Causality Assessment

Causality assessment determines the likelihood that the suspected product caused the adverse event. VIGIA supports two algorithms: Karch-Lasagna and Ficha Técnica (FT).

Overview

VIGIA provides:

Karch-Lasagna Algorithm: ICH-endorsed method using 8 criteria
Ficha Técnica (FT) Algorithm: AI-powered assessment using product technical sheets
Dual-layer Decision: System suggestion + QF (Químico Farmacéutico) final decision
Classification Output: Definida, Probable, Posible, Improbable, Condicional, Inclasificable

Causality Algorithms

Karch-Lasagna
Ficha Técnica (FT)

Karch & Lasagna (1977) - Modified WHO-UMC criteria

8 Evaluation Criteria:

Temporalidad (Temporal relationship)
- Event occurred after drug administration
- Biologically plausible timeframe
Conocimiento Previo (Prior knowledge)
- Event documented in literature or product label
- Known adverse reaction pattern
Dechallenge (Drug withdrawal)
- Event improved after stopping drug
- Clinical recovery observed
Rechallenge (Re-administration)
- Event recurred when drug restarted (rarely done)
- Strongest evidence of causality
Alternativas (Alternative causes)
- Other possible explanations ruled out
- Concomitant medications, underlying disease
Factores Contribuyentes (Contributing factors)
- Patient factors (age, comorbidities, genetics)
- Drug interactions
Exploraciones (Investigations)
- Lab tests, imaging, biopsies
- Objective evidence of drug effect
Plausibilidad (Biological plausibility)
- Known mechanism of action
- Pharmacological rationale

Classification:

Score	Classification	Meaning
≥9	Definida	Highly certain causal relationship
6-8	Probable	Likely causal relationship
3-5	Posible	Possible causal relationship
1-2	Improbable	Unlikely causal relationship
0	Condicional	Insufficient data
N/A	Inclasificable	Cannot be evaluated

Karch-Lasagna Workflow

Open Causality Tab

In case detail view, navigate to Causalidad > Karch-LasagnaURL: /casos/{icsr_id}/causality

Generate AI Suggestion

Click Suggest to trigger AI analysis:

GET /api/v1/causality/karch-lasagna/suggest/{icsr_id}

The system:

Builds payload from case data:

build_karch_payload(icsr) → {
  "paciente": {"edad": 45, "sexo": "F", ...},
  "producto": {"nombre": "Paracetamol", ...},
  "evento": {"descripcion": "...", "gravedad": "Grave"},
  "temporalidad": "Event 2 days after first dose",
  "dechallenge": "Improved after stopping",
  ...
}

Sends to LLM (Gemini 1.5 Pro)

Receives structured response:

{
  "criterios": {
    "temporalidad": {"score": 2, "rationale": "..."},
    "conocimiento_previo": {"score": 1, "rationale": "..."},
    "dechallenge": {"score": 1, "rationale": "..."},
    ...
  },
  "score_total": 7,
  "clasificacion": "Probable"
}

Review System Suggestion

The interface displays:

System Suggestion: Probable (7 points)

Criterios Evaluados:✅ Temporalidad (2 pts): Evento ocurrió 2 días después de iniciar el medicamento. Timeframe biológicamente plausible.✅ Conocimiento Previo (1 pt): Reacción documentada en la literatura para este principio activo.✅ Dechallenge (1 pt): Mejoría clínica tras suspender el medicamento.⚠️ Rechallenge (0 pts): No se realizó re-administración (no indicado).✅ Alternativas (1 pt): Otras causas consideradas pero menos probables.✅ Factores Contribuyentes (1 pt): Paciente con función renal normal, sin interacciones conocidas.✅ Exploraciones (1 pt): Pruebas de laboratorio consistentes con reacción medicamentosa.⚠️ Plausibilidad (0 pts): Mecanismo no completamente establecido.Total: 7 puntos → Probable

QF Review and Adjustment

As the QF, review each criterion:For each criterion, you can:

Accept system score (click ✓)
Modify score (0-3 points depending on criterion)
Edit rationale (add clinical context)

Example adjustment:

Temporalidad:
Score (QF): [2 ▼]  (System: 2)
Rationale (QF): Acepto evaluación del sistema.
                 Timeframe consistente con hipersensibilidad tipo I.

