.935–0.972	<0.0001
Colistin dose	1.011	1.004–1.018	0.002	1.008	1.004–1.012	<0.0001
Colistin days	1.032	0.955–1.113	0.424
A. baumannii	3.365	1.812–6.250	<0.0001	3.809	2.214–6.551	<0.0001
P. aeruginosa	1.242	0.692–2.227	0.468
K. pneumoniae	4.731	2.672–8.374	<0.0001	1.091	0.649–1.833	0.742
E. coli	4.091	2.029–8.245	<0.0001	1.000	0.496–2.010	0.999
Co-infection with colistin-resistant GNB	1.867	0.814–4.282	0.140
Wound culture	0.602	0.208–1.738	0.348
Endotracheal aspirate	12.399	3.858–39.842	<0.0001	5.949	3.358–10.535	<0.0001
Urine culture	0.776	0.218–2.752	0.695
Blood culture	2.043	0.613–6.805	0.245
Fluid culture	1.232	0.327–4.628	0.758
Abscess culture	2.066	0.422–10.093	0.370
SARS-CoV-2 infection confirmed	1.125	0.471–2.683	0.790

Abbreviations: M = male; GNB = Gram-negative bacteria; HR = hazard ratio; CI = confidence interval; p = statistical significance coefficient; significant p-values are highlighted in bold.

).

Post-Pandemic Period: Key mortality predictors: Age (HR 1.069, p < 0.0001) and co-infections with colistin-resistant Gram-negative bacteria (HR 10.087, p = 0.001) were the strongest predictors of mortality; prolonged hospital stays were inversely associated with mortality (HR 0.943), possibly reflecting the selective admission of patients likely to survive. A. baumannii remained the most lethal pathogen (HR 7.998), followed by K. pneumoniae (HR 4.644) (

Table 6. Univariate and multivariate mortality regression analysis of paraclinical and clinical variables in colistin-treated patients during the post-pandemic period.

Variable (Post).	Univariate Analysis			Multivariate Analysis
	HR	95% CI	p-Value	HR	95% CI	p-Value
Gender (M)	1.064	0.469–2.410	0.882
Age	1.072	1.038–1.107	<0.0001	1.069	1.037–1.101	<0.0001
Hospitalization days	0.943	0.910–0.975	0.001	0.943	0.913–0.973	<0.0001
Colistin dose	1.016	1.003–1.027	0.012	1.008	1.002–1.014	0.007
Colistin days	0.911	0.795–1.043	0.177
A. baumannii	11.610	3.255-	<0.0001	7.998	3.368–18.987	<0.0001
P. aeruginosa	1.479	0.452–4.835	0.517
K. pneumoniae	6.368	2.104–19.263	0.001	4.644	1.967–10.959	<0.0001
E. coli	1.800	0.439–7.372	0.414
Co-infection with colistin-resistant GNB	7.860	1.774–34.816	0.007	10.087	2.654–38.329	0.001
Wound culture	0.099	0.024–0.394	0.001	0.105	0.046–0.233	<0.0001
Endotracheal aspirate	0.887	0.229–3.427	0.862
Urine culture	0.505	0.081–3.127	0.463
Blood culture	2.361	0.432–12.890	0.321
Fluid culture	1.437	0.189–10.880	0.726
Abscess culture	2.392	0.135–42.221	0.552
SARS-CoV-2 infection confirmed	6.623	0.834–52.545	0.074

Abbreviations: M = male; GNB = Gram-negative bacteria; HR = hazard ratio; CI = confidence interval; p = statistical significance coefficient; significant p-values are highlighted in bold.

).

3. Discussion

The overuse and improper administration of antibiotics in healthcare systems are major contributors to the development and disseminations of antibiotic-resistant bacteria [32,33,34].

The SARS-CoV-2 pandemic significantly influenced AMR trends, as previously described [35]. Our study confirms that pandemic-related factors, such prolonged hospital stays, higher ventilator use, and excessive empirical antibiotic prescribing, likely contributed to the observed rise in colistin-resistant isolates.

