Discussion

This is the first study assessing HCW perceptions in academic compared with nonacademic public sector hospitals in South Africa and supplements the evidence on AMR and ASP insight locally and globally, including in the One Health context. ^{Reference Balliram, Sibanda and Essack4,Reference Wasserman, Potgieter and Shoul8–Reference Setiawan, Cotta and Abdul-Aziz15,Reference Jha, Mudvari, Hayat and Shankar18–23} The study surveyed a range of ages and work experience in three provinces. Several notable differences were found between academic and nonacademic environments necessitating contextualized approaches.

AMR is widely perceived by HCW as a serious problem globally and nationally, ^{Reference Balliram, Sibanda and Essack4,Reference Gebretekle, Mariam and Abebe7–Reference Baraka, Alsultan, Alsalman, Alaithan, Islam and Alasseri9} with fewer nurses in agreement. ^{Reference Balliram, Sibanda and Essack4,Reference Jha, Mudvari, Hayat and Shankar18} This study also showed that a lower proportion of HCW considered AMR a problem at their facility, pronounced in nonacademic settings, in keeping with other findings. ^{Reference Balliram, Sibanda and Essack4,Reference Gebretekle, Mariam and Abebe7–Reference Baraka, Alsultan, Alsalman, Alaithan, Islam and Alasseri9,Reference Babatola, Fadare and Olatunya14,Reference Setiawan, Cotta and Abdul-Aziz15,Reference Thriemer, Katuala and Batoko20–23} Adequate education on antimicrobials, reported by 36.4% of respondents in this survey, remains crucial. More than one-quarter of doctors and nurses felt that inappropriate antimicrobial use was not linked to AMR and 42.4% overall (mostly nurses and pharmacists) did not agree that poor IPC practices contributed to AMR, a finding that has been noted amongst South African medical students and physicians and pharmacists in Ethiopia. ^{Reference Gebretekle, Mariam and Abebe7,Reference Wasserman, Potgieter and Shoul8} Although national IPC recommendations are in place, successful implementation of these seemingly low-hanging fruit remains a challenge in overburdened and poorly resourced healthcare facilities. ^{23,Reference Magadze, Nkhwashu, Moloko and Chetty24} This cognitive dissonance is striking given that approximately one-third of respondents at academic facilities, mostly nurses, felt it very likely that a patient would develop an infection with a drug-resistant organism during their hospital stay similar to a study in Indonesia and lower than the 66.0% in Ethiopia. ^{Reference Gebretekle, Mariam and Abebe7,Reference Limato, Nelwan and Mudia13} Only 53.0% of nurses overall would change their practice if a patient were known to be infected with a drug-resistant organism.

The tendency to underestimate the impact of individual actions in achieving a greater goal is a logical fallacy. This study showed that a significantly higher proportion of doctors felt that their individual contributions could impact AMR, emphasizing the need to empower nurses and pharmacists to counteract externalization of responsibility. ^{Reference Ngongang, Basera and Mendelson12} Education and active involvement of the hospital IPC team were among the most frequently selected recommendations thought to have the greatest impact on tackling AMR, suggesting a lack of accountability or suboptimal knowledge on the importance of IPC. A comprehensive approach to understanding behavioral aspects and cognitive biases in AMR and AMS is needed. ^{Reference Langford, Daneman, Leung and Langford25}

Diagnostic stewardship, including implementation of rapid diagnostic tests, can provide supplementary or quicker decision support for antimicrobial prescription and can guide microbiology result interpretation and inform appropriate collection of samples. ^{Reference Septimus26} This is highly relevant in both settings; in the academic setting, sample collection changed therapy in just 50.2% of patients, contributing to unnecessary resource consumption. In the nonacademic setting, the lack of availability of rapid diagnostic tests contributed to antimicrobial overuse in 64.1%, similar to the 64.4% of respondents at a tertiary center in Ethiopia but higher than the 52.8% reported in academic settings. ^{Reference Gebretekle, Mariam and Abebe7} Microbiology laboratories are traditionally located in academic settings outlining the need to optimize communication between the laboratory and clinicians, as highlighted previously. ^{Reference Gebretekle, Mariam and Abebe7,Reference Limato, Nelwan and Mudia13}

A greater proportion of staff in academic facilities, especially nurses, felt that ASPs override prescriber autonomy and can negatively affect patient care. These findings may be explained by the presence of more experienced personnel in South African academic settings and underscores the need for synergism between different professionals and education on the goals of AMS. ^{Reference Setiawan, Cotta and Abdul-Aziz15,Reference Sutthiruk, Considine, Hutchinson, Driscoll, Malathum and Botti21} Previous reports from Thailand and Indonesia found that doctors, rather than nurses or pharmacists, objected to interventions limiting prescribing decisions, while only 21.3% of physicians in Ethiopia reported that ASPs impacted decision autonomy. ^{Reference Gebretekle, Mariam and Abebe7,Reference Setiawan, Cotta and Abdul-Aziz15,Reference Sutthiruk, Considine, Hutchinson, Driscoll, Malathum and Botti21}

Exposure to effective ASPs was only reported by 50.2% of respondents in this study. Almost a third of nurses in academic settings did not have time to further invest in ASPs and felt that only clinicians needed to understand AMS; this may be reflective of different professional demands on their time or an alternative perspective on the role of nurses in successful ASPs. ^{Reference Setiawan, Cotta and Abdul-Aziz15} Despite this, greater involvement of nursing staff in ASPs was selected among the top five interventions having an impact on AMR with adequate training reported by 31.0% of nursing staff overall. Significantly fewer nurses felt that ASPs reduce antimicrobial use and can save costs, improve quality of care, reduce AMR and duration of hospital stay, impact nosocomial infection rates, and are currently needed. The important role of nurses in AMS has been recognized, with education, optimization of interprofessional dynamics, and support of nursing leadership being strategic enablers. ^{Reference Setiawan, Cotta and Abdul-Aziz15,Reference Chater, Family and Abraao17}

