Thursday, June 25, 2026
🇺🇸 FDA

Pharmacovigilance Tool · Disproportionality Analysis · Signal Management

Signal Detection Calculator

Free PRR, ROR, and chi-square calculator for pharmacovigilance signal detection. Screen spontaneous reporting data for possible signals of disproportionate reporting (SDR), then validate with case review, seriousness assessment, and causality workflows—not incidence or causality alone.

Quick Answer

Pharmacovigilance signal detection screens spontaneous reports for disproportionate drug–event pairs using a 2×2 count table. PRR (Proportional Reporting Ratio) compares event proportions among reports for the drug of interest versus all other drugs. ROR (Reporting Odds Ratio) compares reporting odds for the pair against the database background. Chi-square tests departure from independence in the table. A frequentist Evans-style screen flags a possible signal of disproportionate reporting (SDR) when PRR ≥ 2, chi-square ≥ 4, and the drug–event count (A) is at least 3—hypothesis-generating only, not causality or incidence.

Core signal detection formulas

PRR = (A / (A + B)) / (C / (C + D))
ROR = (A × D) / (B × C)
χ² = N(AD - BC)² / [(A + B)(C + D)(A + C)(B + D)]

N = A + B + C + D. Possible SDR rule used here: PRR ≥ 2, chi-square ≥ 4, and A ≥ 3.

Enter 2x2 Table Counts

Counts should come from the same spontaneous reporting dataset, analysis period, product scope, and adverse event coding level.

Possible SDR if PRR ≥ 2 χ² ≥ 4 A ≥ 3
2x2 table counts
2x2 count table for disproportionality analysis
Reports Event of interest Other events
Drug of interest Drug + event Drug + other events
All other drugs Other drugs + event Other drugs + other events

Screening interpretation

-

Enter counts and calculate to view the signal screen.

PRR
-
ROR
-
Chi-square
-
A count check
-
PRR ≥ 2 -
χ² ≥ 4 -
A ≥ 3 -

Disproportionality is hypothesis-generating. It does not estimate incidence, exposed-patient risk, relative risk, or causality.

How to Use This Calculator

1
Define one drug-event combination. Use a consistent product grouping and adverse event term, such as a MedDRA Preferred Term.
2
Extract A, B, C, and D from one spontaneous reporting source. Do not mix reporting systems or time windows in the same table.
3
Calculate PRR, ROR, chi-square, and the A count check. A possible SDR screen requires all three criteria shown above.
4
Review the case series, duplicates, seriousness, temporality, known labeling, biological plausibility, reporting bias, and need for further investigation.

Worked Example

Example spontaneous report table

Input: A = 6, B = 94, C = 120, D = 9,880. Total reports N = 10,100.

PRR: (6 / 100) / (120 / 10,000) = 5.00.

ROR: (6 × 9,880) / (94 × 120) = 5.26.

Chi-square: approximately 17.1 using the uncorrected 2x2 chi-square formula.

Interpretation: PRR ≥ 2, chi-square ≥ 4, and A ≥ 3, so this screen is a possible SDR. It remains a hypothesis requiring signal validation and clinical review.

Definitions of A, B, C, and D

A

Drug of interest + event of interest. This is the observed drug-event pair count and the minimum count used in the SDR rule.

B

Drug of interest + all other events. These reports define the background event distribution for the selected drug.

C

All other drugs + event of interest. These reports define how often the same event is reported with comparator products.

D

All other drugs + all other events. This completes the reporting background for the database extract.

Interpreting PRR, ROR, and Chi-Square

PRR

Proportional Reporting Ratio compares the proportion of the event among reports for the drug of interest with the proportion of the same event among reports for all other drugs.

ROR

Reporting Odds Ratio compares reporting odds for the drug-event pair against the rest of the database. It is useful for screening but should not be read as clinical odds or relative risk.

Chi-square

The uncorrected chi-square statistic tests departure from independence in the 2x2 table. In signal detection workflows, it is often used as a strength and stability screen.

Signal Management Workflow for Pharma Professionals

Under EMA GVP Module IX, disproportionality screening is the first statistical layer—not the endpoint. Marketing authorization holders detect signals from spontaneous reports, clinical studies, literature, and other sources; validate them with medical review; prioritize for further evaluation; and document outcomes in PSUR/PBRER cycles and regulator communication.

A typical workflow: (1) extract FAERS or EudraVigilance counts at a consistent MedDRA PT level with deduplication rules; (2) run PRR, ROR, and chi-square screens such as the Evans PRR rule; (3) review the case series for duplicates, masking, and reporting bias; (4) assess seriousness and WHO-UMC causality at case level; (5) confirm or refute the signal in validation before label or risk-minimization action.

Use this calculator for step 2. For downstream steps, link to our WHO-UMC Causality Assessment tool for case-level ADR categories, Seriousness Checker for ICSR seriousness criteria, and MedDRA Lookup to confirm Preferred Term coding before building disproportionality tables.

