Comparative analysis of colistin minimum inhibitory concentration detection methods and risk factors for multidrug-resistant organism in intensive care units and high dependancy units: Implications for infection control practices

INTRODUCTION

Advancements in medical and public health have led to longer life spans, increasing vulnerability to infectious diseases. Infection control is critical for patient safety, as healthcare-associated infections (HAIs) contribute to morbidity, mortality, prolonged hospital stays, and costs.¹ Multidrug-resistant bacteria (MDRB) pose a global challenge, impacting intensive care units (ICU), transplants, cancer therapy, and surgery patients. Common MDRB pathogens include Staphylococcus aureus, Enterococcus faecalis, Streptococcus pneumoniae, Klebsiella pneumoniae, Pseudomonas aeruginosa, Escherichia coli, and Acinetobacter baumannii.^2,3 Overuse of antibiotics drives resistance, with bacterial biofilm formation complicating treatment.³

Colistin, a last-resort polymyxin antibiotic, is essential for treating Gram-negative infections despite its neurotoxic and nephrotoxic effects.^4,5 For ICU patients, Ceftazidime-avibactam (CAZ-AVI) combined with metronidazole serves as an alternative treatment, targeting β-lactamase-producing pathogens.^6,7 By 2050, antibiotic resistance is expected to rise tenfold, with ESKAPE pathogens evading standard treatments.⁸ Many MDR isolates resist carbapenems and β-lactams but remain susceptible to colistin and tigecycline, highlighting the need for rigorous infection control and monitoring.^9,10 Traditional multi-drug-resistant organism (MDRO) surveillance is labor-intensive, requiring extensive manual effort from infection control personnel. The process involves meticulous tracking of resistant organisms, making large-scale monitoring challenging and prone to inconsistencies.¹¹ In India, antibiotic misuse has increased resistance rates, prompting antibiotic stewardship programs (AMSP) to improve prescribing practices.

This study compared four methods, broth microdilution (BMD), Vitek, colistin agar test (CAT), and colistin disc elution (CDE), to detect colistin-resistant organisms. BMD remains the gold standard, providing the most accurate results, but is labor-intensive. Vitek is automated and fast, yet may produce false susceptibility classifications. CAT is highly sensitive (96.92%) but lacks precise minimum inhibitory concentration (MIC) determination, while CDE (70.76%) is easier to use but less reliable. BMD and Vitek are the most effective methods, with CAT as a strong alternative, though CDE requires further validation.

A systematic review and meta-analysis in the Annals of Clinical Microbiology and Antimicrobials confirms that BMD is the gold standard for colistin susceptibility testing, analyzing 55 studies and validating its superiority over Vitek 2, reinforcing its clinical reliability.¹²

Aim and objectives

MATERIAL AND METHODS

Materials required

Colistin agar test

• Colistin stock solution (1 mg/mL): Prepared by dissolving 10 mg of colistin sulfate powder (Sigma-Aldrich, St. Louis, MO, USA) in 6.3 mL of double-distilled water.

• Working solution: Prepared by taking four sterile tubes labeled as growth control (GC), 1, 2, and 4 µg/mL. In each tube, 10 mL of double-distilled water was added. For 1, 2, and 4 µg/mL, 100, 200, and 400 µL of double-distilled water were discarded, respectively, and replaced with 100, 200, and 400 µL of the colistin stock solution, respectively.

• Muller-Hinton agar (MHA): Prepared by dissolving 3.42 g of MHA (Thermo Fisher Scientific, Waltham, MA, USA) in 90 mL of distilled water in separate flasks labeled GC, 1, 2, and 4 µg/mL.

Inoculum preparation

• Around 3-5 colonies were selected from a non-selective agar plate using a loop or swab and transferred to 5 mL of sterile saline.

• The turbidity was adjusted to match the 0.5 McFarland standard (approx. 1.5 x 10⁸ CFU/mL).

• The standardized inoculum was diluted 1:10 with saline.

CAT procedure

• Each colistin agar plate was divided into eight sections for eight different isolates, labeling each section accordingly.

• 10 µL of the 1:10 dilution was transferred to the appropriate section of each colistin agar plate using a sterile pipette and streaked with a loop [Figure 1].

