Help with correlation scenario

jlo5

New Member
#1
Hi everyone, I was hoping to get some guidance on how I should approach data analysis I'm doing for a research project within my medical subspecialty as I have very little stats experience.
Essentially I have collected data for epistaxis (nose bleed) presentations to emergency departments from 2 hospitals in very different climates. I have collected average daily temperature, humidity, dewpoint data for the dates of those patient presentations as well as demographic/medical comorbidity data for those patients. I am trying to analyze whether there is any correlation between frequency of presentation and those climate variables, e.g. does increasing mean daytime temperature correlate with more epistaxis presentations? Additionally I want to do subgroup analysis to see if climate variables have any impact on datapoints like length of hospital stay. My thought process is to simply combine all the data from both hospitals for those analyses and then I can do a subsequent direct comparison between the two hospitals as they relate to their differing climates.

What would be the best type of test for this/these scenario/s?
I really appreciate any help anyone could lend.

Thanks
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
What are the sample sizes and what proportion report exist a is and is it their primary diagnosis or secondary finding? Also are all trauma patient removed from epistaxis grouped or do they have an indicator variable?
 

jlo5

New Member
#3
Sample sizes are 360 patients from hospital A (smaller regional hospital) and 1260 from hospital B (Tertiary metro hospital). I excluded all trauma and post-operative patients from the dataset because I only want to look at spontaneous/idiopathic epistaxis from the point of view of climatic variables. Epistaxis it the primary diagnoses but I have also collected secondary variables such as anticoagulation/antiplatelet status and other risk factors such as pre-existing hematological disorders, hypertension, age, chronic kidney disease and hepatic dysfunction as well.
 

hlsmith

Less is more. Stay pure. Stay poor.
#4
So you only have data for epistaxis patients? If not, you may also need daily ED census data (a denominator for a rate), so that you understand if the uptick are related to patient volume or other reasons. Though, if ED is at capacity you may get a few left without being seen.
 

Miner

TS Contributor
#5
I would think that only a very small percentage of people with a nosebleed actually go to an ER. Most would just put tissue in their nose until it stops. This non-response bias would be very large. How do you intend to deal with that?