topic Pooling results after multiple imputations in Statistical Procedures

Pooling results after multiple imputations

raj00728 — Sat, 24 Feb 2018 01:40:44 GMT

Hi,

I am new to PROC MI. I have 500 unique patient data with 30 variables, of which 15 variables have missing values. Using multiple imputations procedure, I was able to impute 15 different variables (some binary, some ordinal and a few nominal); 30 imputations in total. So my final data set contains 15000 observations. Then I created a few summary variables from the "new imputed variables" to be used as covariates in the regression models. Now, I am unable to conduct univariate, bivariate and multivariate analyses using standard SAS procedures and then combining them using PROC MIANALYZE. I want to do something like the following using the multiple imputed data set (30 imputations):

1. I want to have standard table 1 to describe my sample. I want to report: n (%) for categorical variables and mean (SD) for continuous.

2. I want to have standard table 2 to describe my sample by my exposure (2 level) and conduct a Chi-square for categorical variables and t-test for continuous.

3. I want to have standard table 3 which compares adjusted LS Means and SE for my continuous OUTCOME variable across the exposure, adjusting for several covariates.

Could someone please guide me? Thanks.

-Raj

Re: Pooling results after multiple imputations

Brad19 — Wed, 25 Mar 2020 14:14:24 GMT

Hey,

Did you figured out pooling frequency and means (SD) after imputation?

Brad.

Re: Pooling results after multiple imputations

Rick_SAS — Wed, 25 Mar 2020 18:54:56 GMT

The best way to get a quality response is to post sample data, the SAS code you are using (even if it is wrong), and an explanation of what you are trying to accomplish. Even if you don't know enough to write code, you can describe the problem and someone can point you to documentation and examples.

Re: Pooling results after multiple imputations

Brad19 — Fri, 03 Apr 2020 12:41:43 GMT

I think it does not require to send you the codes or dataset. One who has done it write the syntax to ease others using SAS community.

Re: Pooling results after multiple imputations

ChuksManuel — Mon, 06 Jul 2020 01:40:34 GMT

Hello,

Did you finally succeed in pooling your imputed dataset through sas? It seems SPSS has an easy syntax to pool imputed dataset according to Rubin's rule but i can't find any syntax for this in SAS.

Re: Pooling results after multiple imputations

ChuksManuel — Mon, 06 Jul 2020 01:42:04 GMT

Hello Brad,

Did you make a find a syntax for pooling imputed datasets?

Re: Pooling results after multiple imputations

ChuksManuel — Mon, 06 Jul 2020 02:20:39 GMT

This paper was helpful https://www.lexjansen.com/pharmasug/2013/SP/PharmaSUG-2013-SP03.pdf