Written and maintained by CASRAI Editorial Board
Last updated
A Standardised MedDRA Query (SMQ) is a pre-built grouping of MedDRA terms — mostly Preferred Terms, occasionally Lowest Level Terms — that together represent a defined medical condition or area of pharmacovigilance interest, such as anaphylactic reaction, agranulocytosis, or drug-induced liver injury. SMQs exist so that anyone retrieving case reports for that topic from a safety database is running the same, reproducible search, instead of every reviewer building an ad hoc free-text query and getting a different case set.
This page is about using SMQs, not just defining them: how narrow and broad scope differ, what an algorithmic SMQ adds on top of plain term membership, how hierarchical SMQs let you drill from a wide topic into a narrower one, and where each of these choices actually shows up in a signal-detection run or an aggregate safety report.
The one distinction that matters most in practice: broad scope is not a different topic from narrow scope — it is narrow scope plus additional less-specific terms. Every case a narrow-scope search returns is also returned by the broad-scope version of the same SMQ. Confusing the two, or defaulting to broad scope out of caution, is the most common way a signal-detection run drowns in false positives instead of narrowing in on a real one.
What an SMQ Actually Groups
Each SMQ is built around a defined clinical concept and lists the MedDRA Preferred Terms that can represent it — everything from the textbook presentation down to terms that only sometimes indicate the condition. They are developed under ICH oversight and maintained and distributed by the MSSO with each biannual MedDRA release, which is also why an SMQ’s exact term membership can shift from one MedDRA version to the next — a case count generated against one release is not automatically reproducible against another unless you record which version you ran.
Current MedDRA releases include well over 100 SMQs, covering both well-defined syndromes (torsade de pointes, anaphylactic reaction) and broader safety topics assembled from multiple underlying mechanisms (hepatic disorders, GI perforation). Not every safety topic has a corresponding SMQ — see the company-query alternative below for what to do when one doesn’t exist or doesn’t fit.
Narrow Scope vs. Broad Scope: The Core Design Choice
Most SMQs ship with two term lists:
- Narrow scope — terms with high specificity for the condition of interest. A case retrieved under narrow scope is highly likely to be a genuine instance of what you’re looking for. The trade-off is lower sensitivity: atypical presentations, ambiguous sign/symptom terms, and borderline cases fall outside it.
- Broad scope — the narrow-scope terms plus additional terms that are less specific, could represent the condition, or commonly co-occur with it. Sensitivity goes up (fewer real cases missed); specificity goes down (more manual triage to separate genuine hits from noise).
Because broad scope is a strict superset of narrow scope, running broad scope never loses anything narrow scope would have found — it only adds retrieval volume and review burden on top. That makes the choice less “which is correct” and more “how much manual triage am I prepared to absorb for this specific task.”
Algorithmic SMQs: A Scoring Layer, Not a Third Scope
A subset of SMQs are algorithmic: on top of plain term membership, they apply a defined combination rule — for example, requiring a qualifying term together with a supporting laboratory-value threshold, or requiring two or more terms from different sub-groupings to co-occur in the same case — before a case counts as a hit. The Hepatic Disorders SMQ is a commonly cited example, where liver-enzyme laboratory data is combined with clinical terms to raise specificity beyond what term matching alone could achieve.
Algorithmic scoring is not available for every SMQ — most rely on term membership alone — and where it exists, it sits on top of the narrow/broad scope choice rather than replacing it; you still choose a scope, and the algorithm further filters within it.
Hierarchical (Multi-Level) SMQs
Some broad safety topics are structured as a parent SMQ with several narrower child SMQs nested underneath it, each covering a more specific sub-condition within the parent’s territory. This lets a reviewer start at the wide topic level for an exploratory pass and then drop into a specific sub-SMQ for a more targeted follow-up, without constructing a new search from scratch or trying to guess which terms belong to the narrower question. Whether a given SMQ has this structure, and how many levels deep it goes, is documented per-SMQ in the MSSO’s released SMQ introductory guide and term lists for the version in use — check it rather than assuming a flat single-level list.
Where SMQs Fit in Signal Detection and Aggregate Safety Review
SMQs are a retrieval tool, not the analysis itself — they define which cases you pull, not what you conclude from them. In practice they show up at two different points in the pharmacovigilance cycle:
- Signal detection. A periodic disproportionality screen against a spontaneous-reporting database (a company safety database, EudraVigilance, FAERS) typically runs broad scope first, precisely because the goal is not missing anything worth a closer look. Once a possible signal is identified, the case series gets rebuilt with narrow scope — and, where one exists, the algorithmic filter — to produce a cleaner set for clinical review rather than working the full broad-scope noise by hand.
