Knowledge, Perception, and Antibiotic Prescribing Practice in the Intensive Care Unit: Findings from the Malaysian Public Setting.

INTRODUCTION: Approach to managing infection in the intensive care unit (ICU) often varies between institutions and not many readily adapt to available local guidelines despite it was constructed to suite local clinical scenario. Malaysia already has two published guidelines on managing infection in the ICU but data on its compliance are largely unknown.
OBJECTIVES: A cross-sectional survey was carried out and sent to a total of 868 specialists working primarily in the ICU. The aim of this study was to explore knowledge, perception, and the antibiotic prescribing practice among specialists and advanced trainees in Malaysian ICU.
MATERIALS AND METHODS: A cross-sectional survey was used, consisted of three sections: knowledge, perception, and antibiotic prescribing practice in ICU. Three case vignettes on hospital-acquired pneumonia (HAP), infected necrotizing pancreatitis (INP), and catheter-related bloodstream infection (CRBSI) were used to explore antibiotic prescribing practice.
RESULTS: A total of 868 eligible subjects were approached with 104 responded to the survey. Three hundred eighty-nine antibiotics were chosen from seven different classes in the case vignettes. All respondents acknowledged the importance of pharmacokinetic/pharmacodynamic (PK/PD) in antibiotic optimization and majority (97.2%) perceived that current dosing is inadequate to achieve optimal PK/PD target in ICU patients. Majority (85.6%) believed that antibiotic dose should be streamlined to the organisms' minimum inhibitory concentration (MIC). In terms of knowledge, only 64.4% provided the correct correlations between antibiotics and their respective PK/PD targets. Compliance rates in terms of antibiotic choices were at 79.8%, 77.8%, and 27.9% for HAI, INP, and CRBSI, respectively.
CONCLUSION: Malaysian physicians are receptive to use PK/PD approach to optimize antibiotic dosing in ICU patients. Nonetheless, there are still gaps in the knowledge of antibiotic PK/PD as well as its application in the critically ill, especially for β-lactams. Copyright:

INTRODUCTION

Patients with infection who are admitted into the intensive care unit (ICU) have a substantially increased risk of morbidity and mortality ICU patients commonly show profound pathophysiological changes and are commonly infected with organisms with reduced antibiotic susceptibility.[12] Together, these would hamper the attainment of optimal antibiotic exposure and pharmacokinetic/pharmacodynamic (PK/PD) targets in ICU patients.[3] Antibiotic dosing approach is often variable as different institutions have their perspectives as reported by the survey on antimicrobial dosing and monitoring in ICUs (ADMIN-ICU) survey.[4] The ADMIN-ICU survey reported inconsistent dosing for various classes of antibiotics.

Certain countries, such as France, have developed their own antibiotic guidelines in the ICU. Nonetheless, β-lactam administration and therapeutic drug monitoring among critically ill patients in France (ANTIBIOPERF) study reported that only a few ICUs in France are currently following their antibiotic prescribing guideline despite clinicians’ belief that such a resource will lead to better clinical outcomes.[5]

In Malaysia, guidelines in managing infections in the ICU[6] are available for use by local clinicians. The first guideline, Guide to Antimicrobial Therapy in Adult ICU, was made available for use in 2012, with the second edition published 5 years following this. However, the rates of physicians’ compliance with this guideline remain mostly unknown. Therefore, this study aimed to explore the knowledge, perception, and antibiotic prescribing practice among medical specialists and advanced trainees in Malaysian ICUs.

MATERIALS AND METHODS

This was a nation-wide survey that was carried out for a period of 8 months from December 2017 to July 2018 aiming to explore physicians’: (1) perception and agreeability on PK/PD issues and optimization in critically ill patients; (2) knowledge on antibiotic optimization of commonly used antibiotics in the ICU; and (3) antibiotic prescribing patterns for infections in the ICU. Target population was medical specialists working primarily in the ICU together with their registrars.

The survey was constructed based on previous studies, which comprises three parts designed to evaluate knowledge, perception, and daily practice in the ICU [Supplementary Material 1].[45,7]

For the perception’s part, a Likert scale was used to determine respondents’ perception toward (1) critically ill patients in the ICU; (2) the need for therapeutic drug monitoring (TDM) in the ICU; and (3) antibiotic dosing in the ICU.

