Evaluation of tools to prevent drug incompatibilities in paediatric and neonatal intensive care units

Objective Intravenous drug administration in neonatal (NICU) and paediatric intensive care units (PICU) is critical because of poor venous access, polymedication, fluid restriction and low infusion rate. Risk is further increased by inadequate information on the physicochemical compatibility of drugs. Eight decision-supporting tools were hence evaluated to improve the detection of drug incompatibilities in paediatric wards. Setting NICU and PICU, University hospital. Method Eight tools (Th,riaque 2007, Stabilis 3, Perfysi 2 databases; KIK 3.0 software; Neofax 2007 handbook; King 2008 Guide, CHUV 9.0, pH 2007 cross-tables) were assessed by two pharmacists using 40 drug pairs (20 incompatible; 20 compatible) frequently prescribed in PICUs and NICUs. Trissel's 14th Ed. handbook served as the gold standard. Four criteria were evaluated (each with a maximum of 250 points): accuracy (sensitivity, specificity, positive and negative predictive values), completeness (number of drug pairs documented), comprehensiveness (presence of 16 different items), and applicability (by combining the time needed by 7 pharmacists to classify 5 drug pairs, plus an evaluation of their design, usefulness, reliability and ergonomics, using visual analogy scales). The percentage of non-compliant answers (NCA) was calculated for both the performing pharmacists and the tools. Main Outcome Measure Global score of drug incompatibilities (accuracy + completeness + comprehensiveness + applicability). Results Th,riaque obtained the best global score (840/1000 points), followed by pH (807), CHUV (803), Perfysi (776), Neofax (678), King Guide (642), Stabilis (584) and KIK (523), respectively. The highest scores were reached by Th,riaque for accuracy (234/250); Th,riaque and pH for completeness (200/250); Th,riaque and Perfysi for comprehensiveness (218/250); and pH for applicability (298/250). The range of pharmacists' NCAs was between 9% (4/45 NCAs) and 33% (15/45), whereas that for drug pairs was between 10% (6/63) and 30% (19/63). The range of NCAs for tools was between 6% (2/35, pH) and 49% (18/35, Perfysi). Conclusions Th,riaque proved outstanding as a drug-incompatibility tool. However, all resources showed some shortcomings. The large ranges of pharmacists' NCAs shows that such an assessment is subject to different interpretations. Standard operating procedures for drug-incompatibility assessment should be implemented in drug-information centres. Tools with low NCA percentage, such as the pH or CHUV tables, may be useful for nurses in ICUs.