The Effect of 12 Hour Shifts, Time of Day, and Sleepiness on Emotional Empathy and Burnout in Medical Students
Round 1
Reviewer 1 Report
Interesting and fairly well-written paper, presentation should be improved though. Here are some suggestions:
Introduction (this section needs some work)
30-32: Adding a definition of fatigue would help the reader to see the difference 47-48: Phrasing is unclear (triple the number compared to what?) and misrepresents the study's findings (different odds of fatalities dependent on number of extended duration shifts). 55: There are no sources for this statements, rephrase or introduce later when sources are discussed 63-64: The source is a study about empathy decline in med school, not medical careers as such 72: The introduction doesn't include information on sleepiness and behaviour in general. Add info or rephrase. 85-87 and 91-93: Wrong source [29], none of this information in included in the quoted study. Serious error. 104-6: Rephrase, sentences imply causality between that cannot be established with this study design. Serious error.
Results
Descriptive stats: Was there no sociodemographic information on the participants other than gender? If there was: report it. Drastic lack of any statistical information about the mixed multivariate ANOVA model. At least mention whether the requirements for this method were met. 135-6: The mere calculation of a Pearson correlation tells you nothing about whether the relationship is linear.
Discussion
188-89: Rephrase, causality implied.
Materials and Methods
197: Geographical location of med school is of interest 203-4: Between 1700 and 1000? 204-206: Sentence is irrelevant. If you are interested in med students as a population than the EMTs' sleepiness holds no interest.
Author Response
Thank you very much for the thoroughness of your reading and your comments. I have incorporated them into the paper, and my responses to your suggestions are below (in all caps). In the revised version of the paper, your suggestions are tracked/highlighted in gray. Thank you for your time and suggestions!
Introduction (this section needs some work)
30-32: Adding a definition of fatigue would help the reader to see the difference (DONE, From same paper that defined sleepiness)
47-48: Phrasing is unclear (triple the number compared to what?) and misrepresents the study's findings (different odds of fatalities dependent on number of extended duration shifts). CLARIFIED from original source
55: There are no sources for this statements, rephrase or introduce later when sources are discussed REMOVED 57-58
63-64: The source is a study about empathy decline in med school, not medical careers as such REVISED 65-67
72: The introduction doesn't include information on sleepiness and behaviour in general. Add info or rephrase. CHANGED TO PERFORMANCE
85-87 and 91-93: Wrong source [29], none of this information in included in the quoted study. Serious error. REMOVED; THIS WAS AN ERROR OF READING AND CONSOLIDATING INFORMATION AND NEVER SHOULD HAVE HAPPENED!
104-6: Rephrase, sentences imply causality between that cannot be established with this study design. Serious error. CHANGED THE PHRASING TO USE AFFECT AND RELATED TO.
Results
Descriptive stats: Was there no sociodemographic information on the participants other than gender? If there was: report it.
THANK YOU! We collected more but did not include that in the data analysis, so we did not include that in the paper. We have added the age demographic to the paper. Lines 110 and 111
Drastic lack of any statistical information about the mixed multivariate ANOVA model. At least mention whether the requirements for this method were met. THANK YOU. ALL REQUIREMENTS WERE MET )NORMALITY, HOMOSCEDASTICITY, ETC) AND THAT INFORMATION HAS BEEN INCLUDED. Lines 121-122
135-6: The mere calculation of a Pearson correlation tells you nothing about whether the relationship is linear. STRONGLY DISAGREE ON THIS POINT. IT IS A MEASURE OF LINEAR RELATIONSHIP BETWEEN TWO CONTINUOUS VARIABLES (AS OPPOSED TO CURVILINEAR, ETC). THE SCATTERPLOT AND THE DATA ANALYSIS DEMONSTRATE THAT A LINEAR RELATIONSHIP EXISTS BETWEEN THE VARIABLES.
SOURCES: http://onlinestatbook.com/2/describing_bivariate_data/pearson.html, https://statistics.laerd.com/statistical-guides/pearson-correlation-coefficient-statistical-guide.php
http://www.j-pcs.org/article.asp?issn=2395-5414;year=2018;volume=4;issue=2;spage=116;epage=121;aulast=Yadav
Discussion
188-89: Rephrase, causality implied. CHANGED TO AFFECTS 193-194
Materials and Methods
197: Geographical location of med school is of interest (FIXED added southern , urban, US)
203-4: Between 1700 and 1000? FIXED to 2000
204-206: Sentence is irrelevant. If you are interested in med students as a population than the EMTs' sleepiness holds no interest. REVISED
Reviewer 2 Report
In the introduction, consider also providing a definition of fatigue. For example: "whereas fatigue is defined as... sleepiness is the tendency or increased propensity to fall asleep..."
