A complete statistical-programming deliverable for a Phase 2, single-arm study of an investigational
oral ALK inhibitor in ALK-positive advanced NSCLC: raw CRF-like data → 7 SDTM domains → 4 ADaM datasets →
oncology TLFs → define.xml + Reviewer's Guide. Every output is programmed twice — executed in R and
delivered in SAS — with R independently validating the SAS specification.
Portfolio demonstration built for Senior Statistical Programmer (Oncology) roles · 100% synthetic data, no real patients.
120
Subjects (Safety)
48.3%
ORR — 95% CI 39.1–57.6
81.7%
Disease Control Rate
44 / 65
OS / PFS events
13
CDISC datasets + TLFs
Built from scratch, not from a given spec. SDTM domains (incl. oncology TU/TR/RS),
ADaM derivations (ADTTE censoring rules, ADRS best-overall-response logic, population flags) and the
analysis specification were authored as part of this package — the exact work a senior FSP programmer owns.
Pipeline
Each stage is a versioned program with a paired SAS deliverable.
Generated in R (survminer / ggplot2); SAS equivalents in /tlf/sas.
Kaplan-Meier — Overall SurvivalKaplan-Meier — Progression-Free SurvivalWaterfall — best % change in target-lesion sum (RECIST thresholds)Swimmer — treatment duration & best response (× = death)
Tables & listing
R reference shown; matching FDA-grade RTF (r2rtf) and SAS programs in the package.
Dual programming — R validates SAS
R · executed
R programs run on-server (admiral, tern, rtables, r2rtf, survival), producing the
reference outputs and serving as the independent cell-by-cell validator.
SAS · deliverable
SAS programs (PROC SQL, PROC LIFETEST, PROC FREQ exact binomial, ODS RTF) ship as
the customer-runnable deliverable — the standard for FDA submission environments.
