The IMPRESIVE evidence-preparation pathway, steps 1 to 9
A decision need is framed into questions covering disease burden, treatment use, effectiveness and safety. Data are characterized, then a fit-for-purpose assessment is made for each question and database. Pairs that pass go to the coordinating centre, which distributes one artefact per architecture: a common question, a common protocol, or a common data model with a shared analytical program. Site results return as aggregate output. The deliverables are pooling and assessing consistency, assessing transportability, evidence integration and visualization, and a fit-for-decision judgement. Pairs that fail bypass execution and go to second-best analysis, where the questions are redefined, data partners extended, or a prospective or external control used, feeding back into the decision need.
Deliverables
Identifying the decision need
Framing the question(s)
1
Disease burden
Treatment use
Effectiveness
Safety
Feasibility & Fit-for-purpose assessment
For each question and database
2
PASS
FAIL
Protocol & variable definition
3
Common question
Data Analysis Results
Common protocol
Data Same protocol Results
Common Data Model (CDM)
Data CDM Same analytical program Results
Standardization
4
Quality validation
5
Distributed execution
6
Evidence integration
7
Visualization and interpretation
8
Second-best analysis
Characterization is the deliverable
3bis
Question refined and re-scoped
Mapping extended, sites added
Prospective or external control
Next iteration
Rebuild the common layers
Fit-for-decision
“To what extent findings inform decision?”
The full pathway
Eight steps from decision need to decision
Select any box in the diagram to see what it means, which case exercised it, and where its technical detail lives.
Methods →