Study design

This was a retrospective cohort observational study conducted at Jordan University Hospital (JUH) to assess the rate and outcomes of fluoroquinolone de-escalation in patients who were diagnosed with CAP from December 2022 to March 2024. The population of interest was adults aged ≥ 18 years with a suspected diagnosis of CAP (CAP is defined as an acute infection of the pulmonary parenchyma acquired in the community or within 48 h of admission). We only included those patients who had a positive CAP diagnosis using the ICD-10 diagnostic codes. Other inclusion criteria were: the patients had blood and sputum cultures ordered, received a fluoroquinolone regimen of therapy as the empiric treatment regimen, and were admitted to the hospital until cultures and sensitivity were performed. We excluded patients who were pregnant, were not certain of their diagnosis, and/or had incomplete data. Severity of illness was not assessed in the study.

Sample size calculation

The Raosoft sample size calculator was used to calculate the sample size for this study with a 10% margin of error, a 95% confidence level, and an expected proportion of successful de-escalation of 50%. Based on these parameters, the minimum sample size required was calculated to be 96 patients.

Ethical considerations

The study was conducted in accordance with the principles outlined in the World Medical Association Declaration of Helsinki. It received approval from the JUH Institutional Review Board (approval number 10/2024/1389). To ensure patient confidentiality, all data were de-identified and anonymized. Given the retrospective, observational design, the ethical committee granted a waiver for informed consent.

Data collection

Data collection comprised the gathering of extensive information from the medical records of eligible patients using a planned data collection form. Collected data included demographic data such as age, gender, admission date, discharge date, and LOS. We gathered clinical data, including diagnoses, empiric antibiotic therapy (including name, route of administration, and duration of therapy), and information related to targeted antibiotics after culture results. We also collected laboratory data that included blood and sputum cultures and their sensitivity results to study the susceptibility patterns of the microbiological pathogens that caused CAP.

Primary outcome

The main aim of this research is to assess the rate of antibiotic de-escalation in patients with CAP who were treated with fluoroquinolones. The term de-escalation means moving from broad-spectrum antibiotics (fluoroquinolones) to a narrower spectrum of antibiotics based on final culture and sensitivity results. The narrower-spectrum antibiotics to consider would be cephalosporins, penicillins, trimethoprim/sulfamethoxazole, vancomycin, or linezolid. This approach is used to decrease antibiotic use, and de-escalation might also include discontinuation of antibiotics if cultures do not demonstrate any infection, ideally within 48 h of culture results. Patients were grouped together for analysis into two primary groups based on whether or not they required de-escalation. Patients were down-graded based on final culture and sensitivity results. Patients with positive cultures for atypical bacteria or MRSA were not considered possible candidates for fluoroquinolone de-escalation because fluoroquinolones could likely be indicated as part of the treatment of patients with atypical bacteria or MRSA. Patients categorized as requiring de-escalation were further divided into successful de-escalation, failed de-escalation, and inappropriate de-escalation.

Failed de-escalation pertains to interventions where empiric antibiotics were continued after culture results yielded no definitive evidence to continue antibiotic therapy and/or when broad-spectrum antibiotics were continued after the patient was cultured and a narrower-spectrum antibiotic was more appropriate. Inappropriate de-escalation corresponds to errors or missed opportunities that deteriorate patient care. For example, in the case of negative culture results related to inappropriate de-escalation, transition to narrower-spectrum antibiotics could occur when the very best approach would have been to discontinue antibiotic therapy altogether. In contrast, if positive culture results were obtained, a scenario of inappropriate de-escalation could be where antibiotics are discontinued instead of switched to more directed therapy when more directed therapy would have been of greater benefit for patient care.

Assessment of the adequacy or appropriateness of antibiotics was based on culture and sensitivity, as well as the IDSA guidelines. It should be noted that the decision regarding adequacy or appropriateness of antibiotic therapy did not incorporate patient-specific clinical variables such as the severity of illness or other comorbidities.

Secondary outcomes

The secondary outcomes of the study included assessing various factors among patients who underwent successful de-escalation compared with those who did not successfully de-escalate treatment or performed de-escalation incorrectly after excluding inappropriate de-escalation of antibiotics (e.g., antibiotics were stopped instead of de-escalating towards a narrower spectrum medication following a positive culture). Secondary outcomes assessed included the LOS and in-hospital mortality, which was defined as any death occurring while the patient was still admitted and did not have to be related to the reason for admission. The study also assessed the success rates of fluoroquinolone de-escalation among patients with a positive culture versus a negative culture.

Statistical analysis

Analysis of the data was conducted using version 26 of the IBM SPSS software. Descriptive statistics presented frequencies and percentages for categorical variables and medians and interquartile range (IQR) for continuous variables. Normality of the data was tested using the Shapiro-Wilk test. Fisher’s exact test was used to compare all-cause in-hospital mortality between patients who were successfully de-escalated and those who failed or were inappropriately de-escalated. The Mann-Whitney U test was used to describe the LOS. To assess the association between the microbial culture results and de-escalation practice, the Chi-square test was used. All statistical tests were two-tailed, and significance was determined by p < 0.05.

