Student School of Public Health, Peking University Beijing, China (People's Republic)
Background: International Conference on Pharmacoepidemiology & Therapeutic Risk Management (ICPE), annually organized by ISPE, is one of the most authoritative international academic conferences in the field of pharmacoepidemiology (PE). Since not everyone had the opportunity to attend the meeting, our study can help scholars to know the hot topics of the conference as a whole. Bibliometrics analysis provides a tool for generalization.
Objectives: Quantitatively describe the tracks, hotspot study designs, analytical methods, diseases, drugs, and databases mentioned in the 38th ICPE abstracts.
Methods: Frequency and percentage were used in descriptive statistics. Taking the 38th ICPE documents as the data source for text analysis, we used Python 3.8 to extract the literature title and abstract. R 4.1.2 tidytext package was used to splitting the literature text into tokens (a meaningful unit of text, such as a word or a phrase). Term frequency - inverse document frequency (TF-IDF) value, a weighted statistic, was used to identify the keywords (hotspot topics). IDF, the weight, was calculated by the formula of ln (total number of abstracts / number of abstracts including the token), reflecting the degree of rarity of a token, which was helpful for accurate recognition.
Results: 1280 abstracts in 38th ICPE were included. Top 3 tracks were drug utilization research (192, 18.5%), disease epidemiology/clinical course (113, 10.9%), and methods in PE (110, 10.6%). Top 5 hotspot study designs (TF-IDF value) were cohort study (150.3), systematic review (137.9), case-control study (45.7), cross-sectional study (38.0), and self-control case series (35.6). Top 5 hotspot analytical methods were propensity score (384.2), logistic regression (317.8), Cox regression (211.9), inverse probability of treatment weighting (188.9), and Kaplan-Meier analysis (118.8). Top 5 hotspot diseases were COVID-19 (318.7), type 2 diabetes (124.8), breast cancer (90.8), rheumatoid arthritis (79.0), and Parkinson's disease (65.7). Top 5 hotspot drugs were opioids (146.0), antidepressants (73.3), COVID-19 vaccines (72.7), SGLT-2i (58.9), and statins (56.3). Top 5 Hotspot databases were Clinical Practice Research Datalink (223.3), IBM MarketScan Commercial Claims and Encounters (155.3), FDA Adverse Event Reporting System (109.7), French National Healthcare Database (79.0), and Optum Clinformatics Data Mart Database (70.1).
Conclusions: This study innovatively used informatics methods to provide scholars with a way to quickly understand the hotspots of ICPE 2022. Hot topics can provide PE scholars with reference value on future topic selections. Students can master those commonly used study designs and methods, and the hot databases can be fully utilized for future research.
