Overal statement or summary:

This is an article evaluating the relationship between time-to-antibiotic administration and hospital stay and adverse outcome (hospital mortality or ICU monitoring) in adult cancer patients with febrile neutropenia treated with chemotherapy. Additionally it is investigating possible confounders.

Overall strenghts and impact

The abstract is clear and the background gives a good explanation about what is already known and highlights the missing evidence about the association between time to antibiotics and mortality or length of hospital stay, what will be investigated by the paper. The research question is clear as well. Within the scope of my current knowledge relevant and recent references are used. In total the methods ar well described and replication of the study seems to be possible. The authors take into account various confounders that may influence the choosen outcomes. The MASCC risk index score is a reliable choice to define risk groups. To me outcome and confounding measurements and how they are displayed in the result section is apropriate. The objectives are discussed and restated in the conclusion section and the conclusion made for length of hospital stay is appropriate.

Major points:

1. Analysis of adverse outcome: Is it appropriate to conduct an analysis with only 5 patient having an adverse outcome? As power may not be adequate (as it is also stated by the authors) to my opinion the authors should rather suggest another study with more power investigating this, than come to the same conclusion as for hospital stay in the last sentence of the paper. Maybe a power calculation could be added.

2. For me the section in the discussion (beginns at page 5) about known source of fever ist not clear. To my opinion the written explanation does not explain or evaluate the finding of an association between source of fever and leght of hospital stay. What conclusion do the authors draw from the shown association? Do they justifiy empiric antibiotic therapy with this association?

3. We do not know how some of the variables are defined (e.g. comorbid illness, chonic renal failure) and on what timepoint the blood results were measured. I recommend to give definitions. Also the definition for "known source of fever" could be stated more clear, because it is not absolutely clear whether this includes only bacteremia from blood samples at admission or also later. If later detection of infections are included this should be inverstigated as outcome as well.

Minor points

1. Title: It is simple, but does not attract attention. Maybe one could make a statement about what outcomes are accessed, maybe even something about the results?

2. The description in methods lets suggest that all eligible patients were included, but I do miss a comment about missing data (even if there was none).

3. If there is data, it would be interesting to have a variable for "severety of illness" at presentation of a patient with febrile neutropenia, as this can have quite an influence on time-to-antibiotics. (E.g. shock (although hypotension is included in the MASCC score), general state).

4. At table 3 one could add that this were the results of univariate analysis in the title and delete "patients with febrile neutropenia" because this is clear. (like in table 2)

Things I would discuss with mentor: I am not sure wheaterh Pearson correlation is the correct way to assess the correlation between time-to-anitibiotic administration and length of hospital stay as I suggest that lenght of hospital stay is not normally distirbuted. If log transformation was done this should be stated.

Maybe one could have included more variables into the multivariable analysis (not only p<0.5) although MASCC includes some of them (e.g. age, BUN as a measurement of dehydration).