Citation Bar reference

Article flags

Article metrics

The impact score, in full

Most tools that highlight "high impact" papers won't tell you what that means. Here is the actual function. The default threshold is 5, adjustable from 1 to 6 in the extension's settings. You can disagree with the weights — but you can see them, and you can move the line.

function calculateScore(item) {
    let score = 0;

    // MEDLINE: 2.25 points
    if (item.recordStatus === 'PubMed - indexed for MEDLINE') {
        score += 2.25;
    }

    const citations = parseInt(item.Citations);
    const crc = parseFloat(item.RelativeCitationRatio);

    if (item.PublishCurrent === 'n') {
        // Recent articles (< 2 years)

        // Citations scoring
        if (citations > 30) score += 2.25;
        else if (citations > 15) score += 1.75;
        else if (citations > 5) score += 1;
        else if (citations > 2) score += 0.25;

        // RCR scoring
        if (crc > 2) score += 1.25;

    } else {
        // Older articles (>= 2 years)

        // Citations scoring
        if (citations >= 200) score += 3.5;
        else if (citations >= 50) score += 2.75;
        else if (citations >= 10) score += 1.5;

        // RCR scoring
        if (crc > 10) score += 1.25;
        else if (crc > 5) score += 0.75;
    }

    // SJR: 1.25 points
    if (item.SJR !== '-' && item.SJR > 3) {
        score += 1.25;
    }

    // Trending: 1 point
    if (item.isTrending) {
        score += 1;
    }

    // Influential Citations: 0.5 points
    if (item.s2Url && item.influentialCitations >= 1) {
        score += 0.5;
    }

    return score;
}

Links and lookups

Journal information

Everything to the right of the ||| divider describes the journal rather than the article.

Tools on the bar

Xout — exclusion search

Roughly a quarter of PubMed is not primary research. Xout reopens your current search in a dialog with your query pre-filled and a checkbox for each publication type to exclude. All are checked by default; uncheck any to let that type through. The full query preview updates live, so you can search directly, or copy the string into your methods section.

• Has Abstract — recommended; articles without abstracts are rarely primary research
• No Retracted Publications — excludes retraction notices, not the retracted articles themselves
• No Published Errata · No Letters · No Editorials · No Comments
• No Systematic Reviews · No Meta-Analyses · No Reviews

pR — PubMed reports

Opens the four standing reports: PubMed Summary, Filters, MeSH Counts and Journal Ranking.

/Report — citation report

Every key measure and link for one article in an easy-to-copy format. Two buttons in the extension popup export the whole page of results: Excel gives a tab-separated table of exactly what the bar shows, one row per article, ready to paste straight into a spreadsheet; and Data dumps everything — PMIDs first, then all measures, timestamps and internal fields. Full disclosure of every field the extension holds.

Term Analysis

Enter up to 20 variants of a single concept — vitamin B1, thiamine, B-1 — and it returns PubMed hit counts for each across four field configurations (Title, Title/Abstract, quoted, unquoted), validates each against MeSH, and stars the terms that are both high-volume and MeSH-backed. A "Special" mode handles custom date windows and up to ten term groups at once.

Expand

Takes your current PubMed search and opens the same query against OpenAlex, ClinicalTrials.gov and Google Scholar in three new tabs — a quick check on whether PubMed alone is giving you the whole picture.

What PubMed's Abstract filter checks

PubMed's Abstract filter (the sidebar box, or hasabstract in a query) is not a complete test for an abstract. Checked against the records' XML, it counts a record with a main abstract or an English plain-language summary, but not one whose only abstract was supplied separately by the publisher, in any language. So a record can fail the filter and still have an abstract. The Drift Report labels the filter's two sides has abstract (PubMed) and no abstract (PubMed), and splits the second three ways, for all records and again for MEDLINE:

The two XML-derived groups are counted for the daily cohorts only. The year-to-date and whole-database ranges are too large to read record by record, so they show N/A there. These findings come from testing records in September 2026, not from NLM documentation, and PubMed's behaviour may change.