This was a single-center retrospective observational before–after study managed at the pulmonology department of a tertiary care center in Hungary. At the tertiary care center level, an ASP (titled antibiotic stewardship pilot project) was implemented in June 2019 with the aim of controlling antibiotic use. Restricted antibiotics were listed, which were allowed to be prescribed by physicians and dispensed by the clinical pharmacist only with the permission of infectious disease specialists. Furthermore, written local guidelines for CAP were available at the wards in the pulmonology department, adherence to which was not mandatory but strongly recommended. The COVID pandemic interrupted the ASP, which was resumed in June 2022. However, written guidelines remained available even after the pandemic (10 January 2021). Consequently, data collection in the pre-intervention phase was carried out from 1 January to 30 April 2022, while in the ASP phase, it was from 1 January to 31 March 2023, months with the highest number of CAP cases. Ethics approval was obtained from the Regional Institutional Research Ethics Committee, Clinical Center, University of Debrecen (DE RKEB/IKEB: 6267-2022).

The ASP was introduced in all inpatient care units of the aforementioned center in order to plan the analysis and control of the reasonable and cost-effective use of antibiotics (Center, 2022). The ASP was carried out by the AST (antimicrobial stewardship team) interdisciplinary team consisting of physician specialists (in this study pulmonologists), microbiologists, infectious disease specialists, and pharmacists. The ASP instructions included antibiotic protocols (prophylaxis and empirical therapy), restricted (controlled) antibiotic agents, individual regulation of antibiotic use by infectious disease specialists, elements of infection control, and analysis of antibiotic use. Furthermore, the ASP-guided empirical antibiotic therapy, which was developed based on the local microbiological resistance map and evidence-based antibiotic use. According to these criteria, the narrowest-spectrum agent was recommended in an appropriate dose, adapted to the location and type of infection, for the shortest possible therapeutic period, and in the most optimal route of administration, preferably in monotherapy. Moreover, to reduce the risk of spreading resistance and ensure prudent use, a restriction on antibiotics has been introduced. Restricted antibiotics are as follows: ciprofloxacin, levofloxacin, moxifloxacin, cefiderocol, ceftaroline, ceftazidime/avibactam, ceftolozane/tazobactam, imipenem/cilastatin, meropenem, imipenem/cilastatin/relebactam, meropenem/vaborbactam, linezolid, tedizolid, aztreonam, and colistin. The use of these antibiotics was allowed only after infectious disease specialist approval. An electronic approval request form was filled by the pulmonologist and sent to the infectious disease specialist. Required data consisted of patient data, required agent (based on positive microbiological test results—pathogen and sensitivity—or pulmonologist decision when no microbiological test is available and no clinical improvement), first empirical antibiotic therapy, present clinical outcomes, and pulmonologist contact information. Restricted antibiotics could have been dispensed only by the clinical pharmacist under strict control after the approval of the consultant infectologist (Figure 1).

To predict 30-day mortality in CAP, CURB-65 (NICE—National Institute for Health and Care Excellence guideline) score and PSI (Pneumonia Severity Index) were used and determined (Fine, 2023; Macfarlane, 2023). According to the local guideline, patients with CURB65 score 1–5 or PSI IV–V mostly required hospitalization. The local guideline for CAP for inpatients is presented in Table 1.

Based on the local guidelines, the recommended systemic antibiotic for moderate risk of mortality was mainly monotherapy, while combination therapy was recommended mainly for high risk of mortality (Debrecen, 2023). Empirical antibiotic treatment was recommended to start within 4 to 6 hours after hospital admission. According to the route of administration, starting therapy with intravenous administration was recommended. Switching from intravenous to oral therapy should be considered after clinical stability, in the case of a moderate infection of CAP, as far as absorption and pharmacokinetics allow. Furthermore, contraindications, severe penicillin allergy, and dose adjustment in kidney failure (after an adequate loading dose) should be considered. The proposal for the total duration of empirical antibiotic therapy was 5 days, with prolongation only in particularly justified cases (e.g., immunodeficiency or positive microbiological tests).

Adult (18 years or above) patients hospitalized due to CAP (based on ICD–International Classification of Disease code) who started their first empirical antibacterial therapy at the above-mentioned department were included in the study. All patients with CAP admitted from another hospital/ward having antibiotic therapy or coinfections at admission were excluded from the study.

All patient- and therapy-related data were collected manually by pharmacists. All data were assembled from medication charts and UD-MED Hospital Information System (IT Services, Hungary) and recorded in Microsoft Excel by using a predefined format of the data entity. Data collection forms were designed by the pharmacists. Demographics and clinical characteristics (gender, age, allergies, Charlson Comorbidity Index (CCI), weight, comorbidities, and discharge type), signs and symptoms (chills, fever, cough, dyspnea, chest pain, breathlessness, malaise, collapse, confusion, respiratory rate, blood pressure, presence of sputum, and dehydration) at admission, chest X-ray examination, laboratory test results (blood urea nitrogen (BUN) concentration, sodium, glucose, hematocrit, partial pressure of oxygen, respiratory rate, blood pressure, heart rate, pH, white blood cells (WBC), C-reactive protein (CRP), creatinine, estimated glomerular filtration rate (eGFR), lactase dehydrogenase enzyme (LDH), microbiological tests, and pathogens), empirical antibiotic therapy (agent selection, route of administration, dose, and duration), and clinical outcomes (30-day survival and length of stay (LOS)) were recorded on data collection forms. Empirical treatment is defined as the antibacterial therapy initiated based on the presence of suspected pathogens without any microbiological testing. Patients were anonymized, thus making them unidentifiable in the study.

