How to Track New Medical Evidence with QSEvidence: A Living Update Workflow

Clinical teams do not need another unread stream of article alerts. They need a controlled way to detect new studies, decide what is relevant and credible, compare findings with the current evidence baseline, document whether anything changes, and route important signals to human review. This guide combines database alerts with QSEvidence-assisted triage and synthesis while preserving sources, ownership, and approval.

A New Paper Is a Signal, Not an Automatic Update

A study appears in an email alert and reports a positive result. That does not mean a clinical summary, pathway, or recommendation should immediately change. The paper may study a different population, use a surrogate outcome, conflict with larger evidence, have important risk-of-bias concerns, or add precision without changing the direction of the conclusion.

Evidence surveillance separates two questions:

Is this new information relevant and credible?
Would it change what the team currently says or does?

Cochrane describes evidence surveillance as active monitoring of accumulating evidence and defines a living systematic review as a review that is continually updated as new evidence becomes available. It notes that living approaches are most suitable for high-priority questions when current certainty is low or very low and new research is likely.^[1] Most clinical teams will begin with a lighter surveillance process, not a formal living systematic review, but the same discipline is valuable.

Create the Baseline Before Turning On Alerts

Monitoring without a baseline produces a folder of papers with nothing stable to compare them against. Create an evidence snapshot for each question before the first recurring search.

Question: a structured population, intervention or exposure, comparison, outcomes, setting, and timeframe.
Scope: eligible study designs, languages, dates, and source types.
Current position: the best available estimate, certainty, important harms, and major unknowns.
Anchor sources: the current guideline, high-quality review, pivotal studies, and local policy.
Search record: databases, full search strings, filters, and last search date.
Ownership: who screens alerts, who appraises evidence, who approves changes, and who communicates them.
Update triggers: what kind of evidence would require routine review or urgent escalation.

This snapshot becomes the comparison point. Without it, a team can describe a new study but cannot reliably state whether the study changes the evidence.

Run Every New Record Through Four Queues

Queue	Main question	Minimum record	Owner
Discovery	What new citations, trials, guidelines, safety notices, or official updates appeared?	Source, date, search or alert that found it, and deduplication status.	Information specialist or assigned reviewer.
Relevance	Does the record match the defined question and eligibility criteria?	Include, exclude, or uncertain decision with a reason.	Two reviewers when the stakes or review method require it.
Credibility	How trustworthy are the methods and result for the outcome of interest?	Study design, key bias concerns, effect estimate, uncertainty, and limitations.	Clinical and methods reviewers.
Impact	Does the new evidence alter the direction, magnitude, certainty, applicability, safety, or implementation of the current position?	Change classification, rationale, approver, and effective date.	Named clinical or governance authority.

Use Alerts for Discovery, Not for Judgment

PubMed allows signed-in users to create alerts for saved searches and receive new results automatically. The frequency can be configured, and PubMed also supports RSS feeds.^[2] My NCBI documentation describes daily, weekly, and monthly update options for saved searches.^[3]

A useful alert must be reproducible. Record the complete query rather than a vague topic name. Avoid manually adding changing date ranges to a saved search when the service already identifies newly indexed records. Test the alert against known relevant studies, document deliberate limits, and revisit the strategy when terminology or indexing changes.

Do not rely on a single bibliographic alert for every type of update. Depending on the question, the discovery layer may also need:

clinical trial registries for new or completed studies;
guideline organizations and regulator safety notices;
retraction, correction, or expression-of-concern monitoring;
local policy and formulary updates;
specialty society communications.

Where QSEvidence Fits

QSEvidence presents its evidence process as retrieve, compare, and synthesize, with source links and review checkpoints kept visible.^[4] It can support the middle of a surveillance workflow after records are discovered and before a human change decision is approved.

Normalize the question: keep the same clinical concepts and applicability fields across review cycles.
Prepare a triage view: organize newly identified records by study type, population, intervention, outcome, and likely relevance.
Compare against the baseline: show where the new result agrees, conflicts, or addresses a previous evidence gap.
Extract review fields: draft a table of methods, effect estimates, uncertainty, harms, and limitations for verification.
Draft the impact note: separate what is new, what remains unchanged, and what requires expert resolution.
Preserve source paths: keep each important statement connected to the paper, guideline, registry, or official notice supporting it.

QSEvidence should not be described as a replacement for PubMed, a formal systematic search, duplicate independent screening, risk-of-bias assessment, or governance approval. Its official FAQ positions the product as information retrieval and workflow assistance rather than the responsible medical decision-maker.^[5]

Apply a Four-Level Change Test

Every included signal should end in one of four categories. This is more useful than labeling every new paper “important.”

