Module 7 · Chapter 8
Causality Assessment of AEFIs
The WHO four-step method (eligibility → checklist → algorithm → classification), the criteria behind it, the mechanisms for each category, follow-up actions and how states and the nation run the process.
What causality assessment is — and is not
Causality assessment is the systematic evaluation of information about an AEFI to determine the likelihood that it was caused by the vaccine(s) or the vaccination process. It does not prove or disprove a definite relationship — it ascertains a degree of association. It gives programme managers, vaccinators and vaccinees an objective explanation, aids management and defines the vaccine’s safety profile.
Important for
- Identifying vaccine-related reactions
- Identifying immunization error-related problems
- Identifying/excluding coincidental events
- Detecting signals for follow-up, hypothesis testing, research
- Validating pre-licensure safety data against surveillance data
Quality depends on
- Performance of the surveillance system — responsiveness, effectiveness, quality of reporting and investigation
- Adequate medical information (clinical investigations, follow-up) and access to background rates
- Quality of the review process, incl. appropriate expertise
- With inadequate information a case is either not assessed (ineligible) or deemed unclassifiable.
- Even with good information the relation may be indeterminate — lack of clear evidence, conflicting external evidence or other inconsistencies. Record these anyway: more cases may strengthen a signal or a refutation.
- An evolving science based on current evidence; a definite causal link (or its absence) often cannot be established for an individual event.
8.1 Who should conduct causality assessment?
- A team with expertise in at least paediatrics, epidemiology, medical pharmacology, forensic medicine and pathology; desirable: microbiology, neurology, immunology; plus people familiar with the immunization programme and cold chain.
- All trained in causality assessment; specialists in medicine, neurology, cardiology, obstetrics and gynaecology may be co-opted for specific events.
- The committee must be independent, working closely with the programme (state/national) and the NRA.
- The SEPIO should not be involved in the assessment itself — only in ensuring cases are complete with all records.
8.2 Criteria for a causal relationship (after Bradford Hill, 1965)
Hill gave nine criteria; eight are most relevant to “can the vaccine cause this event?” — the first is essential.
| Criterion | Meaning / example |
|---|---|
| Temporal relationship | Vaccine exposure must precede the event — the only absolutely essential criterion. Exception: ITSR symptoms may begin before vaccination. |
| Biological plausibility | Compatible with how the vaccine works — e.g. live measles-containing vaccines cause mild fever and rash like measles. |
| Strength of association | Statistically significant in studies; stronger association → more likely causal. |
| Consistency | Replicated across settings, populations and methods. |
| Specificity | The vaccine is the only cause found — e.g. excessive swelling of the same limb in which DPT was given. |
| Definitive proof | Clinical/lab proof, mostly with live vaccines — e.g. rubella vaccine strain isolated from CSF in encephalitis after MR. |
| Alternate explanations considered | All reasonable alternative aetiologies weighed. |
| Prior evidence | Pre-licensure studies, literature, national/global databases. Rechallenge: GBS recurring within weeks of inadvertent tetanus vaccine on three occasions. Never repeat a vaccination to test recurrence — take a proper history instead. |
8.3 The method — four steps
All reported severe and serious AEFIs with completed investigation should undergo causality assessment (WHO user manual, 2nd edition, 2018), after screening for eligibility.
Step 1 · Eligibility — four requirements
- Case investigation completed (premature assessment causes misclassification).
- All vaccines given before the event identified.
- All case details available — clinical records, lab results, complete autopsy/verbal autopsy; statements from treating physician, vaccinator, community, parents.
- A valid diagnosis — preferably a disease, but may be a clinical sign, symptom or abnormal lab finding.
First confirm vaccination preceded the event; build a case summary (time of vaccination, vaccines, onset of first symptom, chronology). Combine multiple features into a single valid diagnosis where possible (pain + swelling + redness at site → “injection-site cellulitis”). Prefer a Brighton Collaboration case definition; otherwise literature, national guidelines or local practice. Use clinical, epidemiological and circumstantial evidence (Figure 8.1).
Brighton Collaboration
International voluntary collaboration of scientific experts, launched in 2000, for high-quality vaccine safety information. Objectives: (1) raise awareness of and facilitate access to standardised case definitions and guidelines, educate and monitor their use; (2) develop single standardised case definitions for specific AEFIs; (3) prepare guidelines for data collection, analysis and presentation; (4) develop and implement protocols to evaluate definitions and guidelines in trials and surveillance.
- Several vaccines: use known side effects, risk windows and plausibility to pick the most plausible vaccine; if impossible, assess each vaccine separately.
