Hello,   I have moved your post to the "statistical procedures" board (as it is a question on survival | time-to-event analysis). Higher probability to get a proper answer here.   By the way, I hope this question is different than the one in this post : "Efficacy Time to event" https://communities.sas.com/t5/Statistical-Procedures/Efficacy-Time-to-event/m-p/790028#M38732 If not ..., I want to say you should not post 2 times the same question. If you think the 1st question is not clear enough, then clarify with an additional post in the same topic thread.   I could not judge if these 2 were asking the same , as I have not read your posts in detail.   Thanks, Koen ... View more

Interactions in a modeling procedure are indicated with an asterisk. For example: class genea geneb; model cancer = genea geneb genea*geneb; ... View more

Thank you @Ksharp @SteveDenham @StatDave. Let me correct a mistake I said earlier. My data is not censored. I do not have any loss to follow up. No participants dropped out. I randomized 31 participants to the vaccine and 45 control, all participants completed the study except 1 control. Can all the suggested analysis methods will still works in this case? ... View more

As you say, "adjusting for a variable" (or "controlling for a variable") is related to including that variable in an appropriate regression model. For a good explanation and example, see regression - Explain model adjustment, in plain English - Cross Validated (stackexchange.com)   ... View more

@CathyVI wrote: Am sorry I did not frame my question well. After doing more studies and research I think what am trying to ask is creating a predictor of chronological age from a log.regression, and this will be the biological age. Is that right? You still haven't explained the difference between "chronological age" and "biological age". Furthermore, this seems like a question you should be asking a subject matter expert, not us. ... View more

The "heavy line" at the bottom of the graph is referred to as a FRINGEPLOT and is on of the plot statements in the SAS Graph Template Language. A NEEDLE plot with appropriate data can also make something like that.   Do you have the full document that link references? If you are looking for code to do the same regression then we need to know the exact regression they did and possibly even code used as there are lots of options for regressions and guessing won't help you much. The full paper might even include how the graph was made.   The first step would be getting the data needed for the plot.   ... View more

@CathyVI wrote: Sorry about the confusion my question is specific to the sql, case-when procedure not the put statement. This is a synthesis data. data test; input patient; datalines; -999 12 11 10 -9 -7 -5 9 8 7 6 5 4 3 1 ; run; The data is numeric and I converted into character using this code: data test2; set test; pt_discharge=put(patient,4.); run; Now I will like to rename the observations from -999 to 1 to character values. This is the outcome (table) I hope to achieve.   patient  pt_discharge -999  missing2 12  homeH 11   hospice 10  otherD -9 missing -7  unknown -5 invalid 9   otherH 8   dead 7   court 6   long 5   short 4   LAMA 3  admit 1 home   This pt_discharge=put(patient,4.); creates values that have leading blank characters for any value that can be displayed with 3,2 or 1 character. So your "=" does not match these created text values. Use pt_discharge=put(patient,4. -L); to left justify the resulting text.   You also need to be aware that most of the output will strip the leading space for display an you just don't see ... View more

See the article "Piecewise regression models and spline effects" for a discussion, example, and code.  The example uses PROC GLIMMIX, but also includes a reference to the PROC TRANSREG documentation, which contains a second example.   The example assumes that the two pieces of the model both have the same form (for example, linear or quadratic). If you want to mix-and-match models on separate domains, you can use PROC NLIN to construct a segmented regression model. ... View more

data test; input A BP totchol BMI age ; datalines; 1.222 127 132 22.4 34 1.332 123 133 22.3 44 1.443 234 244 34.2 23 2.344 344 241 54.2 65 3.122 543 234 23.1 44 5.234 353 354 25.5 78 4.344 533 123 45.6 23 2.645 334 234 45.3 16 ; run; proc stdize data=test out=want; var _numeric_; run; proc means data=want mean std; var _numeric_; run; Like this ? ... View more

PROC LOGISTIC is for dichotomous response variables by default. If A is a continuous variable, you might want to use PROC REG. proc reg data=Dataset; Model a= b c; Run; Quit; ... View more

If is not being evaluated correctly because pt_age_code and pt_age_recode are mixed up. ... View more

Proc format; value OneExercise 0 = "More than one exercise" 1 = "One exercise" ; data test; input Name $ run squat jump pushup; array exercise[4] run--pushup; OneExercise=sum(of run--pushup)=1; format OneExercise OneExercise.; cards; Bob 1 0 0 1 Mat 0 1 0 1 Cat 1 0 1 1 Dog 1 0 0 0 Fred 0 1 1 1 Sam 0 0 1 0 Pat 0 0 0 1 Jack 1 1 1 1 Bat 0 0 0 0 ; PROC SORT ; BY OneExercise; RUN; PROC TRANSPOSE DATA=test OUT=Transpose1 PREFIX=Count NAME=Exercise ; BY OneExercise; VAR run squat jump pushup; PROC SORT data=Transpose1 ; BY OneExercise Exercise; PROC TRANSPOSE DATA=Transpose1 OUT=Transpose2 (drop=_NAME_) PREFIX=Count ; BY OneExercise Exercise; VAR Count:; PROC TABULATE DATA=Transpose2 ; VAR Count1; CLASS OneExercise / ORDER=UNFORMATTED MISSING; CLASS Exercise / ORDER=UNFORMATTED MISSING; TABLE /* Row Dimension */ Exercise =" ", /* Column Dimension */ OneExercise=" "* Count1=" "* Sum=" "*f=3.0 ; RUN;   More than one exercise One exercise jump 3 1 pushup 5 1 run 3 1 squat 3 0     ... View more