Skip to content

SAS to R dictionary

For readers who know SAS: each row is a SAS idiom, its R form in this book, and where the book explains it. Use the search of the site (Ctrl+K) to find a SAS keyword.

SAS R Where
DATA new; SET old(KEEP=a b); RUN; new <- old |> select(a, b) ch 1, ch 2
PROC FREQ; TABLES arm; count(arm) ch 1
WHERE / subsetting IF filter() ch 2
assignment statement mutate() ch 2
IF ... THEN ...; ELSE ...; if_else(), case_when() ch 1, ch 2
%LET arm = Placebo; arm <- "Placebo" ch 1
PROC MEANS + merge back (value of all rows) mutate(m = mean(x)) ch 2
MISSING(x) is.na(x) ch 2
missing text " " NA, or "" in data read from SAS files ch 2
DATALINES tibble::tribble() ch 2
PROC IMPORT / INFILE ... DSD readr::read_csv() ch 2
FILE / PUT to a CSV, PROC EXPORT readr::write_csv() ch 2
permanent data set mylib.x saveRDS(x, "path/x.rds"), readRDS() ch 2
WORK library tempdir() ch 2
INPUT(dtc, ?? E8601DA.) as.Date(dtc, format = "%Y-%m-%d") ch 2
impute a partial start date early lubridate::ymd(dtc, truncated = 2) ch 2
INTNX("MONTH", d, 0, "E") lubridate::rollforward(d) ch 2
MDY(12, 31, YEAR(d)) lubridate::make_date(year(d), 12, 31) ch 2
PROC FORMAT for groups and their order case_when() + factor(x, levels = ...) ch 2
the log the console: messages, warnings, errors ch 2
PROC CONTENTS glimpse(), class() ch 3
PUT(x, best.) as.character(x) ch 3
INPUT(x, best.) as.numeric(x) ch 3
SCAN(x, 3, "-") stringr::str_split_i(x, "-", 3) ch 3
LABEL x = "..."; attr(x, "label") <- "..." ch 3
SUBSTR(x, 4, 3) str_sub(x, 4, 6) (start and end) ch 3
FIND(x, "pain", "i") > 0 str_detect(x, regex("pain", ignore_case = TRUE)) ch 3
TRANWRD(x, a, b) str_replace_all(x, fixed(a), b) ch 3
CATS() / CATX() str_c() ch 3
UPCASE() / LOWCASE() str_to_upper() / str_to_lower() ch 3
STRIP(COMPBL(x)) str_squish(x) ch 3
STRIP(x) str_trim(x) ch 3
PUT(x, Z6.) str_pad(x, 6, pad = "0") ch 3
PUT(x, 5.1) sprintf("%.1f", x) ch 3
PUT(d, DATE9.) toupper(format(d, "%d%b%Y")) ch 3
YEAR(d), MONTH(d) lubridate::year(d), month(d) ch 3
YRDIF(a, b, "ACT/ACT") time_length(interval(a, b), "years") ch 3
SUM(a, b) (ignores missing) sum(a, b, na.rm = TRUE) ch 3
ROUND(x, 0.1) (half away from zero) a round-half-up function; round() rounds half to even ch 3
PRXMATCH("/^\d{4}$/", x) > 0 str_detect(x, "^\\d{4}$") ch 3
PROC SORT; BY a DESCENDING b; arrange(a, desc(b)) ch 4
PROC SORT NODUPKEY arrange() + distinct(..., .keep_all = TRUE) ch 4
DUPOUT= filter(duplicated(pick(keys))) ch 4
FIRST.x / LAST.x row_number() == 1 / row_number() == n() in a group ch 4
BY x; IF FIRST.x; slice_head(n = 1, by = x) ch 4
LAG(x) dplyr::lag(x) ch 4
RETAIN base; (carry baseline) mutate(BASE = x[flag %in% "Y"], .by = USUBJID) ch 4
sum statement n + 1; cumsum() ch 4
RETAIN worst; worst = MAX(worst, x); cummax(x) ch 4
RETAIN last non-missing value (LOCF) tidyr::fill(x, .direction = "down") ch 4
IF LAST.x THEN OUTPUT; (one row per group) summarise(..., .by = x) ch 4
SAS R Where
SET a b; bind_rows(a, b) 5.2
SET a(IN=ina) b; with a source variable bind_rows(A = a, B = b, .id = "SOURCE") 5.2
MERGE a(IN=ina) b(IN=inb); BY key; IF ina AND inb; inner_join(a, b, by = join_by(key)) 5.3
MERGE ...; BY key; IF ina; left_join(a, b, by = join_by(key)) 5.3
