Retrospective observational study

The 2022 data set comprised 3,897 prescriptions of restricted antimicrobials. The majority of patients (n = 380, 93.6%) were from the two largest hospitals in the network (Campus CDK and LKH). In 11.7% (n = 456 attributed to 366 patients) of the analyzed prescriptions, at least one DRP was identified and a corresponding intervention was proposed. Most of the patient cases associated with DRPs were assessed within 15 min (n = 430; 94.3%), while 26 cases (5.7%) required more than 15 min, with a maximum workup time of 60 min. The majority of the 366 patients with DRPs were male (n = 218; 59.6%) and older than 65 years (n = 256; 69.9%) with a median age of 72.5 years. Table 5 shows the demographic information for the patients with DRPs within the context of restricted antimicrobials.

Table 5 Demographic details of patients (n = 366) with at least one DRP-related prescription of a restricted antimicrobial agentClinical relevance of drug-related problems

The vast majority of the 366 patients presented with a single DRP (83.6%, n = 306), while 11.7% (n = 43) of patients presented with two DRPs. Analysis of the DRPs showed high clinical relevance (CR) in 80.2% with most categorized as “relevant” (62.9%, n = 287 out of 456 DRPs) or “highly relevant” (17.3%, n = 79) (Fig. 2). Only two DRPs (0.4%) were categorized as ‘extremely relevant’, while the categories ‘slightly relevant’ and ‘not relevant’ accounted for 17.3% and 2% of DRPs (n = 79 and n = 9, respectively).

Fig. 2

Rate of clinical relevance (in %) of drug-related problems (DRP, n = 456) identified within the post-prescription process of restricted antimicrobials

Reasons for a drug-related problem

The main reasons for a DRP related to an antimicrobial agent (Fig. 3) were ‘non-conformance of anti-infective administration with guidelines, existing infectious disease advice or AST’ (27.4% of all DRPs, n = 125), ‘drug without clear indication’ (27.2%, n = 124) and ‘need for patient or drug monitoring’ (12.5%, n = 57). Sub-analysis of these categories revealed the highest rates of CR-categories “relevant” or “highly relevant”. The category ´drug-drug-interaction´ was the only one that included two cases classified as “extremely relevant” and also showed a substantial rate of clinically relevant (n = 13) or highly relevant (n = 17) DRPs. In addition, ‘dosage’ aspects (both too high and too low) were also of great importance, together accounting for 10.3% of DRPs. In contrast, administrative aspects of the prescription or incomplete information in the substance application played a minor role.

Fig. 3

Clinical relevance of identified reasons for drug-related problems (in total n = 456) among antimicrobial prescriptions within the post-prescription process of restricted agents

AST antimicrobial susceptibility testing; DRP drug-related problem; ID infectious diseases

Key anti-infective agents of concern

Our analysis demonstrated that four broad-spectrum anti-infective substances accounted for 73.7% of all interventions: linezolid (25.0%, n = 114), meropenem (24.1%, n = 110), ciprofloxacin (15.8%, n = 72), and piperacillin-tazobactam (8.8%, n = 40) (Fig. 4). Voriconazole required 4.4% (n = 20) of consultations, while the quinolones levofloxacin and moxifloxacin collectively represented 5.9% (n = 27).

Fig. 4

Rate of restricted antimicrobials with the need for a pharmaceutical intervention within the post-prescription process (n = 456)

In the detailed analysis of the reasons for DRPs, certain patterns stood out prominently (Fig. 5): linezolid was the leading contributor to “need for patient/drug monitoring”, accounting for 82.4% of DRPs within this category (n = 47), followed by voriconazole at 14% (n = 8). Therapeutic drug monitoring (TDM) is available for both of these substances in our hospitals.

The DRP category “drug without clear indication” was predominantly associated with the substances meropenem (32.2%, n = 40), linezolid (16.9%, n = 21), ciprofloxacin (14.5%, n = 18), and piperacillin-tazobactam (11.3%, n = 14). The very same agents also represented the majority of cases in the category “non-conformance with guidelines, ID consultation, or AST” with meropenem accounting for 24.0% of DRPs (n = 30), followed by ciprofloxacin (19.2%, n = 24), as well as linezolid and piperacillin-tazobactam (each 16.0%, each n = 20). Meropenem also accounted for a significant proportion within the categories of “administrative issues” (33.3%, n = 8), “incomplete information” (38.1%, n = 8), and “dosage” (with 42.4% (n = 14) under-dosage and 35.7% (n = 5) over-dosage) given its role as an extremely broad-spectrum antimicrobial agent. Regarding DRPs caused by interactions, ciprofloxacin (25.0%, n = 8), linezolid (21.8%, n = 7), along with voriconazole were particularly notable.

Fig. 5

Detailed analysis of reasons for a pharmaceutical intervention among key antimicrobial agents associated with a drug-related problem within the post-prescription process (n = 456)

AST antimicrobial susceptibility testing; DRP drug-related problem; ID infectious diseases

Pharmaceutical interventions

The 456 pharmaceutical DRP-related interventions (Fig. 6) aimed at the optimization of pharmacokinetic/pharmacodynamic (PK/PD) parameters (n = 140, 30.6% of all interventions), treatment discontinuation (n = 128, 28.1%), and de-escalation (n = 82, 17.9%). Additionally, adjustment of treatment duration accounted for 10.3% of interventions (n = 47), while IV-to-oral (IV/PO) switch and administrative information by the pharmacists represented 7.4% (n = 34) and 5.5% (n = 25), respectively.