The interface shows dual columns:

Sistema: AI suggestion
Q.F.: Your final decision

Calculate Final Score

As you adjust scores, the total updates automatically:

Sistema: 7 pts → Probable
Q.F.:    8 pts → Probable

Final Classification (based on QF score):

Score ≥9: Definida
Score 6-8: Probable ✓
Score 3-5: Posible
Score 1-2: Improbable

Add Global Rationale

Provide overall justification:

Justificación Global (Q.F.):

Clasifico como PROBABLE basado en:
1. Temporalidad adecuada (2 días post-inicio)
2. Mejoría tras suspensión (dechallenge positivo)
3. Reacción conocida para AINE
4. Ausencia de causas alternativas más probables

No se realizó rechallenge por contraindicación clínica.

Confirm and Save

Click Confirm Decision. The system:

PUT /api/v1/causality/karch-lasagna/confirm/{icsr_id}
Body: {
  "criterios": {...},
  "score_total_sys": 7,
  "score_total_qf": 8,
  "clasificacion_sys": "Probable",
  "clasificacion_qf": "Probable",
  "rationale_global_qf": "...",
  "version_prompt": "KL-1.0.0",
  "modelo_llm": "gemini-1.5-pro"
}

Creates a CausalityKarch record and updates icsr.ea_causalidad = "Probable"

Caching and Recalculation

The last causality assessment is cached. Clicking “Suggest” again returns cached result unless you:

Edit case data (product, event, dates)
Pass use_cache=0 query parameter
Manually click Recalculate

Ficha Técnica (FT) Workflow

Navigate to FT Tab

In case detail, go to Causalidad > Ficha TécnicaURL: /casos/{icsr_id}/causality/ft

Upload Ficha Técnica

Click Upload FT and select the product’s technical sheet (PDF, max 12MB)Supported: Official Fichas Técnicas with standard EMA/DIGEMID structure

POST /api/v1/causality/ft/upload
Form-data: file, icsr_id

OCR and Section Extraction

The system:

Renders PDF to PNG images (200 DPI)
OCR Processing with Tesseract (spa+eng)
Header/Footer Removal (pattern detection)

Section Detection using regex patterns:

SECTION_PATTERNS = {
  "contraindicaciones": r"4\.?3\s*[:\-]?\s*CONTRA[\s-]?INDICACIONES",
  "advertencias": r"4\.?4\s*[:\-]?\s*ADVERTENCIAS",
  "interacciones": r"4\.?5\s*[:\-]?\s*INTERACCIONES",
  "reacciones_adversas": r"4\.?8\s*[:\-]?\s*REACCIONES\s+ADVERSAS",
  "farmacodinamica": r"5\.?1\s*[:\-]?\s*PROPIEDADES\s+FARMACODIN"
}

Preview shows extracted sections:

✓ 4.3 Contraindicaciones (245 chars)
✓ 4.8 Reacciones adversas (1,824 chars)
✓ 5.1 Farmacodinámica (892 chars)

AI Analysis

Click Analyze. The system sends to Gemini 1.5 Pro:Prompt structure:

Eres un experto en farmacovigilancia. Analiza la siguiente Ficha Técnica
y determina la causalidad para el siguiente caso:

CASO:
- Producto: Paracetamol 500mg
- Evento: Hepatotoxicidad aguda
- Paciente: Mujer, 45 años, sin comorbilidades
- Dosis: 4g/día por 7 días

FICHA TÉCNICA:
[4.8 REACCIONES ADVERSAS]
Trastornos hepatobiliares:
Raros (≥1/10,000 a <1/1,000): Hepatotoxicidad en casos de sobredosis
o uso prolongado a dosis altas...

Evalúa:
1. ¿El evento está listado en la sección 4.8?
2. ¿La dosis/duración es un factor contribuyente?
3. ¿Existe contraindicación en el perfil del paciente?
4. Clasificación de causalidad (Definida/Probable/Posible/Improbable)

Response:

{
  "evento_listado": true,
  "seccion": "4.8",
  "frecuencia": "Raros (<1/1,000)",
  "dosis_relevante": true,
  "contraindicacion": false,
  "clasificacion": "Probable",
  "justificacion": "Hepatotoxicidad por paracetamol es una reacción..."
}

Review and Decision

The interface displays:

Análisis de Ficha Técnica

Resultado: ✅ Evento LISTADO en Ficha TécnicaSección: 4.8 Reacciones AdversasFrecuencia: Raros (≥1/10,000 a <1/1,000)Factores:

Dosis acumulada alta (4g/día × 7 días)
Sin alcohol ni otras hepatotoxinas reportadas
Función hepática normal previa

Clasificación (Sistema): ProbableJustificación: Hepatotoxicidad por paracetamol está documentada en la Ficha Técnica como reacción adversa rara, particularmente en casos de dosis altas o uso prolongado. El cuadro clínico y temporal es consistente. La ausencia de otras causas refuerza la causalidad.

