Bibliometrics

Method

Trials

ClinicalTrials.gov registry, interventional trials, all statuses except withdrawn trials; phases 1 to 3 for therapy, all phases and studies without phase for imaging. A trial is counted in the year it starts ("new") or every year between its start and its end ("active"). The European and national registries are not yet included in these series. Target and radionuclides: the atlas classification.

Publications per target and per radionuclide

Number of articles indexed in PubMed per year of publication, for one query per target (the target's terms combined with a nuclear medicine context) or per radionuclide. An indicative count: an article may mention several targets, and a keyword query is never exhaustive. Recomputed every week.

PubMed corpus

A descriptive bibliometric study of the publications indexed in PubMed (MEDLINE, PubMed Central and publisher records), queried through the E-utilities of the National Library of Medicine. The indicators are record counts obtained by queries, except journals, which are collected record by record. Recomputed every month. A publication belongs to the corpus if it is indexed under the MeSH heading "Radiopharmaceuticals" (or carries this pharmacological action), or if its title or abstract mentions a radiopharmaceutical, a radiotracer, theranostics, radioligand or radionuclide therapy, or radiopharmacy:

Domains

Indicators

Output

Number of records per year of publication ([dp] field); the last full year is the reference.

Compound annual growth

(N_end / N_start)^1/n − 1 over the last ten full years.

Share of PubMed

Corpus publications per 10,000 PubMed records of the same year: corrects for the overall growth of the literature.

Open access

Share of records with a free full text ("free full text" filter); underestimates actual open access (unlinked repositories).

Article types

Publication types assigned by the NLM (Review, Clinical Trial, Randomized Controlled Trial, Meta-Analysis, Case Reports).

Countries and cities

Full counting: a publication counts once for each country (or city) present in its authors' affiliations ([ad] field). PubMed has indexed the affiliations of all authors since 2014: the country series start in 2015. Cities are searched among the publications with a French affiliation; a city counts all its institutions (university hospitals, cancer centres, universities, research organisations).

Journals

All records of the last five full years are collected (esummary) and counted per journal; Bradford zones are computed from this list.

Topics

Query of the topic in the title or abstract, combined with the corpus; an article may belong to several topics.

Trend table

Yearly average of the last three full years compared with that of the three previous years; a change of at least 10% is coloured.

Limitations

• PubMed only: no Web of Science or Scopus, hence no citations or impact factors. Coverage of conference proceedings and of non-biomedical journals (radiochemistry, physics) is partial.
• Title and abstract keyword queries pick up noise (laboratory radiotracers, radiolabelling outside nuclear medicine) and miss articles without these words; MeSH indexing of recent years is not complete, which slightly underestimates the last year.
• Affiliations: a mistyped or ambiguous address ("Nice", "Tours", "Georgia") can create false positives; country names are searched in English. Full counting inflates the totals of countries that collaborate a lot.
• Year of publication: the print or electronic publication date depending on the record; an article published online in December may be counted the following year.

Reproducibility

All the queries are listed here and in the data files (biblio.json, publications.json). Each can be pasted into PubMed advanced search with the desired year range; counts change as PubMed indexes new records.