Background: Previous studies have determined that a small subset (~10%) of the results of tests pending at discharge (TPADs) require action. Challenges for hospitalists in dealing with actionable TPADs include identifying TPADs that are truly actionable and ensuring appropriate follow-up. Little is known about the factors that predict whether a TPAD is actionable. The goal of this study is to identify the types of tests, patient characteristics, and physician characteristics that predict whether a TPAD is actionable. Knowledge of this could support campaigns, such as Choosing Wisely, which address wasteful or unnecessary medical test ordering in hospital settings.
Methods: We performed a secondary analysis of 3374 TPADs on a randomly selected population of 1522 patients discharged from general medicine and cardiology services at a major academic center. After excluding normal and near normal TPADs, two board-certified internists independently determined whether the remaining TPADs were actionable or non-actionable by reviewing the electronic discharge summary. All discrepancies were adjudicated by the two reviewers. Odds ratios were calculated for the association of different factors with having an actionable TPAD, and the chi-squared test was used to determine statistical significance.
Results: Of the 3374 TPADs, we excluded 2554 TPADs that were normal or near normal. Of the remaining 820 TPADs, 253 (30.85%) were determined to be actionable by independent review. Overall, 7.50% of all TPADs were actionable. Among test types (Table), pathology and laboratory/chemistry TPADs were significantly more likely to be actionable compared to TPADs of all other test types (OR 95% [CI]: 1.86 [1.35-2.57]; and 1.49 [1.11-2.00], respectively). Microbiology and radiology TPADs were significantly less likely to be actionable compared to TPADs of other test types. Among physician characteristics, TPADs of patients cared for by internal medicine (IM) physicians were significantly more likely to be actionable compared to TPADs of patients cared for by subspecialty [non-IM] physicians. Among patient characteristics, TPAD results of patients with lengths of stay ≥ 3 days and annual incomes < $47,000 were significantly more likely to be actionable compared to TPAD results of patients with LOS < 3 days and annual incomes ≥ $47,000, respectively.
Conclusions: We have confirmed that a small percentage of TPADs are truly actionable and would therefore alter management. Furthermore, we have identified several factors that may predict “actionablity” of TPADs. Understanding factors associated with whether TPADs are actionable is a first step toward addressing wasteful testing and can be used to inform the design of systems to manage TPADs more effectively.
