Development and Piloting of an Algorithm to Select Older Patients for Different Types of Medication Review

Aim: To develop and pilot an algorithm to select older people for different types of medication review based on their case complexity. Methods: Experts rated complexity of patient cases through a Delphi-consensus method. The case characteristics were included in a regression model predicting complexity to develop a criteria-based algorithm. The algorithm was piloted in four community pharmacies with 38 patients of high and low complexity. Pharmacists conducted medication reviews according to their personal judgment and rated the patients' complexity. Time needed for reviewing and number of interventions (proposed and implemented) were assessed. Feasibility was evaluated with in-depth interviews. Results: We developed the algorithm with 75 cases proceeding from patients in average 79 years old and using 10 prescribed medications. The regression model (adjusted R(2) = 0.726, P < 0.0001) resulted in the following criteria for the algorithm: “number of medications” × 1 + “number of prescribers” × 3 + “recent fall incident” × 7 + “does not collect own medication” × 4. The pharmacists performed advanced medication reviews with all patients. The time needed to perform the medication review did not differ significantly according to case complexity (76.9 min for high complexity; 66.1 min for low complexity). Agreement between the algorithm scores and the pharmacists' ratings on complexity degree was slight to moderate (Kappa 0.16–0.42). The pharmacists had mixed opinions about the feasibility of applying the algorithm, particularly regarding the criterion “fall incidents.” Conclusion: We developed an algorithm with four criteria that distinguished between high and low complexity patients as rated by experts. Additional validation steps are needed before implementation.