test for difference between matched pairs

[Merdock]

I performed a 1:1 case-control match (i.e., one control for each case) and I want to compare cases with controls on a continuous, as well as categorical outcome variable. The cases are people with a certain genetic disease, while the controls are matched on various criteria (age, gender, race) thought to be relevant to the outcome variables (markers of poor kidney health) as it is suspected that the genetic disease increases the risk of kidney disease.

My question is: for the matched analysis, what should I use to evaluate differences between matched pairs (i.e., comparing cases with controls) on specific characteristics (I want a table like shown below)? Is it ok to use signed ranked test (if variable is continuous) and McNemar test (if categorical)? If yes, does anybody have any examples/sample code for how to apply these tests on my specific data?  

Characteristic	Disease (Case)	No Disease (Control)	P-value
Serum Creatinine Value, (mg/dL)
N
Mean (SD)
Median (IQR)
CMV Infection, n (%)
Yes
No

My final dataset looks like this, where disease=1 denotes presence of disease (case), and group links the case with its matched control:

data exam;
	input id $ age race $ gender $ disease $ group;
cards;
data exam;
	input id $ age race $ gender $ creat cmv $ disease $ group;
cards;
0001 19 2 2 23 1 1 1 
0017 19 2 2 28 0 0 1 
0002 10 2 1 43 1 1 2 
0005 10 2 1 26 1 0 2 
0060 15 2 2 54 1 1 3 
0010 15 2 2 43 0 0 3 
0018 14 2 2 120 1 1 4 
0105 14 2 2 29 1 0 4 
0008 18 2 1 36 1 1 5 
0022 18 2 1 57 0 0 5 
0548 15 2 1 49 0 1 6 
0052 15 2 1 100 1 0 6 
0059 13 2 1 95 0 1 7 
0982 13 2 1 65 1 0 7 
0047 12 2 1 20 1 1 8 
0084 12 2 1 39 0 0 8 
0680 17 2 2 78 0 1 9 
0042 17 2 2 110 0 0 9 
0984 15 2 2 66 1 1 10 
0007 15 2 2 85 0 0 10 
0021 16 2 1 73 0 1 11 
0873 16 2 1 62 0 0 11 
0193 17 2 1 71 1 1 12 
0178 17 2 1 76 0 0 12 
;
run;
proc print data=exam; run;

Thanks for any input/help!

[PaigeMiller]

I don't see anything that looks to me like "matched pairs" in your data or in your description.

Typically, such a "matched pairs" study has the same subject receiving both treatments, such as vehicles (the subject) with two different types of tires installed (the treatment).

Please explain further about what are the "matched pairs" in your study.

[Merdock]

@PaigeMiller, sorry I should have clarified. The two observations in a matched pair need not refer to the same subject. For example, case-control studies that match a single control with each case yield these matched-pairs data. By "matched pairs" I mean that the cases and controls were matched 1:1 on age, gender, race. So each case got one control that was similar to the case in terms of age, gender, race. For example, a matched pair would be IDs that have the same group number, i.e. id#0001 and 0017 (group#1), or 0002 and 0005 (group#2), or 0060 and 0010 (group#3) etc. Hope this makes sense. Basically, all I know is that we have some registry data (retrospective) on outcomes of patients with this specific disease and the investigator is interested in matching one non-diseased patient to every disease patient and then comparing the two groups on specific outcomes/variables (table with n, mean, median, count (%) and pvalue).

I don't have any experience with matched case-control or cohort studies but from what I've been reading, it seems that any statistical analysis that you do on the data after this matching has been performed, should account for the matching?

[PaigeMiller]

But in this case you would not do a statistical "matched pairs" t-test. I would do an ordinary t-test to see if the means are the same. Is that what you want?

[Merdock]

I guess that's the question and what I'm trying to figure out. Because I don't have any prior experience with this and from what I've been reading, it seems like opinions are divided when it comes to whether ignoring the matching or not. Same argue you should account for the paired nature of the matched data. Others disagree. It's still very much an area of debate. If I am to ignore the matching and do ordinary t-test and chi square test, I know how to do that but I don't know what to do if I need to account for the matching.

[PaigeMiller]

I don't have experience with this type of matching either, nor do I have any understanding of what leads to poor kidney health. My thought though is that unless you control for other factors, such as weight and ethnicity and diet and probably dozens of other things (which really isn't realistic to do), these are not treated in statistics as matched pairs for a matched pair t-test. But I'm going to wait until to hear from others who might have experience with this type of matching.

[Merdock]

Got it, thank you for your input, I appreciate it. Looking forward to hear what others might advise.