Demographic and medical characteristics of the study sample

A total of 125 patients met eligibility criteria after screening for inclusion criteria during the study. The median age of participants was 73 years old (IQR = 24). The number of female patients was 44 (35.2%) and 81 male patients (64.8%). The median LOS for all patients was 13 days (IQR = 11). In addition, 93 patients (74.4%) survived the hospital stay, while 32 patients (25.6%) died while in the hospital. Among the patients, 20 (16%) were transferred to the ICU, 100 (80%) stayed in the general ward, and 5 patients (4%) were in the ICU upon admission. Table 1 summarizes the demographic and medical characteristics of the study sample.

Antibiotic prescription patterns among the study sample

All patients in the study received therapy utilizing fluoroquinolones, with the great majority receiving levofloxacin (n = 124; 99.2%). A small number of patients received fluoroquinolone monotherapy (n = 12; 9.6%), but most patients received combinations of antibiotics along with fluoroquinolone therapy. A total of 42 patients (33.6%) received two antibiotics, 49 patients (39.2%) received three antibiotics, and 22 patients (17.6%) received four antibiotics. Vancomycin was prescribed most frequently (n = 72; 57.6%) among adjunct antibiotics. The median number of prescribed empiric antibiotics was 3 (IQR = 1.0), with a total of 331 empiric antimicrobial agents prescribed, including fluoroquinolones.

Blood or sputum culture and susceptibility testing

The microbial culture results indicated that 96 of the patients (76.8%) were negative for culture, and 29 of the patients (23.2%) were positive for culture (Table 2). The majority of the positive samples (28/29, 96.6%) underwent susceptibility testing, while one positive sample (3.4%) did not undergo susceptibility testing. The majority of specimens contained pathogens that were resistant to fluoroquinolones (23/28, 82.1%), and the minority of samples were sensitive to fluoroquinolones (3/28, 10.7%). Interestingly, two samples (7.1%) did not undergo susceptibility testing for fluoroquinolones but did undergo antibiotic susceptibility testing (Table 2).

The positive culture specimens predominantly showed 1 microorganism (28/29, 96.6%); there was 1 specimen (3.4%) that had two pathogens, which had a total of 30 isolated pathogens. The most commonly reported pathogens were Staphylococcus hominis (6/30, 20.0%), Klebsiella pneumoniae (5/30, 16.7%), and Staphylococcus epidermidis (3/30, 10.0%) (see Table 2).

Assessing the appropriateness of fluoroquinolone de-escalation

Of the patients suspected to have CAP, 76.8% had a negative blood culture. All patients were then classified into one of three groups depending on the appropriateness of fluoroquinolone de-escalation. There were those who incorrectly de-escalated (42/96, 43.8%), those who had a failed de-escalation (29/96, 30.2%), and those who successfully de-escalated (25/96, 26.0%).

On the contrary, in the sample, 29 patients (23.2%) had a positive culture finding for otherwise suspected CAP (25 positive blood cultures and four positive sputum cultures). All patients required de-escalation. The 29 patients were then stratified into three categories. There were those who incorrectly de-escalated (n = 3, 10.3%), those who failed de-escalation (n = 10, 34.5%), and those who successfully de-escalated (n = 16, 55.2%). These results appear in Figure 1. Likewise, Table 3 includes examples demonstrating the decision process surrounding de-escalation for patients with both positive and negative cultures.

Display full size

Figure 1. Decision-making scheme for the de-escalation process.

Clinical outcomes and demographics by de-escalation status comparison

Before assessing the study’s secondary objectives, we duly attempted to ensure that cases were included that were relevant to our goals. Consequently, we excluded cases in which de-escalation occurred incorrectly, with particular reference to cases where antibiotics were stopped instead of switching to a narrower spectrum after a positive culture result (3 patients). After excluding these 3 patients, the reduced sample size was 122 patients. Of the 122 patients, 41 were successfully de-escalated and 81 either improperly de-escalated or did not de-escalate.

The study noted that patients who were successfully de-escalated had a median LOS of 12 days (IQR = 8 days) compared to patients who lacked successful de-escalation or proper de-escalation [median LOS 18 days (IQR = 11 days)], with a statistically significant p-value of 0.004 via Mann-Whitney U test.

Regarding mortality at discharge, there was no difference between the two study groups. Among the patients who had successful de-escalation, 29 of 41 (70.7%) survived, whereas of those who had either failed de-escalation or were inappropriately de-escalated, 62 of 81 (76.5%) survived. Fisher’s exact test yielded no significant difference in in-hospital mortality between the two groups for this variable (p = 0.514).

Age, gender, and ICU admission status were not significantly different between the two groups (p ≥ 0.05 for all). Further details can be found in Table 4.

Association between culture results and de-escalation practice

Among patients with a positive culture, 16 of 26 patients (61.5%) had successful de-escalation. Of those patients with negative cultures, 26.0% (n = 25) had successful de-escalation. Chi-square analysis demonstrated a significantly greater rate of de-escalation success with positive cultures compared to negative cultures (p = 0.002). The details are displayed in Table 5.