← SASnRConditional Logic in Macros

Introduction

Many clinical macros need to behave differently depending on the type of analysis — use means and standard deviations for continuous endpoints, counts and percentages for categorical ones.
SAS handles this with %IF / %THEN / %ELSE inside macro code. R uses standard if / else or switch() inside a function.
This lesson builds a macro/function that picks the right summary method based on a "type" parameter: continuous or categorical.
We use a small ADSL-like dataset and demonstrate both paths through the conditional logic.
After running, we confirm that the continuous path produces means and the categorical path produces frequency counts.

Conditional Logic in Macros

SAS (Base SAS)+

*==============================================================================;
* Create sample ADSL data;
*==============================================================================;

data adsl;
    input usubjid $ trt01p $ age sex $ race $12.;
    datalines;
SUBJ001 Placebo 67 M White
SUBJ002 DrugA   59 F Black
SUBJ003 Placebo 72 F White
SUBJ004 DrugA   65 M Asian
SUBJ005 Placebo 60 F White
SUBJ006 DrugA   55 M Black
;
run;

*==============================================================================;
* Macro with conditional logic based on variable type;
*==============================================================================;

%macro analyse(ds=adsl, var=, type=);

    %if %upcase(&type.) = CONT %then %do;
        *--- Continuous: means, std, min, max ---;
        proc means data=&ds. n mean std min max;
            var &var.;
            title "Continuous summary: %upcase(&var.)";
        run;
    %end;

    %else %if %upcase(&type.) = CAT %then %do;
        *--- Categorical: frequency counts ---;
        proc freq data=&ds.;
            tables &var. / nocum nopercent;
            title "Categorical summary: %upcase(&var.)";
        run;
    %end;

    %else %do;
        %put ERROR: [analyse] type=&type. is not valid. Use CONT or CAT.;
    %end;

%mend analyse;

*==============================================================================;
* Call with different types;
*==============================================================================;

%analyse(var=age,  type=cont);
%analyse(var=sex,  type=cat);
%analyse(var=race, type=cat);
%analyse(var=age,  type=other);

To branch inside a SAS macro, we use %IF condition %THEN %DO; ... %END; followed by optional %ELSE %IF and %ELSE blocks.
%upcase(&type.) converts the parameter to uppercase before comparing, so the call works whether the user passes cont, CONT, or Cont.
The continuous branch runs PROC MEANS to produce N, mean, std, min, and max — the standard descriptive statistics for a numeric variable.
The categorical branch runs PROC FREQ with TABLES to produce frequency counts for each level of the variable.
The final %ELSE block is a safety net: it writes an error to the log if the user passes an unrecognised type, preventing silent failures.
%PUT ERROR: writes to the log with the red ERROR prefix, making it easy to spot in a long log file.
After running, we see a PROC MEANS table for AGE, PROC FREQ tables for SEX and RACE, and an error message for the invalid type call.

R (tidyverse)+

library(tidyverse)

#==============================================================================
# Create sample ADSL data
#==============================================================================

adsl <- tribble(
  ~usubjid,  ~trt01p,    ~age, ~sex, ~race,
  "SUBJ001", "Placebo",  67,   "M",  "White",
  "SUBJ002", "DrugA",    59,   "F",  "Black",
  "SUBJ003", "Placebo",  72,   "F",  "White",
  "SUBJ004", "DrugA",    65,   "M",  "Asian",
  "SUBJ005", "Placebo",  60,   "F",  "White",
  "SUBJ006", "DrugA",    55,   "M",  "Black"
)

#==============================================================================
# Function with if/else branching
#==============================================================================

analyse <- function(ds = adsl, var, type) {
  type <- toupper(type)

  if (type == "CONT") {
    #--- Continuous: means, sd, min, max ---
    ds %>%
      summarise(
        n    = n(),
        mean = mean({{ var }}),
        sd   = sd({{ var }}),
        min  = min({{ var }}),
        max  = max({{ var }})
      )

  } else if (type == "CAT") {
    #--- Categorical: frequency counts ---
    ds %>%
      count({{ var }}) %>%
      mutate(pct = round(n / sum(n) * 100, 1))

  } else {
    stop(paste0("type='", type, "' is not valid. Use CONT or CAT."))
  }
}

#==============================================================================
# Call with different types
#==============================================================================

analyse(var = age,  type = "cont")
analyse(var = sex,  type = "cat")
analyse(var = race, type = "cat")

R uses standard if / else if / else blocks inside a function — the same syntax we use everywhere in R, not a separate macro language.
toupper(type) normalises the input to uppercase, mirroring the SAS %upcase() for case-insensitive matching.
The continuous branch uses summarise() with {{ var }} (curly-curly) to pass the column name through tidy evaluation.
The categorical branch uses count({{ var }}) to tally each level, then mutate() to add a percentage column.
The error branch uses stop() to halt execution with a clear message — the R equivalent of SAS %PUT ERROR:.
Unlike SAS, there is no distinction between "macro-level" and "data-step-level" conditional logic — it is all the same if/else in R.
After running, we confirm: a one-row summary tibble for AGE, a frequency tibble for SEX, and a frequency tibble for RACE.

Happy learning!

Search lessons

Introduction

Conditional Logic in Macros