[REQ] Database schema, naming convention and documentation

[peterdutey]

Overview

The Ramses data model follows a fact constellation schema design to accommodate the various many-to-many relationships that exist between medication prescriptions, hospital stays, and the many associated dimensions, chiefly time (particularly date/month/quarter/year), antibiotic classes (ATC, AWaRe, and other groupings).

Table naming convention

Every table generated by the Ramses package should be named prefix_domain_item_

Prefixes (mandatory component):

• fact: to designate fact tables storing Fact tables, which are the database representations of clinical processes such as medication prescription, medication administrations, or hospital encounters.
• dimension: to designate tables storing Dimensions, which characterise clinical processes
• bridge: to designate tables for facilitating many-to-many relationships between Facts and Dimensions.

Domain (optional component):

• medication for medication-related facts and dimensions
• microbiology for tables related to antimicrobial culture and susceptibility testing
• inpatient for tables related to inpatient wards
• outpatient for tables relating to outpatient department

Item (mandatory component):

• one or more words designating what a table row corresponds to.

Documentation

Users will need:

• a vignette presenting a database model diagram linking facts and dimensions and a demo of dplyr joins
• a data dictionary for each table
• signposting to training resources about database modelling

Related issues

#109 #73 #104