Methods for assessing the preventability of adverse drug events: a systematic review.

BACKGROUND Preventable adverse drug events (ADEs) are common in both outpatient and inpatient settings. However, the proportion of preventable ADEs varies considerably in different studies, even when conducted in the same setting, and methods for assessing the preventability of ADEs are diverse. OBJECTIVE The aim of this article is to identify and systematically evaluate methods for assessing the preventability of ADEs. DATA SOURCES Seven databases (Cochrane, CINAHL, EMBASE, IPA, MEDLINE, PsycINFO and Web of Science) were searched in September 2010 utilizing the databases' index terms and other common terminology on preventable ADEs. No limits for the years of publication were set. Reference lists of included original articles and relevant review articles were also screened. STUDY SELECTION After applying predetermined inclusion and exclusion criteria on 4161 unique citations, 142 (3.4%) original research articles were included in the review. One additional article was included from reference lists. Outcome measures of included studies had to include the frequency of ADEs and the assessment of their preventability. Studies were excluded if they focused on individuals with one specific type of treatment, medical condition, medical procedure or ADE. DATA EXTRACTION Measurement instruments for determining the preventability of ADEs in each article were extracted and unique instruments were compared. The process of assessing the preventability of ADEs was described based on reported actions taken to standardize and conduct the assessment, and on information about the reliability and validity of the assessment. DATA SYNTHESIS Eighteen unique instruments for determining the preventability of ADEs were identified. They fell under the following four groups: (i) instruments using a definition of preventability only (n = 3); (ii) instruments with a definition of preventability and an assessment scale for determining preventability (n = 5); (iii) instruments with specific criteria for each preventability category (n = 3); and (iv) instruments with an algorithm for determining preventability (n = 7). Of actions to standardize the assessment process, performing a pilot study was reported in 21 (15%), and use of a standardized protocol was reported in 18 (13%), of the included 143 articles. Preventability was assessed by physicians in 86 (60%) articles and by pharmacists in 41 (29%) articles. In 29 (20%) articles, persons conducting the assessment were described as trained for or experienced in preventability assessment. In 94 (66%) articles, more than one person assessed the preventability of each case. Among these 94 articles, assessment was done independently in 73 (51%) articles. Procedures for managing conflicting assessments were diverse. The reliability of the preventability assessment was tested in 39 (27%) articles, and 16 (11%) articles referred to a previous reliability assessment. Reliability ranged from poor to excellent (kappa 0.19-0.98; overall agreement 26-97%). Four (3%) articles mentioned assessing validity, but no sensitivity or specificity analyses or negative or positive predictive values were presented. CONCLUSIONS Instruments for assessing the preventability of ADEs vary from implicit instruments to explicit algorithms. There is limited evidence for the validity of the identified instruments, and instrument reliability varied significantly. The process of assessing the preventability of ADEs is also commonly imprecisely described, which hinders the interpretation and comparison of studies. For measuring the preventability of ADEs more accurately and precisely in future, we believe that existing instruments should be further studied and developed, or that one or more new instruments should be developed, and the validity and reliability of the existing and new instruments be established.