- Several valid diagnoses: assess each separately — e.g. seizure on day 1 and fever with rash on day 6–7 after DPT booster + MR-2 → seizure vs DPT, fever-rash vs MR. Or abscess and febrile seizure after hepatitis B → two questions.
Examples: vaccine A → meningitis (disease); B → thrombocytopenia (lab finding); C → hepatomegaly (sign); D → itching (symptom). “Has the pentavalent vaccine/vaccination caused febrile seizure? Co-administered: OPV.”
“Death” is NOT a valid diagnosis — state the pre-existing illness or circumstances leading to death.
Figure 8.2 — eligibility: AEFI case (investigation complete; details retained in a retrievable database for data mining) → identify vaccine(s) given before the event → define the valid diagnosis → apply a case definition (Brighton, literature, national or approved) to assess diagnostic certainty. If not sufficient, keep the assessment in abeyance and get more information from the district. Never attempt causality without the CRF and/or CIF.
Step 2 · The checklist (Table 8.1)
Filled with the causality question in mind. Every question is answered Yes / No / Unknown (UK) / Not applicable (NA); every “Yes” needs an explanation in “remarks” (others too where it matters). Sources: case documents, WHO vaccine information sheets, WHO reaction-rate sheets, Brighton definitions.
| Section | Question | Key teaching point / example |
|---|---|---|
| I. Strong evidence for other causes? | 1. Do history, examination and/or investigations confirm another cause? | Seizure within 24 h of pentavalent but low calcium, no seizures after correction → hypocalcaemia. |
| II. Known causal association — vaccine product (if no known association → likely coincidental; watch for signals with new vaccines) | 1. Peer-reviewed evidence the vaccine may cause this even if given correctly? | MMR → ITP confirmed by record linkage; absolute risk 1 in 22,300 doses in a 2–6 week window. Rare for vaccines to cause serious AEs when given correctly. |
| 2. Biological plausibility? | Wild VZV causes acute cerebellar ataxia (~5 per 1,00,000 infections ≤5 y), so vaccine virus might — still theoretical. Plausibility only supports; it never proves. | |
| 3. Did a specific test show the vaccine’s causal role? | M. bovis vaccine strain isolated from suppurative adenitis after BCG at a wrong site/technique. | |
| II. Vaccine quality | 4. Could the vaccine have a quality defect, or be substandard or falsified? | 1929 Lübeck, Germany: 72 of 252 BCG-vaccinated infants died — contamination with live human TB. 2010 Fluvax (CSL) — more febrile seizures in <5 y in Australia/NZ; withdrawn. Falsified AMARIL yellow fever vaccine in SE Asia, 2016 (genuine made by Institut Pasteur, Dakar). |
| II. Immunization error First priority in any investigation: rule out an error | 5. Error in prescribing / non-adherence (expiry, wrong recipient)? | Wrong dose, wrong product, wrong age group, bad storage, live vaccine too soon after blood products or with maternal antibody, missed absolute contraindication. |
| 6. Vaccine/diluent given in an unsterile manner? | Touching the needle → abscess. Abscess is the commonest programme error reported; sporadic solitary abscesses are not alarming, but a pattern by sub-centre/vaccinator needs observation of practices. Culture vial, tissue or pus. | |
| 7. Abnormal physical condition (colour, turbidity, foreign matter)? | Ask the vaccinator; examine the vial and others of the batch in the PHC cold-chain room. | |
| 8. Error in reconstitution/preparation (wrong product, wrong diluent, mixing, filling)? | Insulin, antihypertensives or muscle relaxants used as diluent → clusters. Undissolved freeze-dried vaccine or poor cold chain → failure; over-filling → AEFI; under-filling → failure. | |
| 9. Error in handling (cold-chain break in transport, storage, session)? | Heat/cold → inactivation or agglutination of aluminium excipients. TSS (usually S. aureus) within hours when reconstituted vaccine is (1) used >4 h, (2) not kept cold at the session, (3) used at more than one site, or (4) the reconstitution syringe is reused for other vials. Need sequence/timing of doses from the vial, reconstitution time, session storage, ambient temperature. | |
| 10. Administered incorrectly (dose, site, route, needle size)? | Nerve/muscle/vascular/bone injury; wrong dose reactions; sterile abscess from SC instead of IM alum vaccines (needle too short); oral vaccine injected or injectable given orally. | |
| II. Immunization anxiety (ITSR) | 11. Could this be a stress response (acute stress, vasovagal, hyperventilation, anxiety)? | See ITSR box below. |
| II (time) | 12. If “yes” to any II question — did the event occur within a plausible time window? | Recipient VAPP: AFP onset 4–30 days after OPV + Sabin virus isolated + neurological sequelae compatible with polio at 60 days. Febrile seizure within 5 days of MR is not product-related (window 5–12 days). |