1. No material change

The study is consistent with the baseline and does not meaningfully change effect size, certainty, safety, applicability, or implementation. Record it and leave the approved summary unchanged.

2. Monitor

The result is potentially important but imprecise, early, indirect, at risk of bias, or awaiting related trials. Add it to the evidence map and define what future information would trigger reassessment.

3. Amend the evidence summary

The study changes context, adds a relevant subgroup, clarifies harms, improves precision, or modifies certainty without requiring an urgent pathway change. Update the evidence note after expert approval and record exactly what changed.

4. Urgent clinical or governance review

A credible safety signal, major regulator action, important guideline change, decisive high-quality result, or clear contradiction with current practice may require prompt escalation. The surveillance team should notify the named authority; it should not autonomously rewrite policy.

A Repeatable 30-Day Operating Cycle

Day 1: Run or receive scheduled searches, import records, remove duplicates, and log the search date.
Days 2–5: Screen titles and abstracts against the predefined scope. Record exclusion reasons for records that appeared plausibly eligible.
Days 6–12: Obtain full texts and related materials. Use QSEvidence to prepare structured extraction drafts and locate linked guidelines or conflicting evidence.
Days 13–18: Verify extracted data, appraise credibility, and classify potential impact.
Days 19–23: Submit material changes to the clinical, methods, safety, or governance reviewer named in the protocol.
Days 24–27: Update the approved summary, evidence table, references, and change log where required.
Days 28–30: Communicate the decision, confirm the next search, and review any delayed or unresolved records.

The interval should match the decision. A rapidly evolving, high-stakes safety question may need continuous or weekly review; a stable topic may only justify quarterly monitoring. Frequency should be specified rather than chosen after a striking paper appears.

Keep an Evidence Change Log

Field	What to record
Version and date	Previous and new version identifiers, search date, review date, and effective date.
New evidence	Full citation, persistent identifier, source link, study type, and reason for inclusion.
Verified finding	Effect estimate, uncertainty, important harms, population, and follow-up.
Impact judgment	No change, monitor, amend, or urgent review, with a concise rationale.
Text changed	The exact conclusion, table, pathway, or communication that was revised.
Human accountability	Screeners, appraisers, clinical approver, conflicts of interest, and unresolved disagreement.

The World Health Organization describes evidence-informed decision-making as identifying, appraising, and mobilizing the best available evidence for safe and effective policy and practice.^[6] A change log makes that process inspectable: readers can see not only the latest conclusion but also why it changed.

Know When This Is Not a Living Systematic Review

A recurring departmental evidence scan is useful, but it should not automatically be labeled a living systematic review. A formal review requires an explicit protocol, reproducible searches, transparent eligibility criteria, systematic screening and extraction, risk-of-bias assessment, appropriate synthesis, versioned reporting, and a predefined approach to updating or leaving living mode.

If the question will determine a guideline, policy, formulary decision, or other high-stakes action, involve information specialists and review methodologists early. QSEvidence can support evidence organization, but formal methods and accountability remain essential.

Frequently Asked Questions

Can QSEvidence automatically monitor PubMed?

Use PubMed or another approved database service to create and document recurring alerts. QSEvidence can help structure the question, triage retrieved records, compare findings, and prepare source-linked review material. Do not claim an automated integration unless it is configured and verified in your environment.

How often should a clinical evidence search run?

Frequency depends on urgency, research volume, uncertainty, and the cost of being outdated. Define the interval in advance and allow safety alerts or regulator actions to trigger off-cycle review.

Can a new randomized trial overturn a systematic review?

Possibly, but not simply because it is newer. Review its population, size, methods, effect, uncertainty, harms, and consistency with the existing evidence. Then assess whether adding it would materially change the synthesis or certainty.

Should AI decide whether a guideline needs updating?

No. AI can flag potential changes and organize the evidence. The decision to update guidance requires the responsible clinical and governance process.

What should happen when reviewers disagree?

Record the disagreement and its reason, obtain an additional methods or clinical review when specified, and do not hide uncertainty inside a single generated conclusion.

References

Medical and research disclaimer: This article describes an evidence-monitoring workflow. It is not a systematic-review protocol, clinical guideline, medical recommendation, or substitute for qualified appraisal and governance. Teams must verify original sources and follow applicable institutional, ethical, legal, and clinical requirements.