MERGE ...; BY key; (all records) full_join(a, b, by = join_by(key)) 5.3
WHERE key IN (SELECT key FROM b) semi_join(a, b, by = join_by(key)) 5.4
IF ina AND NOT inb; / NOT IN (SELECT ...) anti_join(a, b, by = join_by(key)) 5.4
RENAME= to avoid an overwrite in MERGE .x / .y, or suffix = c("_a", "_b") 5.5
NOTE: MERGE statement has more than one data set with repeats of BY values Warning: many-to-many relationship 5.6
PROC SORT NODUPKEY check before a MERGE relationship = "many-to-one" 5.7
PROC TRANSPOSE of SUPPDM, then MERGE pivot_wider() of SUPPDM, then left_join() 5.8
PROC CONTENTS glimpse(), names(), sapply(df, attr, "label") 6.1
PROC PRINT DATA=x(OBS=5); print(x, n = 5) 6.1
PROC FREQ; TABLES a*b / LIST OUT=; count(a, b) 6.2
TABLES x / MISSING; count(x) (counts NA by default) 6.2
PROC MEANS; CLASS g; VAR x; OUTPUT OUT=; group_by(g) |> summarise(mean = mean(x)) 6.3
Statistics skip missing values mean(x, na.rm = TRUE) 6.3
Q1 with PCTLDEF=5 quantile(x, 0.25, type = 2) 6.3
SELECT COUNT(DISTINCT usubjid) INTO :n1-:n3 count(TRT01A, name = "BIGN") 6.4
Remerge of a summary statistic in PROC SQL group_by(g) |> mutate(stat = mean(x)) 6.5
COUNT(DISTINCT usubjid) n_distinct(USUBJID) 6.6
UNION, INTERSECT, EXCEPT union(), intersect(), setdiff() 6.8
WHERE x LIKE '%TEXT%' filter(str_detect(x, fixed("TEXT"))) 6.9
ARRAY + DO i = 1 TO DIM() mutate(across(cols, rule, .names = "{.col}N")) 7.1
SELECT; WHEN ...; OTHERWISE ...; END; case_when(cond ~ value, .default = other) 7.2
DO adt = start TO end; OUTPUT; END; rowwise() |> reframe(ADT = seq(start, end, by = "day")) 7.3
PROC TRANSPOSE; BY id; ID name; VAR value; pivot_wider(id_cols, names_from, values_from) 7.4
PREFIX= / SUFFIX= of PROC TRANSPOSE names_prefix = / names_glue = 7.4
PROC TRANSPOSE wide to long, NAME=, RENAME=(col1=aval) pivot_longer(cols, names_to, values_to) 7.6
PROC FORMAT CNTLIN= lookup a lookup table and left_join() 7.7
VALUE format for a few codes named vector: c(Pbo = 0, Xan_Lo = 54)[code] 7.7
VALUE format with NOTSORTED, ORDER=DATA factor(x, levels = ..., labels = ...) 7.8
INVALUE that turns text into a sort number as.integer(factor(x, levels = ...)) 7.8
Hash object FIND() table$value[match(x, table$key)] 7.9
%MACRO name(param=default); name <- function(param = default) { } 8.1
&var. as a variable name in a macro {{ var }} in a function 8.2
A variable name in a macro variable .data[[var]] 8.4
%PUT ERROR: ...; %GOTO finish; cli::cli_abort("..."), stopifnot() 8.5
%INCLUDE of a macro file source("R/file.R") 8.5
%DO i = 1 %TO %SYSFUNC(COUNTW(&list.)); map(list, f), walk(list, f), for (v in list) 8.6
SET out_1-out_3; after a macro loop list_rbind(map(...)) 8.6
%PUT to the log cat(line, file = log, append = TRUE) 8.7
SAS R Where
PROC MEANS ... MIN= with CLASS usubjid summarise(TRTSDT = min(EXSTDT), .by = USUBJID) 9.3
MAX= skips missing values; NMISS= counts them max(x) is NA when one value is NA; max(x, na.rm = TRUE) skips it 9.3
COALESCE(a, b) coalesce(a, b) 9.3
%m_merge(data=, add=, out=) left_join(data, add, relationship = "one-to-one") in a small function 9.4
PROC TRANSPOSE ... ID qnam; VAR qval; tidyr::pivot_wider(names_from = QNAM, values_from = QVAL) 9.4
PROC COMPARE BASE= COMPARE=; ID usubjid; diffdf::diffdf(base, compare, keys = "USUBJID") 9.6
