topic Re: Correct SAS procedure to compare tumor sizes for two groups with gold standard in SAS Procedures

Correct SAS procedure to compare tumor sizes for two groups with gold standard

smunigala — Mon, 21 Nov 2016 18:32:25 GMT

Hi all!

I have a data set with tumor sizes (in centemeter) measured by physicians for different patients. The data set is divided into two groups (randomized). So each patient is randomized to either group 1 or group 2. After the tumor size is predicted by the physician, it is compared to the gold standard (automated computer generated size).

I need to determine which group has a better accuracy/ or prediction closer to gold standard.

Can anyone suggest me a SAS procedure to compare two groups relative to the gold standard tumor size prediction?

Thank you so much!

Sat

Re: Correct SAS procedure to compare tumor sizes for two groups with gold standard

ballardw — Mon, 21 Nov 2016 18:41:56 GMT

Do you have the result of the "gold standard" comparison? How is the result recorded?

It may be that a good starting point would be a simple t-test of the mean difference percentage with the "gold standard".

Re: Correct SAS procedure to compare tumor sizes for two groups with gold standard

smunigala — Mon, 21 Nov 2016 18:44:22 GMT

Thanks for the quick response. Yes, I do have the "gold standard" as size of tumor (in centemeter). Tumor size is measured in centemeter for group 1, group 2 and for "gold standard".

Re: Correct SAS procedure to compare tumor sizes for two groups with gold standard

Reeza — Mon, 21 Nov 2016 18:44:45 GMT

You have continuous data.

I would assume you would be creating a variable is the difference between the tumour and gold standard. And then evaluate the two groups by a variety of methods. A simple T-Test or proc NPAR1way to look at the difference between the distribution of the two groups.