turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Data Mining
- /
- Dynamic Summary Table

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-14-2015 02:29 PM

Hi All,

I never worked on a Dynamic Summary Table. Can anyone help me with some example code/reference/suggestion on how to produce a Summary table like this:

Significant variables | ||||||

Age | Number of Patients (%) | p value | Unadjusted odds ratio | 95% Confidence interval | ||

65 | 92 (30.5) | p<.01 | ||||

65-75 | 65 (45.4) | 0.25-0.48 | ||||

75-84 | 45 (41) | |||||

85+ | 30 (23.8) | |||||

No. of Visit | ||||||

1 | ||||||

2 | ||||||

3 | ||||||

4 or more | ||||||

mean (standard deviation) | ||||||

median (range) | ||||||

Diagnosis | ||||||

Breast Cancer | ||||||

Lung cancer | ||||||

Cancer of the pancreas | ||||||

Congestive heart failure | ||||||

Copd | ||||||

Cancer of the cervix | ||||||

Others | ||||||

Admission | ||||||

Yes | ||||||

No | ||||||

Day of Visit | ||||||

Monday | ||||||

Tuesday | ||||||

Wednesday | ||||||

Thursday | ||||||

Friday | ||||||

Saturday | ||||||

Sunday |

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-14-2015 02:34 PM

Several procs will give you results similar to that.

I would start with proc freq or proc univiriate.

Personally I like univariate because you can output the results and then print them in a convenient way.

Curious, did you design that output? Are you converting it from some report, or what you up to?

Thanks,

M

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-14-2015 02:41 PM

Thanks for your help. Yes I just designed that output the way I want in an excel spreadsheet. I just don't know how to get that output from my data. Your suggestion would be a good place to start with.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-14-2015 02:46 PM

When I want to convert my Excel reports to a SAS proc I use proc report. It is kind of old school and has a little bit of steep learning curve, but I found it really useful.

Proc report with a custom format (through proc format) can do 90% of the things you have on that report.

For the confidence interval you might have to use a stats proc like univariate or freq.

Google proc report examples and see if that suits your report needs.

Here a paper that seems like a tood start: Proc Report beyond the Basics

Good luck!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-14-2015 11:16 PM

Hi **MiguelMaldonado**,

I actually want to import excel dataset to sas, analyse the data in SAS, and output in pdf or in other format which will look like what I posted.

Thanks,

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-15-2015 10:30 AM

That's a bunch of different steps just a quick outline below. It can be broken down further as well.

1. Import data and verify data is imported correctly - e.g. numeric variables are numeric and chars are chars

2. Format the data for the analysis required

3. Running analysis for variables to determine results required

4. Converting code from step 3 into macro

5. Running a macro

Where are you in the process?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-14-2015 03:39 PM

Including a p-value indicates some sort of statistical test is to be performed. What do want to test? Compared to what population?

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-14-2015 04:09 PM

It looks like you'll need a macro - I'm assuming you're running a univariate logistic regression since you state unadjusted odds ratio -though you could obtain that from proc freq as well.

Search Lexjansen.com for clinical reporting macros and you should find something pretty close. A lot of it depends on how your data is structured and what you want as reference variables and such.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-14-2015 04:10 PM

Also, what does dynamic refer to here? I'm not sure this belongs in Data Mining but ODS reporting instead.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-14-2015 11:12 PM

Hi Reeza,

I am not quite sure if I should call it 'dynamic'... I just thought this type of report is called 'dynamic summary report'....may be I am wrong!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

05-15-2015 10:14 AM

proc sql + proc transpose to do such table/listing .