Need help to figure out a study design

#1
Hello everyone!
I'm currently writing a systematic review and I just can't figure out the study design of one of the included studies. The study is from Davis et al. (2002) with the title "Swallowing with a tracheostomy tube in place: does cuff inflation matter?". The aim of the study was to assess the frequency of aspiration with an inflated versus a deflated cuff and to see if the texture swallowed made any difference.
The authors included 12 patients and evaluated swallowing of four viscosities (e.g. thin liquid, pureed food) with the cuff inflated and with the cuff deflated (so eight evaluations in total). The assessor who evaluated aspirations was blinded to cuff inflation status. The order of testing is not described in the text. They used logistic regression analysis and the McNemar test for analysis of the two groups with (p<0,05).

I think it should be a cohort study with the patients being their own controls. Is that possible?
In the study there's only written that's it's a prospective, blinded analysis.

I'm very thankful for any help!
If there are informations missing let me know :)
 

hlsmith

Less is more. Stay pure. Stay poor.
#4
I don't think there is any ambiguity - it just doesn't fit into the pretty little randomized control trial design.
 
#5
@hlsmith thank you for your reply! There's another study by Suiter et al. (2003) where they tested two consistencies (in random order) two times in three conditions (so: two trials with liquids with inflated cuff, two trials with liquids with deflated cuff, two trails with liquids with deflated cuff and speaking valve and the same procedure for pureed food). The study is pretty much as the other study by Davis et al. (2002): one group where all patients participate in all trials. There are two systematic reviews which classify this study by Suiter et al. as a cohort study. But shouldn't this be a cross-over study as well?
 

hlsmith

Less is more. Stay pure. Stay poor.
#6
Yeah, you are going to see people describe things in varied ways. I start by saying was there an intervention/action or not. If not it is observation, if so - I use the blanket term interventional. Observational can be retrospective or prospective, etc. I typically group 'cohort' into observational, where a cohort is determined and examined, but no intervention. In that group you would also have case-control, etc. The intervention group you tease out by randomization, blinding, arms, and control groups, etc.

Per your description, I would call it a Multi-armed full cross-over trial with randomized treatment ordering. But there are many descriptives that can be used.
 
#7
Oh okay that's good to know! So there are basically multiple designs that would be "right"? Which tool would you use for assessing risk of bias? ROBINS-I because the study isn't randomized (but that's not even possible if there's only one group) or RoB2 for crossover studies? The study by Davis is (probably) not randomized but it's not described and in the study of Suiter they only randomized the order of the drinks/food given but not of the cuff pressure status (inflated/deflated). Thank you for your help! :)