A hospital-wide cross-sectional point prevalence survey (PPS) of antimicrobial prescribing was performed in two teaching hospitals in Klang Valley, Malaysia (Jamaluddin et al., 2021). Hospital Canselor Tuanku Muhriz or HCTM (1,054 beds, 63 wards) and University Malaya Medical Centre or UMMC (1,617 beds, 44 wards) are university-affiliated hospitals with multidisciplinary AMS teams. PPS was conducted for each facility on designated days between 16 April to 30 April 2019. Auditors were assigned a specific day to complete a standard Hospital NAPS protocol, completing a data collection form for each patient prescribed with an antimicrobial on the designated audit day (Supplementary Material). A detailed description of the Hospital NAPS antimicrobial prescribing surveys is described in previous publications (James et al., 2014; James et al., 2022). Survey and assessment were executed by fourteen pharmacists and two infectious disease (ID) physicians in HCTM, while one pharmacist and four ID physicians undertook the exercises in UMMC. The Australian NAPS support team provided training, technical and clinical support throughout the survey period. All surveyors received online webinar training on the audit protocol before the survey day. Data collected during the survey were compiled and submitted through a secure web-based online platform. Data on antimicrobial prescribing among patients admitted to surgical-practice units were analyzed for this report. The study was approved and ethics approval from each institution was obtained before the commencement of this study.

All adult patients admitted to the obstetrics and gynecology (OBGYN), trauma and orthopedic and surgical specialties, before or at 8 a.m. on the day of the survey were audited once (denominator). Patients admitted after 8 a.m., outpatients, as well as patients undergoing same-day treatment and surgery in daycare or at emergency unit, were excluded. The following information were retrieved from medical records and associated documents for patients who were prescribed with at least an antimicrobial (numerator) regardless of route of administration: demographics, diagnosis, antimicrobial data (including indications, dose, route, frequency, duration, start and review/stop date) and any additional clinical variables (cultures, biomarkers) relevant for the assessment. The survey also included patients who were prescribed a stat dose of antimicrobial or SAP since 8 a.m. the previous day. A unique, non-identifiable survey number was assigned to every de-identified patient data. Aligning with established protocols by Hospital NAPS (James et al., 2014; James et al., 2022) and WHO (World Health Organization, 2018), setting 8 a.m. as the cut-off time for patient inclusion ensures comprehensive representation of all admitted patients while minimizing variability of different time points and the capture of diverse sample encompassing individuals who have undergone consultations and received treatment. Additionally, corresponding with the facilities’ operational day, this timing facilitates efficient data collection by the survey team. This method strikes a balance between practical considerations and the imperative to obtain a representative patient group, plus ensuring consistency and comparability with existing literature. The calculated minimum sample size, determined by the Krejcie and Morgan formula, was 256 subjects. This estimation was based on a preliminary survey conducted in the hospital, which reported a prevalence of antimicrobial use at 78.9%, considering type 1 error rate of 5% and a precision of 5% (Krejcie and Morgan, 1970).

Antibiotics were classified as “Access,” “Watch” and “Reserve” (AWaRe) according to the 2021 WHO AWaRe classification (World Health Organization, 2021). Antimicrobials not included in the AWaRe classification were listed as “unclassified.” Details on AWaRe classification for the type of treatment are shown in Supplementary Material. Continuous data were presented as the mean and standard deviation (SD) for normally distributed data. If the distribution was not normal, continuous data were presented as the median and interquartile range (IQR). Other descriptive statistics, such as minimum and maximum values were reported when necessary. Normality of the data was examined using histogram (approximately bell-shaped), skewness (within −1 to 1) and kurtosis (within −3 to 3). The difference between hospitals was analyzed using the Chi-square test or Fisher’s exact test (if minimum expected count was less than 5) for categorical variables. For continuous age variables, independent t-test was used to analyze the mean difference between hospitals. Compliance with guidelines and appropriateness were treated as dichotomous variables. The associations of each potential factor with compliance and appropriateness were examined through the Chi-square test or Fisher’s exact test. Multiple logistic regressions were used to evaluate significant factors. Odds ratio and 95% confidence interval for each potential factor were calculated, where a p-value of less than 0.05 was considered significant. All analyses were carried out using SPSS (IBM Corp. released 2011 IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp).

