WHICH Test Analysis Can I use ?

#1
I am kindly requesting your kind advise. I am doing an analysis for collected data of experimental design using control group, but my participants in the pre-test group are not the same participants in the post-test (1 group for pre-test & 1 group for post-test & they are both don't overlap) because they are patients admitted to ICU (intensive care unit).

So, I have 4 different groups (1 pretest intervention group(182 pt), 1 posttest intervention group(188pts), 1 pretest control group(169pts) and 1 posttest control group(148pts)) and I have to test the differences within groups, I mean within intervention and within control (between pre and post). First I tested them with chi-square but I thought that I have to use McNemar test but McNemar is not applicable in this case because the participants are not the same in pre and post test. So I am thinking now about regression analysis ?


My question is: which test can I use in this case ? & how can I apply it in the SPSS if I have to use regression analysis ?

Note: the data is normal distributed and I have categorical and numerical variables.

Really I am appreciating your kind advice & help.
 

Karabiner

TS Contributor
#3
Could you explain with a little more detail how the experiment
was actually performed? E.g. are these 2 different wards?
Since the groups don't overlap, I suppose that the statistical
analysis is not too complicated to perform; but how to reach
valid conclusions about the impact of the experiment might
perhaps be a little bit more demanding. For example, do you
already have a concept for which conditions/groups you want to
compare?

With kind regards

K.
 
#4
First of all thank very much for your help.

My main goal is to study the impact of nurses training on patients outcomes as adverse events. So, I assess the occurrence of adverse events in patients admitted to ICUs (for 6 weeks) in both groups (intervention and control) before making the training for the nurses. Then again I assess the patients after the training is done for the nurses caring of the patients admitted to the ICUs (for 6 weeks). Because they are inpatients (critically ill patients) I could not admit them again after training. So I took the patients admitted to ICUs in that time (after training time).

Note: I took in my account the homogeneity of the four groups (intervention pre & post and control pre & post) for example, by evaluate the disease severity and antibiotics use.

Best regards
 

Karabiner

TS Contributor
#5
Not completely sure, but perhaps one could consider to perform
binary logistic regression with a patient's outcome as dependent
variable and 3 predictors: nurse group (experimental versus control) and
time (baseline versus post-intervention), and their interaction.
If the interaction was significant, then the difference between
pre and post was different between nurse groups (or, to put it
another way, the difference between nurse groups was different
after intervention compared with baseline).

With kind regards

K.