WHAT PHYSICIANS WANT- SEARCH AND EVIDENCE SOURCE TRENDS IN THE TRIP HEALTHCARE DATABASE
Author(s)
Davis JA, Saunders R
Coreva Scientific, Freiburg im Breisgau, Germany
Presentation Documents
OBJECTIVES: Evidence-based practice is the target of healthcare providers, but this goal is increasingly difficult as the evidence base grows. The TRIP search engine is designed to facilitate providers' finding the most relevant, highest-level clinical evidence. As an iterative process, understanding current trends in evidence requirements and sources is required. This study analyzes a physician-dominated search engine for topics searched, sources accessed, and documents returned. METHODS: Anonymized 2017 TRIP clickstream data were analyzed to assess the search habits, documents clicked, and information sources accessed. Results were stratified by country to assess differences in clinical interests and evidence sources, and to construct an evidence network. Analyses were performed in R. Only unique browser sessions consisting of at least 2 documents were retained, with country determined from the IP address. For the top 10 represented countries, stemmed search terms and evidence-provider domain names were subjected to frequency analysis. RESULTS: User searches focused on pain (median rank 2), cancer (5.5), and diabetes (5.5), with obesity notable for its much lower median rank of 40. Searches including “treatment” were significantly (p<0.05) more common than those for “prevention”. Of the top 5 accessed evidence sources (ncbi.nlm.nih.gov, crd.york.ac.uk, guideline.gov, clinicaltrials.gov, emedicine.medscape.com) four showed higher tendency to be the last click in the datastream versus the first click, while only crd.york.ac.uk showed a greater propensity to be the first rather than last. Popular evidence sources were generally consistent across countries, but ncbi.nlm.nih.gov and guideline.gov showed significant differences in first/last position (p<0.05). CONCLUSIONS: Popular search terms may reflect complex areas of medicine where physicians need to check progress more regularly, or there remains uncertainty regarding treatment. Training resources could be targeted to these topics. Country differences may highlight local concerns, reflect national resource availability, and allow targeting of the most trusted data sources for publication of important evidence.
Conference/Value in Health Info
2018-11, ISPOR Europe 2018, Barcelona, Spain
Value in Health, Vol. 21, S3 (October 2018)
Code
PRM76
Topic
Real World Data & Information Systems
Topic Subcategory
Reproducibility & Replicability
Disease
Multiple Diseases