As QF, you can:

Accept classification
Override with different classification
Add your rationale

Save FT Decision

Click Confirm:

PUT /api/v1/icsr-caso-prueba/{icsr_id}/ftqf
Body: {
  "ea_causalidad_ft": "Probable",
  "ea_causalidad_ft_justificacion": "Acepto análisis del sistema...",
  "ft_evento_listado": true,
  "ft_seccion": "4.8",
  "ft_frecuencia": "Raros"
}

FT vs Karch-Lasagna

You can perform both assessments:

Karch-Lasagna: Mandatory for regulatory reports
FT: Complementary, useful for product labeling decisions

If both are done, the Karch-Lasagna classification is used as the official ea_causalidad value.

Special Cases

Rechallenge

Rechallenge (re-administration after dechallenge) is the strongest evidence of causality but is rarely performed due to ethical concerns.When rechallenge happens:

Unintentional (patient restarts medication without medical advice)
No alternative treatment available
Event was mild and resolved

Scoring:

Positive rechallenge (event recurred): +3 points (Karch-Lasagna)
Classification often becomes Definida

Documentation:

Rechallenge: Sí
Resultado: Positivo
Detalles: Paciente reinició paracetamol por cuenta propia.
          Urticaria reapareció a las 4 horas.
          Se suspendió definitivamente.

Multiple Suspected Products

When multiple products are suspected:

Perform separate causality assessments for each product
Create multiple CausalityKarch records linked to same ICSR

Use producto_evaluado field to distinguish:

CausalityKarch(
  icsr_id=123,
  producto_evaluado="Paracetamol",
  clasificacion_qf="Probable"
)
CausalityKarch(
  icsr_id=123,
  producto_evaluado="Ibuprofeno",
  clasificacion_qf="Posible"
)

Mark the most likely product as producto_sospechoso

Drug-Drug Interactions

If the event is likely an interaction between drugs:

Primary assessment: Evaluate the triggering drug
Secondary assessment: Note interaction in Factores Contribuyentes

Document:

Factores Contribuyentes: Interacción medicamentosa.
Paciente recibió Warfarina + Paracetamol (altas dosis).
Aumento de INR consistente con interacción conocida.

Classification usually: Probable or Definida (if well-documented interaction)

Insufficient Information

If critical data is missing:

Classification: Condicional or Inclasificable

Document missing elements:

Información Faltante:
- Fechas exactas de inicio/fin del evento
- Dosis y duración del tratamiento
- Medicamentos concomitantes
- Resultado de dechallenge (medicamento aún en uso)

Next action: Initiate followup → Followup Management
Re-evaluate after receiving additional data

Quality Assurance

Before finalizing causality:

Narrative reviewed completely

All events coded in MedDRA

Temporal relationship documented (dates)

Concomitant medications listed

Dechallenge outcome documented (if applicable)

Alternative causes considered and excluded

Rationale clear and clinically sound

Classification consistent with score

API Reference

# Karch-Lasagna
GET /api/v1/causality/karch-lasagna/last/{icsr_id}
GET /api/v1/causality/karch-lasagna/suggest/{icsr_id}?debug=1&use_cache=0
PUT /api/v1/causality/karch-lasagna/confirm/{icsr_id}

# Ficha Técnica
POST /api/v1/causality/ft/upload
POST /api/v1/causality/ft/analyze/{icsr_id}
PUT /api/v1/icsr-caso-prueba/{icsr_id}/ftqf

See API Reference for complete endpoint documentation.

Get Started

Core Features

User Guides

Modules

Regulatory

Administration

Causality Assessment

Overview

Causality Algorithms

8 Evaluation Criteria:

Classification:

Workflow:

Karch-Lasagna Workflow

System Suggestion: Probable (7 points)

Caching and Recalculation

Ficha Técnica (FT) Workflow

Análisis de Ficha Técnica

FT vs Karch-Lasagna

Special Cases

Quality Assurance

API Reference

Build docs developers (and LLMs) love

Get Started

Core Features

User Guides

Modules

Regulatory

Administration

​Overview

​Causality Algorithms

​8 Evaluation Criteria:

​Classification:

​Workflow:

​Karch-Lasagna Workflow

System Suggestion: Probable (7 points)

​Caching and Recalculation

​Ficha Técnica (FT) Workflow

Análisis de Ficha Técnica

​FT vs Karch-Lasagna

​Special Cases

​Quality Assurance

​API Reference

Build docs developers (and LLMs) love

Overview

Causality Algorithms

8 Evaluation Criteria:

Classification:

Workflow:

Karch-Lasagna Workflow

Caching and Recalculation

Ficha Técnica (FT) Workflow

FT vs Karch-Lasagna

Special Cases

Quality Assurance

API Reference