• The original inoculum was subcultured onto a blood agar plate using a 10 µL loop to check for purity and incubated at 37°C for growth.^13,14

Close

Figure 1:

Comparative workflow of four colistin susceptibility testing methods. Stepwise schematic of CAT, Vitek MIC, BMD, and CDE methods for Gram-negative isolates. Highlights procedural differences in setup, incubation, and result interpretation for resource-based method selection. CAT: Colistin agar test, MIC: Minimum inhibitory concentration, BMD: Broth microdilution, CDE: Colistin disc elution.

Export to PPT

Vitek MIC

• For this, 3-5 colonies were selected from a non-selective agar plate using a loop or swab and transferred to 5 mL of sterile saline.

• The turbidity was adjusted to match the 0.5 McFarland standard.

• The standardized inoculum was loaded into Vitek cards (bioMérieux, Marcy-l’Étoile, France) and processed in the Vitek instrument. Results were recorded the following day.^13,14

Broth microdilution

• For this, 3-5 colonies were selected from a non-selective agar plate using a loop or swab and transferred to 5 mL of demineralized water. The turbidity was adjusted to match the 0.5 McFarland standard.

• BMD tubes were prepared by transferring 30 µL of the inoculum to 11 mL of Muller-Hinton broth (Thermo Fisher Scientific, Waltham, MA, USA) [Figure 2 and 3].

• 50 µL was inoculated per well into a microdilution plate, covered with an adhesive seal, and incubated aerobically at 34-36°C for 18-24 hours.^13,14

Close

Figure 2:

BMD Test: In a 96-well plate with pre-coated colistin, add each isolate to a 12-well row. Interpret the maximum MIC where no button growth occurs; use the 12th well for control. Interpretation: >0.12 to ≤2 μg/ml (sensitive), ≥4 to <128 μg/ml (resistant) , BMD: Broth microdilution, MIC: Minimum inhibitory concentration.

Export to PPT

Close

Figure 3:

BMD uses a 96-well plate pre-coated with colistin in increasing concentrations as depicted in this picture , BMD: Broth microdilution.

Export to PPT

CDE method

• Four glass test tubes were labeled for each test strain as control, 1, 2, and 4 µg/mL.

• Around 10 mL of Cation-Adjusted Mueller-Hinton Broth (Thermo Fisher Scientific, Waltham, MA, USA) was added to each tube.

• 0, 1, 2, and 4 colistin discs were aseptically added to the respective tubes, achieving final concentrations of 0, 1, 2, and 4 µg/mL. Tubes were vortexed and incubated at room temperature for 45 minutes.^13,14

• Then, 50 µL of the standardized inoculum was added to the respective tubes, vortexed, and incubated at 33°C for 18-24 hours [Figure 4].

• The original inoculum was subcultured onto a blood agar plate to check for purity and incubated at 37°C for growth.

• The MIC was interpreted as the lowest concentration at which no growth was observed.

Close

Figure 4:

Standardized Inoculum Testing: Add 50 μL to control, 1, 2, and 4 μg/ml tubes. Vortex, loosen caps, and incubate at 33°C for 1-20 hours. Subculture on blood agar, incubate at 37°C. Interpret MIC as lowest concentration with complete inhibition. Valid if GC is turbid. Interpretation: ≥2 μg/ml (sensitive), ≤4 μg/ml (resistant).

Export to PPT

Ethical Committee Waiver: As this was supplementary routine testing for Colistin resistance and only retrospective data from Hospital Infection Surveillance System records for quality improvements, a waiver from the ethical committee was sought.

RESULTS

Interpretation criteria with results:

Broth microdilution

BMD uses a 96-well plate pre-coated with increasing concentrations of colistin. Each isolate is added to a 12-well row, and the MIC is interpreted based on the last well where no bottom formation occurs. The MIC ranges were >0.12 to ≤2 µg/mL for sensitive (S) and ≥4 to <128 µg/mL for resistant (R) isolates. Colistin MIC values were compared using Vitek, CAT, and CDE methods, with BMD serving as the gold standard reference.

Vitek MIC

All isolates tested by Vitek showed MIC > 4 µg/mL and were thus considered colistin resistant. The sensitivity was 100%.

