Xpert Carba-R to identify people carrying carbapenemase-producing organisms

Search strategy

The search strategy was designed to identify evidence on the clinical and cost effectiveness of the Xpert Carba‑R to identify patients carrying CPOs, specifically KPC, NDM, VIM, IMP‑1, OXA‑48, OXA‑181 and OXA‑232.

The strategy was developed for MEDLINE (Ovid interface). The strategy was devised using a combination of subject indexing terms and free text search terms in the title, abstract and keyword heading word fields. The search terms were identified through discussion within the research team, scanning background literature, browsing database thesauri and use of the PubMed PubReMiner tool (http://hgserver2.amc.nl/cgi-bin/miner/miner2.cgi). The strategy reflected the nature of the MIB assessments as rapid evidence reviews, with a pragmatic, focused search approach being used.

The main structure of the search strategy comprised 2 concepts:

The search concepts were combined as follows: CPO AND PCR.

The strategy also included an additional focused approach which combined CPO terms with screening / diagnosis terms in the title field, 2 stand‑alone search lines which co‑ordinated CPO terms and non‑specific diagnosis / screening terms in close proximity, and 4 standalone lines which searched on the manufacturer and device names. These lines were designed to retrieve studies which might be missed by the 2 concept approach.

The strategy excluded animal studies using a standard algorithm. Non‑English language publications were also excluded from the search results. The strategy was limited to studies published from 2013 to date; this reflected the date when the device was first developed.

The MEDLINE strategy was translated appropriately for the other databases searched. The PubMed search was limited to all records not fully indexed for MEDLINE apart from In‑Process records and PubMed‑not‑MEDLINE records, which were also excluded.

The following databases were searched:

Evidence selection

A total of 1705 records were retrieved from the literature search. After de‑duplication, 1036 records remained. Records were grouped into 2 batches, 1 batch with search results from a focussed search strategy (n=452) and a second batch which included the remainder of the search results.

Records from batch 1 were sifted independently by 2 researchers against the inclusion criteria at title and abstract level. Any disagreements were discussed and agreement was reached in all cases, so a third independent arbiter was not required. The first sift removed 432 records based on the following exclusion criteria:

Full articles were retrieved for 20 of the remaining studies. Full text assessment was done independently by 2 researchers to identify relevant primary research addressing the key outcomes of interest. At this stage, 17 papers were excluded:

Three studies remained from batch 1, which included 1 comparative diagnostic accuracy study and 2 diagnostic accuracy studies. These studies became the focus of the evidence review.

The second batch of studies was reviewed by 1 researcher to ensure that no additional relevant studies had been overlooked. No additional studies were identified from this review.