Data collection

Medical records of outpatients were reviewed in the hospital pharmacy, where patients came to pick up their new medication after outpatient department (OPD) visits. If patients had a treatment window of more than six months, data of the last treatment episode was used for the study. Medical records of inpatients were reviewed on the two surgical wards and the intensive care unit of the hospital, the day after the surgery. Information was obtained on demographic characteristics, co-morbidity, co-medication, the dose of the NSAID, chronic use (defined as NSAIDs described ≥15 days during the last month or a new prescription of ≥15 days) and prescription of a GPA. Side effects of GI-bleeding, other bleeding and pain in stomach area, including symptoms of epigastric pain, GI ulcer and gastritis, were recorded. For inpatients, data was collected on the day after the surgery and therefore GI side effects after initiation of therapy were not included in the analysis.

Medical records were reviewed in the hospital pharmacy, where patients came to pick up their new medication after OPD visits. If patients had a treatment window of more than six months, data of the last treatment episode was used for the study. Information was obtained on demographic characteristics, co-morbidity, co-medication, measurements of serum creatinine and potassium within 6 months before start of therapy, and measurements of serum creatinine and potassium after start of therapy, within and after one month. Side effects of high creatinine (serum creatinine level >125.0 mmol/L) and hyperkalemia (serum potassium level > 5.4 mmol/L) were recorded. If biochemical monitoring was done more than once, the date of the last value before start of therapy was used, and the date of the first value after start of therapy was used for defining the time within which monitoring was done.