- Aggregate safety review. In periodic reports built to the ICH E2C(R2) PSUR/PBRER format and in DSURs, SMQs are the standard mechanism for defining and reporting cumulative case counts for a topic of special interest across the reporting period. Using the standard query — and recording its scope and MedDRA version — is what makes the resulting case count reproducible across report cycles, across reviewers, and comparable against another sponsor’s report on the same topic, instead of resting on a bespoke search nobody else can rerun.
Both uses assume MedDRA-coded case data to search against in the first place — see MedDRA coding for trial safety data for how cases get coded to Preferred Terms before an SMQ can retrieve them, and pharmacovigilance in clinical research for how AE/SAE/SUSAR data feeds that pipeline. GVP Module VI and EudraVigilance reporting cover the case-collection side that ultimately populates the database an SMQ searches.
Choosing a Scope Without Over-Retrieving
A few practical rules keep an SMQ search from becoming a triage backlog:
- Start narrow when the question is already specific — you’re investigating a defined, named concern (a suspected hepatotoxicity signal on a known compound, for instance) rather than fishing for anything unexpected. Fall back to broad only if the narrow-scope case count looks implausibly low, or you have a specific reason to suspect coding inconsistency is masking cases.
- Start broad only when you’re genuinely hypothesis-generating — an exploratory periodic screen where the point is surfacing something not yet on anyone’s radar. Budget the triage time that comes with it; broad scope’s extra sensitivity is not free.
- Record the SMQ version, scope, and (if applicable) whether the algorithmic form was used alongside the case count. Because term membership changes release to release, a case count without that context can’t be reproduced or defended later.
- Don’t treat a hit as a confirmed case. An SMQ, at either scope, is a retrieval filter — it still requires the same causality judgement (see the Naranjo algorithm for one structured approach to that step) that any other case-identification method would.
SMQs vs. Company MedDRA Queries (CMQs)
When no standard SMQ exists for the specific concern, or the standard term list is a poor fit for a particular product or indication, sponsors build a Company MedDRA Query (CMQ) using the same term-grouping mechanics — but a CMQ is company-specific and not standardised or shared. It solves the immediate retrieval problem, but the resulting case count isn’t automatically comparable to another sponsor’s numbers on the same nominal topic, unlike an SMQ. Use an SMQ whenever one adequately covers the topic; reach for a CMQ only when it genuinely doesn’t, and document why.
Frequently Asked Questions
Is an SMQ the same for every company that uses it?
Yes — that’s the point of “standardised.” SMQs are built and maintained centrally by the MSSO under ICH oversight and distributed as part of each MedDRA release, so any two organisations running the same SMQ, scope, and MedDRA version against their own coded data are applying an identical term list, even though their underlying case data differs.
Should a signal-detection search always start with broad scope, to be safe?
Not by default. Broad scope maximises sensitivity at the cost of specificity, which is the right trade-off for genuine hypothesis-generating screens but turns into unnecessary manual triage when you already have a defined, specific question. Match the scope to how well-defined the question already is, not to a blanket “broad is safer” assumption.
Are all SMQs algorithmic?
No. Most SMQs rely on plain term membership alone. The algorithmic form — combining term hits with additional criteria such as lab-value thresholds — exists for a subset of SMQs where term matching alone doesn’t achieve adequate specificity, and it’s documented per-SMQ rather than being a universal feature.
What’s the difference between an SMQ and a Company MedDRA Query?
An SMQ is standardised, MSSO-maintained, and shared across every organisation using that MedDRA release. A CMQ is a company-built query using the same term-grouping approach for a topic that has no adequate standard SMQ — useful, but not standardised or cross-comparable the way an SMQ is.
Do SMQ term lists change between MedDRA versions?
Yes. SMQs are maintained and released alongside each biannual MedDRA update, and term membership can change. A case count generated with one version’s SMQ isn’t guaranteed to reproduce exactly against a later version — record the MedDRA version used alongside any case count you report.
For the surrounding coding and reporting infrastructure, see the lab compliance hub, or the specific pieces this guide builds on: MedDRA, MedDRA coding for trial safety data, and PSUR/PBRER format.