A total of 229 admissions in HCTM from twenty wards plus one burn unit, and 337 in UMMC from thirteen wards were identified. Admissions to the surgical and burn units accounted for 51.1% (289) of patients, followed by OBGYN with 24.4% (138), trauma and orthopedic with 22.6% (128) and mix ward with 1.9% (Browne et al., 2021). Among 566 patients, 250 (44.2%) received at least one antimicrobial prescription at the time of the survey, for a total of 339 prescriptions (median 1 per patient, range 1–5), with 171 (68.4%) receiving one antimicrobial agent, 71 (28.4%) receiving two and 8 (3.2%) receiving three or more. Demographic data is presented in Table 1.

Common types of treatment and route of antimicrobials administration were seen to be prescribed in both hospitals (p > 0.05), but UMMC demonstrated better rates (>95%) for documentation practice (p < 0.001) (Table 2). Of all agents prescribed empirically, 51.2% (86/168) were in the Watch group [piperacillin/tazobactam (31.4%) and cefuroxime (29%)], while Access antibiotics accounted for 42% (74/168) of prescriptions. More than half (59.6%; 62/104) of all antimicrobials prescribed prophylactically were Watch antibiotics constituted mainly by cefuroxime (54.8%). Directed therapy was largely entailing antibiotics of Watch by 71.6% (48/67), where meropenem (19%, 13), cefepime (13%, 9) and vancomycin (13%, 9) were prescribed. Access antibiotics were higher in HCTM (49.6%, 59), while the use of Watch antibiotics was found to be higher in UMMC (64.1%, 141) (p = 0.005).

Antimicrobial were mostly prescribed for surgical prophylaxis (27.1%, 92), followed by cystitis (4.7%, 16), necrotizing fasciitis (4.4%, 15) and acute cholecystitis (4.1%, 14). There was a significant difference in the choice of antimicrobial between hospitals (p < 0.001). Cefuroxime (25.5%, 56) and metronidazole (12.3%, 27) were the most commonly used antimicrobials at UMMC, while HCTM recorded the most frequent use of amoxicillin/clavulanic acid (23.5%, 28). From 92 antimicrobial prescriptions for surgical prophylaxis, cephalosporins (53.3%, 49) accounted for predominant choices. The five most used SAP in both hospitals were cefuroxime (37%, 34), metronidazole (18.5%, 17), amoxicillin/clavulanic acid (12.0%, 11), ceftriaxone (7.6%, 7) and vancomycin (5.4%, 5). UMMC mainly utilized cefuroxime (48.5%, 33/68), metronidazole (23.5%, 16/68) and vancomycin (7.4%, 5/68), while HCTM’s preferred choice was amoxicillin/clavulanic acid (41.7%, 10/24). A remarkable use of ceftriaxone (29.2%, 7/24) for SAP in HCTM was observed.

The study revealed a compliance rate with guidelines was at 32.8% and an appropriateness level at 55.2%. Both indicators displayed no statistically significant difference between the two hospitals (Table 2). Of 146 (44.8%) prescriptions that were assessed as inappropriate, 72 (22.1%) were classified as suboptimal while the remaining 74 (22.7%) were classified as inadequate. The percentage of prescriptions judged suboptimal and inadequate did not differ between hospitals with p = 0.219 and p = 0.056, respectively. Inappropriate prescribing varied by subspecialties, overall ranging from 40.9% to 58.3%. A group of units inclusive of plastic surgery, oral and maxillofacial surgery and ENT (others) had the highest percentage of inappropriate orders at 58.3% (7/12), along with cardiothoracic at 57.1% (8/14), ophthalmology at 52.9% (9/17) and OBGYN at 46.9% (23/49). HCTM recorded inappropriateness ranging from 30.8% to 100%, with high rates in cardiothoracic, neurosurgery and others. Meanwhile, the tabulation in UMMC revealed ophthalmology, urology and others as among the units with a high percentage of inappropriate orders ranging from 35.3% to 100%.

Prophylaxis (medical and surgical) prescriptions had the highest inappropriateness (n = 69/146, 47.3%) compared to empirical and directional therapy. The greatest percentage of inappropriate prescriptions was SAP with 40 (43.5%) of 92 prescriptions classified as inadequate and 27 (29.3%) as suboptimal. Both hospitals recorded a high number of inappropriate SAP orders presenting 83% (20/24) in HCTM and 69% (47/68) in UMMC. Unnecessary prolongation ≥24 h was the most common reason for inappropriate prescribing of SAP prescriptions, respectively; 50% (12/24) in HCTM and 38.2% (26/68) in UMMC.

