The National Institute for Health and Care Excellence (NICE) guidelines manual recommend that MEDLINE, Embase and Cochrane Central Register of Controlled Trials should be prioritized for searching for reviews of the effectiveness of pharmacological interventions. Additionally, searching trial registries and conference abstracts are recommended to identify ongoing or unpublished research. However, the approaches to searching conference abstracts have not been previously studied. The aim is to analyze searches of conference abstracts reported in NICE Technology Appraisal (TA) company submissions for cancer interventions from 2013 until September 2018.

The company submissions were searched and obtained via the NICE technology appraisal guidance website. The sources used to find conference abstracts were identified from the company clinical effectiveness review search methods and appendices. Conference abstract searching in both database and website sources were compared.

Of all 394 TAs, 124 (31 percent) were cancer TAs. Between 2013 and 2018, 91 TAs were completed or updated, which covered 18 cancer categories and 52 different named technologies. Technologies to treat non-small-cell lung cancer was the most frequently appraised in the last five years. Nivolumab was the most frequently appraised technology. Searches for conference abstracts were reported in 70 (77 percent) out of 91 company submissions. Supplementary searching was reported in 59 (84 percent), compared with 11 (16 percent) searching either/both Embase and the Web of Science Conference Proceeding Index (WoS-CPCI). A total of 54 supplementary website sources were searched which ranged from one to 11 per TA (average four sources). The American Society of Clinical Oncology (ASCO) and the European Society of Medical Oncology were the most frequently searched sources.

Whilst the WoS-CPCI has better coverage of cancer conference abstracts than Embase, searching databases alone are inadequate. Supplementary conference websites should be searched for reasons such as access to the most recent abstracts and incomplete indexing of titles within databases. A wide range of cancer specific sites exists although the impact of broad (e.g. ASCO) versus condition specific sites is unclear.