Review of 'Cardiac events after macrolides or fluoroquinolones in patients hospitalized for community-acquired pneumonia: post-hoc analysis of a cluster-randomized trial'

  The study is relevant and may be of practical important.
   Tittle of work clearly reflects its content.
   The abstracts contains information about the materials, methods and main results of the research.
   References without notice.
   Information on sources of financing and the absence of conflicts of interest are given separately.

   Cardiotoxic effects of antibiotics are well studied. However, studies of the frequency of cardiac complications with the combined use of antibiotics for the treatment of adult patients with community-acquired pneumonia are not sufficient and ambiguous. References to the results of studying the long-term effects of antibiotic therapy can be excluded from the introduction as the authors studied its complications during the inpatient treatment period.

      The study is based on a  secondary analysis of data obtained in the course of a previously conducted (2011-2013) cluster-randomized cross-prospective study of the effectiveness of empirical treatment of CAP, beta-lactams, fluoroquinolones or beta-lactams in combination with macrolides.
       The article under review does not contain a detailed description of the methodology, but there is a link to a previously published protocol. However, the protocol does not contain a detailed description of the procedure of medical observation of patients, the ECG regulation of observation, the volume of intravenous administration of erythromycin.
       The principal is the absence of clear criteria for the diagnosis of heart failure, which presents known difficulties in patients with pneumonia. The authors refer to an article by American researchers, but it also lacks a clear description of the identification of this outcome.

The study protocol provides for the probability of the physician deviating from a randomly established empirical antibiotic therapy method for clinical or organizational reasons, and this was not an exclusion criterion. There is no information about the providence of the statistical evaluation of this opportunity confounding factor.

   When considering the data presented in the text and tables 1 and 2, the inconsistency of the number of patients who received different antibiotic therapy regimens attracts attention. The text and table 1 indicate that 650 patients received macrolides, whereas the sum of such patients in table 2 is 734. For fluoroquinolones, this difference is 954 and 994, respectively. This discrepancy may be due to the fact that a significant number of patients received several different macrolides or fluoroquinolones. In this case, the comparison results for stratification by different antibiotics may be unreliable.
    There is a mismatch between the severity of pneumonia in accordance with PSI or CURB criteria and a high mortality rate (Table 1).
     Figure 2 shows the exact days of detection of various cardiac complications. Such accuracy, especially with regard to heart failure, is possible with daily observation by a physician. It follows from the protocol that patients were treated in general wards under conditions of routine medical practice. It is advisable to describe in the methodology how everyday  observation ву physician was carried out. .
    There is no information on the number of cases of deviation from the established treatment scheme by a random method.
     In a secondary analysis of data from a previously conducted study, the authors revealed a previously unknown phenomenon — a significant increase in the hazard ratio of  heart failure when using beta-lactam in combination with erythromycin.
     The authors do not find this clear clinical and pharmacological explanations. The direct toxic effect of these drugs on the myocardium is not described. The presence of hidden arrhythmia is unlikely with such careful medical observation, there are no signs of QT lengthening or a combination in one patient and heart failure and arrhythmia. The effect of the infusion load cannot be estimated, since data on erythromycin dosages are not given.The amounts of concomitant therapy are also unknown and therefore it is impossible to assess the role of drug interactions and the contribution of intravenous infusions volume.
      The authors of the study applied adequate statistical methods to exclude possible confounding factors, but did not reveal their obvious effect on the result.

In our opinion, researchers underestimate the importance of system errors. First of all, the impact of selection biases is not evaluated. Unknown number of patients who were prescribed treatment is not in accordance with the protocol (broken randomization). Patients who received several antibiotics of the same group are not excluded from the study.   The authors point to the likelihood of a more thorough identification by the doctor of cardiac complications in the group of patients treated with beta-lactam in combination with erythromycin. Another reason for increased attention may be the impression of intravenous infusion of the drug. Unfortunately, the design flaws of the study, which was originally planned for another purpose, did not reduce the influence of these factors.  We believe that the weakest link in this work is the choice of heart failure syndrome as an outcome. Physical methods for its detection in acute pneumonia are unreliable, and targeted laboratory and instrumental monitoring was not included in the protocol. With stratification of samples by type of antibiotic therapy, the same patient could fall into different groups. Therefore, the internal validity of their comparison is questionable. Based on the above, we consider it necessary to re-analyze the data, taking into account both cases of violation of randomization, and the appointment of several antibiotics.   The most difficult is to resolve the issue of a more objective diagnosis of heart failure. Assess the possibility of using the Framingham criteria or select only a clear symptom (pulmonary edema e.g.).