ASA Connect

I recommend talking with the clinician about how they intend to use the information from the study.  They may be looking at decrease in heparin dose (continuous outcome) with increasing amounts of the other drug, and may not be interested in heparin as a binary outcome.

Hi,

The question asks if use of the drug results in an overall reduction in heparin infusion rate. The data you received were continuous. I think the heparin data must have at least pre-drug and post-drug values so that a change from pre-drug heparin infusion can be calculated (post-drug - pre-drug). If that's so, then any changes < 0 indicate a reduction in heparin use while changes >= 0 indicate no reduction in heparin use. You now have a binary response. Of course, there are other considerations you'll need to be aware of. Before starting though, have a good talk with the clinicians to make sure they understand what it is they want and that you agree with their idea. Clinicians are not always the most statistically savvy people out there, although there are exceptions. Good luck!

I will suggest you to talk with your clients.  The terms that physicians expressed sometimes was not what a statistician thought.  You will clarify what the "heparin infusion rates" means.  It might mean the percentage of patients who used certain dose (or above) heparin.  It also might mean the dosage of the heparin infusion (rate might mean per min).  Then based on the clinical definition, you will choose methods to analyze the data. 

In addition, I don't think median will be a good way to define the cutoff.  A clinical meaningful cutoff will work better. 

Hi Karl,

This is one of those questions that seems straightforward until you dig into the details. I agree with the comments of others suggesting you clarify exactly what this clinician is expecting to see in terms of ‘decreased infusion rates.’  I’ll add a data consideration, having worked on studies with infused drug previously.  Look carefully at your dataset.  Infused drugs may be provided in different concentrations and it is not unusual for a patient to start with one concentration but change to another concentration at a different infusion rate.  It is also not uncommon to have dosing changes as part of the standard of care, so be careful to address this possibility, too.  Good luck with the project!

The question is a bit fuzzy with regard to design and data collected.  Is there a control group with no intervention? Does the intervention being given have different doses? Do you have data for both baseline and followup for your outcome variable (which I assume is continuous)--in which case you need to decide how to treat the baseline value (as a covariate) or using it in a change from baseline, or repeated measure. Is the time between baseline and followup a constant?  So sitting down with the clinician is a must!