A multicentre point prevalence survey of patterns and quality of antibiotic prescribing in Indonesian hospitals

Abstract Background The global emergence of antimicrobial resistance is driven by antibiotic misuse and overuse. However, systematic data in Indonesian hospitals to adequately inform policy are scarce. Objectives To evaluate patterns and quality indicators of antibiotic prescribing in six general hospitals in Jakarta, Indonesia. Methods We conducted a hospital-wide point prevalence survey (PPS) between March and August 2019, using Global-PPS and WHO-PPS protocols. The analysis focused on antibacterials (antibiotics) for systemic use. Results Of 1602 inpatients, 993 (62.0%) received ≥1 antimicrobial. Of 1666 antimicrobial prescriptions, 1273 (76.4%) were antibiotics. Indications comprised community-acquired infections (42.6%), surgical prophylaxis (22.6%), hospital-acquired infections (18.5%), medical prophylaxis (9.6%), unknown (4.6%) and other (2.1%). The most common reasons for antibiotic prescribing were pneumonia (27.7%), skin and soft tissue infections (8.3%), and gastrointestinal prophylaxis (7.9%). The most prescribed antibiotic classes were third-generation cephalosporins (44.3%), fluoroquinolones (13.5%), carbapenems (7.4%), and penicillins with β-lactamase inhibitor (6.8%). According to the WHO AWaRe classification, Watch antibiotics accounted for 67.4%, followed by 28.0% Access and 2.4% Reserve. Hospital antibiotic guidelines were not available for 28.1% of prescriptions, and, where available, guideline compliance was 52.2%. Reason for the antibiotic prescription, stop/review date and planned duration were poorly documented. Culture-guided prescriptions comprised 8.1% of community-acquired infections and 26.8% of hospital-acquired infections. Conclusions Our data indicate a high rate of empirical use of broad-spectrum antibiotics in Indonesian hospitals, coupled with poor documentation and guideline adherence. The findings suggest important areas for antimicrobial stewardship interventions.


Introduction
Drug-resistant infections have been estimated to account for 700 000 deaths per year globally, cumulating to 10 million by 2050, higher than cancer (8.2 million) and diabetes (1.5 million) combined. 1 The overuse and misuse of antimicrobial agents has been well recognized as one of the key drivers of emerging antimicrobial resistance (AMR), 2,3 with antimicrobial consumption projected to rise further globally. 4 In response to the emerging public health crisis of AMR, the WHO has launched a global action plan, including strategies for surveillance and mitigation of antimicrobial overuse. 5 Indonesia, a populous (271 million) and diverse middle-income country, is potentially an AMR hotspot, due to persistently high infectious disease burdens, including respiratory infections, V C The Author(s) 2021. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. diarrhoeal diseases and TB among others, 6 coupled with liberal antibiotic practices and fragile health systems. 7,8 The Indonesian government is increasingly supporting antimicrobial stewardship (AMS), through the national action plan on AMR launched in 2014, 9,10 and as part of hospital accreditation. 11 National antibiotic guidelines were released in 2011, 12 but have not been updated since. However, inappropriate or unnecessary antibiotic prescribing is believed to be widespread, although systematic data are lacking to adequately inform AMS policies.
In global datasets reporting point prevalence surveys (PPS) of antibiotic use in hospitals, 13,14 low and middle-income countries (LMICs) remain underrepresented. 15,16 The recently introduced WHO AWaRe (Access, Watch, Reserve) antibiotic classification framework, based on accessibility versus AMR potential, is a useful metric to provide an indication of the appropriateness of antibiotic consumption. 17,18 We performed a hospital-wide PPS across six acute-care, general hospitals in Jakarta, the capital city of Indonesia, with the aim of evaluating patterns and quality indicators of antibiotic prescribing. We assessed community and hospital-acquired infections as well as medical and surgical prophylaxis, by hospital, ward type and diagnosis.

Study design and population
We conducted a hospital-wide PPS of antimicrobial use in a purposive sample of six hospitals across Jakarta, between March and August 2019. We followed Global PPS (2018) 19 and WHO (2019) 13 protocols. Briefly, a PPS is a 'snapshot' survey to collate medical record data on antimicrobial prescriptions in hospitalized patients. Eligible patients were all hospitalized patients who received 1 active (i.e. currently ongoing) antimicrobial by 8 a.m. on the survey day or surgical prophylaxis 24 h prior to the survey. We excluded emergency and day-care wards, outpatient clinics and inpatients who were discharged before or admitted after 8 a.m.

Ethical considerations
The study was approved by the Research Ethics Committee of the Faculty of Medicine of the University of Indonesia (1364/UN2.F1/ETIK/2018) and the Oxford Tropical Research Ethics Committee (559-18). The requirement for individual patient consent was waived. Permission was obtained from the hospital management or research/medical committee in each participating hospital.

