Interaction plot

[kristiepauly]

Hello All,

I am having difficulty with an interaction plot.  I am trying to see if there is a difference in the change in pulmonary function in people with diabetes vs those without over, a 10 year time period.  I wrote this to calculate the change.

data working4;
	set working3; 
	change_fev1pp = fev1pp_Post_p3 - fev1pp_Post_p1;
	change_fev1_FVC = fev1_FVC_post_p3 - Fev1_FVC_post_p1;
	run;

I used proc glm to give me the type III table and the interaction plot for change in fev1pp. 

proc glm data=working4;
	class diabetes_P1(ref='No') finalgold_P1 (ref='GOLD 0');
	model change_fev1pp = diabetes_P1 finalgold_p1 diabetes_P1*finalgold_p1;
	lsmeans diabetes_P1 finalgold_p1 diabetes_P1*finalgold_p1 / pdiff;
	format finalgold_p1 baseline_gold_stage. diabetes_P1 diabetes_baseline.;
	run;

I get the type III table and an interaction plot.   

I need this same output but after I adjust for several variables (age, gender, bmi, high bp, high cholesterol, corticosteroid use, smoking status). I first tried to do this using Proc glm

proc glm data=working4;
	class diabetes_P1(ref='No') finalgold_P1(ref='GOLD 0') Gender(ref='1') BMI_P1(ref='Healthy')
	highbloodpres_P1(ref='No') highcholest_P1(ref='No')
  		cortsterinhal_P1(ref='No') cortsteroral_P1(ref='No') smokcignow_P1(ref='No');
	model change_fev1pp = diabetes_P1 finalgold_P1 diabetes_P1*finalgold_P1
	 BMI_P1 highbloodpres_P1 highcholest_P1 age_P1
  		cortsterinhal_P1 cortsteroral_P1 smokcignow_P1 gender;
	lsmeans diabetes_P1 finalgold_p1 diabetes_P1*finalgold_p1 / pdiff;
	format finalgold_P1 final_gold_stage. diabetes_P1 diabetes_final. BMI_P1 
  		BMI_group. highbloodpres_P1 highbloodpres_group. highcholest_P1 highcholest_group. smokcignow_P1 smokcignow_group.
  		cortsterinhal_P1 cortsterinhal_group. cortsteroral_P1 cortsteroral_group.;
	run;

and I get the type III table but not the interaction plot. Is this interaction plot not possible when including variables I'm adjusting for or did I do it wrong?  Is there a better visual representation to show the change in fev1pp by GOLD stage over 10 years time in people with diabetes vs without? 

I also tried to adjust for the variables in proc logistic, but that wasn't right either. 

proc logistic data=working4;
  	class diabetes_P1 (ref ='No') finalgold_P1 (ref='GOLD 0') race (ref='1') Gender(ref='1') BMI_P3(ref='Healthy')
	highbloodpres_P3(ref='No') highcholest_P3(ref='No') 
  		cortsterinhal_P3(ref='No') cortsteroral_P3(ref='No') smokcignow_P3(ref='No');
  	model change_fev1pp = diabetes_P1 finalgold_p1 diabetes_P1*finalgold_p1
  		age_P3  BMI_P3 highbloodpres_P3 highcholest_P3
  		cortsterinhal_P3 cortsteroral_P3 smokcignow_P3 gender/ link=glogit;
  		lsmeans diabetes_P1 finalgold_p1 diabetes_P1*finalgold_p1 / pdiff;
  	format finalgold_p1 final_gold_stage. diabetes_P1 diabetes_baseline. BMI_P3 
  		BMI_group. highbloodpres_P3 highbloodpres_group. highcholest_P3 highcholest_group. smokcignow_P3 smokcignow_group.
  		cortsterinhal_P3 cortsterinhal_group. cortsteroral_P3 cortsteroral_group. finalgold_P3 Baseline_GOLD_Stage.;
run;

I get this error.

 ERROR: Computations are terminated because the number of response levels, 614, exceeds MAXRESPONSELEVELS=100.
 NOTE: The SAS System stopped processing this step because of errors.
 NOTE: There were 2864 observations read from the data set WORK.WORKING4.
 NOTE: PROCEDURE LOGISTIC used (Total process time):
       real time           0.19 seconds
       user cpu time       0.17 seconds
       system cpu time     0.02 seconds
       memory              33575.15k
       OS Memory           61600.00k
       Timestamp           04/21/2024 10:06:27 PM
       Step Count                        525  Switch Count  0
       Page Faults                       0
       Page Reclaims                     8318
       Page Swaps                        0
       Voluntary Context Switches        0
       Involuntary Context Switches      0
       Block Input Operations            0
       Block Output Operations           40
       
 81         
 82         OPTIONS NONOTES NOSTIMER NOSOURCE NOSYNTAXCHECK;

Thanks.

[Ksharp]

Check @Rick_SAS blogs:
https://blogs.sas.com/content/iml/2019/05/30/visualize-interaction-effects-regression.html
https://blogs.sas.com/content/iml/2016/06/22/sas-effectplot-statement.html