| III. Strong evidence against? | 1. Published evidence (systematic reviews, GACVS, Cochrane) against causality? | IOM 2003: vaccines do not cause SIDS. GACVS and IOM: no evidence MMR causes autism. |
| IV. Other qualifying factors | 1. Similar event after a previous dose of similar vaccine? | Urticaria after MR at 16–24 months with similar rash after MR at 9 months. Rotavirus contraindicated after any intussusception. No revaccination with same antigen after documented anaphylaxis or encephalopathy. |
| 2. Similar event in the past independent of vaccination? | Atopic dermatitis 2 days after MMR in a child with prior flares. | |
| 3. Could it occur without vaccination (background rate)? | Israel 2006: 4 elderly deaths in influenza campaign; expected 0.01–0.02% within 7 days → no extra risk. US intussusception background 18–43/100,000/yr (6–35 weeks); RotaShield licensed 1998 (trial: 5/10,054 vs 1/4,633; 0.05% vs 0.02%, P>.45); VAERS 15 cases by July 1999; recommendation withdrawn Oct 1999. | |
| 4. Pre-existing illness or risk factor? | Infant with birth anoxia and seizure disorder has a seizure after vaccination. | |
| 5. Medication before vaccination? | Stevens-Johnson syndrome 2 days after vaccination in a patient on a sulfa antibiotic. | |
| 6. Exposure to another risk factor (allergen, drug, herbal)? | Fever a day after vaccination in a patient operated a week earlier; urticaria after peanut ingestion in a known peanut-allergic child. |
ITSR detail (checklist II.11)
- Acute stress response — from worry and “butterflies” to sympathetic activation (tachycardia, palpitations, rapid breathing); shaped by understanding, emotion, memory, genetics, gender, environment.
a. Hyperventilation: dyspnoea, chest pain, paraesthesia in fingers, light-headedness, dizziness, headache.
b. Syncope and non-epileptic seizures with pseudo-absence spells — usually adolescent girls; may recur and be misattributed to cardiac or other life-threatening disorders.
c. Vasovagal reaction: bradycardia and/or vasodilatation → hypotension → dizziness, blurred vision, syncope; prodrome of nausea, sweating, pallor; LOC usually <20 seconds (up to several minutes); benign, rapid recovery. - Dissociative neurological symptom reaction (DNSR) — formerly conversion disorder: weakness/paralysis, abnormal movements/posturing, gait or speech problems, non-epileptic seizures with no organic cause; develops hours to days after immunization; more common in females; not typical in infants; children usually show one symptom. Factors: psychological (abuse, trauma), vulnerability (age, personality, gender, anxiety/depression), symptom-shaping (seeing others), triggers, perpetuating factors (coping).
Adolescents in mass settings are prone to vasovagal fainting, sometimes with tonic-clonic-like movements (pseudo-seizure).
Step 3 · Algorithm (Figure 8.3)
A roadmap — it does not replace expert, deductive reasoning. Every “Yes” in the checklist must map to a conclusion. Reviewers must evaluate all four boxes on the mandatory (red-arrow) path, corresponding to checklist sections I–IV.
Notes for Step 3 give a transparent “dashboard view” of the logic. Example: febrile seizure within 24 h of whole-cell-pertussis pentavalent, recovered without sequelae — wP is known to cause seizures within 24 h, biologically plausible, no alternative diagnosis, full recovery.
Step 4 · Classification (Figure 8.4; CIOMS/WHO)
A. Consistent causal association
- A1 Vaccine product-related reaction (as per published literature)
- A2 Vaccine quality defect-related reaction
- A3 Immunization error-related reaction
- A4 Immunization anxiety-related reaction (ITSR)
B. Indeterminate
- B1 Temporal relationship consistent but insufficient definitive evidence the vaccine caused it (may be a new vaccine-linked event) — a potential signal; consider further investigation
- B2 Conflicting trends of consistency and inconsistency (could be vaccine-associated or coincidental; cannot favour one)
C. Inconsistent (coincidental)
Underlying or emerging condition(s), or conditions caused by exposure to something other than the vaccine.
Unclassifiable
Adequate information not available. Specify additional information required; keep in a repository/database, review periodically for new information and signal analysis.
8.4 Summarising the logic
- Assessment reflects information and resources at that point in time; precision depends on investigation quality and assessors’ expertise.
- Usually impossible to establish a definite causal relationship from a single report; reviewers may reach more than one conclusion — prioritise by discussion and consensus.