Colistin agar test

CAT detected colistin-resistant isolates effectively, with different concentrations: GC, 1, 2, and 4 µg/mL. Out of 65 isolates, two had MIC ≤2 µg/mL (one Klebsiella pneumoniae and one Acinetobacter baumannii). Sensitivity of CAT was calculated as follows:

• Sensitivity = True Positive/(True Positive + False Negative) × 100

• Sensitivity = 63/(63 + 2) × 100 = 96.92%

Colistin disc elution

CDE was performed according to CLSI guidelines. The results showed that out of 65 isolates, 46 were resistant. Sensitivity of CDE was calculated as follows:

• Sensitivity = True Positive/(True Positive + False Negative) × 100

• Sensitivity = 46/(46 + 19) × 100 = 70.76%

Comparison of methods

CAT proved more accurate with a sensitivity of 96.92% compared to CDE in resource-limited settings as shown in Table 2.

Table 2: Sensitivity and specificity of methods.

Method	Sensitivity (%)	Specificity (%)	True positive	False negative	True negative	False positive	VME	ME
Broth microdilution	100	-	65	0	-	-	0	0
Vitek	100	-	65	0	-	-	0	0
Colistin agar test	96.92	-	63	2	-	-	2	0
Colistin disc elution	70.76	-	46	19	-	-	19	0

VME: Very major errors, ME: Major errors

Distribution of colistin resistance

Colistin-resistant isolates showed resistance to carbapenems, cephalosporins, fluoroquinolones, and aminoglycosides, with MIC values ranging from ≥16 to ≥320. Klebsiella pneumoniae was the most affected species, associated with prolonged hospital stays and increased mortality [Figure 5].

Close

Figure 5:

Distribution of colistin resistance. The above pie chart shows the prevalence of various organisms showing colistin resistance: Klebsiella pneumoniae 51 (78%), Pseudomonas aeruginosa 9 (14%), Acinetobacter baumannii 3 (5%), and Escherichia coli 2 (3%).

Export to PPT

Specimen-wise distribution of colistin-resistant isolates

Displays the frequency of colistin-resistant isolates across different clinical specimen types, aiding in source-specific surveillance [Figure 6].

Close

Figure 6:

Specimen wise distribution of colistin resistant isolates - The above bar graph depicts the specimen wise distribution of colistin resistant isolates having the maximum in tracheal samples (83%) followed by pus (12%) and external ventricular drain (EVD) Tips (5%).

Export to PPT

Area-wise dispersion of resistant isolates

Illustrates the geographic or departmental spread of resistant strains, supporting targeted infection control interventions [Figure 7].

Close

Figure 7:

Area-wise dispersion of resistant isolates - The above bar graph represents the area-wise dispersion of resistant isolates having the maximum in emergency ICU (49%) followed by Step Down ward (29%) and neurosurgical ICU. (22%). ICU: Intensive care unit.

Export to PPT

These analyses showed Klebsiella pneumoniae (n=51) was the most common organism, with Emergency ICU (EICU) having the maximum isolates. Tracheal aspirates yielded most colistin-resistant isolates.

Discrepancies in colistin susceptibility results

Interpretation

Long-term hospital stay

Patients with long-term hospital stays were significantly more likely to develop MDRO infections. The odds ratio (OR) of 6.24 indicates that such patients are over six times more likely to have MDRO infections compared to those with shorter stays.

Immunocompromised state

Immunocompromised patients were also at higher risk, with an OR of 3.61, indicating a 3.61 times greater likelihood of MDRO infection.

Use of broad-spectrum antibiotics

The use of broad-spectrum antibiotics was strongly associated with MDRO infections, with an OR of 7.14. Patients receiving these antibiotics were over seven times more likely to develop MDRO infections.

Recent surgery

Patients who had undergone recent surgery had an OR of 4.26, suggesting they were more than four times as likely to develop MDRO infections.

ICU admission

ICU admission was also a significant risk factor, with an OR of 2.58, indicating these patients were over twice as likely to develop MDRO infections compared to non-ICU patients.

All the variables included in the model were statistically significant, with p-values less than 0.05 as depicted in Table 4.

Table 4: Risk factor analysis using logistic regression.

Variable	Coefficient (β)	Standard error	Odds ratio (OR)	95% CI for OR	p-value
Long-term hospital stay	1.83	0.42	6.24	2.75 - 14.14	<0.001
Immunocompromised state	1.28	0.37	3.61	1.75 - 7.41	<0.001
Use of broad-spectrum antibiotics	1.97	0.45	7.14	3.03 - 16.81	<0.001
Recent surgery	1.45	0.40	4.26	1.97 - 9.21	<0.001
ICU admission	0.95	0.30	2.58	1.42 - 4.69	0.002

p-values less than 0.05 statistically significant, CI: Confidence interval, ICU: Intensive care unit.