The data obtained in both (pre-intervention phase and ASP phase) periods were analyzed and compared. We compared empirical antibiotic use regimen (types: mono- or combination therapy of the first empirical therapy, agent selection, dosage, and duration), the rate of guideline adherence, need for antibiotic change, antibiotic exposure (DDD/patient), antibiotic costs, as well as clinical outcomes (30-day survival and length of stay), and the need for antibiotic prescription at discharge.

Antibiotic exposure was determined using the World Health Organization’s ATC/DDD index (version 2023). The defined daily dose (DDD) is the assumed average maintenance dose per day for an agent used for its main indication in adults. Direct empirical antibiotic costs were calculated based on actual prices obtained from the central hospital pharmacy and expressed in HUF/patient. Antibiotic selection and duration of antibiotic use recommended for CAP are included in Table 5. One treatment followed by another was considered consecutive therapy. Any switch from an intravenous to oral regimen was considered sequential therapy. Empirical antibiotic therapy was considered guideline-adherent when all members were used appropriately regarding the severity of CAP. LOS indicated the number of days that the patient spent in the hospital.

The obtained data were compared to evaluate the effects of ASP implementation on empirical antibiotic use. We applied interrupted time-series analysis (ITSA) to express antibiotic prescription patterns in CAP in pre-intervention and ASP periods.

Fisher’s exact test was applied to compare categorical variables, while Pearson’s correlation coefficient and t-test were used to compare continuous variables between the two study periods. P values below 0.05 were defined as significant.

Demographic and clinical characteristics of patients hospitalized with CAP are described in Table 3. No significant differences were found regarding their gender, age, comorbidities, discharge types, outcomes, PSI, and CURB-65 scores between the two periods. In both periods, more than half of the patients (61–59.2% and 112–57.7%, respectively) were men, aged between 65 and 84 years, and had a CCI score above 4. The most common comorbidities included chronic obstructive pulmonary diseases and cardiovascular diseases. The majority of patients were discharged home. The 30-day mortality rates were 27.2% and 21.6%, with 21.4% and 19.1% being in-hospital deaths (Table 3).

Signs and symptoms at admission are listed in Table 4. No significant differences were found regarding signs and symptoms between the two periods.

The characteristics of empirical antibiotic therapy in CAP and outcomes are summarized in Table 5. Combination therapy was used for CAP in both the periods in the vast majority of cases (pre-intervention period: 74.8% vs ASP period: 76.8%). In the ASP period, the guideline-adherent agent selection increased significantly by 35% (from 58.3% to 93.3%; p = 0.015). However, the overall (agent selection, route of administration, dosage, and duration) guideline adherence increased only by 6.2% (p > 0.05). The inappropriate use of clarithromycin and doxycycline decreased by 38.2% (pre-intervention period: 5.8% vs., ASP period: 3.6%; p > 0.05), while respiratory fluoroquinolones were not used in the ASP period. At the same time, significant decreases in the use of guideline non-adherence combinations with metronidazole (p < 0.05) were observed between the two periods. In both periods, overdosing was relatively high (7.8% and 8.2%, respectively), while underdosing was not frequent. In the ASP period, shorter-duration antibiotic therapy (˂7 days) was significantly more frequent (47.0% vs. 22.3%; p = 0.004), while the total duration of in-hospital antibiotic therapy decreased significantly by 16% (from median 8 to 6 days; p < 0.001). Moreover, the total duration of antibiotic therapy, including antibiotic prescriptions at discharge, decreased by 4.5% (from 10.74 ± 4.78, median 11 days to 9.87 ± 4.95, median 10 days; p > 0.05). Furthermore, in the ASP period, a significant increase in the number of consecutive therapies (by 13.2%, from 14.6% to 27.8%; p = 0.045) was observed, whereas switching from intravenous to oral route of administration resulted in an increase by 10.5% (from 3.9% to 14.4%; p = 0.010). However, in the majority of cases, there were no significant differences in changes between the first empirical therapy (80.6% and 72.2%, respectively) and escalation (14.6% and 11.3%; p > 0.05).

Antibiotic exposure in the ASP period decreased significantly by 23.6% (from 19.89 ± 11.66 to 14.52 ± 9.55 DDD/patient; p < 0.001). As expected, the decrease in the CAP duration led to a lower costs of direct antibiotic therapy after ASP implementation (by 33.2%, from 19,334.10 ± 46,040.22 to 10,582.25 ± 11,124.98 HUF/patient; p < 0.001) (Table 5).

The correlation coefficient between the two periods was −0.19 DDD/patient/day (95% CI –0.3831 to 0.0096), suggesting that the ASP has a sustained long-term effect (Figures 2, 3).

In our study, comparing the pre-intervention period and ASP period revealed that the mean LOS decreased significantly by 13.5% (from 8.85 ± 6.10, median 8 days to 7.09 ± 5.84, median 6 days; p = 0.016). Guideline non-adherent combination therapies of beta-lactams with metronidazole were associated with prolonged LOS in both periods (pre-intervention period: 13.33 ± 9.79, median 11 days and ASP period: 9.5 ± 5.68, median 7 days) compared to the average LOS. Furthermore, we found that the 30-day survival rate increased by 5.9% (from 72.5% to 78.4%; p > 0.05) (Table 5).