The significance of this study lies in the establishment of goals and detailed methods for conducting qualitative assessment of antibiotics, both today and long-term. Given the absence of standard recommendations for qualitative antibiotic assessments, the experts who participated in the survey were presented with various approaches currently being used in different countries. By identifying the strengths and weaknesses of these approaches, a more objective application of qualitative antibiotic assessments in the ROK can hopefully be achieved.

The consensus reached regarding the long-term goal for qualitative assessment of antibiotics involved conducting annual assessments by the KDCA for antibacterial and antifungal agents. Antibiotics prescribed for specific periods will be assessed through random sampling, and evaluations will be conducted for particular antibiotics or diseases as needed. The short-term goals largely align with the long-term goals, except that primary hospitals, clinics, specialised medical institutions, and hospitals with fewer than 300 beds will not be included in qualitative antibiotic assessments. These findings indicate that infectious disease or ASP experts in the ROK acknowledge the significance of qualitative assessments of antibiotics, even at present, and agree on the necessity of expanding the assessments to eventually include clinics. There is currently no consensus regarding the frequency of antibiotic evaluations, with experts agreeing on an annual evaluation for the long term. In the short term, annual evaluations are deemed unfeasible owing to resource constraints. It may be necessary to initiate a 2–5 years cycle, but the long-term goal is to gradually increase the frequency of evaluations to an annual basis, as is currently done in Australia [6].

Determining how to select the antibiotics to be evaluated is crucial, as it is connected to the required resources. In this study, there was a long-term consensus on using random sampling while also incorporating time periods. Evaluating all patients would be labour-intensive; therefore, random sampling presents a sustainable assessment method in the long run. In the ROK, qualitative assessments of antibiotic prescriptions were conducted in 2018 and 2019 using a 1-day complete enumeration and a 2-days random sampling method, respectively. Despite including more institutions in the second survey, the rates of inappropriate antibiotic prescriptions were similar in both surveys [2, 9].

Evaluating the duration of antibiotic use is an important aspect of qualitative assessment but is difficult to conduct using point surveillance methods [2]. In 2021 and 2022, disease-specific evaluations for urinary tract infections and bacteraemia were conducted, and the appropriate antibiotic prescription period was determined by evaluating the entire period. This approach overcame the challenge of assessing many quality indicators with other point surveillance methods [2]. It was agreed that specific quality indicators should be used for assessment, which cannot rely solely on expert judgement, and continuous updates are necessary [18].

The lack of expert consensus on using HIRA data for qualitative assessment of antibiotic prescriptions to minimise the labour input required from individual institutions is noteworthy, as only 47.4% of respondents agreed to its use. This could be due to concerns surrounding code shifting and the unavailability of clinical data in the HIRA database, as previously reported [19]. The survey results emphasised the need for significant human resources for qualitative assessments of antibiotics and the importance of actively seeking funding to recruit personnel for this purpose. However, since manpower alone would not cover all institutions, evaluation based on a national database method should be considered as a supplementary measure for some areas [20,21,22].

It was agreed that after assessments are conducted, the feedback method should involve anonymising and publicly disclosing the data, as well as including it in the antibiotic use/resistance report. In the ROK, KONAS conducts quantitative assessments of antibiotics, and annual antibiotic resistance reports are published through Kor-GLASS, the Korean national antimicrobial resistance surveillance system based on the GLASS platform [23,24,25]. The integration of reported data is expected to identify the impact of ASP activity on the antibiotic resistance rate through both quantitative and qualitative antibiotic assessments. If the same analysis is also performed on healthcare-associated infection rates, as is done in the US, it may serve as an indicator of the effect of the ASP [12].

It is noteworthy that the entire expert panel agreed on the utilization of the results of the qualitative assessment for the national ASP. The qualitative assessment results enable physicians to identify antibiotics that are being prescribed incorrectly and can be used as evidence to improve the use of those antibiotics. The qualitative assessment implemented at a national level provide an opportunity to reduce the amount of antibiotics used inappropriately, which may be related to decrease in antibiotic resistance. It is expected that the integration of antibiotic qualitative assessment into healthcare quality assessment or the introduction of incentive systems will increase the interest and participation of medical institutions in ASPs. However, three experts in this Delphi survey took a neutral attitude to public disclosure and publicizing the results of the qualitative assessment. Therefore, the strategy for linking the results of the qualitative assessment and the national ASP needs to be discussed further.

This study had some limitations. Firstly, there are several unanswered questions about conducting a large-scale, population-based assessment of antibiotic appropriateness. We did not discuss what criteria should be used to assess antibiotic appropriateness, how many cases should be sampled from each hospital, or who should actually perform the antibiotic appropriateness audits. Further research is warranted to address these questions. Secondly, the international applicability of this study is limited because the survey was only administered to experts in the ROK. However, the study framework itself could be used as a template in other countries, and the methodology presented could also be used to make country-specific adjustments. Lastly, consensus/closing criteria applied in studies using Delphi procedure are varied because there is no standard for that criteria [26]. In this study, we used CVRs, which are widely used to quantify content validity, and tried to effectively reflect the opinions of experts with consensus criteria, considering the number of experts.