Mapping the Delirium Literature Through Probabilistic Topic Modeling and Network Analysis: A Computational Scoping Review.

BACKGROUND

Delirium is an acute confusional state, associated with morbidity and mortality in diverse medically-ill populations. Delirium is recognized, through both professional competencies and instructional materials, as a core topic in consultation psychiatry.

OBJECTIVE

Conduct a computational scoping review of the delirium literature to identify the overall contours of this literature and evolution of the delirium literature over time.

METHODS

Algorithmic analysis of all research articles on delirium indexed in MEDLINE between 1995 and 2015 using network analysis of citation Medical Subject Headings (MeSH) tags and probabilistic topic modeling of article abstracts.

RESULTS

The delirium corpus included 3591 articles in 874 unique journals, of which 95 were primarily psychiatric. The annual delirium publication volume increased from 40 in 1995 to 420 in 2015 and grew as a proportion of total indexed publications from 8.9 to 38.6 per 100,000. The psychiatric journals published 720 of the delirium publications. Articles on treatment of delirium (806) outnumber articles on prevention of delirium (432). Abstract topic modeling and Medical Subject Headings graph community analysis identified similar genres in the delirium literature, including: delirium in geriatric, critically ill, palliative care, and postsurgical patients as well as diagnostic criteria or scales, and clinical risk factors. The genres identified by topic modeling and community analysis were distributed unevenly between psychiatric journals and nonpsychiatric journals.

CONCLUSION

The delirium literature is large and growing. Much of this growth is outside of psychiatric journals. Subtopics of the delirium literature can be algorithmically identified, and these subtopics are distributed unevenly across psychiatric journals.