A new medication-based prediction score for postoperative delirium in surgical patients: Development and proof of feasibility in a retrospective patient cohort

Structured risk screening for postoperative delirium (POD) considering prehospital medication is not established. We aimed to develop a POD-risk prediction score based on known risk factors and delirium-risk increasing drugs to be used by pharmacists during medication reconciliation at hospital admission, and to test for feasibility in a retrospective cohort of surgical patients. Therefore, established POD-risk factors and drugs were extracted from the literature and a score was generated. Following this, the score was tested for feasibility in a retrospective 3-month-cohort of surgical patients. For patients with higher scores suggesting higher probability of POD, patient charts were screened for documentation of POD. For development of the score, the following POD-risk factors were defined and points assigned for score calculation: age (≥65 years=1 point/≥75 years=2), male sex (1), renal insufficiency (RI; 1), hepatic impairment (HI; Model-of-endstage-liver-disease (MELD) 10-14=1/≥15=2), delirium-risk increasing drugs (1 point per drug class), anticholinergic drug burden (ACB; ≥3=1). In the retrospective test cohort of 1174 surgical patients these factors concerned: age ≥65 years 567 patients (48%)/≥75 years 303 (26%), male 652 (55%), RI 238 (20%), MELD 10-14 106 (9%)/≥15 65 (5%), ≥ 1 delirium-risk increasing drug 418 (36%), ACB ≥3 106 (9%). The median POD-risk prediction score was 2 (range 0-9). Of 146 patients (12%) with a score ≥ 5, POD was documented for 43 (30%), no evidence for POD for 91 (62%) and data inconclusive for 12 (8%). For scores of ≥ 7, POD was documented for 50% of the patients with sufficient POD documentation. Overall, POD documentation was poor. To summarize, we developed and successfully tested the feasibility of a POD-prediction-score assessable by pharmacists at medication reconciliation at hospital admission.