Data collection
We developed a paper data collection form (DCF) comprising ward, patient and treatment sections, modified from Global-PPS 19 and WHO-PPS (Appendix S1, available as Supplementary data at JAC-AMR Online). 13 Data collection was conducted by one or two medical doctors from the study team, joined by one to four junior hospital doctors, who received 1 day of training and a DCF completion guideline. Each ward was completely surveyed within 1 day (to minimize the effect of patient movements) and all wards of a single hospital were surveyed within 4 weeks. De-identified data were extracted from medical notes, drug chart and/or laboratory records; if crucial data were missing or unclear (e.g. unclear writing, mismatch between diagnosis and antimicrobial treatment, missing culture result), the responsible ward nurse or clinician was asked for clarifications. The completed DCFs were entered into a study database. The data coding was verified at two stages: (i) queries during DCF completion were directly resolved with a senior team member (R.L., R.L.H.); and (ii) database inconsistencies were checked against the source data in the medical records as needed.
We included systemic antimicrobials coded on the basis of the WHO Anatomical Therapeutic Chemical (ATC) classification system as follows: antibacterials (J01), antimycotics (J02), antifungals (D01BA), antimycobacterials (J04), antivirals (J05), nitroimidazole derivatives (P01AB), intestinal anti-infectives (A07A) and antimalarials (P01B). We recorded the diagnosis/ reason for the prescribed antimicrobial (what the clinician aimed at treating or preventing), according to a diagnostic code list 19 (Table S1). Antimicrobial indications were classified as: (i) community-acquired infection (CAI) if symptoms were present on admission or started ,48 h after admission; (ii) hospital-acquired infection (HAI) if symptoms started 48 h after admission; (iii) medical prophylaxis; (iv) surgical prophylaxis, categorized as single-dose, 1 day or longer than 1 day; (v) other; and (vi) unknown. We recorded the following five quality indicators of prescribing: (i) documentation of diagnosis/reason for antimicrobial use, stop/review date and treatment duration in the patient records; (ii) hospital antibiotic guideline availability (i.e. based on a review of all local guidelines by the study team) and compliance with regards to drug choice; if not available, this item was recorded as 'not assessable'; (iii) parenteral administration; (iv) culture sample taken in therapeutic use; and (v) targeted (antibiotic prescribed in response to microbiology results) or empirical treatment.

Statistical analysis
We used descriptive statistics to summarize the data, expressed as counts or percentages, by hospital, ward type, indication and diagnosis. The analysis focused on antibacterials (antibiotics) for systemic use (ATC code J01). Antibiotics were reported by drug names, chemical class (according to the fourth level WHO ATC classification) and AWaRe groups. We used RStudio Version 1.3.1093 for all analyses.

Hospital characteristics
The six participating hospitals varied by care level (four secondary, two tertiary); sector (three private, three public); availability of hospital antibiotic guidelines (five yes, one no); inclusion in the national health insurance scheme (four yes, two no). All hospitals had an antibiotic stewardship team. All 238 inpatient wards surveyed included 87 medical, 31 surgical, 95 mixed medical-surgical wards and 25 ICUs, of which there were 123 adult, 51 paediatric-neonatal and 64 mixed adult-paediatric-neonatal wards (