A sub-analysis of 146 inappropriate prescriptions is shown in Figure 1. Total rates of SAP ≥24 h (41.3%, 38/92) contributed mainly to the incorrect duration of antimicrobials in overall prescriptions (15%, 49/326). The extensive use of broad-spectrum antimicrobials in the overall prescribing was depicted at 15.6% (51/326). A higher rate of a broader spectrum of antimicrobials was noted in UMMC (17.1%, 36/211), while incorrect dosage/frequency (13.9%, 16/115) was more commonly seen in HCTM.

The results of univariate and multivariate models for both hospitals are presented in Tables 3, 4. Non-compliance and inappropriate antimicrobial prescriptions were more frequently associated with prophylaxis indications compared to empirical and directed therapy. The likelihood of antimicrobial prophylaxis prescriptions being non-compliant was 4.5 times higher (OR 4.55, 95% CI 1.40–14.78, p = 0.012), and 4.2 times more likely to be found as deemed inappropriate (OR 4.22, 95% CI 1.61–11.10, p = 0.003) in HCTM. Conversely, UMMC showed 2.4 times (OR 2.37, 95% CI 1.21–4.65, p = 0.012) higher likelihood of inappropriateness in prescribing antimicrobial prophylaxis. General surgery (OR 12.56, 95% CI 1.82–86.48, p = 0.010), OBGYN (OR 29.89, 95% CI 3.78–236.49, p = 0.001) as well as trauma and orthopedic (OR 8.06, 95% CI 1.25–52.11, p = 0.028) had significantly higher odds of non-compliance with guidelines compared to cardiothoracic unit. Additionally, prescribing cephalosporins was significantly associated with higher likelihood of non-compliance with guidelines (OR 8.57, 95% CI 2.89–25.39, p < 0.001) compared to penicillins.

The overall usage of antimicrobials in our surgical-practice units at 44.2% was relatively lower compared to rates reported in African hospitals (Bediako-Bowan et al., 2019; Nnadozie et al., 2020), Asia (Limato et al., 2021; Sheng et al., 2019; de Guzman Betito et al., 2021), Italy (Macera et al., 2021), and Serbian hospitals (Šuljagić et al., 2021) (ranged 55.7%–97.6%). Conversely, other surveys, such as PPS in German (Aghdassi et al., 2018) and Belgian hospitals (Vandael et al., 2020), observed a lower prevalence of antimicrobial use, at approximately 30%. The variability in antimicrobial prescribing prevalence, both between our two hospitals, and in comparison to previous reports could be related to differences in the surgical-based case-mix, or structural characteristics unique to each hospital, including the type and proportion of surgical-based specialties. Moreover, significant differences were observed in the patterns of antimicrobial prescribing between the two hospitals, indicating the nature of using local guidelines, which provide various recommendations in accordance with each hospital policy, as well as considerations related to institutional antibiograms and costs, including administrative expenses. Our data also showed a higher usage of antibiotics classified as Watch antibiotics, particularly in UMMC. In response to the global concern of AMR, the AWaRe classification was developed as a general guide to antibiotic prescribing patterns aimed at promoting rational prescribing (World Health Organization, 2021). The WHO recommends at least 60% of all antibiotics prescribed nationwide to be from the Access group. Access antibiotics exhibit a wider range of activity against commonly susceptible pathogens, while sustaining lower resistance potential compared to antibiotics in the other groups. Watch group contain generally broader spectrum antibiotics, pose a higher risk of selecting antimicrobial resistance and are primarily used in patients with more severe conditions. Their use should be vigilantly monitored to prevent overuse. Integrating AWaRe index into our hospital policies shall be an essential measure, as it has been associated with improved usage of Access antibiotics (Budd et al., 2019), highlighting its potential benefits in promoting responsible antimicrobial use and combating AMR.

