Start of project: pretty much right after D4D formed; gather Medicare drug spending data and analyze to find interesting trends, anomalies, etc.
- CMS.gov (for Medicare drug spending data)
- FDA (for FDA drug codes, which are IDs assigned to each individual drug)
- Genome.jp (for ATC codes, which map drugs to their usages)
- Centerwatch.com (for mappings between drug names and their usages)
All data now housed either at data.world (mostly there) or in the project’s Git repository:
- https://github.com/data4democracy/drug-spending
- https://data.world/data4democracy/drug-spending
- Several files annotated with URL(s) where data was originally sourced
- For a short while, both co-leads had to step away at roughly the same time; this made work on the project a bit less cohesive and resulted in some confusion
- Lack of specific questions: we knew Medicare drug spending was worthy of our attention, but we didn’t know where to start or what had already been explored
- Couldn’t find any subject matter experts until later in the project’s life; by that time, activity had already died down considerably
- Led to a lack of actionable tasks, even with available volunteers
- Led to lack of interest/activity in the project’s Slack channel
- Difficult to cross-reference drug names with other information (e.g., treatments, conditions)
- Drug names and authorized usages differ by brand, locality, and method of use
- Many different sources of truth
- (Minor) Multiple programming languages in use
- Some volunteers knew one, some knew another
- # of available reviewers for any given PR depended on the language the submitter used