Population
During the study period, there were 4251 admissions at the ED. Among these patients, 605 (14.2%) received antibiotics, and 484 were finally included in the study (Figure 1).
Two-hundred and fifty-five (52.7%) patients were men and median age was 79.9 years (extremes 18-105 years) (Table 1). The most frequent underlying conditions were chronic cardiovascular diseases (250, 51.7%), high blood pressure (240, 49.6%), dementia (131, 27.1%), and diabetes mellitus (111, 22.9%). One hundred patients (20.7%) were immunocompromised.
Clinical presentation in the ED
Respiratory symptoms were the most frequent cause of consultation to the ED (150, 31.0%), before abdominal symptoms (110, 22.7%) or “isolated” fever (49, 10.1%) (Table 2). Most of the patients had a baseline qSOFA < 2 (409, 84.5%), whereas a septic shock was diagnosed in 11 (2.3%) patients.
Blood cultures and urine analysis were performed in the ED in 372 (76.9%) and 259 (53.5%) patients, respectively. Most of the blood cultures were negative (324 patients, 87.1%). Among the 48 positive ones, Escherichia coli (19, 39.6%), other Enterobacteriaceae (15, 31.3%), Staphylococcus aureus (9, 18.8%) and anaerobic bacteria (5, 10.4%) were the most frequently observed pathogens. Out of the 92 (35.5%) positive urine analyses, Escherichia coli (61, 66.3%), other Enterobacteriaceae (20, 21.7%), Pseudomonas aeruginosa (8, 8.7%) and Enterococci (7, 7.6%) were the most frequently observed pathogens.
Antibiotics prescribed in the ED
Among the 484 treated patients, antibiotic therapy was most often prescribed as monotherapy (324, 66.9%). Third generation cephalosporin (244, 50.4%), amoxicillin-clavulanic acid (167, 34.5%), metronidazole (88, 18.2%) and spiramycine (47, 9.7%), were the most often prescribed antibiotics. The antibiotic therapy in the ED was graded as optimal in 328 (67.8%), as adapted in 110 (22.7%) and as inadequate in 46 (9.5%).
The main reasons justifying that the antibiotic therapy was just considered as adapted at ED were mainly a too wide spectrum (45, 40.9%), a non-adapted dose (33, 30.0%), and a non-adaptation to the clinical severity (14, 12.7%). The main reasons justifying that the antibiotic therapy was inadequate at ED were an absence of indication to treat with antibiotics (24, 52.1%) and a prescribed antibiotic which was not active for the type of infection (11, 23.9%).
Non-optimal antibiotic therapy was prescribed in 51 (33.3%) patients with lung infection, in 29 (34.9%) patients with UTI, in 13 (17.1%) patients with abdominal infections and in 9 (39.1%) patients with cutaneous infections.
Primary outcome
The primary endpoint was reached for 74 (15.3%) patients: 38 deaths (7.9%), 22 transfers to ICU (4.6%) and 14 new hospitalizations within 30 days after discharge (2.8%).
Compared to an optimal antibiotic therapy, an adapted antibiotic therapy at ED was significatively associated with unfavorable outcome (HR = 1.95, Confidence Interval (CI) 95% (CI95%) [1.18-3.22]; p = 0.01). On contrary, no significant association was found with an inadequate antibiotic therapy (HR = 0.58, CI95% [0.21-1.61]; p = 0.29) (Figure 2).
This significant association between an adapted (compared to an optimal) antibiotic therapy in the ED and an unfavorable event was also found when considering only patients with a bacterial infection (HR = 2.16, CI95% [1.25-3.75]; p = 0.006) and those with a pulmonary infection (HR = 2.96, CI95% [1.25-7.03]; p = 0.01).
Factors associated with the main outcome in the univariate and multivariate analyses are summarized in Table 3. The candidates as adjustment variables for the multivariate model were cardiovascular diseases, chronic respiratory failure, diabetes mellitus, autonomy of the patient, community infection, qSOFA, septic shock, haemoglobin, CRP, creatinine, adequation of antibiotics’ prescription at ED, type of infection, bacterial infection, adequation of antibiotics’ prescription after the reevaluation.
Reevaluation of antibiotic therapy at D2
At D2, 358 patients (74%) were still hospitalized, whereas 103 (21.2%) have been discharged and 23 (4.8%) underwent an unfavorable event (14 deaths, 4 transfer to ICU, 5 discharges followed by re-hospitalization for the same infection).
The reevaluation of antibiotic therapy at D2 was performed in 436 patients (90.1%). Three-hundred and twenty-three of 358 patients (90.2%) still hospitalized at D2 underwent the reevaluation. Among the 103 patients discharged at D2, most of them (101, 98.1%) underwent a reevaluation before they were discharged.
The infectious diagnosis was modified between the ED and D2 for 56 patients (11.6%). Most of the diagnoses’ modifications concerned pulmonary infections at ED (24, 42.9%) and UTI at ED (15, 26.8%).
Only 26 (6.0%) had their initial antibiotic therapy unchanged. The antibiotic therapy was stopped in 106 patients (24.3%). The reasons of discontinuation were no retained diagnosis of bacterial infection (84, 79.2%), complete duration of antibiotic therapy (17, 16.1%) and comfort care only (5, 4.7%). The antibiotic therapy was de-escalated in 150 patients (34.4%), had its administration way or dosage modified in 113 (25.9%), or was escalated in 41 (9.4%).
Out of the 330 (75.7%) patients still on antibiotic therapy following reevaluation, monotherapy was used in 278 (84.2%) patients. Amoxicillin-clavulanic acid (148, 44.8%), third generation cephalosporins (69, 20.9%), amoxicillin (42, 12.7%) and metronidazole (26, 7.9%) were the antibiotics the most often prescribed.
Among the 436 patients who underwent the reevaluation at D2, the antibiotic therapy was graded as optimal in 326 (74.8%) patients, including 104 (31.9%) with justified antibiotic therapy discontinuation. Antibiotic therapy was graded as adapted in 64 (14.7%) and as inadequate in 46 (10.5%).
The main reasons justifying that the antibiotic therapy was just considered as adapted after reevaluation were mainly a not-adapted proposed duration (28, 43.8%), a too wide spectrum (22, 34.4%), and the use of an antibiotic not recommended by the guidelines (7, 10.9%). The main reasons justifying that the antibiotic therapy was inadequate after reevaluation were mainly an absence of indication (24, 54.2%) and a not-adapted molecule (12, 26.1%).
Compared to an optimal antibiotic therapy, an inadequate prescription after reevaluation was significantly associated with an unfavorable outcome: HR = 3.52, CI95% [1.42-8.72]; p = 0.003.