Skip to content

Appendix: Which OHDSI Tool Answers Each Check

Reference. Not part of the 30-minute clock.

Each feasibility check on the checklist is answered by a specific tool in the OHDSI stack, or by a conversation with your data steward. This appendix maps them, using the same numbering as the checklist.

The mapping

# Feasibility check Primary tool What you do with it
1 Concepts exist (and are standard) Athena and ATLAS → Search Search the vocabularies; confirm a standard concept, its domain, and its class before you build anything
2 Concepts are present in the data Achilles and ATLAS concept set → Included Concepts Achilles profiles record and person counts per concept in a source; the concept set panel shows what your rules resolve to, with those counts
3 Population fits Achilles (age, sex, and observation profiles), shown in ATLAS → Data Sources Read the age and sex distribution and the domain reports for the source; the domain reports show which kinds of records a source holds
4 Time can be anchored ATLAS → Cohort Definition, then Cohort Diagnostics Define the index event and windows; Cohort Diagnostics breaks down index-event timing and observation time
5 Outcome is capturable ATLAS → Search counts for the outcome concepts, and your data steward Check that the outcome is recorded as structured data in this type of source and in which domain; for the ALSFRS-R, count the records and ask where the scores are kept
6 Sample size is sufficient ATLAS cohort generation (attrition), then Cohort Diagnostics (incidence, cohort counts) Generate to see counts and attrition locally; Cohort Diagnostics is a common check before a network study
7 Governance clears Your data steward, the IRB, and any data use agreement Governance is a question for people and policy, not for a tool

Data quality is a separate question that runs alongside these checks. The Data Quality Dashboard (DQD) reports whether the CDM meets a set of conformance, completeness, and plausibility checks, which informs how much weight to put on the counts you see.

Notes on the workhorses

Achilles is the one that answers the most feasibility questions before you build anything. It is a characterization run over a data source that produces counts by domain, concept, age, sex, and calendar time. If your steward has run it and can show you the results, checks 2 and 3 can often be answered from those reports.

Cohort Diagnostics is the bridge from a single-site feasibility check to a network study. It runs a battery of diagnostics on one or more cohort definitions (incidence, index-event breakdown, cohort characterization, concept overlap) and is commonly reviewed before a network study is distributed. When you are ready to move from "feasible here" to "feasible across the network", this is the tool.

The Data Quality Dashboard (DQD) does not tell you whether your question is answerable. It reports whether the CDM passes a library of data quality checks. A source can contain your population and still fail checks that would undermine the study, so read the DQD results for the tables your question uses. DQD is not a governance tool, and it does not compare your counts with outside benchmarks.

How this connects to the curriculum

The daily curriculum teaches each of these tools in depth. This module's contribution is the discipline of reaching for them in the right order, before a protocol exists, so a dead end costs minutes instead of months.