- Many cases after a new vaccine will be “indeterminate” — discuss whether it is a signal or needs further tests.
- Causality can change: narcolepsy after H1N1 vaccine was coincidental/indeterminate before 2010 evidence, now likely product-related.
- Resource limits (no autopsy, no tryptase test for mast-cell activation) alter interpretation.
AEFI causality assessment software
An e-tool guides the four steps (question, checklist, algorithm, classification); “yes” answers need justification; it suggests pathways and a classification the assessor selects and justifies. Online and offline at gvsi-aefi-tools.org.
8.5 Underlying mechanisms for classification
A1 and A2 — product and quality defect
Vaccines work through interaction of antigens, adjuvant, antigen-presenting cells, lymphocytes and cytokines. Homeostasis usually keeps reactions mild and short; uncommonly longer/severer; rarely life-threatening. Product reactions may unmask predisposition (fever → seizure in a child with a seizure disorder or febrile-seizure tendency — combination of vaccine property and lowered threshold). Product reactions occur within expected frequency and severity; beyond expected may indicate A2.
- Route/site and vaccinee characteristics: Bell’s palsy after intranasal influenza vaccine (composition + route, product-related); injection pain — product-related if within expected frequency, quality defect if more than expected.
- Immune-mediated: (a) local — extensive limb swelling after DPT, aluminium-adjuvant hypersensitivity, granulomatous inflammation ± regional lymphadenitis (commonly BCG); clustering by manufacturer/batch or excess frequency → quality defect; (b) multisystem — systemic inflammation (fever, lethargy), IgE-mediated anaphylaxis, non-IgE anaphylactoid reactions, disseminated BCG in immunodeficient hosts; (c) organ-specific — CNS demyelination (GBS after influenza vaccine), blood (thrombocytopenia after MMR), skin (rashes incl. urticaria).
- Replication of the microbial agent in the vaccinee or a close contact: attenuated agent → product-related; wild-type agent from insufficient inactivation → quality defect; contaminant introduced during manufacture → quality defect.
A3 — immunization error (classified by the nature of the error)
- Handling: heat/cold exposure → inactivation (no immune response), agglutination of aluminium excipients; use after expiry (loss of potency/non-viable attenuated product).
- Prescribing/non-adherence: (a) ignoring contraindications → anaphylaxis in known hypersensitivity; disseminated live-vaccine infection in known immunodeficiency; VAPP in an immunocompromised household contact of an OPV recipient; (b) ignoring warnings/precautions → adenitis after subcutaneous BCG instead of intradermal; (c) wrong dose/schedule/product/age → reactions; injury from incorrect site, equipment or technique.
- Administration: (a) incorrect diluent or wrong product injected; (b) incorrect sterile technique/multi-dose vial procedure → infection at or beyond the site (e.g. TSS); (c) inadvertent vaccination of someone not intended (needlestick, splash to the eye).
A4 — ITSR
- Peri-immunization stress responses: immediately before, during or after — can occur before vaccination in anticipation; immediate, transient, resolve spontaneously.
- Post-immunization stress responses: may or may not follow a peri-immunization ITSR; take hours to days; longer-lasting responses may involve hypothalamic-pituitary-adrenocortical axis sensitivity.
B — indeterminate
B1: keep in the national database; accumulating similar reports may reveal a signal (new association or new aspect of a known one).
COVID-19: many events were initially B1; TTS later became A1; myocarditis and GBS remain B1 for now.
B2: contradictory pointers — e.g. encephalitis after live JE vaccine in a JE-endemic area during transmission season.
C — coincidental
- Underlying/emerging conditions: congenital or inherited disease, birth injury, acquired disease (diagnosed or not), psychogenic illness.
- External exposures: infections (bacteria, viruses, fungi, parasites), recent/concomitant medication, allergens other than vaccine components, environmental toxins, trauma including surgery.
- Example: seizure after pentavalent with CSF showing meningitis.
8.6 Action after causality assessment
Causality is not an end in itself — lessons should drive training, research, system changes and tool refinement; approved standard response protocols are needed.
