The most important finding of the study was that the presence of comorbidities was associated more with potential multidrug-resistant (MDR) pathogens as a cause of CAP than was age. Thus, the authors concluded that “comorbidities rather than age should be considered in the selection of antibiotic treatment.” However, the outcome (in this case, a microbial cause) was not assessed uniformly in all included patients, which is a well-known cause of bias in predictive research.2 Apparently, microbial testing was left to the discretion of the treating physician. This is at least suggested by the pattern of microbial testing: serologic tests in 1,537 patients (44%), sputum cultures in 1,913 patients (54%), and blood cultures in 2,753 patients (78%).3 By ignoring the fact that microbial tests could be different across patients, the authors implicitly assumed that the tests were missing at random. However, in clinical practice, the choice for microbial testing is often influenced by patient and disease characteristics. Therefore, more extensive diagnostic testing in patients with comorbidities may well explain the higher prevalence of potential MDR pathogens in this patient group. Demonstration of comparable diagnostic procedures in patients with and without comorbidities would reduce the likelihood of bias, although even that would not rule out bias completely. It is hardly possible to adjust for this analytically, even when information on why certain microbial tests were (or were not) obtained was available.