During the first two years of the COVID-19 pandemic (2020–2021), there was a significant increase in the antimicrobial prescriptions and total days of therapy, largely due to increased concerns over secondary bacterial infections, as well as the use of empirical antibiotics, including broad-spectrum agents [36,37]. It is estimated that 80% of hospitalized COVID-19 patients received antibiotics between March and October 2020, according to U.S. reports [2,38].

In our study, each pathogen demonstrates a unique distribution pattern across sample types, reflecting its biological niches and clinical manifestations. Wound cultures and endotracheal aspirates dominate nosocomial pathogens like A. baumannii and P. aeruginosa, while urine cultures are the main source for E. coli. The 2019–2021 period in our study shows notable shifts in the distribution for A. baumannii and K. pneumoniae, particularly in respiratory and blood cultures. This aligns with increased ICU admissions, invasive procedures, and ventilator use during the pandemic.

The present study highlights critical trends in the management of GNB infections over a decade, focusing on the impact of colistin therapy during and after the COVID-19 pandemic. Significant findings include the increased frequency of colistin use during the pandemic, with notable spikes in the Delta and Omicron phases, reflecting the growing prevalence of multidrug-resistant (MDR) infections; persistent challenges with A. baumannii and K. pneumoniae, which were the most frequent pathogens associated with colistin resistance and mortality, particularly during the pandemic. The proposed mathematical Model_1 demonstrated strong predictive capabilities for mortality across pandemic phases, with the highest accuracy during the post-pandemic period (AUC = 0.866).

The findings underscore the critical role of colistin in managing MDR GNB infections when therapeutic options are limited. The significant increase in colistin use during the pandemic highlights its essential role in treating severe infections caused by pathogens such as A. baumannii and K. pneumoniae. However, the rise in colistin-resistant infections during this period presents an alarming challenge for clinicians; during the Delta phase, the prevalence of A. baumannii and associated mortality (HR = 5.548, p < 0.0001) was particularly concerning, reflecting its role in severe cases. The Omicron phase saw an increase in K. pneumoniae-related infections, with colistin resistance further complicating management.

The study of Medrzycka-Dabrowska et al. on global antimicrobial resistance from 2021 has identified A. baumannii and K. pneumoniae as major contributors to MDR infections worldwide [39]. Consistent with these findings, our study also observed significant mortality associated with these pathogens, particularly among critically ill patients.

A. baumannii exhibited the highest mortality rates at various time points, highlighting the need for greater attention to bloodstream infections caused by this antibiotic-resistant bacterium. Additionally, a study reported 30-day all-cause mortality rates of 54.7%, 80.5%, and 63.8% for bacteremia caused by carbapenem-resistant A. baumannii, P. aeruginosa, and K. pneumoniae, respectively [40].

We observed a higher incidence of A. baumannii infections during the pandemic, similar to the findings reported by Cogliati Dezza et al., thus aligning with other reports that found carbapenem-resistant A. baumannii as the dominant pathogen. Its ability to persist in hospital environments and thrive in immunocompromised patients contributed to its increased prevalence. Their study further highlighted a notable shift in the epidemiology of MDR GNB, with a rise in CR A. baumannii infections in hospital admissions, while bloodstream infections caused by carbapenem-resistant K. pneumoniae (CR-KP) declined [35].

Our results similarly suggest that colistin-resistant isolates were predominantly A. baumannii, reinforcing concerns about its persistence as a difficult-to-treat pathogen.

Russo et al. further confirmed that COVID-19 patients had a higher incidence of MDR-AB infections, attributing this increase to prolonged mechanical ventilation and excessive antibiotic use [41].

Alenazi et al. demonstrated that COVID-19 patients co-infected with MDR-AB had significantly higher mortality rates than non-COVID-19 patients. They concluded that MDR-AB infections worsen prognosis in critically ill patients, which is consistent with our findings that colistin-resistant GNB infections were associated with prolonged hospitalization and poor clinical outcomes [42].

