Incremental yield differences between two cultures

[Hatim Tareq]

Hi All,

I would like to see if there is a significance difference in the Incremental yield percentage between the two culture?

culture with standard media
	Specimen 1 N= 236	Specimen 2 N= 236	Specimen 3 N= 236
Number and % culture positive	19(8.1%)	15(6.4%)	18(7.6%)
Additional number positive	19	3	2
Incremental yield	79.2%	12.5%	8.3%

culture with supplemented media
	Specimen 1 N= 236	Specimen 2 N= 236	Specimen 3 N= 236
Number and % culture positive	21 (8.9%)	18(7.6%)	16(6.8%)
Additional number positive	21	2	3
Incremental yield	80.8%	7.7%	11.5%

I was a bit confuse how to do this, and which test should I use. so I have no clear idea, or code to show what I have done.

Thank you in advance for your help.

Regards,

[Hatim Tareq]

I did a ttest, but not sure if its the correct way of answering this question (I re-arrange the data in one data set), and did the following:

ttest IY, bu(Sup)

[Carlo Lazzaro]

Hatim:
were the specimens 1,2,3 for both coltures taken from the same patients in different points in time?
Kind regards,
Carlo

[Svend Juul]

To get an informed response, you must explain more about the study. To Carlo's question I would add:

Do the N=236 correspond to 236 different patients/persons/guinea pigs or something else?

What is "incremental yield percentage"? I have seen it mean the extra cases identified by a second test.

Do the two cultures mean that the same sample (of, for example, sputum) is split in two, and cultured by different techniques? If so, we have a matched design. Is there an ID that identifies the origin of each sample.

There may be other information that is useful. Not knowing the basics about the study gives too much room for guessing. That is probably the reason why you did not get a response sooner.

[Hatim Tareq]

Hi Carlo,

Yes, the specimens taken from the same patients in different points in time.

[Hatim Tareq]

Hi Svend,

Yes, there are 236 different patients.

Yes, the two cultures mean that the same sample is split in two. and there is an ID that identifies the origin of each sample.
There are many other information (variables) in the study.

Regards

[Svend Juul]

I have not worked with this before, but the incremental yield concept is especially used in the bacterial diagnosis of tuberculosis infection. The definition of a case is that the person has been diagnosed with any method. In this case 24 persons were diagnosed after 3 examinations with method A and 26 with method B. We need to know how many were diagnosed with any method; it is at least 26, but it could be more. Let's assume it is 28. I think it must be wrong to use 24 as denominator for method A and 26 for method B; it gives a cumulated yield of 100% after 3 examinations for both methods, and that cannot be right.

Generate a table like this for the first examination:

                 | Method A               |
Method B         |  Positive    Negative  |      Total
-----------------+------------------------+------------
        Positive |        18           3  |         21
        Negative |         1           6  |          7
-----------------+------------------------+------------
           Total |        19           9  |         28

This is a matched design; the same persons have been examined by two methods. Use McNemar's test; it is available in mcc and mcci. The language is that of case-control studies, but it corresponds to the first table.

. mcci 18 3 1 6
                 | Controls               |
Cases            |   Exposed   Unexposed  |      Total
-----------------+------------------------+------------
         Exposed |        18           3  |         21
       Unexposed |         1           6  |          7
-----------------+------------------------+------------
           Total |        19           9  |         28

McNemar's chi2(1) =      1.00    Prob > chi2 = 0.3173
Exact McNemar significance probability       = 0.6250

You probably want to repeat this for the cumulated number of cases discovered after 2 and after 3 examinations.

Hope this helps