Sensitivity analyses
To perform sensitivity analysis for the outcome of mortality rate, we excluded the study by Fouda et al. as this RCT was not explicitly indicating the infection status of patients in the exclusion criterion. This analysis did not change the results significantly, suggesting they were relatively stable. (Supplementary Materials Figure S3).
For the proportion of patients with mRS ≤ 2 points, the heterogeneity was decreased (P = 0.31, I2 = 15%) and the results were not changed by excluding the Lampl et al. study. (Supplementary Materials Figure S4).
Discussion
This systematic review and meta-analysis revealed that prophylactic antibiotics could not reduce the mortality rate and pneumonia of stroke patients, nor could they improve functional outcomes, and length of hospital stay. However, we find a significant reduction in infection and urinary tract infection after stroke without increasing the incidence of adverse events.
Several factors might influence the result. First of all, the duration of prophylactic antibiotics was insufficient. Finlayson et al. found pneumonia was associated with longer hospitalization.25 And a retrospective study revealed that the length of hospitalization was a strong predictor for post-stroke pneumonia, the incidence of pneumonia for patients with hospitalization for more than two weeks was 3.9 times more than that for patients hospitalized for four days to one week [OR 3.90 (95%CI: 3.73-4.08, P < 0.0001)]. 26 Among the RCTs included in this meta-analysis, 9-11 the mean length of hospital stay in the prophylactic antibiotics group was 23.5 days, but the use of prophylactic antibiotics was only two to seven days. Therefore, it was difficult to avoid the occurrence of pneumonia in stroke patients throughout the treatment. In particular, the length of the hospital stay was too long. Second, in most studies, the time window before the start of prophylactic antibiotic therapy was quite long-up to 24h after stroke. To evaluate the effects of different administration times, we performed a subgroup analysis and found a decreasing trend in pneumonia in patients with early prophylactic use of antibiotics within 24 hours after admission [RR 0.81 (95%CI: 0.62-1.07,I2 = 0%)] as compared with those using prophylactic use of antibiotics within 48 hours after admission [RR 0.94 (95%CI: 0.79-1.11, I2 = 0%)]. Third, the severity of stroke is a risk factor for infection after stroke.27 Thus, patients with severe stroke who are at high risk for infection should be considered for prophylactic antibiotics. However, an RCT with a large sample size included in our meta-analysis involved patients with mild stroke, and the average hospital stay of patients was only 6.35 days. 10 Considering the low incidence of nosocomial infection, 26 the role of prophylactic antibiotics in reducing post-stroke infection was not fully reflected.
Stroke-related infections, particularly pneumonia, are regarded as an independent risk factor associated with mortality after stroke.1 The results of this study did not find a reduction in pneumonia, so the mortality rate was not significantly reduced.
In two studies, 12,22 fluoroquinolones were used as prophylactic antibiotics for post-stroke infection. There concerns about the adverse events of fluoroquinolones on the nervous system. However, no serious adverse events were reported in these two studies. And several preclinical studies had described the neuroprotective effects of moxifloxacin after transient focal brain ischemia. 28Nevertheless, attention still should be paid to the adverse events in the clinical application of fluoroquinolones.
There are several strengths in this meta-analysis. First, compared with previous studies, 7,8,29 this meta-analysis included more RCTs and sample size; thus, the confidence interval of the study and the heterogeneity between studies for each outcome were reduced. Second, we performed a subgroup analysis by time to the first dose of antibiotic drugs. We found that patients with early prophylactic use of antibiotics within 24 hours after admission had a decreasing trend in mortality and pneumonia compared with those with prophylactic use of antibiotics within 48 hours. Third, we assessed the overall evidence according to the GRADE quality of evidence approach.
However, this study still has some limitations. First, the outcomes of patients with post-acute stroke infection might be impacted by stroke severity, chronic conditions, dysphagia, age, invasive procedure, different timings, types, and duration of antimicrobial use.30,31 The invasion procedures, such as urinary catheterization or mechanical ventilation, could increase the risk of infection by facilitating the entry of a pathogen; However, we could not perform these subgroup analyses due to unavailable data. For example, only one RCT in our meta-analysis described the use of urinary catheterization, 12 so we could not evaluate the effect of urinary catheterization on prophylactic antibiotics. Secondly, clinical heterogeneity is inevitable due to different criteria used for infection in RCTs; therefore, the results of this meta-analysis should be interpreted with caution. Thirdly, most of the RCTs included in this study were performed in developed countries with standard stroke units. Due to the differences in medical resources and nursing levels between developing and developed countries, standard stroke unit management is unconditionally carried out in some places, and the medical environment is poor, 32,33 which increases the incidence of infection in patients. Evidence for prophylactic antibiotics in stroke patients is lacking in developing countries or remote areas. Therefore, whether to recommend prophylactic antibiotics in patients with post-acute stroke infection needs to be further confirmed by large-scale RCTs.