Adding confidence limits to line graph using sgplot

[scarrico]

Hello,

I am working on generating line graphs using proc sgplot vline and I am having some difficulty assigning confidence intervals to my data points.  I am able to generate the line but no confidence intervals for each data point.

Below is the code I have been using:

proc sgplot data=CAUTI2;
vline SummaryYM / response=SIR stat=mean limitstat=clm;
yaxis grid values=(0 to 2) valueshint;
refline 0.6963750 0.3762621 1.0164879 / label = ('SIR(mean)' 'UCL' 'LCL');
yaxis label='SIR';
title 'SIR Trend of CAUTI in Acute Care Hospitals in 2018';
run;

 SAS Output

GRAPH1 is the desired output.

Thank you for your help!

Savannah

[ballardw]

---

@scarrico wrote:

Hello,

I am working on generating line graphs using proc sgplot vline and I am having some difficulty assigning confidence intervals to my data points.  I am able to generate the line but no confidence intervals for each data point.

Below is the code I have been using:

proc sgplot data=CAUTI2;
vline SummaryYM / response=SIR stat=mean limitstat=clm;
yaxis grid values=(0 to 2) valueshint;
refline 0.6963750 0.3762621 1.0164879 / label = ('SIR(mean)' 'UCL' 'LCL');
yaxis label='SIR';
title 'SIR Trend of CAUTI in Acute Care Hospitals in 2018';
run;

 SAS Output

GRAPH1 is the desired output.

Thank you for your help!

Savannah

---

Data example? Do you actually have multiple values of SIR for values of SummaryYM? If there are not multiple values, or they are all the same, then there is no variability and the confidence interval is +/- 0 from the mean.

[scarrico]

Attached are my data

[ballardw]

---

@scarrico wrote:

Attached are my data

---

Which shows single values of SIR (as well as all the other variables) for each and every value of summaryYM. So there is NO variability  within the data you are graphing.

Since that PDF shows values like SIR95CI of 0.592, 3.040 then go back to the data set used to create that summary and graph from there.

OR instead of a not-very-friendly to use value like you have for the SIR95CI create and leave the upper and lower limits in separate variables and use a BAND plot with the UPPER and LOWER values the variables with the limits

[Ksharp]

It seems you need VBAR not VLINE statement.