Access to essential medicines of assured quality is a key component in optimizing antimicrobial use. ² When indicated, it is imperative that the appropriate antimicrobials are accessible. Almost half of this survey’s respondents reported that availability drove selection compared with 68.9% in Ethiopia, 63.8% in the Democratic Republic of Congo, and 14.0% in Peru. ^{Reference Gebretekle, Mariam and Abebe7,Reference García, Llamocca and García11,Reference Thriemer, Katuala and Batoko20} Timely administration is also fundamental and reduces mortality in patients with sepsis; ^{Reference Kumar, Roberts and Wood27} almost half of the nursing staff reported interruptions in therapy due to nonadministration of prescribed agents and delays between prescription and administration of longer than one hour. Staff shortages may play a role; utilization of technology to implement more rapid delivery of results or electronic prescribing garnered substantial support in this survey and may ease the effects of staffing challenges.

Although the overwhelming majority of doctors reported routinely assessing for de-escalation and switching to oral formulations, in agreement with data from junior doctors in France, this was not corroborated by nursing staff nor pharmacists. ^{Reference Lévin, Thilly and Dousak19} Possible reasons include recall or performance bias or the fact that pharmacists and nurses have exposure to prescriptions written by a variety of doctors including those who were not survey participants. Prescription practices should be targeted as a quality improvement initiative as identified by doctors and nurses in Gabon, as 70.0% of pharmacists in this study often noted errors in antimicrobial prescriptions. ^{Reference Adegbite, Edoa, Schaumburg, Alabi, Adegnika and Grobusch10} Poor adherence to antimicrobial prescription guidelines (45.0%) has been previously reported in South Africa. ^{Reference Gasson, Blockman and Willems28}

Additional pharmacists, infectious disease physicians, or clinical microbiologists were not viewed as having the greatest impact on AMR, contrary to previous studies. ^{Reference Wasserman, Potgieter and Shoul8,Reference Adegbite, Edoa, Schaumburg, Alabi, Adegnika and Grobusch10,Reference Babatola, Fadare and Olatunya14,Reference Han and Ramsamy30} These skilled personnel are, however, essential in enabling some of the interventions chosen as most impactful in the present and previous studies, such as education, multidisciplinary rounds (with prospective audit and feedback), and institutional guideline development. ^{Reference Gebretekle, Mariam and Abebe7,Reference Limato, Nelwan and Mudia13,Reference Setiawan, Cotta and Abdul-Aziz15,Reference Jha, Mudvari, Hayat and Shankar18,Reference Han and Ramsamy30} Pharmacists approach antimicrobial therapy from a systems perspective and are integral to successful ASPs; ^{Reference Chetty, Reddy, Ramsamy, Naidoo and Essack6,Reference Gebretekle, Mariam and Abebe7,Reference Setiawan, Cotta and Abdul-Aziz15,Reference Sutthiruk, Considine, Hutchinson, Driscoll, Malathum and Botti21,Reference Chetty, Reddy, Ramsamy, Dlamini, Reddy-Naidoo and Essack29} in this survey, 42.9% of all respondents did not feel that sufficiently trained pharmacists were available. The vast majority of pharmacists (80.0%) felt that they were not supported by their hospital to implement ASPs, while over 60% reported patient expectations as a contributor to antimicrobial overuse. The vital role of financial, information technology, and management support in effective ASPs was echoed in this survey. ^{Reference Chetty, Reddy, Ramsamy, Dlamini, Reddy-Naidoo and Essack29}

Limitations to this study include volunteer bias, which is difficult to mitigate in this type of research, and survey availability only in English. Lack of dedicated time, the absence of incentives, and restrictions in mobile data may have influenced participation; paper-based surveys were also distributed on request. The length of the survey may have dissuaded potential participants. There was a substantial female predominance; it is unclear whether this represents the total population of HCW. The modal age groups differed notably between professional groups. Generational differences, varying durations of working experience, and differing training/knowledge (reflective of undergraduate and postgraduate curriculum trends) may have influenced the responses as previously reported. ^{Reference Limato, Nelwan and Mudia13,Reference Babatola, Fadare and Olatunya14,Reference Chater, Family and Abraao17,Reference Jha, Mudvari, Hayat and Shankar18,Reference Padigos, Reid, Kirby and Knowledge31} The small sample size overall prohibited firm conclusions with no category reaching the desired participant number; while poor response rates are a recognized limitation of surveys, stronger support from hospital management and unit managers may have improved participation and facilitated collection of more representative data.

Perceptions on AMR and ASPs differed somewhat between academic and nonacademic settings in this multicenter survey although many similarities were noted. AMR was of more concern in academic settings. In nonacademic settings, staff were willing to allocate more time to ASPs. Diagnostic stewardship, prescribing practices, and antimicrobial access must be optimized in both settings. IPC practices are underestimated as a contributor to AMR. Education, involvement of the hospital IPC team, and inclusion of nursing staff in ASPs were chosen as having the greatest impact on AMR. A multidisciplinary behavioral science-driven approach is needed to address the complex issue of AMS among HCW.