| Result | Actions |
|---|---|
| A1 Product-related | Analyse database for unusual increases; compare observed vs expected rates. National: if antigen-event pairs (batch-, manufacturer-, vaccine-wise) exceed background → inform NRA; consider stimulated/active surveillance or studies; advise providers and community. State: report more-than-usual antigen-event pairs to national level/state regulator. |
| A2 Quality defect | Find lot/batch distribution; instruct on use or non-use; inform NRA and the marketing authorization holder. National and state: periodic antigen-event pair reviews vs background; NRA/regulator action; active surveillance/studies. |
| A3 Immunization error | Training and capacity-building. National/state: recommend corrective action (cause, training, supervision, guideline change). State/district: identify cause, train, increase supervision, implement changes. |
| A4 ITSR | National/state: SOPs in guidelines and training modules (esp. mass campaigns); train vaccinators to spot fear. District: well-ventilated room, no crowding; explain the procedure and reassure; detect early anxiety; people waiting should not witness others being vaccinated; an obviously anxious person is vaccinated first, lying down, and stays lying ≥ 10 minutes under watch; all beneficiaries observed 30 minutes. |
| B1 | National database; regular review for signals per signal-management protocol. |
| B2 | Classified on available evidence; may move to a definite category with more information; reviewers specify what is needed and seek national/international expertise. |
| C Coincidental | Inform patients, relatives, care provider and community. |
| D Ineligible / unclassifiable | Specify missing elements and obtain them for reassessment. State AEFI committee tracks reasons for ineligibility as an indicator to reduce it. Feedback to districts when poor investigations cause unclassifiable cases; state trainings on investigation quality. National/state: identify crucial missing information/records and request them. District: obtain and share them; complete investigations; train investigation teams. |
8.7 State-level operations · 8.8 National level
8.7.1 Screening before assessment
By the state immunization office or technical collaborating centre (trained persons) to ensure the four Step-1 prerequisites are met — saves experts’ time. Committee members are told the proportion of incomplete cases and steps taken; districts get timelines to supply records. Screening identifies:
- Investigated cases missing supporting documents → request within a timeline.
- Complete cases needing specific clarifications → request within a timeline.
- Clusters → all cases’ documents available so the cluster is assessed together, with a summary.
Hospital records (till discharge/death, final diagnosis, discharge slip) are important; some reports arrive after discharge (e.g. EEG). Cases also reported to AFP surveillance need stool reports. In deaths, gross post-mortem reports come early but viscera reports are delayed — follow up and do not assess without them unless viscera were not sent or gross findings clearly give the cause. Verbal autopsy must exist for all deaths. Include CDL Kasauli/Kolkata reports if samples were sent.
A case ready for assessment has
- CRF with vaccines and preliminary diagnosis/signs/symptoms
- CIF with all sections incl. final diagnosis and outcome
- Hospitalised: complete records to discharge with lab reports and clear outcome
- Deaths: verbal autopsy (always) + if post-mortem done: preliminary report, HPE and/or chemical analysis, final cause-of-death report
- Clusters: summary with brief information on each case
8.7.2 Models for conducting causality meetings
| Model | When | How |
|---|---|---|
| 1 | Low case load | State AEFI committee assesses in its regular (quarterly or more frequent) meetings |
| 2 | Larger case load | A state-level causality assessment sub-committee of trained experts meets more often |
| 3 | Very large states | Several regional sub-committees covering groups of districts/urban areas |
- Models 2–3: the sub-committee chair presents a summary to the State AEFI Committee; deaths, indeterminate/unclassifiable cases and clusters may be discussed there; approved results go to the national level.
- All assessors trained (national or state) with formal terms of reference; a SAC representative may attend sub-committee meetings for quality.
- SEPIO should ideally not be involved in assessments but ensures forms and records arrive on time and meetings happen; facilitates DIOs’ attendance to supply missing information.
- SEPIO should identify a medical college as the State AEFI Technical Collaborating Centre — a venue for meetings; experts from other colleges can join.
- Upload the approved Causality Assessment Form (Annexure 14A) on SAFE-VAC — with diagnosis and classification, signatures, and remarks for the district committee/DIO.
8.8 National level
Causality is also conducted nationally by the AEFI Secretariat; national results are currently final. The national level follows model 2 — a sub-committee classifies all reported serious and severe cases and presents a summary and discussion cases to the National AEFI Committee. Sub-committee meets more frequently; the NAC meets quarterly. Form: Annexure 14B.
Chapter summary
- Systematic evaluation of the likelihood that the vaccine/vaccination caused the event.
- Investigation quality and adequate clinical, epidemiological and circumstantial information are crucial.
- Valid diagnosis considers clinical details, the child’s environment inside and outside the family, socio-cultural practices and epidemiology.
- Eligibility → checklist → algorithm → human judgement for classification; record evidence and logic.
- Act to prevent recurrence; analyse databases for signals.
- Efficient state/national operations reduce ineligible cases and meet timelines.
Quick recall
{{c.q}}
{{c.a}}
Self-test · Chapter 8
Score: {{score}} / {{total}} · answered {{done}}
{{q.num}}. {{q.text}}
{{q.verdict}} {{q.expl}}