Patients with CR A. baumannii infections demonstrated significantly worse survival rates, as reported in our study as well as in the above-mentioned studies by Cogliati Dezza et al. and Alenazi et al. These findings highlight the urgent need for alternative therapeutic options beyond colistin. In a meta-analysis of 24 studies examining bacterial co-infections in hospitalized COVID-19 patients, co-infection rates were reported at 3.5% (95% CI: 0.4–6.7%) and secondary infections at 14.3% (95% CI: 9.6–18.9%) [43]. Bacterial co-infections were observed in 6.9% of cases, ranging from 5.9% in general hospitalized patients to 8.1% in critically ill individuals [43]. Common co-pathogens included A. baumannii, K. pneumoniae, E. coli, P. aeruginosa, Streptococcus pneumoniae, S. aureus, and others, including fungal species like Candida and Aspergillus, and viral pathogens such as influenza and rhinovirus [23,44,45,46,47]. These findings highlight the diversity of co-pathogens complicating COVID-19 cases, particularly in critically ill patients.

The findings of our research indicate that Gram-negative isolates were predominant among hospitalized COVID-19 patients, aligning with the study by Bazaid A.S. et al., which identified Gram-negative bacteria as the most common pathogens in COVID-19 cases [48].

Furthermore, the analysis reveals a rising trend in A. baumannii co-infections over the years, with the highest frequencies recorded between 2017 and 2021. The differences were statistically significant (p < 0.0001). These results highlight the prominent role of A. baumannii, a highly resistant pathogen, in severe polymicrobial infections, particularly during the pandemic years when antibiotic usage and hospitalization rates surged.

The results of our study demonstrate that A. baumannii and K. pneumoniae were the most frequently isolated bacterial species, particularly from sputum and blood samples. Supporting these results, A. baumannii has been identified as a primary pathogen in the respiratory tracts of COVID-19 patients, accounting for approximately 10% of all cultured samples [23]. Furthermore, another study reported its presence in 90% of COVID-19 patients [41]. Coinfection with A. baumannii in COVID-19 patients has been significantly associated with the development of systemic infections and an increased risk of mortality [22,49]. These findings are also consistent with the results of our study, highlighting the critical impact of A. baumannii on patient outcomes.

Similar trends are seen with other co-infections, especially during the years surrounding 2020. P. aeruginosa and K. pneumoniae combinations are particularly frequent, with statistical significance (p < 0.0001). This highlights the burden of multidrug-resistant GNB as co-pathogens during periods of increased healthcare strain, such as the pandemic.

Medrzycka-Dabrowska et al. reported in a recent review that the prevalence of carbapenem-resistant K. pneumoaniae isolated among COVID-19 patients ranged from 0.35% to 53% across six countries—Italy, China, Egypt, the United States, Spain, and Peru [39]. Additionally, P. aeruginosa has been frequently identified as a common co-infecting pathogen in COVID-19 patients, contributing to the exacerbation of illness [50,51,52,53,54].

In a study documenting antimicrobial use (AU), the consumption of colistin, imipenem, and meropenem significantly increased in the ICU and general wards after the onset of COVID-19 while remaining stable in the emergency department [55]. Our study similarly observed an increase in colistin use during the pandemic, particularly in critically ill patients. However, unlike the referenced study, our data further highlight a distinct spike in colistin use during the Omicron phase, reflecting the ongoing reliance on this antibiotic in managing multidrug-resistant infections during the pandemic’s later stages. This comparison underscores the consistent pattern of heightened colistin use in response to the pandemic’s pressures on healthcare systems, especially in settings managing severe infections.

Our findings expand on prior research by providing a decade-long perspective that incorporates the unique pressures of the COVID-19 pandemic. Notably, the significant increase in colistin use during the pandemic is consistent with studies that have reported heightened antimicrobial consumption due to prolonged hospital stays, increased ventilator usage, and co-infections. Our study also provides novel insights into the differential impacts of pandemic phases on pathogen prevalence in each sample type.

The longitudinal design of our study is a significant strength, providing a comprehensive analysis of trends over a decade and capturing both pre-pandemic and pandemic-specific dynamics. Additionally, the use of predictive modeling (e.g., Model_1) offers valuable insights into mortality risks and clinical outcomes, with important implications for patient management.