Discussion
This was the first contemporary hospital-wide survey in Indonesia that systematically evaluated patterns and quality of antibiotic prescribing, using the recommended PPS methodology. 13,19 We demonstrated the feasibility of PPS in this low-resource setting, and generated useful data to guide local AMS interventions. We found proportions of inpatients in Indonesian hospitals receiving antibiotics to be substantially higher (62%) than reported in global PPS datasets (27%-39%), which were dominated by data from high-income countries in Europe, North America and Asia. 15,16 In our survey, antibiotic use varied between hospitals (53%-79% of patients), and was highest in ICUs (86.8%). Consistent with other surveys in Asia 20,21 and globally, 15 lower respiratory tract infections were the predominant reason for antibiotic prescribing in Jakarta hospitals. In our survey, the most-used antibiotic classes were third-generation cephalosporins (mainly ceftriaxone), fluoroquinolones (mainly levofloxacin) and carbapenems (mainly meropenem), all predominantly used for pneumonia, among several other diagnoses. Ceftriaxone was the mostused antibiotic across all major indications (i.e. CAI, HAI, surgical and medical prophylaxis). These findings are consistent with the widespread use of broad-spectrum antibiotics, predominantly third-generation cephalosporins and fluoroquinolones, in Indonesia, 22 other Asian countries 20,21,23-25 and globally, 15,16 which may suggest that at least a proportion of these prescriptions are unnecessary or inappropriate. Moreover, empirical use of meropenem for CAI and HAI represented nearly 10% of all antibiotics for therapeutic use; this was similar to a globally reported rate of 12.2%, 15 but substantially higher than the overall 4.1% reported in European countries. 26 Substantial use of carbapenems in our survey could partially be explained by the fact that two of the six hospitals were tertiary referral centres attending to complex patients, as well as high reported rates of AMR in Indonesian hospitals, particularly in common Gram-negative organisms. 27 Nonetheless, culture-guided prescribing for CAI (8%) and HAI (27%) was low in comparison to a global study (12%-27% and 20%-44%, respectively) 15 , suggesting underutilization of microbiological diagnostics as well as overuse of broad-spectrum antibiotics.
Antibiotic prescriptions for HAI (18.5% of total), predominantly for pneumonia but also intervention-related and post-operative surgical site infections, were comparable to recent surveys in India (19%) 21 and Thailand (34%), 20 but considerably higher than in reports from high-income settings, e.g. ECDC survey (6%) 26 and the GLOBAL-PPS survey (8.4%). 15 These data confirm the significantly higher burden of HAI in LMICs compared with high-income countries.
A high proportion of antibiotic prescriptions were for surgical (23%) and medical prophylaxis (10%), for a range of indications. Prophylactic prescribing was unusually high for gastrointestinal infections. Prolonged (.1 day) surgical prophylaxis was very common (76%) in our survey, as has also been observed in other countries in Asia (Pakistan 97%, 23 India 77%, 21 Thailand 90% 20 ) as well as in Europe. 15,26 Prolonged antibiotic prophylaxis for more than 24 h for most surgical indications does not prevent development of postoperative infections, compared with ,24 h, but increases the risk of AMR and side-effects. 28 Further research is warranted to explain the reasons for these patterns.
We investigated five basic quality indicators, which could be used to set benchmarks for quality improvement of antibiotic use 29 and AMS programmes. 30 Documentation of the reason of prescribing (64%) was lower than reported across studies in Europe, Asia, Africa and America (70%-85%). 15,31 Stop or review date was poorly documented (15%) across indications and ward types. Post-prescription review of a prescribed antimicrobial within 48-72 h of the initial order ensures appropriate choice and route of administration and optimal de-escalation (IV to oral switch) practices and prevents unnecessarily long antibiotic courses. The high (85%) proportion of parenteral route of administration, coupled Limato et al.
with high rates of empirical therapy and suboptimal use of microbiological cultures, suggests lack of de-escalation protocols in the participating hospitals. Proactive IV to oral switching policies are recognized as a key metric for AMS processes, and can reduce catheter-related complications, healthcare costs and duration of hospital stays. 32 A systematic review and meta-analysis showed that guidelineadherent empirical therapy was associated with a relative risk Antibiotic use in Indonesian hospitals JAR reduction for mortality of 35%. 32 The reason for poor guideline compliance (52%) in our survey is uncertain and probably multifactorial, including local resistance patterns, ineffective guideline dissemination and clinical uncertainty with fear of treatment failure. Our findings should trigger further detailed investigations at hospital and country level.
The WHO AWaRe framework offers an attractive metric for LMICs in the absence of validated quality indicators for antibiotic appropriateness, 17,33,34 and includes a . 60% national target of total antibiotic consumption in the Access category by 2023. 35 However, a recent assessment of antibiotic consumption data from 76 countries in 2000-15 found that the global per-capita   Hospital antibiotic guidelines were not available to assess compliance. c Only applicable to therapeutic use.
consumption of Watch antibiotics increased by 90.9%, compared with an increase of 26.2% in Access antibiotics, with disproportionate increases in Watch antibiotic consumption in LMICs (165% compared with 27.9% in high-income countries). 18 Although Indonesia national-level data have not been included in the AWaRe reports to date, 35 our survey found hospital consumption of Access antibiotics at 28% to be below the 60% target, mostly driven by ceftriaxone and levofloxacin use for CAI and HAI. Although these findings could partially be explained by the national health insurance scheme which determines available antibiotics based on the national formulary, 36 they also highlight significant challenges for AMS.
Limitations of this study are inherent to the cross-sectional PPS design, providing a mere snapshot of the antibiotic situation in the hospital surveyed. Moreover, given that we used a convenient sample of six hospitals in metropolitan Jakarta, which are potentially better resourced than many other hospitals in Indonesia, data are not necessarily representative for all hospitals in Indonesia, urging caution in extrapolating the observed patterns. Indeed, antibiotic prescribing can be influenced by many factors, e.g. patient case-mix, prevalence of different types of infections, AMR patterns and institutional factors.
In conclusion, we observed high levels of parenteral, empirical use of broad-spectrum antibiotics in Indonesian hospitals, and inadequate performance on key quality indicators of prescribing. Despite important progress in AMS, supported by national policies, 10,11 the study findings highlighted the need to strengthen AMS to increase use of narrower-spectrum antibiotics through culture-guided, targeted treatment and hospital guideline compliance. Further research is needed to understand the complex drivers of antibiotic prescribing, and to develop context-specific and feasible quality improvement strategies to strengthen existing AMS programmes.