In this study, we identified appropriateness as the key measure of antimicrobial prescribing quality, moving beyond mere guideline compliance. This approach allowed us to consider various contexts in which non-compliance with the guidelines may not necessarily be deemed as inappropriate prescribing, but rather a case-specific approach that may still be adequately appropriate (Ierano et al., 2019a). However, it is important to note that due to variations in definitions of appropriateness and compliance across the literature, comparisons can be challenging and should be interpreted with caution. Ideally, the target for appropriate antimicrobial prescribing rates in surgical-based units should be above 90%, aligning with general goals for hospital-wide antimicrobial prescribing and SAP prescribing (Vandael et al., 2020; National Centre for Antimicrobial Stewardship and Australian Commission on Safety and Quality in Health Care, 2021). Alarmingly, our study revealed that both the rates of compliance with guidelines and appropriateness fell below this recommended threshold in the surveyed population. Only a small number of PPS studies reported the findings on surgical-practice units specifically, demonstrating compliance with guidelines that ranged from 70% to 92.7% (Elhajji et al., 2018; Singh et al., 2019; Vandael et al., 2020; Macera et al., 2021).

One of the main reasons for inappropriate prescribing in this study was the incorrect duration of antimicrobial prescriptions (15%), predominantly reflecting the extended use of SAP following surgery (41.3%). Best practice guidelines typically recommend a total SAP duration of less than 24 h for most procedures (Bratzler et al., 2013) and NAPS setting targets for this quality indicator at less than 5% (National Centre for Antimicrobial Stewardship and Australian Commission on Safety and Quality in Health Care, 2018). Unfortunately, there has been a persistent pattern of non-compliance and inappropriate prescribing for SAP documented in the literature (Kaya et al., 2016; Mousavi et al., 2017; Alemkere, 2018; Satti et al., 2019; Vicentini et al., 2019; Alahmadi et al., 2020; Bunduki et al., 2020; Khan et al., 2020; Prévost et al., 2020; Bardia et al., 2021; Cabral et al., 2023) ranging from a complete non-compliant to the national guideline (Alemkere, 2018) to 64% compliant (Prévost et al., 2020), while other reports varied from 40.9% inappropriate (Kaya et al., 2016) to 9.5% appropriate SAP use (Khan et al., 2020). Notably, Australian hospitals have monitored the key indicators of antimicrobial appropriateness using NAPS since 2013. While improvements have been observed in certain indicators (i.e., documentation), the proportion of SAP prescriptions exceeding 24 h has remained consistently high, at approximately 30% and has been static since 2015 (Australian Commission on Safety and Quality in Health, 2021). This persistence underscores the challenges of addressing this widespread issue. A comprehensive systematic review across various surgical subspecialties has also highlighted that extending prophylaxis duration does not confer additional reduction on the risk of surgical site infection when best practice (appropriate timing, dosage and re-dosing) is applied (de Jonge et al., 2020). In addition, prolonged SAP duration has been linked to increased risk of adverse events, including acute kidney injury and Clostridiodes difficile infection, contributing to the risk of acquired AMR (Harbarth et al., 2000; Bell et al., 2014; Bernatz et al., 2017; Branch-Elliman et al., 2019).

The proportion of SAP prescriptions in the study (27.1%) was higher compared to Australian Hospital NAPS reports for 2018 and 2019, ranging from 13.9% to 12.6% (National Centre for Antimicrobial Stewardship and Australian Commission on Safety and Quality in Health Care, 2021). Surveys conducted in Europe, Canada, Belgium and Thailand have reported a common preference of cefazolin for SAP (Versporten et al., 2018; Vandael et al., 2020; German et al., 2021; Anugulruengkitt et al., 2022). In contrast, our study observed a high usage of cefuroxime, which depicted similar therapeutic efficacy to cefazolin in preventing surgical site infections (Ahmed et al., 2022). This unconventional choice was influenced by local guidelines recommending cefuroxime, with or without metronidazole, and amoxicillin/clavulanic acid for most procedures due to the unavailability of cefazolin in our centers during the audit period, resulting in non-standard cefazolin use among prescribers. Similar antimicrobials were commonly employed in several lower-middle-income countries (LMIC) (Labi et al., 2018; Saleem et al., 2019; Umeokonkwo et al., 2019; de Guzman Betito et al., 2021), in accordance with their standard treatment guidelines (Bediako-Bowan et al., 2019). In line with global standards and recommendations, our recent guidelines have designated cefazolin as the first-line agent for the majority of procedures (Ministry of Health Malaysia, 2019; University Malaya Medical Center, 2020). While narrow-spectrum antimicrobial is recommended for SAP, inappropriate broad-spectrum antimicrobials were observed, with a dominance of third-generation cephalosporins (ceftriaxone and cefoperazone) and unnecessary anaerobe coverage with metronidazole. Ceftriaxone, a WHO Watch group antibiotic, is not recommended for SAP in our settings due to its lack of significant advantages over the first and second-generation cephalosporins, and its potential for resistance selection. Its usage is limited to cases of contamination or treatment for infection (Bratzler et al., 2013; Ministry of Health Malaysia, 2019). The preference for ceftriaxone in SAP can be attributed to its easy availability and long half-life, which eliminates the need for additional intra-operative doses. An extensive use of ceftriaxone as SAP in this study and various studies globally (Alemkere, 2018; Satti et al., 2019; Rachina et al., 2020; Limato et al., 2021; Fentie et al., 2022) ranging from 26.4% to 84%, poses another significant challenge for AMS.

