Series and reference percentiles based on the same dataset

[hengophone]

Hi all,

My goal is to produce several parameter-by-age series for each ID and have several reference percentiles on the same plot. Reference percentiles' data are provided for the age range of 24-240:

Age (in months)	3rd Percentile BMI Value	5th Percentile BMI Value	10th Percentile BMI Value	25th Percentile BMI Value	50th Percentile BMI Value	75th Percentile BMI Value	85th Percentile BMI Value	90th Percentile BMI Value	95th Percentile BMI Value	97th Percentile BMI Value
24	14.52095	14.73732	15.09033	15.74164	16.57503	17.55719	18.16219	18.60948	19.33801	19.85986
24.5	14.50348	14.71929	15.07117	15.71963	16.54777	17.52129	18.11955	18.56111	19.2789	19.79194
25.5	14.46882	14.68361	15.03336	15.67634	16.49443	17.45135	18.03668	18.4673	19.16466	19.66102
...	...	...	...	...	...	...	...	...

The actual data is much narrower by the age and also very scarce in datapoints for each ID. They have to be on the same dataset, so both setting and merging them together were tested, but desired output can't be achieved either way using SGPLOT. What is the key to have these series be plotted correctly?

Current code:

PROC SGPLOT DATA = plotdata;  
styleattrs datacontrastcolors=('red' 'green' 'blue' 'black' 'maroon');

SCATTER Y=chaaval X=ageass / MARKERATTRS = (SYMBOL=circlefilled size=5px  color="GREEN") legendlabel='Group 1' name='cohorta' ;
SERIES  Y=chaaval X=ageass / group=USUBJID datalabel=subjnum LINEATTRS = (THICKNESS = 2.25 COLOR="GREEN" PATTERN = Solid) legendlabel='Group 1' name="linea";
SCATTER Y=chbaval X=ageass / MARKERATTRS = (SYMBOL=circlefilled size=5px  color="RED") legendlabel='Group 2' name='cohortb' ;
SERIES  Y=chbaval X=ageass / group=USUBJID datalabel=subjnum LINEATTRS = (THICKNESS = 2.25 COLOR="RED" PATTERN = Solid) legendlabel='Group 2' name="lineb";

SERIES Y = _3RD_PERCENTILE_VALUE X = agem_htab / LINEATTRS = (THICKNESS = 1 COLOR=gray PATTERN = Solid) curvelabel='3rd percentile' CURVELABELATTRS=(SIZE = 5 COLOR=gray weight=NORMAL) CURVELABELLOC=OUTSIDE;
SERIES Y = _5TH_PERCENTILE_VALUE X = agem_htab / LINEATTRS = (THICKNESS = 1 COLOR=gray PATTERN = Solid) curvelabel='5th percentile' CURVELABELATTRS=(SIZE = 5 COLOR=gray weight=NORMAL) CURVELABELLOC=OUTSIDE;
SERIES Y = _10TH_PERCENTILE_VALUE X = agem_htab / LINEATTRS = (THICKNESS = 1 COLOR=gray PATTERN = Solid) curvelabel='10th percentile' CURVELABELATTRS=(SIZE = 5 COLOR=gray weight=NORMAL) CURVELABELLOC=OUTSIDE;
SERIES Y = _25TH_PERCENTILE_VALUE X = agem_htab / LINEATTRS = (THICKNESS = 1 COLOR=gray PATTERN = Solid) curvelabel='25th percentile' CURVELABELATTRS=(SIZE = 5 COLOR=gray weight=NORMAL) CURVELABELLOC=OUTSIDE;
SERIES Y = _50TH_PERCENTILE_VALUE X = agem_htab / LINEATTRS = (THICKNESS = 1 COLOR=orange PATTERN = Solid) curvelabel='50th percentile' CURVELABELATTRS=(SIZE = 5 COLOR=gray weight=NORMAL) CURVELABELLOC=OUTSIDE;
SERIES Y = _75TH_PERCENTILE_VALUE X = agem_htab / LINEATTRS = (THICKNESS = 1 COLOR=gray PATTERN = Solid) curvelabel='75th percentile' CURVELABELATTRS=(SIZE = 5 COLOR=gray weight=NORMAL) CURVELABELLOC=OUTSIDE;
SERIES Y = _90TH_PERCENTILE_VALUE X = agem_htab / LINEATTRS = (THICKNESS = 1 COLOR=gray PATTERN = Solid) curvelabel='90th percentile' CURVELABELATTRS=(SIZE = 5 COLOR=gray weight=NORMAL) CURVELABELLOC=OUTSIDE;
SERIES Y = _95TH_PERCENTILE_VALUE X = agem_htab / LINEATTRS = (THICKNESS = 1 COLOR=gray PATTERN = Solid) curvelabel='95th percentile' CURVELABELATTRS=(SIZE = 5 COLOR=gray weight=NORMAL) CURVELABELLOC=OUTSIDE;
SERIES Y = _97TH_PERCENTILE_VALUE X = agem_htab / LINEATTRS = (THICKNESS = 1 COLOR=gray PATTERN = Solid) curvelabel='97th percentile' CURVELABELATTRS=(SIZE = 5 COLOR=gray weight=NORMAL) CURVELABELLOC=OUTSIDE;
REFLINE 15 20 25 30 35 49/LINEATTRS = (THICKNESS = 1 COLOR=gray PATTERN = Solid) transparency=.3;
XAXIS label="BMI at Visit (months)" values=(24 to 132 by 12);
YAXIS label= "BMI" values=(15 to 40 by 5);
KEYLEGEND 'cohorta' 'cohortb' /title="Group:" location=OUTSIDE position=bottom autoitemsize;
run;

Data example:

subjnum	AGEASS	chaaval	chbaval	AVAL	agem_htab
1	99.876796715	17.6	.	17.6	99.5
2	68.73100616	16.4	.	16.4	68.5
2	73.659137577	16.9	.	16.9	73.5
3	87.260780287	.	19.3	19.3	86.5
3	91.86036961	.	19	19	91.5
4	72.542094456	.	19.2	19.2	72.5
5	56.509240246	.	16.7	16.7	56.5

Expected kind of plot:

[ballardw]

How do we know what "series plotted correctly" would look like?

I am afraid that without actual working data step example data including some for the variables you show no values for such as Chaaval Chbaval Ageass Agem_htab

If USUBJID is individual identifier how many values are there? That has the possibility of making any graph next to useless if the number is excessive. With the number of different "series" plots you have that number might be excessive at a pretty small number.

The example "data" percentiles looks a bit like standard growth charts, is it? You may want to investigate the stuff at https://www.cdc.gov/nccdphp/dnpao/growthcharts/resources/sas.htm for what I think comes close.

[hengophone]

Yes, that is a CDC growth chart. 

The number of IDs that is expected on the plot is not too big, so it is not too cluttered.

Edited the post to include data example.

upd.: another edit to include desired output.