Francesco Dentali1, Andrea Fontanella2, Alexander T Cohen3, Monica Simonetti4, Luca Parretti5, Ettore Marconi4, Damiano Parretti6, Gualberto Gussoni7, Mauro Campanini8, Giancarlo Agnelli9, Claudio Cricelli6, Francesco Lapi4. 1. Department of Internal Medicine, Hospital of Luino, ASST-Sette Laghi, University of Insubria, Varese, Italy. 2. Department of Internal Medicine, Hospital "Buon Consiglio-Fatebenefratelli," Naples, Italy. 3. Department of Haematological Medicine, Guy's and St Thomas' NHS Foundation Trust, King's College London, London, United Kingdom. 4. Health Search, Italian College of General Practitioners and Primary Care, Florence, Italy. 5. Institute of Gerontology and Geriatrics, Department of Medicine, University of Perugia, Perugia, Italy. 6. Italian College of General Practitioners and Primary Care, Florence, Italy. 7. Clinical Research Department, FADOI Foundation, Milan, Italy. 8. Department of Internal Medicine, Hospital "Maggiore della Carità," Novara, Italy. 9. Department of Internal and Cardiovascular Medicine-Stroke Unit, Hospital "S. Maria della Misericordia," University of Perugia, Perugia, Italy.
Abstract
BACKGROUND: Most episodes of venous thromboembolism (VTE) occurred in primary care. To date, no score potentially able to identify those patients who may deserve an antithrombotic prophylaxis has been developed. AIM: The objective of this study is to develop and validate a prediction model for VTE in primary care. METHODS: Using the Health Search Database, we identified a cohort of 1,359,880 adult patients between 2002 and 2013. The date of the first General Practitioner's (GP) visit was the cohort entry date. All VTE cases (index date) observed up to December 2014 were identified. The cohort was randomly divided in a development and a validation cohort. According to nested case-cohort analysis, up to five controls were matched to their respective cases on month and year of cohort entry and duration of follow-up.The score was evaluated according to explained variance (pseudo R2) as a performance measure, ratio of predicted to observed cases as model calibration and area under the curve (AUC) as discrimination measure. RESULTS: The score was able to explain 27.9% of the variation for VTE occurrence. The calibration measure revealed a margin of error lower than 10% in 70% of the population. In terms of discrimination, AUC was 0.82 (95% confidence interval: 0.82-0.83). Results of sensitivity analyses substantially confirmed these findings. CONCLUSION: The present score demonstrated a very good accuracy in predicting the risk of VTE in primary care. This score may be therefore implemented in clinical practice so aiding GPs in making decision on patients potentially at risk of VTE. Georg Thieme Verlag KG Stuttgart · New York.
BACKGROUND: Most episodes of venous thromboembolism (VTE) occurred in primary care. To date, no score potentially able to identify those patients who may deserve an antithrombotic prophylaxis has been developed. AIM: The objective of this study is to develop and validate a prediction model for VTE in primary care. METHODS: Using the Health Search Database, we identified a cohort of 1,359,880 adult patients between 2002 and 2013. The date of the first General Practitioner's (GP) visit was the cohort entry date. All VTE cases (index date) observed up to December 2014 were identified. The cohort was randomly divided in a development and a validation cohort. According to nested case-cohort analysis, up to five controls were matched to their respective cases on month and year of cohort entry and duration of follow-up.The score was evaluated according to explained variance (pseudo R2) as a performance measure, ratio of predicted to observed cases as model calibration and area under the curve (AUC) as discrimination measure. RESULTS: The score was able to explain 27.9% of the variation for VTE occurrence. The calibration measure revealed a margin of error lower than 10% in 70% of the population. In terms of discrimination, AUC was 0.82 (95% confidence interval: 0.82-0.83). Results of sensitivity analyses substantially confirmed these findings. CONCLUSION: The present score demonstrated a very good accuracy in predicting the risk of VTE in primary care. This score may be therefore implemented in clinical practice so aiding GPs in making decision on patients potentially at risk of VTE. Georg Thieme Verlag KG Stuttgart · New York.