The limitations of our study include its retrospective design, which may introduce selection bias due to the focus on colistin-treated patients in specific settings, and its restricted temporal and geographic scope, which may not fully capture global trends. Additionally, the analysis is limited to a single center in Romania, which while representative of the northeastern region of the country, may not fully capture variations in clinical practices, healthcare infrastructure, and patient demographics across different regions. Expanding the study to include multiple centers would enhance the robustness of our findings, allowing for a more comprehensive understanding of GNB infections, treatment responses, and potential regional differences in healthcare delivery. This broader approach would ultimately strengthen the validity and generalizability of our conclusions.

4. Materials and Methods

We conducted a retrospective observational study to evaluate trends in colistin resistance and clinical outcomes among colistin-treated patients over a 13-year period (2011–2023).

The study was conducted at “Saint Spiridon” County Clinical Emergency Hospital, a tertiary care facility in Iasi, Romania, and focused on Gram-negative bacterial (GNB) infections, particularly those caused by Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Escherichia coli. Notably, “Saint Spiridon” Emergency Clinical County Hospital serves as a university hospital that, according to internal statistical data, provides medical care to 67% of patients from Iasi County and 33% of patients from the northeastern region of Romania. This region has a population of over 3.7 million residents, accounting for more than 17% of the country’s total population [24]. The study period was further divided into three phases, namely the pre-pandemic (2011–2019), pandemic (Delta and Omicron waves, 2020–2022), and post-pandemic (2023).

In Romania, the use of colistin is guided by national and international antimicrobial stewardship recommendations, particularly in the treatment of multidrug-resistant (MDR) Gram-negative infections. The national guidelines, issued by the Romanian Society of Infectious Diseases, align with the European Committee on Antimicrobial Susceptibility Testing (EUCAST) and European Society of Clinical Microbiology and Infectious Diseases (ESCMID) guidelines, which emphasize colistin as a last-resort antibiotic for carbapenem-resistant A. baumannii, P. aeruginosa, and K. pneumoniae infections [56,57]. The national guidelines recommend intravenous colistin use only in cases where other therapeutic options are ineffective, with dosing adjusted based on renal function to minimize nephrotoxicity. The final therapeutic decision was left at the discretion of the attending physicians.

4.1. Isolate Characterization

Bacterial isolates were identified using standard microbiological techniques and using matrix-assisted laser desorption ionization–time-of-flight mass spectrometry (MALDI-TOF MS) (Bruker Daltonik GmbH, Bremen, Germany).

Colistin susceptibility testing was performed by the broth microdilution method in an automated system on the MICRONAUT-S (Merlin, Dortmund, Germany) and by broth microdilution strips. For the other antibiotics, testing was conducted using both the MICRONAUT-S system and the disk diffusion method. The interpretation of the results was performed according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) standard, applicable at the time [57]. For colistin, The Clinical and Laboratory Standards Institute (CLSI) and EUCAST have established colistin susceptibility testing through the determination of the minimal inhibitory concentration (MIC) using broth microdilution, a standardized method worldwide accepted [57,58].

4.2. Statistical Analysis

Data analysis was conducted using Graph Pad Prism, v5 (Graph Pad Software, San Diego, CA, USA) and SPSS, v25 (IBM SPSS Software, Chicago, IL, USA). The significant associations between distinct categorical variables were determined using the chi-squared test. Univariate and multivariate analyses were performed to identify factors associated with patient mortality. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated for key clinical and paraclinical variables. Receiver operating characteristic (ROC) curves were generated to assess the predictive accuracy of the models for mortality across different phases of the pandemic. The pandemic period was temporally split in four phases, namely the pre-pandemic (from 1 January 2017 until the 25 February 2020), the Delta pandemic time (from the start of SARS-CoV-2 pandemic on the 26 February 2020 until the 3 December 2021), the Omicron period (from the 4 December 2021, the date when the first two cases of Omicron infection were officially reported in Romania, until the 5 May 2023), and post-pandemic (starting with 5 May 2023 until the end of 2024). Statistical significance was defined as p < 0.05.

5. Conclusions

This study highlights the evolving landscape of GNB infections and colistin resistance over the past decade, with the COVID-19 pandemic serving as a pivotal factor in shaping treatment and resistance patterns. The findings point to the importance of continued antibiotic resistance monitoring, along with trend analysis, to better understand the evolution of MDR infections.