Perhaps outside the scope of this paper, but it seems that the literature on compassion fatigue (as a potential risk of too much empathy) could be relevant here.
Be careful not to introduce acronyms before spelling them out (SSS etc.)
In table 1 clarify what pre and post mean
What were the shift hours? Set (e.g. 7-7, or variable?)
Where any other control variables included in the model?
Why is it surprising that there was no gender difference for empathy? If one was anticipated please explain why.
Time of day is a complicated variable as it will mean very different things based on time of last sleep period (i.e. a participant who is coming on shift will have a very different experience than a participant coming off shift). Where interactions between shift and time of day explored?
Author Response
Thank you so much for your comments! I tried to elaborate as much as possible and incorporate your suggestions that you made. I really appreciate your comments!
Your comments are pasted, and my reply is in all caps (your revisions are yellow on the revised version of the paper).
In the introduction, consider also providing a definition of fatigue. For example: "whereas fatigue is defined as... sleepiness is the tendency or increased propensity to fall asleep..." I REALLY LIKED YOUR PHRASING AND WAS HAPPY TO INCORPORATE IT!
Perhaps outside the scope of this paper, but it seems that the literature on compassion fatigue (as a potential risk of too much empathy) could be relevant here. THIS IS ANOTHER HUGE ISSUE THAT I HOPE TO INCORPORATE IN THE FUTURE, BUT WE DIDN'T SPECIFICALLY MEASURE COMPASSION FATIGUE, SO I AM HESITANT TO USE IT HERE. THAT IS SOMETHING WE ARE CURRENTLY STUDYING.
Be careful not to introduce acronyms before spelling them out (SSS etc.) THANK YOU! FIXED.
In table 1 clarify what pre and post mean FIXED (LINES 113-116)
What were the shift hours? Set (e.g. 7-7, or variable?) FIXED (207-208)
Where any other control variables included in the model? THERE WERE NO CONTROL VARIABLES FOR THIS STUDY; WE JUST COMPARED PARTICIPANTS TO THEMSELVES. HOWEVER, WE ARE WORKING ON THE NEXT STEP FROM THIS STUDY AND ARE INCORPORATING A CONTROL GROUP.
Why is it surprising that there was no gender difference for empathy? If one was anticipated please explain why. I REMOVED THIS INFO, BUT THERE ARE SOME STUDIES THAT SHOW WOMEN AS MORE EMPATHETIC. HOWEVER, AFTER REVISITING THIS I SAW THAT IN MEDICAL STUDENTS THERE DOES NOT APPEAR TO BE A SIGNIFICANT DIFFERENCE.
Time of day is a complicated variable as it will mean very different things based on time of last sleep period (i.e. a participant who is coming on shift will have a very different experience than a participant coming off shift). Where interactions between shift and time of day explored?I DID EXPLORE THIS, AND IT WAS VERY COMPLICATED, BUT WE FOUND THAT TOD HAD AN EFFECT ON EMPATHY, BUT ONLY PRIOR TO THE SHIFT IN THE EVENING. AFTER THE DAY SHIFT THERE WAS NO EFFECT. I ADDED A BIT ON THIS (129-130)
Round 2
Reviewer 1 Report
135-6: The mere calculation of a Pearson correlation tells you nothing about whether the relationship is linear. STRONGLY DISAGREE ON THIS POINT. IT IS A MEASURE OF LINEAR RELATIONSHIP BETWEEN TWO CONTINUOUS VARIABLES (AS OPPOSED TO CURVILINEAR, ETC). THE SCATTERPLOT AND THE DATA ANALYSIS DEMONSTRATE THAT A LINEAR RELATIONSHIP EXISTS BETWEEN THE VARIABLES.
Dear authors,
I think this is a simple misunderstanding. Yes, the Pearson correlation is of course the appropriate measure of association for linear relationships. The problem is your phrasing: "A Pearson correlation was conducted to determine if there were a linear relationship between burnout, sleepiness, and empathy." You make it sound like you can DETERMINE whether there is a linear relationship by calculating the Pearson correlation, when in fact it is the other way around (a linear association is the reason for choosing Pearson). You look at the association pattern first (e.g. scatterplots, which you have done) to assess whether you're dealing with a linear association, then you choose the appropriate measure (Pearson). In terms of figures it is absolutely possible to calculate Pearson for a curvilinear association (your statistics programme won't mind and just give you the number), but it would be the wrong parameter for this type of association, so you have to make sure that the relationship is linear BEFORE you calculate your measure of association.
Please consider rephrasing along the lines of "A Pearson correlation was conducted to measure the strength of the linear association between..."
Author Response
Thank you for the explanation. That makes much more sense, and I will definitely adopt the language you suggest (lines 140-141). Thank you for taking the time to go through that point!
Reviewer 2 Report
Nice work! I have no further comments.
Author Response
Thank you for your time and expertise!
