Value of Perspective: Quantifying Decision Discordance in Health Economic Evaluation

A proof-of-concept framework for quantifying the Value of Perspective (VoP) in health economic evaluation. This project uses published economic evaluations from the Aotearoa New Zealand context to measure the decision impact of choosing between health system and societal perspectives.

The Value of Perspective Concept

When decision-makers evaluate health interventions, the choice of analytical perspective can lead to different conclusions:

• Health System Perspective: Includes only direct healthcare costs (hospitalisations, medications, clinical visits)
• Societal Perspective: Additionally includes productivity losses, caregiver burden, and broader social costs

The Value of Perspective (VoP) quantifies the potential loss from decision discordance—when the two perspectives yield conflicting cost-effectiveness conclusions. This serves as:

1. A trigger threshold: Identifying when perspective choice materially affects recommendations
2. A quantitative measure: Expressing opportunity cost in monetary terms (NZ$/QALY lost)

Key Features

Case Studies from Aotearoa New Zealand

This framework was developed using published economic evaluations relevant to the NZ healthcare context:

1. HPV Vaccination - Societal benefits include productivity gains from cancer prevention
2. Smoking Cessation - Broader costs captured through absenteeism and family impact
3. Hepatitis C Treatment - Reduced transmission creates societal value beyond direct healthcare

Quick Start

Installation

pip install vop-poc-nz

Basic Usage

from vop_poc_nz import run_cea

# Run analysis from both perspectives
health_system = run_cea(params, perspective="health_system")
societal = run_cea(params, perspective="societal")

# Calculate Value of Perspective
from vop_poc_nz.discordance_analysis import calculate_decision_discordance

discordance = calculate_decision_discordance(
    "HPV Vaccination",
    params,
    wtp_threshold=50000
)

if discordance["discordant"]:
    print(f"Decision discordance detected!")
    print(f"Value of Perspective: ${discordance['loss_from_discordance']:,.0f}")
    print(f"Equivalent to {discordance['loss_qaly']:.2f} QALYs lost")

Snakemake Pipeline

For reproducible full analysis:

snakemake -c1

Directory Structure

vop_poc_nz/
├── src/vop_poc_nz/               # Source package
│   ├── cea_model_core.py         # CEA Markov model
│   ├── discordance_analysis.py   # Value of Perspective calculations
│   ├── dcea_equity_analysis.py   # Distributional CEA
│   ├── value_of_information.py   # EVPI/EVPPI analysis
│   ├── sobol_analysis.py         # Global sensitivity
│   ├── dsa_analysis.py           # Deterministic sensitivity analysis
│   ├── bia_model.py              # Budget impact analysis
│   ├── visualizations.py         # Publication-quality plots
│   └── parameters.yaml           # Model parameters
├── tests/                        # Unit tests (pytest)
├── docs/                         # Documentation
├── output/                       # Generated outputs
└── Snakefile                     # Workflow definition

Key Outputs

Tables (CSV)

• comparative_icer_table.csv - Health System vs Societal perspective comparison
• value_of_perspective_table.csv - Discordance metrics by intervention
• parameters_assumptions_sources_table.csv - Full transparency documentation
• dcea_equity_results.csv - Subgroup analysis with equity metrics

Figures (PNG/PDF/SVG)

• Perspective comparison planes
• Decision discordance heatmaps
• CEAC/CEAF curves by perspective
• Tornado diagrams (1-way, 2-way, 3-way DSA)
• Lorenz curves and equity impact planes

Methodological References

• CEA: Drummond et al. (2015) - Methods for the Economic Evaluation of Health Care Programmes
• Perspectives: Sanders et al. (2016) - Recommendations for Conduct of Cost-Effectiveness Analysis (Second Panel)
• DCEA: Cookson et al. (2017) - Distributional Cost-Effectiveness Analysis
• VOI: Claxton et al. (2001) - The Value of Information
• CHEERS: Husereau et al. (2022) - CHEERS 2022 Reporting Guidelines

For detailed mathematical specifications, see FORMULAE.md.

Testing

pytest                              # Run all tests
pytest --cov=src --cov-report=html  # With coverage

Test coverage: 184 tests passing with 95%+ coverage

CI/CD

GitHub Actions workflows:

• CI (.github/workflows/ci.yml): Testing across Python 3.10-3.13, linting, type checking
• Release (.github/workflows/release.yml): PyPI publishing via Trusted Publishers (OIDC)

License

Apache License 2.0 - see LICENSE file for details.

Citation

If you use this framework in your research, please cite:

@software{mordaunt_2024_17759272,
  author       = {Mordaunt, Dylan A},
  title        = {Value of Perspective: Quantifying Decision Discordance in Health Economic Evaluation},
  year         = 2024,
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.17759272},
  url          = {https://doi.org/10.5281/zenodo.17759272}
}

Research Data & Materials

• Zenodo Archive: 10.5281/zenodo.17759272 - Citable software archive with DOI
• OSF Project: osf.io/unq76 - Supplementary materials and data

Acknowledgments

• Developed using published economic evaluations from the Aotearoa New Zealand context
• Inspired by ISPOR and PHARMAC guidelines for health economic evaluation