RT Journal Article SR Electronic T1 External validation of the DIVA and DIVA3 clinical predictive rules to identify difficult intravenous access in paediatric patients JF Emergency Medicine Journal JO Emerg Med J FD BMJ Publishing Group Ltd and the British Association for Accident & Emergency Medicine SP 762 OP 767 DO 10.1136/emermed-2020-209658 VO 37 IS 12 A1 Cristian Girotto A1 Marta Arpone A1 Anna Chiara Frigo A1 Matteo Micheletto A1 Alessandro Mazza A1 Liviana Da Dalt A1 Silvia Bressan YR 2020 UL http://emj.bmj.com/content/37/12/762.abstract AB Background Intravenous (IV) peripheral access is often a difficult procedure in the paediatric ED, causing pain and significant distress. Clinical prediction tools including reproducible variables have been developed to help clinicians identify children at risk of difficult IV access, likely to need additional resources/interventions to maximise success at first attempt. We aimed to externally validate the Difficult IntraVenous Access (DIVA) and DIVA3 scores developed for this purpose.Methods Cross-sectional study of children undergoing IV cannulation by nurses in a tertiary-care paediatric ED. Data were collected at the time of the procedure in a clinical report form.Results Of 440 children included (56.8% males; median age 4.7 years (IQR 1.5–9.5)), 23.4% had a difficult IV access (defined as requiring >1 attempt). Diagnostic accuracy measures for a DIVA cut-off ≥4 and their 95% CIs were sensitivity 24.3% (16.4% to 33.7%), specificity 92.6% (89.2% to 95.1%), positive and negative predictive value 50.0% (35.3% to 64.5%) and 80.0% (75.7% to 83.9%), respectively. The same measures for the DIVA3 were 22.3% (14.7% to 31.6%), 93.5% (90.3% to 95.9%), 51.1% (35.8% to 66.3%) and 79.8% (75.4% to 83.6%). The area under the receiver operating characteristic curve was 0.652 (95% CI 0.591 to 0.712) for the DIVA and 0.649 (95% CI 0.589 to 0.709) for the DIVA3 score. In patients with DIVA and DIVA3 <4, nurses’ prediction of greater difficulty in IV placement and moderate/severe dehydration were common independent predictors of difficult IV at multivariate analysis. Only nurses’ prediction of greater difficulty in IV placement were associated with higher odds of difficult cannulation for both DIVA/DIVA3 scores ≥4.Conclusion We externally validated the DIVA and DIVA3 showing a similar accuracy compared with the DIVA derivation cohort and between DIVA and DIVA3. We identified factors that can help refine further the risk of difficult IV access and support decision making on the best strategy to maximise the chances of cannulation success on first attempt.