Glossary
| Word | Meaning |
|---|---|
| ADaM | The analysis data sets of a study (CDISC standard), made from the SDTM data; the tables are computed from them (12). |
| ADaMIG | The ADaM Implementation Guide of CDISC: the rules of names, labels and structures of the ADaM data sets (13). |
| agent | A language model that runs commands in a terminal in a loop, such as Claude Code, Codex or Gemini CLI (6). |
| agent score | Sapling’s measure (0 to 100) of what an agent can improve in a plan by writing blocks (5). |
| agreement | Your yes to the text of a block, pinned to a hash of that text (9). |
| ARD | The analysis results data: every number of every table, with the analysis that made it (CDISC ARS) (15). |
| block | One decision of the plan in Sapling’s Markdown: a population, an estimand, a table, a test (4). |
| co-primary endpoints | Two or more primary endpoints that must all show an effect; no alpha adjustment is needed, but power is lost (7). |
| codespace | A computer in the cloud with VS Code in your browser (GitHub) (2). |
| dashboard gate | The check that every number the dashboard shows is equal to the ARD (19). |
| decision | A place where the protocol leaves a choice to the statistician; only a person answers it (10). |
| designed case | A patient of the dummy data made to test one rule of the plan, such as a dropout or a missing baseline (21). |
| dossier | The page out/eval/judge.html: each block to read, with its protocol sentence and your two commands (9). |
| dummy data | Made-up SDTM data to test a plan; never a result of a study (21). |
| dummy treatment | A made-up randomization on the real data, so that the programs are proven before the unblinding (18). |
| estimand | What a trial estimates, by ICH E9(R1): treatment, population, variable, intercurrent events with their strategies, and the summary (7). |
| FINAL | The state of a product made, fresh, from the approved SAP of now with every block agreed (20). |
| indication library | Sapling’s medians of the posted results of published trials, per therapeutic area, with their NCT ids (21). |
| intercurrent event | An event after the start of treatment that changes the meaning or the existence of the measurements, for example stopping the drug (7). |
| LOCF | Last observation carried forward: a missing value takes the patient’s last observed value (7). |
| MAR, MNAR | Missing at random: whether a value is missing depends only on data that were observed, so similar patients predict it. Missing not at random: it depends also on the missing value itself (7, 8). |
| omakase default | A conventional block that Sapling writes into the first draft, marked as a default for you to keep or change (4). |
| readiness | Your state of the plan: open decisions, agreed blocks, approval (5). |
| round | One evaluation of the plan, saved and compared with the last one (5). |
| SAP | The statistical analysis plan: what is analysed and how, final before the blind is broken (1). |
| SDTM | The data of a study as collected, in the CDISC standard (12). |
| sensitivity analysis | An analysis of the same estimand under other assumptions, to see whether the conclusion holds (8). |
| shell | A table without its numbers, reviewed before the unblinding (14). |
| TLF | Tables, listings and figures: the outputs of the analysis (15). |
| tipping point | The smallest shift of the missing values at which the conclusion of the primary test changes (8). |
| trace | The links from a failed line to the SAP block, the specification rule or the hook that made it (17). |
| type I error | The chance that a trial declares an effect when there is none (7). |
| unblinding | The moment the treatment of each patient is revealed for the analysis; a person’s act (18). |
| USDM | The structured protocol of CDISC (Unified Study Definitions Model) (2). |
Found an error or have a question? Write to datastep.studio@gmail.com· The plan is written in Sapling; every R program runs in DataStep Studio for R.
Data: the CDISC pilot study (cdiscpilot01). The protocol as CDISC USDM from the CDISC 360i project (MIT licence); the SDTM data from the PhUSE scripts repository (MIT licence). Public example data: no real patient.
R is a project of the R Foundation. SAS is a registered trademark of SAS Institute Inc. CDISC, SDTM, ADaM and USDM are trademarks of the Clinical Data Interchange Standards Consortium. Claude is a trademark of Anthropic. This book is not affiliated with or endorsed by any of them.
© 2026 DataStep Labs. All rights reserved.