Important Caveats for Spontaneous Reporting Data

  • Disproportionality analysis does not calculate incidence, prevalence, exposure-adjusted risk, or population risk.
  • Reporting can be distorted by notoriety bias, stimulated reporting, media attention, litigation, new-product monitoring, and regional reporting practices.
  • Duplicate reports, vague case narratives, missing dates, and inconsistent MedDRA coding can materially change counts.
  • A high PRR or ROR can occur from sparse data. The minimum count check helps but does not eliminate small-number instability.
  • A negative screen does not exclude a safety issue, especially for underreported events or narrow patient subgroups.

Evidence & Sources

Frequently Asked Questions

A signal of disproportionate reporting (SDR) is a statistical association between a medicinal product and an adverse event in a spontaneous reporting database such as FAERS or EudraVigilance. It indicates the drug–event pair is reported more often than expected relative to background reporting. An SDR is a screening hypothesis for signal validation under EMA GVP Module IX, not proof of causality, incidence, or clinical risk.
The Evans proportional reporting ratio (PRR) rule, described by Evans et al. (2001), flags a possible disproportionality signal when three criteria are met simultaneously: PRR ≥ 2, chi-square ≥ 4 (uncorrected 2×2 test), and at least three reports for the drug–event pair (A ≥ 3). This rule is widely cited in EudraVigilance screening guidance but organizations may apply database-specific or product-specific thresholds in their signal management procedures.
Both ROR and PRR screen disproportionality in spontaneous reporting data but use different mathematics. PRR compares proportions of the event among drug reports versus among other-drug reports. ROR compares the odds of the drug–event pair to the odds in the rest of the database. They often rank pairs similarly but can diverge with sparse counts or extreme background rates. Neither is relative risk; both use report counts, not exposed-patient denominators.
Frequentist methods such as PRR, ROR, and chi-square apply fixed thresholds to observed counts (e.g., Evans PRR rule). Bayesian approaches—common in EudraVigilance—use shrinkage estimators or information component (IC) metrics that borrow strength from the full database and reduce false positives from rare events. Both are disproportionality screens; choice depends on database tooling, regulatory expectations, and validated organizational procedures under signal management SOPs.
FAERS (U.S. FDA) and EudraVigilance (EU) are separate spontaneous reporting systems with different report volumes, duplicate-handling rules, MedDRA versions, masking policies, and statistical screening tools. Counts for the same drug–event pair will not match across systems. Disproportionality tables must be built from one database, one analysis window, and one coding level—never mix FAERS and EudraVigilance counts in the same 2×2 table.
MedDRA Preferred Terms (PTs) are the standard adverse event coding level for regulatory safety reporting and most disproportionality workflows. Screening at PT level balances specificity with manageable case groupings. Higher levels (HLGT, HLT) aggregate events and can mask signals; lower levels (LLT) may split clinically similar events. Signal detection extracts should document the MedDRA version, hierarchy level, and whether grouped PTs or custom groupings were applied.
Duplicate or linked reports for the same patient and event inflate count A and can artificially raise PRR, ROR, and chi-square. Robust signal detection workflows deduplicate or link cases before building 2×2 tables, apply company-specific duplicate rules, and review whether duplicates cluster on one product. A high disproportionality screen driven by duplicates is a data-quality issue, not a validated safety signal.
Masking occurs when a drug–event disproportionality signal is hidden because the event is also reported frequently with many other products, diluting the comparator background. Conversely, notoriety bias can inflate signals when media attention stimulates reporting for one product. Masking and stimulation are reporting-bias phenomena; disproportionality alone cannot distinguish them—case review, epidemiology, and exposure context are required.
After a possible SDR screen, EMA GVP Module IX requires signal validation: assess case quality, seriousness, labeling, medical plausibility, duplicates, confounding, temporal patterns, and whether epidemiologic studies are needed. Validation may confirm, refute, or close the signal without regulatory action. Use our causality assessment and seriousness checker tools alongside clinical review—not PRR alone—to support validation documentation.
EMA GVP Module IX (Signal management) defines how marketing authorization holders detect, validate, prioritize, and document safety signals from all sources—not only disproportionality statistics. It links signal detection to risk minimization, PSUR/PBRER outputs, and regulator communication. Statistical screens such as PRR and ROR are inputs to the workflow; Module IX requires medical judgment and traceable signal documentation.
Very small counts are unstable: one or two spontaneous reports can produce extreme PRR or ROR values that fail to replicate. The A ≥ 3 minimum in the Evans rule reduces small-number false positives but does not guarantee clinical importance—a serious event with three reports may still warrant urgent review. Conversely, A ≥ 3 with a negative screen does not exclude rare but important risks.
A safety signal is a hypothesis that a product may be associated with an adverse event, often arising from disproportionality screening, case series, or epidemiology. Causality is the judgment whether the product actually caused the event in individual cases (e.g., WHO-UMC categories) or at the population level after integrated evidence review. A validated signal may lead to label updates or studies; causality assessment applies to case-level ICSR review—see our WHO-UMC causality tool for case workflows.

Related Pharmacovigilance Tools

WHO Causality Assessment SAE Seriousness Checker PT MedDRA Lookup SUSAR SUSAR Assessment ICSR ICSR Case Processing PV Pharmacovigilance Hub

Drug approvals, clinical trials, M&A, event intelligence, and company updates — weekly.

Customize topics & region
Topics
Region

No spam. Unsubscribe anytime. See our privacy policy.