Finally, three different case vignettes, namely hospital-acquired pneumonia (HAP), infective acute pancreatitis (IAP), and catheter-related bloodstream infection (CRBSI), were used in the third part of the survey, to explore respondents’ prescribing patterns in managing such infections. Each response was then compared with the Malaysian Guide to Antimicrobial Therapy in the Adult ICU 2017.[6]

The target population was specialists working primarily in the ICU. On the basis of the Malaysian National Specialist Register database, it is estimated that the whole study population is around 700. For determining the appropriate sample size, the equation for finite population was used.[8] The equation for determining the sample size was as follows:

where n′ is the sample size with finite population correction, N is the population size, Z is the Z-statistic for a level of confidence, P is the expected proportion (in proportion of one), and d is the precision (in proportion of one).

The values for N, Z, P, and d in this study were 700, 1.96, 0.5, and 0.05, respectively,

From the equation above, the calculated sample size, n′, will be 248. With this sample size, the study will be able to generate a 95% confidence level with 5% margin of error.

The survey was disseminated to all major ICUs in Malaysia via Google Forms. Prior to distribution, the survey was validated by five panels consisting of an intensivist, an infectious disease specialist, and three lecturers from local university who are accomplished in the field of PK/PD antibiotics. Data were analyzed using Statistical Package for the Social Sciences software program for Windows version 22.0 (IBM, Armonk, New York). This survey study was approved and registered by the Malaysian National Medical Research Register (NMRR) with identification number, NMRR-18-336-40425.

RESULTS

Overall, 104 respondents from both public and teaching hospital ICUs in Malaysia completed the survey with the majority coming from public hospitals (51.0%). Anesthesiology and critical care specialist made the bulk of the respondents (41.3%).

Results on knowledge

All respondents acknowledged that critically ill patients have profound pathophysiological changes and that each antibiotic has its own distinctive PK/PD characteristics for optimal antibacterial activity. Nonetheless, only 52.9% of respondents were able to correctly identify antibiotics exhibiting time-dependent activity. Moreover, when asked about the correlation of a given antibiotic with its own PK/PD, only 65.0% of respondents answered correctly.

Only 13.5% of respondents were able to answer correctly for the best PK/PD exposure of β-lactam antibiotics in critically ill patients which states that the concentration should be above the minimum inhibitory concentration (MIC) of the infecting pathogen for 100% of the dosing interval (100% fT>MIC) with the majority coming from anesthesiology and critical care specialist. Most (32.7%) believed that the target should be 70% fT>MIC.

Results on perception

The majority agreed with all the statements presented except on the statement “Current antibiotic dosing in critically ill patients may likely lead to inadequate antibiotic exposure” with only 35.0% of respondents who agreed. Respondents acknowledged that antibiotic dosing should be based on the infecting pathogens and its PK/PD characteristics. But only 51.0% concurred that TDM is vital in the ICU.

Results on antibiotic prescribing practice

A total of 17 different antibiotics from seven different classes were empirically prescribed across the three case vignettes. Monotherapy was the favored choice in all three cases, 83.7%, 80.8%, and 60.6% for HAP, IAP, and CRBSI, respectively.

Case 1: hospital-acquired pneumonia

The number of antibiotics that were empirically prescribed in Case 1 was 121 with piperacillin–tazobactam being the preferred antibiotic (62.0%). Only 17 respondents (14.0%) opted for combination therapy with azithromycin being the preferred add-on antibiotic (88.2%). Antibiotic prescribing compliance with the guideline was 79.8%.

Of those who showed compliance, 80.7% chose the appropriate daily dosing regimen for the chosen agents. The median duration of days of antibiotic being prescribed was 7 days (7.0–10.0). Table 1 shows the respondents’ antibiotic prescribing patterns and their compliance with the guideline.

