How to adjust for age as a cofounder

[Dylannn3]

Question prompt:

examine the effect of alcohol use on CKD, adjusting for age as a potential confounder (included as a dichotomous variable representing 50+ compared to 20-49 years).

Here is my code:
libname cesd4 '/home/u61727912/bios5204';
data logis;
set cesd4.logis;
if age >= 50 then agegrp2 = 1;
else if age >= 20 and age <= 49 then agegrp2 = 0;
run;

proc logistic data=logis descending;
model CKD=EtOH agegrp2 / clodds=wald;
run;

Error from SAS:
There are no valid observations

I’m not sure why it is saying no observations because that shouldn’t be true?

[Patrick]

Normally such SAS messages are correct. To verify what you've got you could run:

proc freq data=logis;
  table agegrp2;
run;

[ballardw]

Or show the LOG from running the data step. Copy the data step code and all notes, messages, warnings or errors. Then on the forum open a text box and paste all that text.

[Dylannn3]

Here is the LOG: 

1 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;

68

69 libname cesd4 '/home/u61727912/bios5204';

NOTE: Libref CESD4 was successfully assigned as follows:

Engine: V9

Physical Name: /home/u61727912/bios5204

70 data logis;

71 set cesd4.logis;

NOTE: Data file CESD4.LOGIS.DATA is in a format that is native to another host, or the file encoding does not match the session

encoding. Cross Environment Data Access will be used, which might require additional CPU resources and might reduce

performance.

72 if age >= 50 then agegrp2 = 1;

73 else if age >= 20 and age <= 49 then agegrp2 = 0;

74 run;

NOTE: Variable age is uninitialized.

NOTE: There were 3000 observations read from the data set CESD4.LOGIS.

NOTE: The data set WORK.LOGIS has 3000 observations and 6 variables.

NOTE: DATA statement used (Total process time):

real time 0.00 seconds

user cpu time 0.00 seconds

system cpu time 0.00 seconds

memory 1214.65k

OS Memory 23972.00k

Timestamp 05/02/2023 02:27:30 PM

Step Count 67 Switch Count 2

Page Faults 0

Page Reclaims 163

Page Swaps 0

Voluntary Context Switches 10

Involuntary Context Switches 0

Block Input Operations 0

Block Output Operations 520

75

76 proc logistic data=logis descending;

77 model CKD=EtOH agegrp2 / clodds=wald;

78 run;

ERROR: There are no valid observations.

NOTE: The SAS System stopped processing this step because of errors.

NOTE: There were 3000 observations read from the data set WORK.LOGIS.

NOTE: PROCEDURE LOGISTIC used (Total process time):

real time 0.00 seconds

user cpu time 0.00 seconds

system cpu time 0.00 seconds

memory 744.34k

OS Memory 23972.00k

Timestamp 05/02/2023 02:27:30 PM

Step Count 68 Switch Count 0

Page Faults 0

Page Reclaims 52

Page Swaps 0

Voluntary Context Switches 0

Involuntary Context Switches 0

Block Input Operations 0

Block Output Operations 0

79

80 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;

90

[StatDave]

The log says that your AGE variable is uninitialized. As a result, all of the AGEGRP2 values will be missing so that none of the observations are usable in LOGISTIC. Perhaps your age variable has a different name?

[Dylannn3]

Hi so you were correct about the "age" variable being missed labeled. I updated it to "agegrp" which is the correct variable name but I still can't get the code to run. 

I also ran this proc:

proc freq data=logis;
table agegrp2;
run; 

To see the frequencies and it came up blank. 

After updating the variable here is the new log (with error still):

1 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;

68

69 libname cesd4 '/home/u61727912/bios5204';

NOTE: Libref CESD4 was successfully assigned as follows:

Engine: V9

Physical Name: /home/u61727912/bios5204

70 data logis;

71 set cesd4.logis;

NOTE: Data file CESD4.LOGIS.DATA is in a format that is native to another host, or the file encoding does not match the session

encoding. Cross Environment Data Access will be used, which might require additional CPU resources and might reduce

performance.

72 if agegrp >= 50 then agegrp2 = 1;

73 else if agegrp >= 20 and agegrp <= 49 then agegrp2 = 0;

74 run;

NOTE: There were 3000 observations read from the data set CESD4.LOGIS.

NOTE: The data set WORK.LOGIS has 3000 observations and 5 variables.

NOTE: DATA statement used (Total process time):

real time 0.00 seconds

user cpu time 0.00 seconds

system cpu time 0.00 seconds

memory 1140.84k

OS Memory 23204.00k

Timestamp 05/06/2023 09:59:45 AM

Step Count 75 Switch Count 2

Page Faults 0

Page Reclaims 186

Page Swaps 0

Voluntary Context Switches 18

Involuntary Context Switches 0

Block Input Operations 0

Block Output Operations 264

75

76 proc logistic data=logis descending;

77 model CKD=EtOH agegrp2 / clodds=wald;

78 run;

ERROR: There are no valid observations.

NOTE: The SAS System stopped processing this step because of errors.

NOTE: There were 3000 observations read from the data set WORK.LOGIS.

NOTE: PROCEDURE LOGISTIC used (Total process time):

real time 0.00 seconds

user cpu time 0.01 seconds

system cpu time 0.00 seconds

memory 603.43k

OS Memory 23204.00k

Timestamp 05/06/2023 09:59:45 AM

Step Count 76 Switch Count 0

Page Faults 0

Page Reclaims 52

Page Swaps 0

Voluntary Context Switches 0

Involuntary Context Switches 0

Block Input Operations 0

Block Output Operations 0

79

80 OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;

90

[StatDave]

Run this to see the patterns of missing values that exist in the data and post the results:

   proc mi data=logis nimpute=0;
      var CKD EtOH agegrp2;
      ods select MissPattern;
      run;

[ballardw]

Probably another variation of not knowing your data.

Your code in the log assumes AGEGRP has values of 50 or greater and then between 20 and 49.

72 if agegrp >= 50 then agegrp2 = 1;
73 else if agegrp >= 20 and agegrp <= 49 then agegrp2 = 0;

I bet that someone already reduced something of an age, or used a response to a question so that your AGEGRP variable is already categories of some sort and very likely to have many fewer values.

Run this and show us the result:

Proc freq data=logis;
   tables agegrp * agegrp2 / list missing;
run;

[Dylannn3]

Ran the proc freq and this is what it is showing me 

[ballardw · Accepted solution]

---

@Dylannn3 wrote:

Ran the proc freq and this is what it is showing me 

---

Which tells use that your Agegroup variable has already been collapsed to represent some range of ages and does not have any values greater than 1. You need to go to the source of the data to find out what 0 and 1 represent.Likely the 1 means in some specific range of ages and 0 all the other ages. But without documentation somewhere the range is unknown.

Or find an actual age variable somewhere (might have really different name) or possibly a variable containing date of birth and calculate an age at a given point in that that you can used to make your Agegrp2.

[Dylannn3]

Hi thank you!!! you were correct the age group variable was collapsed. I was able to finish the rest of my analysis. Thank you once again for everyones help in this thread!