The impact of PIPs on mortality and readmissions in older adults: a retrospective cohort study.

PURPOSE: Our aim was to determine the impact of potentially inappropriate prescriptions (PIP), according to "Screening Tool of Older Persons' Prescriptions" criteria version 2 (STOPP-2), on mortality and hospital admissions. METHODS: Monocentric retrospective cohort study. Patients over 65 years of age and who were consecutively discharged from internal medicine at a Spanish university hospital in 2016 were included. The mortality and hospital admissions of the cohort of patients were analysed using their electronic health records within two years from the time of discharge. Analysis was done based on the type and number of STOPP-2 criteria as well as taking into account the total number of medications. The subdistribution hazard ratios (SHR) were estimated through a competing proportional hazards model. RESULTS: A total of 270 patients with a median age of 82 years (interquartile range/IQR 76-86 years), and 152 (56.3%) women were studied. It was found out that 28.3% of patients with PIP died compared to 17.2% of patients without it. Digoxin (B1 STOPP-2 criterion) with a subdistribution hazard ratio (SHR) 2.40 (95% CI 0.63-9.18), selective serotonin re-uptake inhibitors/SSRIs (D4) with a SHR 1.76 (95% CI 0.52-5.96) and neuroleptic drugs (K2) with a SHR 2.01 (95% CI 0.82-4.95) non-significantly increased the risk of death. Dementia (SHR 5.45; 95% CI 2.76-10.78) was then the only statistically significant risk factor for death. Sixty percent of patients with a PIP had shown at least one hospital admission compared to 51% of patients without it. The number of drugs at discharge (SHR 1.03; 95% CI 1.01-1.05) and having 1-2 STOPP-2 criteria (SHR 1.17; 95% CI 1.02-1.35) significantly increased the risk of hospital admission. CONCLUSION: The number of drugs at discharge and having any STOPP criteria significantly increased the risk of hospital admission in this cohort. PIP, only according to some specific STOPP-2 criteria involving digoxin, neuroleptics and SSRIs, might associate with a statistically non-significantly higher risk on mortality.