Implementation of an antimicrobial stewardship program in the Australian private hospital system: qualitative study of attitudes to antimicrobial resistance and antimicrobial stewardship

This was a single-centre mixed methods study conducted at a 766-bed private hospital in Melbourne. The hospital provides a broad range of emergency and elective medical and surgical services, excluding obstetrics and paediatrics. Before the implementation of the AMS program in 2013, pre-implementation studies were conducted which involved a staff survey and focus group discussions (Fig. 1) [13, 14]. Between 2013–2015 the hospital implemented an AMS program, in keeping with mandatory accreditation standards [10]. Key elements of the program included establishment of a governing AMS committee, implementation of formulary restriction processes, development of surgical antimicrobial prophylaxis guidelines and initiation of post-prescription feedback and review by the AMS team, consisting of a full-time senior pharmacist and three infectious diseases physicians. AMS physicians in the study hospital have a fractional appointment. The AMS team provides post-prescription review for broad-spectrum antimicrobials (such as ceftriaxone, piperacillin/tazobactam or meropenem) or high-cost, highly restricted antimicrobials (such as ceftazidime-avibactam) or, in the event of a significant positive blood culture result (such as isolation of Staphylococcus aureus in blood cultures). Feedback on the prescription is provided if the antimicrobial therapy deviated from guideline recommendations. AMS implementation at this hospital included direct feedback to prescribers provided by AMS pharmacists and clinicians, and reporting of point prevalence survey results and antibiograms to clinicians and hospital executives.

The program was initially resourced to launch only at the studied hospital, however, it was expanded to two other hospitals in the private hospital group without change to the staff resourcing. Additionally, the size of the study hospital has grown since the program launched leading to further resource spreading. This study was approved by the hospital’s research ethics committee. Invitations to participate in the focus groups and the survey were emailed from executive officers to doctors, nurses and pharmacists who undertake direct or supervised clinical care of patients at the hospital. The hospital has 1090 visiting medical officers, 1200 nurses and 21 pharmacists.

Participants were purposively sampled, having been identified as key stakeholders of the AMS program, following methods employed by Cotta et al. [13]. The focus groups consisted of a range of health professionals as outlined in Table 1, none were AMS team members. Participants who attended signed a consent form.

Three focus group discussions were undertaken between December 2018—January 2019. Two researchers (JB & DA) facilitated the discussions using a semi-structured interview guide (Supplementary Material 1) pertaining to participants’ attitudes and perceptions towards AMR and AMS including AMS policies and guidelines. Discussions were audio recorded, deidentified and transcribed verbatim. Data was managed in NVivo™.

The 26-item survey (Supplementary Material 2) was based on the pre-implementation survey [14] with some questions altered to reflect post-implementation analysis. The initial survey was developed by an expert, multidisciplinary team consisting of infectious diseases physicians, clinical microbiologists, AMS pharmacists and nurse practitioners [14].

The survey was conducted online using Qualtrics™. Email invitations contained a link to the survey which was available for 9 weeks from November 2018. Agreement to the introductory statement served as implied consent. Questions were multiple choice or used a Likert scale. Personal information was collected on profession, clinical speciality and years of experience. The survey covered topics including factors contributing to AMR at the hospital, and participants’ awareness of, and agreement with the AMS program at the hospital. All questions referred specifically to the study hospital. Missing or incomplete survey responses were not retained.

Descriptive analysis was undertaken for the survey, with categorical data presented as proportions for each profession. For Likert scale responses, participants were considered “in agreement” or to have viewed AMR as a “serious problem” if they indicated a response of ‘6’ or ‘7’. Pre- [14] and post-implementation response differences were tested using Pearson’s chi-squared test, with a p-value < 0.05 considered statistically significant.

Two researchers (EW and JB) independently coded and recoded transcripts using NVivo. Analysis of focus group data was continuous with deductive coding applied from the pre-implementation research, and Michie et al.’s behaviour change wheel [13, 15]. The Behaviour Change Wheel (BCW, also referred to as the COM-B model) was the guiding framework for this research. The BCW consists of three interacting constructs to explain behaviour − capability, opportunity and motivation. Capability is an individual’s knowledge and skills (eg. psychological and physical activity) to engage with the program [15]. Motivation is the reflective and automatic brain action that triggers and directs behaviour (eg. choice and intention) [15]. Opportunity is defined as external factors (social and environmental) that influences behaviour [15]. The BCW has been widely adopted in implementation and health services research. Mapping themes to these conditions allows for an understanding of the areas where an intervention is succeeding or failing to change behaviour. All authors reviewed coding and discrepancies were resolved by discussion.

Survey and focus group data were analysed separately with a process of triangulation applied at the interpretation stage of the analysis to determine whether the findings were convergent, complementary or contradictory [16]. This process involved mapping the survey analysis to the themes identified from the BCW and determining if the qualitative themes were consistent with the survey, whether the qualitative themes expanded or provided nuance to the survey results, or whether the survey and focus group results differed.

