Detection of adverse events through hospital administrative data

Abstract

This study aims to estimate and describe the incidence of adverse events (AE) registered in Portuguese public hospitals and consequently to determine the feasibility of using hospital administrative data as a tool for AE surveillance. A retrospective observational study using hospital administrative data was performed to detect the incidence of AE based on a selection of ICD-9-CM codes (diagnoses and external causes). All episodes in public hospitals in the period 2000–2010 were included. AE were divided in three main categories: complications of surgical or medical procedures, misadventures of surgical and medical care, and adverse drug events (ADE). The ADE subgroup was further subdivided in: poisoning, late effect, and adverse drug reaction. Over the studied period, the algorithm was able to detect 543,242 episodes with AE events (3.7% of all episodes), with an in-hospital mortality rate of 6.3%, and a median length-of-stay of 8 days. In a scenario of underreporting of AE, this administrative data approach in an important complement to the other existing surveillance techniques.