INPUT(xxdtc, E8601DA.) lubridate::ymd(xxdtc) or admiral derive_vars_dt() 9.3
IF adt >= trtsdt THEN ady = adt - trtsdt + 1; ELSE ady = adt - trtsdt; as.numeric(ADT - TRTSDT) + if_else(ADT >= TRTSDT, 1, 0), or derive_vars_dy() 10.1
BY usubjid; IF LAST.usubjid; among the rows before the first dose mutate(ABLFL = if_else(row_number() %in% last(which(can_be_base)), "Y", NA), .by = USUBJID) 10.1
merge a one-row-per-subject baseline back with BY usubjid mutate(BASE = first(AVAL[ABLFL %in% "Y"]), .by = USUBJID), or derive_var_base() 10.1
%m_baseline(by=usubjid paramcd atptn, ...) derive_baseline(by = c("USUBJID", "PARAMCD", "ATPTN"), order = c("ADT", "VSSEQ")) 10.2
IF aval < anrlo THEN anrind = "LOW"; ELSE IF ... case_when(AVAL < ANRLO ~ "LOW", AVAL > ANRHI ~ "HIGH", !is.na(AVAL) ~ "NORMAL"), or derive_var_anrind() 10.5
astdt = MDY(mm, 1, yy); astdtf = "D"; ymd(paste0(AESTDTC, "-01")), or derive_vars_dt(highest_imputation = "M", min_dates = exprs(TRTSDT)) 11.2
PROC SORT; BY usubjid astdt aeseq; then IF FIRST.usubjid arrange(ASTDT, AESEQ), then mutate() with row_number() == 1 and .by = USUBJID, or derive_var_extreme_flag(mode = "first") 11.2
IF astdt >= trtsdt THEN trtemfl = "Y"; if_else(ASTDT >= TRTSDT, "Y", NA), or derive_var_trtemfl() 11.2
%m_tte(events=, enddt=, censordt=) derive_param_tte(event_conditions = list(event_source(...)), censor_conditions = list(censor_source(...))) 12.3
PROC LIFETEST; TIME aval*cnsr(1); STRATA trt01a; survival::survfit(Surv(AVAL, 1 - CNSR) ~ TRT01A) 12.5
SAS R Where
CALL SYMPUTX for the big N count(TRT, name = "N"), then left_join() 13.1
OUTPUT; trt01an = 99; OUTPUT; (Total column) bind_rows(data, mutate(data, TRT01A = "Total")) 13.1
PROC MEANS for a demographics table group_by(), summarise() with sprintf() 13.2
PROC TRANSPOSE of the layout pivot_wider(names_from = TRT, values_from = value) 13.2
grid DATA step, IF MISSING(n) THEN n = 0 complete(..., fill = list(n = 0)) 13.2
ROUND(x, 0.1), format 5.1 round_half_up(x, 1) (not round()) 13.3
PROC SORT NODUPKEY; BY trtan aebodsys usubjid; distinct(TRT, AEBODSYS, USUBJID) 13.4
study table macros (%m_desc, %m_freq) gtsummary::tbl_summary() 13.5
AE table macro gtsummary::tbl_hierarchical(id = USUBJID, denominator = adsl) 13.5
PROC REPORT COLUMN / DEFINE, PROC TABULATE rtables: basic_table() |> split_cols_by() |> analyze(), build_table() 13.6
PROC COMPARE of two tables join the cells and count the same text 13.7
ODS HTML gt::gtsave() 13.8
ODS RTF with PROC REPORT flextable + officer::body_add_flextable() 13.8
PROC REPORT DEFINE x / ORDER in a listing flextable::merge_v(j = "x") 14.1
PROC REPORT FLOW WIDTH=30 flextable::width() (text wraps in the cell) 14.1
PROC MEANS STDERR= sd(x) / sqrt(n()) 14.2
PROC SGPLOT SERIES, SCATTER YERRORLOWER= YERRORUPPER=, REFLINE geom_line(), geom_errorbar(), geom_hline() 14.3
ODS GRAPHICS image file ggsave("f.png", fig, width = 8, height = 5) 14.3
PROC LIFETEST TIME aval * cnsr(1); STRATA trtan; survfit(Surv(AVAL, 1 - CNSR) ~ TRT) 14.4
PROC LIFETEST PLOTS=SURVIVAL(ATRISK=...) ggsurvfit() + add_risktable(times = ...) 14.5
ATTRIB from the specification, LIBNAME x XPORT + PROC COPY xportr_type(), xportr_length(), xportr_label(), xportr_write() 14.6
LIBNAME x XPORT to read haven::read_xpt() 14.6