Participants had varied antimicrobial knowledge and experience (Table 2), ranging from extensive to little involvement with antimicrobial prescribing and AMS. Participants discussed changes in prescribing habits and number of presentations of AMR organisms over time. The survey results reflected this with 73% of respondents noticing increased AMR presentations in the previous 10 years (Table 3). Similarly, the proportion of survey respondents previously involved in the care of patients with AMR significantly increased (84% to 98%, p < 0.0001) from pre- to post-implementation (Table 3). There was a significant increase in awareness of AMS programs from 2013 when only 41% of respondents had heard of AMS (p < 0.0001) (Table 3). The clinical microbiologists, infectious diseases physician and pharmacist had the most accurate understanding of the hospital’s program. The clinicians interviewed who were not directly involved in AMS were comfortable knowing when to refer to those with more expertise.

Most participants understood the rationale and general approaches for AMS programs but had limited knowledge of the hospital’s program (Table 2). Almost 90% of survey respondents stated they had heard of AMS, but only 67% were sure the hospital had an AMS program (Table 3). The clinical microbiologists and the Nurse Unit Managers (NUMs) felt professional education for nurses and junior doctors about AMS and antimicrobial prescribing was lacking.

The survey showed that 60% of respondents agreed that the AMS team at the hospital should continue. However, appreciation and perceived importance of the AMS program differed between participants (Table 4). Some viewed the program as “a very critical part” (CM1, FG1) of infection control in the hospital. However, some were doubtful of the effectiveness of the program, indicating that stewardship does not require “major formalisation” (Endocrinologist, FG2). Participants across the three focus groups agreed on the importance of monitoring and governance of antimicrobial prescribing data to encourage appropriate prescribing and review of technique.

The cost of patient care, antimicrobials and hospital services was highlighted as a “driving force” (NUM 3, FG 3) for the private hospital sector, with participants noting the focus on “getting people out of bed so you can put someone in it” (Pharmacist, FG 3). The high price of some antimicrobials was seen a factor influencing treatment choices, with participants saying the cost can be “very scary” (Executive 1, FG 3) due to concern that private health insurance funds will not cover it. While others noted that the high cost may “improve practice” by encouraging less use of “the scary drugs” (Pharmacist, FG 3).

Staff who had less involvement with antimicrobial prescribing showed little desire to engage with the program, indicating they were unlikely to access guidelines and resources. Surveyed staff believed antimicrobial resistance was significantly less of an issue at the surveyed hospital (40%) compared with hospitals Australia-wide (58%) (p < 0.001). The implementation of the AMS program did not have a significant impact on attitudes of respondents, with 62% of respondents from the 2018 survey agreeing that improving antimicrobial prescribing at the hospital would decrease AMR, compared to 58% of respondents in 2013 (p = 0.450) (Table 5).

Some clinicians in the focus groups expressed concern they may be prescribing incorrectly but had not been approached about it and felt they should be. Staff were not aware of the progress of the AMS program, and felt that receiving feedback could highlight its impact.

All three focus groups expressed an opinion that all staff have a responsibility to ensure appropriate antimicrobial prescribing (Table 6). Knowledge and input from specialist staff was valued, with staff frequently referring to the expertise of infectious diseases physicians. The AMS pharmacist was considered a key player in the program, providing valuable input and guidance. The high level of knowledge that nursing staff have of their patients and the wards was also considered important for ensuring appropriate prescribing, and some participants noted that increasing their role in AMS could be beneficial for the success of the program.

Participants suggested that due to the hospital being private, clinicians have a high level of autonomy in their prescribing choices and frequently did not follow guidelines, instead maintaining their historical prescribing practices. Only 17% of survey respondents frequently consulted the antimicrobial guidelines. Some felt that staff at lower or equal hierarchical levels to prescribing doctors could not question others’ prescribing habits, even if it was against guidelines “because they know they’ll start a fight” (Pharmacist, FG 3). Similarly, one executive (FG 3) stated that there are no processes to empower junior staff to speak up and be protected if they question practice. Participants recognised the importance of the hospital executives’ championship of AMS programs to help drive further system changes, stating that “the only way to fix [antimicrobial prescribing] is from the top” (Pharmacist, FG 3) (Table 6). The survey showed that significantly more respondents were willing to participate in initiatives involving antimicrobial use in 2018 (68%) compared with 2013 (50%) (p = 0.003) (Table 7).

