Appraisal of: "Gurung P, Makineli S, Spijker R, Leeflang MMG. The EMTREE term "Diagnostic Test Accuracy Study" retrieved less than half of the diagnostic accuracy studies in EMBASE. J Clin Epidemiol. 2020;126:116-21."

Reviewer(s): 
Short description: 

The authors tested the retrieval of the EMTREE terms ‘diagnostic test accuracy study’ (checktag), ‘diagnostic test accuracy’ and ‘sensitivity and specificity’.  Two sets of records were obtained by randomly selecting records from Embase and screening them to identify reports of diagnostic test accuracy studies (DTAs).  The second set, which tested the recall of all three terms, comprised 2000 records added in 2016. Following exclusions and deduplication, 1223 records were screened to identify DTA studies. 33 of the records were considered to be DTA studies, but only 14 were labelled with the EMTREE checktag (sensitivity=42.4%). Of the non-DTA studies, 6 were labeled with the EMTREE checktag. 9 studies were assigned the term ‘diagnostic test accuracy’ (8 accurately, giving a sensitivity of 24.2%) and 30 studies were assigned the term ‘sensitivity and specificity’ (21 accurately, giving a sensitivity of 63.6%). Using all three terms gave a sensitivity of 72.7%.

The EMTREE checktag has only 42.4% sensitivity and its use alone risks missing more than half of the relevant studies. Using all three terms together might provide adequate sensitivity for a scoping search only.  Imaging studies were particularly lacking in terms of accurate assignment of  indexing terms.

The Embase indexing is not currently reliable enough to use for identifying DTA studies for systematic review purposes and new techniques such as machine learning may have increasing potential to identify studies.

 
Limitations stated by the author(s): 

The authors acknowledged that the two sets of records might not reflect a real review situation where the search would usually not limit to DTA studies and subject terms would also be used. Also, records that were excluded by two assessors were not all checked in detail.

 
Limitations stated by the reviewer(s): 
No further limitations noted.
Study Type: 
Single study