Parts IV and V: quality and advanced functions

Section titled “Parts IV and V: quality and advanced functions”
SAS R Where
PROC COMPARE BASE=prod COMPARE=qc; ID usubjid; RUN; diffdf::diffdf(prod, qc, keys = "USUBJID") 15.2
&SYSINFO. after PROC COMPARE diffdf_has_issues(res), res$NumDiff 15.4
PROC COMPARE ... CRITERION=1E-8 diffdf(..., tolerance = 1e-8) (the default is about 1.5e-08) 15.5
%logcheck on a log file withCallingHandlers(source(file), warning = ..., message = ...) 15.9
NOTE: MERGE statement has more than one data set with repeats of BY values the dplyr warning Detected an unexpected many-to-many relationship 15.9
PROC SQL checks with NOT IN (SELECT ...) anti_join() and a tribble() of counts 16.3
PROC FREQ of a decision variable count(data, WTBLSRC) 16.4
%LOCAL macro variable a name made inside a function: always local 17.2
&var. in a macro that writes MEAN(&var.) {{ var }} in a function, or enquo() and !! 17.5
&var._gt&limit. as a variable name "{{ var }}_GT{limit}" := 17.6
%STR, %NRSTR, %BQUOTE, %SUPERQ not needed: a value is never code 17.7
CALL SYMPUTX(CATS("pcd", _N_), paramcd) and &&pcd&i. codes <- distinct(advs, PARAMCD)$PARAMCD, then codes[i] 18.1
%DO i = 1 %TO &nparam.; with WHERE paramcd = "&&pcd&i." split(data, data$PARAMCD) |> map(f) 18.2
cont=age|Age (years)|0#... with %QSCAN a tribble() of settings and pmap() 18.4
PROC CONTENTS ... OUT= as a specification map_chr(data, \(x) attr(x, "label")) 18.6
CALL EXECUTE, DOSUBL map(); code as text with eval(parse(text = )) only as the last choice 18.7
%STR(ERR)OR: [MACRO] ... and %GOTO finish cli::cli_abort("...") 19.1
an autocall library and SASAUTOS a package folder and pkgload::load_all() 19.2
a test harness that counts ERROR lines per test case testthat::test_that() with expect_error(), testthat::test_dir() 19.4
OPTIONS MPRINT MLOGIC SYMBOLGEN; debugonce(f) or browser() 19.6