Table 1

Drug choice and dosing compliance for each case

Compliance of prescribing with “Guide to Antimicrobial Therapy in the Adult ICU 2017”
Case	Guideline categorya	Drug choice			Dosing regimenb
		NC	C	C + NC	BR	C	AR
Case 1: HAP	HAP: High risk of multidrug-resistant organisms	16 (15.4%)	83 (79.8%)	5 (4.8%)	16 (19.3%)	67 (80.7%)	0
Case 2: IAI	Acute infective pancreatitis	11 (10.6%)	81 (77.9%)	12 (11.5%)	17 (21.0%)	59 (72.8%)	5 (6.2%)
Case 3: CRBSI	Catheter related blood stream infection	75 (72.1%)	29 (27.9%)	0	6 (20.7%)	19 (65.5%)	4 (13.8%)

AR = above recommendations, BR = below recommendations, C = compliant, C + NC = compliant, but with additional noncompliant antibiotics, HAP = hospital-acquired pneumonia, IAI = intra-abdominal infection, NC = noncompliant

Drug choice compliance is reported as N (%). The total numbers for drug choice are numbers of responses for each case

aGuideline category used as basis for comparison

bDosing regimen compliance reported as N (%) for patients with compliant drug choice

Case 2: acute infective pancreatitis

A larger variation of antibiotic selection was seen in Case 2, with carbapenems being the preferred antibiotic (55.2%). Of these, 40.8% chose meropenem as the carbapenem of choice, whereas 14.4% opted for imipenem. Combination therapy only constituted 20.2% of respondents.

Drug choice compliance for Case 2 was 77.9% with only 72.8% of them met the proposed dosing regimen. The median duration of days of antibiotic prescribed in Case 2 was 7 days (7.0–10.0).

Case 3: catheter-related bloodstream infection

The initial choice of empiric therapy against CRBSI showed the largest variation, with vancomycin being the predominant agent (35.2%), followed by piperacillin–tazobactam and cloxacillin at 19.3% and 15.9%, respectively. Monotherapy was the preferred choice in Case 3, (60.6%). Overall, low compliance rate to guideline recommendation was seen, with the rate of 27.9%.

Of 29 respondents who did comply, 19 (65.5%) chose the recommended dosing regimen. The mean duration days of antibiotic prescribed in Case 3 were 9 days (7.0–14.0).

Mode of administration

Prolonged infusion defined as either extended infusion (EI) or continuous infusion (CI) was the favored choice in β-lactam antibiotics as shown in Table 2. Meropenem was the antibiotic being likely to be prescribed as either extended or CI with 90.1% of respondents opted to do so.

Table 2

Summary on the mode of administration for antibiotics prescribed

Antibiotic	Intermittent bolus dosing, N (%)	Extended infusion, N (%)	Continuous infusion, N (%)
β-lactam (non-carbapenem)
Amoxicillin-clavulanate	5 (55.6%)	4 (44.4%)	0
Cefazolin	1 (50.0%)	1 (50.0%)	0.0
Cefepime	7 (36.8%)	12 (63.2%)	0.0
Cefoperazone	2 (40.0%)	3 (60.0%)	0.0
Cefotaxime	1 (50.0%)	1 (50.0%)	0.0
Ceftazidime	3 (60.0%)	2 (40.0%)	0.0
Ceftriaxone	11 (78.6%)	3 (21.4%)	0.0
Cefuroxime	1 (33.3%)	1 (33.3%)	1 (33.3%)
Cloxacillin	11 (47.8%)	7 (30.4%)	5 (21.7%)
Piperacillin-tazobactam	30 (23.1%)	90 (70.0%)	9 (6.9%)
β-lactam (carbapenem)
Imipenem	3 (14.3%)	16 (76.2%)	2 (9.5%)
Meropenem	7 (9.9%)	55 (77.5%)	9 (12.7%)
Nitroimidazole
Metronidazole	11 (78.6%)	3 (21.4%)	0.0
Macrolide
Azithromycin	13 (86.7%)	2 (13.3%)	0.0
Fluoroquinolone
Ciprofloxacin	1 (100.0%)	0.0	0.0
Lincosamide
Clindamycin	1 (100.0%)	0.0	0.0
Glycopeptide
Teicoplanin	1 (100.0%)	0.0	0.0
Vancomycin	31 (56.4%)	19 (34.5%)	5 (9.1%)