The expertise of specialist staff was seen as a benefit for promoting AMS in our study. Appreciation of AMS pharmacists increased post-implementation, and participants had a greater understanding of the importance of their clinical knowledge and guidance than pre-implementation [13]. The benefits of AMS pharmacist input is supported by previous studies [17‐20]. Similarly, both our study and the pre-implementation study highlighted the important role that nurses have in AMS due to their contact with patients and clinicians [13]. Nurses are vital in AMS programs through their role in close monitoring of patients, advocating for best patient care and liaising between staff to ensure appropriate antimicrobial prescribing [18, 20‐23]. Gillespie et al. [24] demonstrated that increasing AMS education for nurses improves knowledge and may influence antimicrobial management practices. Continued engagement with pharmacists and nurses is necessary for AMS success, and it is vital that their important roles are not overlooked or undermined by hospital hierarchies.

While awareness of the AMS program increased, participants expressed doubts about its effectiveness in reducing AMR in the hospital, similar to the pre-implementation survey [14]. The National Antimicrobial Prescribing Survey results for the hospital reflected these results, showing almost no change in rates of appropriate prescribing and compliance with national prescribing guidelines for the five years between 2012–2017 for our study hospital [8].

Participants in our study showed varied understandings of the mechanisms and benefits of the program. Furthermore, staff in both the pre- and post-implementation studies believed that AMS is more of an issue in all Australian hospitals compared with the surveyed hospital [14]. Previous studies have indicated that staff may have difficulty in committing to change actions and behaviours if there is little knowledge of the benefits and outcomes the changes can produce [4, 25, 26]. Other studies have highlighted that this may indicate externalisation of the issue and be a barrier to appropriate antimicrobial management [20, 27‐30]. Our results support staff concern about a lack of AMS education, with a shortfall in staff understanding of the importance of AMS in reducing AMR within the institution. Hence staff education may be needed to increase awareness of the AMS program and highlight the benefits.

Feedback and monitoring of behaviour has been highlighted as a factor in promoting behavioural change [25, 31, 32]. Specifically, feedback and auditing of clinical performance has been shown to improve effectiveness of AMS programs [3, 33]. Participants reinforced this, with several calling for more explicit governance of prescribing practices to improve AMS compliance. Based on the participant discussions, AMS auditing and reporting processes may not have been readily available or accessed by those end-users who would benefit from it. Research has demonstrated that feedback which provides actionable targets, is tailored to the prescribing group, and delivered monthly by a senior colleague is most effective at promoting behaviour change in AMS settings [33‐35]. Understanding clinicians’ desired feedback models could improve uptake of AMS programs [35‐37]. Some participants stated that they were not being approached regarding their prescribing practices, but felt they should be. At present, feedback on the prescription is only provided in the instance of significant deviation from antimicrobial guideline recommendations. It is therefore possible that these hospital staff were prescribing in keeping with the guidelines and as such, were not contacted by the AMS team. However, studies have shown that providing positive reinforcement for correct behaviour can improve outcomes including appropriate antimicrobial prescribing [38, 39].

In private hospitals in Australia, medical practitioners are granted clinical responsibilities in the hospital by a Medical Advisory Committee, rather than being employees [11, 40]. The structure of private hospitals encourages clinical autonomy, limiting the ability of hospital managers to directly influence behaviour change because the clinician is not their employee [11]. Additionally, clinicians who choose to work in private hospitals are more inclined to do so because of the increased autonomy and agency over how they treat their patients [41].

Staff in the private sector have previously highlighted difficulties in enforcing AMS guidelines on medical practitioners [20, 42]. Research has shown that familiarity between the AMS team and prescribing clinicians is an important aspect of AMS and providing feedback to clinicians [19, 20]. Creating an environment where the AMS team are well known to all medical practitioners may improve delivery of feedback and uptake of actionable targets. Staff hierarchy is also a common issue in preventing inappropriate antimicrobial prescribing in both private and public hospitals [18, 20, 43]. Clinician autonomy was highlighted as a key theme in our study, with clinicians indicating infrequent referrals to guidelines and continued use of potentially unsupported historical prescribing practices [18, 42]. Studies have found it is common for doctors to be unwilling to change their prescribing practices and be resistant to guidelines, as they can prevent clinical freedom [44‐47].

Executive engagement may be needed to drive further system changes across a complex network. Participants in our study indicated that this view remains. The championship of hospital executive to support and promote AMS is a major determinant of success for the implementation of AMS programs [19, 37, 45, 47].

Cost was seen as an influential factor on antimicrobial prescribing in the private hospital setting. However, similar to previous studies there are mixed views on whether costs have a positive or negative impact on prescribing practices [42]. It is possible that it can lead to over-prescribing of prophylactic antimicrobials to curb any risk of readmission, or it could mean more thought is taken to prescribe the correct antimicrobials and reduce cost [42]. Despite clinicians favouring private practice due to increased autonomy, there are external factors which impact antimicrobial decision making [41, 42].

This study had several limitations. The survey had fewer participants than the pre-implementation study survey [14]. Additionally, the focus groups did not have the same participants as the pre-implementation study and therefore cannot offer a direct comparison of opinions [13]. However, both the survey and focus groups in this study included participants from a range of different healthcare professions and therefore provided an appropriate variety of responses.