Part VI: DataStep Studio for SAS and for R

Section titled “Part VI: DataStep Studio for SAS and for R”
SAS R Where
SAS log (NOTE: The data set WORK.ADSL has 254 observations ...) the DataStep Studio for R log (NOTE: adsl has 254 rows and 6 columns) ch 20
NOTE: Missing values were generated New: WGTGR (1 missing) in the log; rule missing_values ch 20
NOTE: MERGE statement has more than one data set with repeats of BY values the suspicious log rule join: rows went up in a join ch 20
NOTE: Invalid numeric data the suspicious log rule coercion: NAs introduced by coercion ch 20
Run To Line (Ctrl+Alt+Up) Run to Line (Cmd/Ctrl+Alt+T), dsr run --to-line ch 20
a new SAS session, empty WORK dsr stop --session, dsr start ch 20
Run in Batch dsr batch a.R b.R ch 26
viewer WHERE trtemfl = "Y" AND aesev = "SEVERE" viewer WHERE TRTEMFL == "Y" & AESEV == "SEVERE" ch 21
viewer SQL button viewer R code button, dsr query --code ch 21
. missing, smaller than every number NA, not a value: is.na() ch 21
keyboard macro + Tab snippet + Tab ch 22
Format SAS Program Format R Program (Ctrl+Shift+A), dsr format ch 22
Enlarge / Optimize Macro Call (Ctrl+Shift+E / Ctrl+Shift+O) Enlarge / Optimize Function Call (same keys), dsr call ch 22
Open Macro Definition (Cmd/Ctrl+click) Go to Definition (Cmd/Ctrl+click), dsr def ch 22
DATA step debugger: watch the PDV dsr pipe file.R:line: the rows at each step of a pipeline ch 23
IN= flags of a MERGE @pipe[k]:unmatched:x, @pipe[k]:unmatched:y ch 23
breakpoint condition usubjid = "01-701-1015" dsr pipe ... --follow 01-701-1015 ch 23
%pry pry() ch 24
%pry(when=&i. = 3) pry(when = sev == "SEVERE") ch 24
%pry(rescue=Y) nothing to write: an error keeps its frames open (dsr pry stack) ch 24
vt work._sev, freq trta at a %pry stop dsr pry eval 'dplyr::count(sub, TRT01A)', dsr query @pry:sub ch 24
watch &nsub watch nsub ch 24
PROC SORT BY usubjid visit; (visit names sort as text) arrange(USUBJID, VISIT): the same trap; sort by VISITNUM ch 24
macro debugger: F5, F10, F11, Shift+F11 R debugger: F5, F10, F11, Shift+F11 ch 25
Edit Condition &i. = 3 Edit Condition any(age == 65) ch 25
%assert_eq tests of a macro testthat tests, dsr test ch 26
macro coverage ×N dsr test --coverage, dsr coverage ch 26
SAS Pry terminal: vt sdtm.cm, profile dsr describe cm, dsr profile cm ch 26
dups usubjid dsr dups adae --key USUBJID ch 26

Found an error or have a question? Write to datastep.studio@gmail.com· Every program runs in DataStep Studio for R, the R extension for VS Code and Positron.

Data: the CDISC pilot study (cdiscpilot01), from the R packagespharmaversesdtm andpharmaverseadam (Apache License 2.0).

R is a project of the R Foundation. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. This book is not affiliated with or endorsed by SAS Institute Inc. or the R Foundation. CDISC, SDTM and ADaM are trademarks of the Clinical Data Interchange Standards Consortium.

© 2026 DataStep Labs. All rights reserved.