Both hospitals also displayed a tendency to choose broader-spectrum coverage antimicrobials across all types of treatment (empiric, prophylaxis and directed therapy). In general, the antimicrobial sensitivity testing (AST) results serve as a valuable tool in determining the optimum antimicrobial therapeutic option, highlighting narrow-spectrum agents whenever possible and keeping in check broad-spectrum antimicrobials that exert higher selective pressure for AMR (Gajic et al., 2022). However, the accurate and timely AST performance is challenged by several factors in our hospitals. Proper interpretation of AST results with regard to efficacy and sensitivity among susceptible categories should be counselled by experts to provide individualized or personalized targeted treatment, as selecting antimicrobials based upon a direct comparison of susceptibility values obtained through in vitro testing could be misleading and inaccurate (Gajic et al., 2022). The absence or delay of laboratory data and AST in empiric therapy decisions often leads to the use of broad-spectrum antimicrobials, and at times, polypharmacy, inadvertently encouraging drug resistance (Chokshi et al., 2019).

In UMMC, a noteworthy pattern of non-compliance was identified, with significantly higher occurrence observed in general surgery, OBGYN and trauma and orthopedic units. Evidence of guideline compliance has yielded diverse outcomes in various prospective observational studies. NAPS reports on antimicrobial use in Canada indicated a commendable rate of appropriate prescription, notably in gynecology unit at 86.2% (CARSS, 2022). Conversely in Nigeria, an audit in OBGYN wards painted a different picture, highlighting excessive and inappropriate antimicrobial usage, with similar output including high incidence of redundant anaerobic coverage with metronidazole (Abubakar et al., 2018). Meanwhile, Thomas et al. (2022) found higher compliance in both gynecology (88.6%) and orthopedic (86.3%) compared to surgery (67.9%). Our study also highlighted that although documentation practices were significantly higher in UMMC, which utilizes electronic medical records (EMR) and electronic prescribing (e-prescribing), compared to HCTM, where paper-based health records are used, this criterion did not significantly influence the rates of compliance and appropriateness. However, King et al. (2017) and Hand et al. (2017) have outlined the potential of digital platforms and electronic health information technology in aiding prescribers throughout the antimicrobial lifecycle encompassing initiation, review, stopping and supplying of discharge medications. The technology is anticipated to have a positive impact on documentation and compliance in the surgical unit (Charani et al., 2017).

Despite these observations, the precise causes of the high non-compliant of prescribers in this study were uncertain and unexplored; thus, the explanation for this finding warrants further investigation. Insights drawn from an ethnographic study shed light on surgeons’ priorities, which primarily revolve around surgical procedures, surgical care and patient outcomes. Surgeons often place a strong emphasis on starting antimicrobials than on reviewing or stopping them, while rarely discussing the choice of antimicrobial (Charani et al., 2018). This potentially leading to prolonged and unnecessary use of these drugs. A review by Hassan et al. identified a common barrier to compliance with guidelines stemming from prescribers’ inadequate knowledge and unfamiliarity of guideline content (Hassan et al., 2021). However, Ierano et al. (2019b) highlighted that prescriber preferences and autonomy are often considered more important than strict compliance with guidelines, even when prescribers are well-informed about the guidelines. Moreover, guidelines are often viewed as general recommendations that lack the necessary details to address the diverse array of surgical procedures and various patient characteristics and environmental factors that complicate decision-making in complex situations. A recent survey conducted among Surgical Infection Society (SIS) members, experts in surgical infections, revealed that surgeons hold varying opinions regarding the appropriate duration of prophylaxis and therapeutic antimicrobials for inpatients across common indications (Delaplain et al., 2022). It is evident that heightened prescribers’ awareness regarding their prescribing practices is a crucial component of AMS efforts.