DISCUSSION

To the best of our knowledge, this study is the first to explore clinician’s perception, knowledge, and practice of antibiotic use in Malaysian ICU. Majority of recruited physicians were receptive of PK/PD concept in antibiotic optimization for ICU infections, with close to halve agreed that current dosing was insufficient. These results are congruent with the findings of ANTIBIOPERF, a survey performed in French, with 77% of respondents agreeing to β-lactams optimization via prolonged infusion. Furthermore, about half of respondents perceived the value of routine TDM for all ICU patients despite majority agreeing that dosing should be based on PK/PD factors and pathogen’s MIC. TDM is crucial in the ICU as patients warded there often have deranged pathophysiological changes necessitating individual dosing.[49] In a similar study done by Tabah et al.,[4] the authors found that TDM was routinely applied but only limited for glycopeptides and aminoglycosides. A Belgian survey study in 2013 also documented regular TDM practice in the ICU but as in the ADMIN-ICU Study, it is only limited to certain antibiotic which in this case vancomycin.[10] As our study did not differentiate on the type of antibiotics being used for TDM, it might explain on the low number of respondents agreeing on the usage of TDM in ICU.

Moreover, there is a sizeable gap between respondent’s perception and their knowledge on antibiotic PK/PD. Our findings echoed the results of a recent survey performed in Amsterdam, whereby more than half of intensivists failed to identify the relevant PD target for β-lactams.[11] These trends call for an improved education on antibiotic PK/PD among physicians in managing infections, especially in critically ill patients. This is especially essential, given that sepsis is the leading cause of ICU admissions in Malaysian ICU.

Wide variability in prescribing practices was found throughout the case vignettes. Recommendations for empirical antibiotic treatment were deemed appropriate only for HAP and AIP as CRBSI showed poor compliance. Majority opted for monotherapy vancomycin as their initial empiric therapy when the guideline proposed otherwise. Combination therapy consisting of broad-spectrum antibiotics together with either cloxacillin or vancomycin (if there is evidence of MRSA) is the suggested empirical therapy in CRBSI since Malaysian Registry of Intensive Care Report that in 2015 and 2016, gram-negative bacteria predominate at 68.2% and 54.6%, respectively. Only 26.9% of respondents adhere to the ICU guideline.

In addition, patterns of antibiotic prescribing were found to be variable, relative to the recommendations as set in the local guidelines. Management of CRBSI showed the lowest rates of compliance, whereby majority of respondents preferred the use of non-recommended agents, as well as monotherapy in the initial empiric therapy of CRBSI. Vancomycin was the most preferred agent for CRBSI, despite guidelines promoting its use only when patients are found to be colonized or previously infected with MRSA. Recent registry on the causative organisms of CRBSI for local ICU showed that MRSA was seen only in 1.4% of the cases.[12] These deviations from guidelines could potentially lead to higher rates of negative outcomes, as guideline compliance had been shown to reduce mortality in blood stream infections.[1314] Despite Gram positive microorganisms usually predominate in CRBSI, Gram negative microorganisms also need to take into consideration especially in sepsis patients as outlined by Infectious Disease Society of America (IDSA).

Comparing to previous survey studies which investigate on use of β-lactam antibiotics in the ICU, our study found that respondents are more likely to prolong the infusion time instead of using conventional intermittent bolus infusion.[47,10] Nonetheless, pattern of CI saw low rates of use even for β-lactams that are known to be stable after reconstitution, as well as its use for the ones that are unstable like imipenem. This reflects poor knowledge on antibiotic stability as captured in previously mentioned study, ANTIBIOPERF.[5] Correct responses for antibiotic stability for specified antibiotics ranged between 3% to 33%, with a third of the respondents stated that they could not answer these questions.

Limitations of study includes low response rate (12%), which limits the generalizability of its results and conclusions. The second limitation is the simplicity of the case vignettes that did not test respondents on the impact of augmented or impaired organ function, as well as complex morbidities on antibiotic dosing. Nonetheless, we were still able to capture the variable dosing and pattern of antibiotic prescription with these cases.

CONCLUSION

Malaysian physicians are receptive to PK/PD approach in antibiotic optimization among ICU patients. Nonetheless, there is still a gap in the knowledge of antibiotic PK/PD as well as its application in the critically ill, especially for β-lactams. We also found variable compliance to guidelines for antibiotic choice, as well as antibiotic dosing, for commonly encountered infections in ICU. This warrants further study and interventions in order to improve guideline compliance among local physician.

Financial support and sponsorship

This study was funded by the Fundamental Research Grant Scheme (FRGS) under ID: FRGS16-048-0547.

Conflicts of interest

There are no conflicts of interest.

Questionnaire that was used for the study