Overview of four colistin susceptibility testing workflows

Stepwise comparison of CAT, Vitek MIC, BMD, and CDE methods for detecting colistin resistance in Gram-negative bacteria as shown in Figure 8.

Close

Figure 8:

Schematic workflow of colistin susceptibility testing methods used for comparative analysis. MIC: Minimum inhibitory concentration.

Export to PPT

DISCUSSION

The increasing incidence of multidrug-resistant Gram-negative bacteria (MDR-GNB) in ICU and HDU is a significant concern for patient safety. Colistin (polymyxin B) is often the last-resort antibiotic against gram-negative ESKAPE pathogens, including Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species. A study conducted at NIMHANS identified 65 colistin-resistant MDR isolates, primarily in the EICU, where prolonged hospital stays and lapses in infection control may have facilitated resistant organism transmission¹⁵

Colistin-resistant isolates exhibited resistance to multiple antibiotic classes, including tetracyclines, carbapenems, cephalosporins, fluoroquinolones, fosfomycin, amikacin, and aminoglycosides, with MIC values ranging from ≥16 to ≥320. This high degree of resistance complicates treatment options and contributes to increased hospital stays and mortality, particularly in cases involving Klebsiella pneumoniae.^16,17

Comparisons among methods for colistin MIC detection followed CLSI guidelines (2022) and evaluated BMD, CAT, CDE, and the Vitek® automated system. BMD, as the reference standard, with Vitek® showing equivalent sensitivity.¹⁸ CAT was a cost-effective and reliable alternative with 96.92% sensitivity, making it suitable for resource-limited settings. However, CDE had lower sensitivity at 70.76%, highlighting discrepancies possibly linked to genetic mutations in resistant strains. Further research is needed to validate and refine the CDE method.¹⁹

Robust infection control practices remain critical in preventing MDR organism transmission. Adherence to CDC guidelines for standard and contact precautions, combined with antimicrobial stewardship programs (AMSP), can mitigate MDR infection risks. Enhanced surveillance and rapid resistance pattern identification are essential for effective management.²⁰

Discrepancies between CDE and BMD results can be attributed to several factors. Interpretation differences play a role, as criteria for categorizing isolates as resistant or susceptible may vary between methods.²¹ Technical variability, including differences in inoculum size, incubation times, or media used, can affect test outcomes.²² Method sensitivity also contributes, with CDE showing a lower sensitivity of 70.76% compared to BMD, leading to false negatives when CDE indicates susceptibility while BMD shows resistance.²³ Additionally, heteroresistance, where a subpopulation of bacteria is resistant while the majority remain susceptible, may lead to discrepancies in results.²⁴ Finally, colistin’s cationic properties can impact its diffusion and activity in different testing environments, further influencing variations between CDE and BMD.²⁵

This study reinforces the importance of selecting appropriate and cost-effective methods for the detection of colistin resistance, particularly in resource-constrained settings. Reliable techniques such as CAT can support the timely identification of resistant strains and inform antibiotic therapy decisions. Additionally, promoting alternative treatments like CAZ-AVI can improve patient outcomes and reduce MDR-related mortality.²⁶

Its novelty lies in several aspects. Isolates were primarily obtained from EICUs, representing a high-risk population that has been less frequently studied. The study employed a broad methodological approach, comparing four distinct colistin MIC testing methods, and included 65 colistin-resistant isolates, enhancing statistical reliability. Compared to existing studies focusing on single-method evaluations or smaller sample sizes, this research provides a comprehensive perspective on susceptibility testing reliability and resistance mechanisms, particularly in high-burden clinical settings.

CONCLUSION

The emergence of MDR-GNB is a critical global health issue, particularly in ICU and HDU settings. Emerging resistance to colistin, one of the last-resort drugs for MDR-GNB infections, is posing a serious hazard. BMD and VITEK® systems, while highly sensitive, are costly and not feasible for routine use. CAT offers a cost-effective alternative with 96.92% sensitivity, suitable for resource-limited settings. CDE requires further validation due to inconsistent results. A multifaceted approach, including accurate diagnostics, robust AMSPs, stringent IPC measures, and continuous research, is essential to manage and mitigate the threat of colistin-resistant infections.