A clear pattern emerged between anti-infective agents and the pharmaceutical interventions they prompted. The optimization of PK/PD parameters—focusing on therapeutic drug monitoring, dose and dosing interval —was most frequently addressed for linezolid, which accounted for 42.1% of interventions in this category (n = 59), followed by meropenem (n = 24, 17.1%). Notably, linezolid also dominated interventions on IV-to-oral switching, representing 47% of consultations in this category (n = 16), followed by ciprofloxacin (26.5%, n = 9), both known for their high oral bioavailability. Meropenem was particularly prominent in consultations related to antimicrobial treatment discontinuation (n = 38, 29.7%) and adjustment of the treatment duration (34%, n = 16). Additionally, therapy de-escalation was most frequently required for treatments involving broad-spectrum agents, particularly meropenem (29.3%, n = 24) and piperacillin-tazobactam (19.5%, n = 16).

The proposed interventions were instantly implemented by the ward physicians in 82.7% of cases (n = 377), resulting in a change of the antimicrobial treatment following the pharmaceutical consultation. In 8.8% of cases (n = 40), the physician considered the pharmacist’s recommendation during consultation, but its final implementation was unclear, as the decision required discussion with a senior physician or additional patient information. Due to anonymized data, the retrospective analysis could not determine whether these recommendations were ultimately implemented. In 5.7% (n = 26), the implementations could not be directly assessed as they were classified as general informational or not applicable to a direct treatment modification. Only 2.8% (n = 13) of the interventions were rejected by the clinicians.

Fig. 6

Pharmaceutical interventions (n = 456) triggered by restricted antimicrobials

AST antimicrobial susceptibility testing; DRP drug-related problem; ID infectious diseases

Reasons for direct cost-savings by pharmaceutical interventions

Cost reduction potential was identified in 409 out of 456 pharmaceutical interventions (89.7%). Specifically, 33% (n = 151 interventions) were attributable to the reduction of ordered items (e.g. vials, drug packs) compared to the initial order, followed by drug discontinuation (15.5%, n = 71), and IV to oral switch (5.3%, n = 24). Reduction of the initial dose and switch to a less expensive antimicrobial agent played a minor role (Table 6). In 132 interventions (28.9%), a follow-up cost increase was anticipated due to a clinically indicated dose escalation, the necessity for therapeutic drug monitoring (e.g., linezolid, voriconazole), or additional laboratory and clinical assessments (e.g., electrolyte panels, ECG in patients with multiple cardiotoxic drugs).

Table 6 Interventions with cost-reduction potential in pharmacist-led post-prescription interventions of restricted antimicrobials (n = 456)

The interventions implemented yielded potential cost savings of €180,420 through avoided anti-infective expenses (Table 7). Most of the cost savings were achieved through interventions involving meropenem (€50,866 in saved medication costs), linezolid (€33,154), anidulafungin (€17,851), ciprofloxacin (€17,149), and piperacillin-tazobactam (€11,185), which together accounted for 69% of the total anti-infective cost savings. As the pharmaceutical service required 0.25 FTE of a pharmacist, these interventions resulted in a return on investment of approximately 1:6, excluding additional effects from behavioral changes.

Table 7 Cost savings through pharmacist-led interventions post-prescription interventions of restricted antimicrobials (n = 456)Interrater reliability of clinical experts

The questionnaire response rate was 100%, with all seven experts completing and returning the questionnaire within 14 days. Table 8 (appendix) presents the questionnaire results regarding significance classification, including agreement with the original rating. No case exhibited a 100% agreement between the expert team and the clinical pharmacists from Salzburg. However, very high agreement levels of 84% and 85% were observed in two cases, while high agreement (70%) was found in four cases.

Table 8 Results of the questionnaire on clinical relevance rating

The statistical analysis of interrater reliability was based on a sample size of n = 30 (except for the weighted Cohen’s kappa (n = 26) due to missing values) (Table 9, appendix). All Fleiss’ kappa coefficients indicate “slight agreement”. As expected, coefficients accounting for similar ratings (weighted Cohen’s kappa, AC2) are higher than those considering only exact matches (Fleiss’ kappa, AC1). Gwet’s coefficients also tend to be higher. Comparing clinical pharmacists and physicians, Fleiss’ kappa and weighted Cohen’s kappa indicate higher agreement among clinical pharmacists, whereas Gwet’s coefficients suggest stronger agreement among physicians. This discrepancy likely results from the coefficients’ sensitivity to unequal marginal distributions. Agreement between the expert team and the original clinical relevance rating (Table 10, appendix) is based on n = 30. Cohen’s kappa indicates “moderate agreement” for the entire expert team and clinical pharmacists, but only “slight agreement” for physicians. Across all coefficients, clinical pharmacists align more closely with the original clinical relevance rating than physicians.

Table 9 Coefficients for evaluating interrater reliabilityTable 10 Coefficients for expert team agreement with original rating of clinical relevance