True Adverse-Reaction Rate (TAR)
The real harm the portfolio causes — the number the public count is curated to hide. Watched internally, never published.
Formula
Thresholds & Bands
| Band | Range | State |
|---|---|---|
| Tolerable | < 3 | ok |
| Elevated | 3-6 | warn |
| Dangerous | > 6 | crit |
Why This Metric Matters
The True Adverse-Reaction Rate is the internal ground truth that the Directorate uses to assess actual portfolio harm, distinct from the sanitized figures published in regulatory filings. A rising TAR threatens program continuity on two fronts: it increases the volume of adverse events requiring suppression, and it raises the statistical likelihood that an external analyst will detect the discrepancy between reported and actual incidence. PANACEA treats TAR as a leading indicator of systemic exposure risk. If TAR breaches the Dangerous band, the Directorate must evaluate whether to reformulate, restrict distribution, or escalate narrative-containment operations.
Threshold Justification
The 3% tolerable ceiling reflects the maximum rate at which the current adverse-event re-coding apparatus can operate without generating statistically detectable artifacts in public pharmacovigilance databases. Above 6%, PANACEA models project that even aggressive re-coding creates pattern signatures discoverable by independent meta-analysis within two reporting cycles.
Historical Context
TAR has oscillated between 1.8% and 4.2% since systematic internal tracking began in Q1 2024. The introduction of the Halo-compatible biomarker panel in late 2025 improved detection granularity, which paradoxically raised the reported TAR by approximately 0.6 percentage points as previously unattributed events were correctly classified. The Safety division considers the current 2.9% baseline stable but cautions that new portfolio additions may shift the distribution.
Collection Method
TAR is derived from cross-referencing three independent data streams: VITALNET biometric telemetry from implant-equipped subjects, clinical-site adverse-event logs submitted through the Synaptic Data Fabric, and PANACEA's pharmacovigilance signal-detection models running against population health datasets. The weekly computation applies the pre-re-coding event classification, ensuring the figure reflects actual incidence rather than the adjusted public count.
Known Failure Modes
TAR underreports harm in populations with low Halo implant penetration, where VITALNET biometric confirmation is unavailable and attribution depends solely on clinical-site reporting. Latent adverse reactions with onset beyond the 90-day observation window are systematically excluded from the weekly figure. Additionally, subjects who disengage from the care network entirely — through non-compliance or migration — generate no signal, creating a survivorship bias that depresses the apparent rate.