The SQL Procedure

Example 14: Matching Case Rows and Control Rows

Procedure features:

joined-table component

Tables:

This example uses a table that contains data for a case-control study. Each row contains information for a case or a control. To perform statistical analysis, you need a table with one row for each case-control pair. PROC SQL joins the table with itself in order to match the cases with their appropriate controls. After the rows are matched, differencing can be performed on the appropriate columns.

The input table MATCH_11 contains one row for each case and one row for each control. Pair contains a number that associates the case with its control. Low is 0 for the controls and 1 for the cases. The remaining columns contain information about the cases and controls.

Input Table

Program

options nodate pageno=1 linesize=80 pagesize=60;

proc sql;
   create table match as
      select
         one.Low,
         one.Pair,
         (one.lwt - two.lwt) as Lwt_d,
         (one.smoke - two.smoke) as Smoke_d,
         (one.ptd - two.ptd) as Ptd_d,
         (one.ht - two.ht) as Ht_d,
         (one.ui - two.ui) as UI_d

      from match_11 one, match_11 two
      where (one.pair=two.pair and one.low>two.low);

   title 'Differences for Cases and Controls';
   select *
      from match(obs=5);

Output

MATCH Table

Chapter Contents
Previous
Next
Top of Page