ADDIS: aggregate data drug information system

What is it?

ADDIS is a data management and analytical tool used to assist in evidence-based decision-making in healthcare. ADDIS aims to enhance efficiency and transparency throughout the decision-making process, and make objective evidence and subjective value judgements explicit.

ADDIS uses a structured database of clinical trial results for:

  • combining and summarising the results of individual trials (evidence synthesis capabilities)
  • assessing trade-offs between favourable and unfavourable effects of treatments (‘benefit‑risk analysis’).

Addis Aggregate Data Drug Information System diagram

GetReal has focused on expanding ADDIS’s evidence synthesis capabilities and has also improved its data management capabilities.

The analytical components of ADDIS are also available in separate stand-alone web applications:

  • GeMTC : evidence synthesis
  • MCDA : benefit-risk analysis.

These stand-alone applications do not include the data management functionality, and do not allow evidence synthesis results to be automatically carried over into a benefit-risk analysis. The analysis functionality in ADDIS and the stand-alone applications are powered by open source packages for the R statistical software, partly developed by GetReal.

Addis Web app for data management and analysis diagram

Why is it useful?

  • State-of-the-art methods for evidence synthesis of aggregated data: including with network meta-analysis and network meta-regression.
  • Greater re-use of data between analysis projects: ADDIS includes features for sharing extracted study data, and datasets are publicly available. This enhances efficiency when updating or re-analysing existing datasets.
  • Access to study design information and results: these can be imported directly from ClinicalTrials.gov.
  • Structured benefit‑risk analyses: structured decision models make explicit both the underlying evidence and the subjective value judgements that lead to a decision. In ADDIS, these can be constructed from individual studies or from the results of network meta-analyses.

When is it useful?

  • Combining estimates of treatment effects through (network) meta-analysis: when the available evidence consists of aggregate data, that is, summary level estimates. Analyses can include randomised controlled trials as well as non-randomised comparative studies.
  • Benefit‑risk analysis.
  • For making data publicly available.
  • For using publicly available data.

What are the limitations?

  • Some analytical techniques are not supported by ADDIS:
    • analysis of individual participant data (IPD; as used by GetReal – see case studies on depression here and schizophrenia here). However, it may be possible to use GeMTC to do this – see here.
    • analysis of contrast-based data (comparative treatment effects)
    • understanding the limitations and practical feasibility of using patient-powered research networks for relative or comparative effectiveness research.
    • combining trial evidence on a new treatment with real-world evidence on a similar treatment to predict real-world effectiveness (as used by GetReal – see case study on schizophrenia here)
    • analysing datasets with missing values – methods of adjusting meta-analysis of aggregate data are not possible in ADDIS.
  • Data entered into ADDIS are public: there is no way to make data private. However, it may be possible for data to be private if the underlying R package is used on a local computer. For in-house or private use of ADDIS, please contact the developers.

What do stakeholders say?

  • A survey of stakeholders in academia, industry and health technology assessment (HTA) agencies was carried out to determine development priorities. Versions of the software were circulated for stakeholder feedback. Feedback for targeted user testing, which is incorporated in the new version of the software is reported here [insert link to report].

Key contributor

Gert van Valkenhoef, University Medical Center Groningen