pubmed-database

Use the [mh] field tag with the controlled vocabulary term, such as diabetes mellitus, type 2[mh]. Add subheadings before the tag, like diabetes mellitus, type 2/drug therapy[mh], and combine with free-text [tiab] synonyms for newer topics.

Call esearch.fcgi with db=pubmed, your query term, and retmode=json to get PMIDs, then use esummary or efetch for metadata and abstracts. Pass an email parameter, load the API key from environment variables, and add delays to respect rate limits.

Use [pt] for publication types like randomized controlled trial[pt] or systematic review[pt], and [dp] for dates such as 2020:2026[dp]. Other useful tags include [la] for language and free full text[sb] for availability.

An API key is optional but recommended for production scripts since it raises rate limits. Store it in an environment variable like NCBI_API_KEY and never commit it to files or command history.

Use the NCBI history server by passing usehistory=y in esearch, then reference the returned WebEnv and query_key in subsequent efetch or esummary calls. This avoids passing very long PMID lists through URLs.

Use [majr] only when the topic must be central to the paper, as it improves precision but can miss relevant work. For broader recall, use [mh] combined with title/abstract free-text terms.