July 12, 2017




Antimicrobial resistance is a growing problem in medicine. Previously, resistant microorganisms were combatted with the development of new antibiotics (1). The evaluation of their effectiveness through RCTs was based on several endpoints, which have not been evaluated extensively for their validity.


Timsit et al. have reviewed the various endpoints that randomised controlled trials (RCTs) can employ as markers for effectiveness of antibiotic therapy in the context of severe infections. They identify that there are currently no well-validated measures. They discuss the advantages and disadvantages of examples of such endpoints.

Mortality on the surface is the most easily measured, but there are many confounding factors in critically-ill patients, with lack of improvement not always down to lack of cure of the infection per se. Current standards for demonstrating non-inferiority between drugs may be too high and that small decreases in treatment effects required may lead to more manageable numbers needed to achieve adequate study power. Using clinical cure has problems due to lack of standard definition and level of subjectivity. Critically ill patients have so many confounding variables to ‘cure’, leading to bias and variability. Biochemical markers and microbiological cure is advantageous in its objectivity, however there are difficulties in isolating organisms and correlating clinical improvement against microbiological clearance. Days free of antibiotics or organ-failure are not sensitive indicators. Using safety endpoints are difficult in the critically ill due to the patients’ high propensity for adverse events.

Emerging antimicrobial-resistance is an additional confounding factor in itself as it is impossible to determine susceptible organisms prior to randomisation and if done so, critical time frames may be missed.
The authors go on to discuss some possible solutions. Composite endpoints combine several different elements, however there are difficulties in their construction and interpretation. They cover the desirability of outcome ranking (DOOR), whereby higher scores are assigned to patients with more desirable outcomes, and the response adjusted for the duration of antibiotic risk (RADAR) based on for how long the antibiotic was taken for (2).  Advantages include a reduction required in sample sizes. They recommend caution, and clear definition of a statistical analysis plan prior to analysis to avoid manipulation through choice of categories. Hierarchical Nested Designs are specifically for multi-drug-resistant organisms (MDROs) and involve a dichotomous endpoint, whereby initially non-inferiority is tested for organisms susceptible to the control drug (4). Then superiority testing for resistant organisms is carried out. So far it has not been used for any RCTs.
The authors further identify that the underlying high mortality rate makes analytical evaluation of non-mortality endpoints challenging, including death before cure. Using competing risk methods can aid when death results in being a competing event for cure. Using such a multistate model allows for patients dying from an underlying illness whereby infection may or may not be contributory. Finally, they suggest that secondary endpoints include level of resistance development, impact on the microbiome and super-infections. Few RCTs have been carried out so far, and authors identify this need to be incorporated during design phase of trials.
Strengths and weaknesses 
• As a review paper, it is useful as it is addressing a crucial element in the structure of the RCTs evaluating the effectiveness of antibiotics in severe infections.
• It identifies both advantages and disadvantages for each endpoint clearly.
• The study paper is accessible and understandable, with useful summary tables.
• It does not include any mention of endpoints that have been useful in seminal trials with less-severely ill patients that may potentially be developed to accommodate those in the ICU setting.
Investigators need to agree on what is appropriate. The most useful suggestion is combining several endpoints in a multi-level study.
Article review prepared and submitted by EJRC members Drs Bryony Shelton and Hatem Soliman Aboumarie.


1. Norrby SR, Nord CE, Finch R. (2005). Lack of development of new antimicrobial drugs: a potential serious threat to public health. Lancet Infectious Diseses; 5(2):115-119.
2. Evans SR et al. (2015). Desirability of Outcome Ranking (DOOR) and Response Adjusted for Duration of Antibiotic Risk (RADAR). Clin Infec Dis; 61(5):800-6.
3. Huque MF et al. (2014). Hierarchical Nested Trial Design (HNTD) for demonstrating treatment efficacy of new antibacterial drugs in patient populations with emerging bacteriall resistance. Stat Med; 33(25): 4321-4336.

Comment on this news