LABORATORY QUALITY, DIAGNOSTIC ACCURACY, AND WORKFLOW OPTIMIZATION IN CLINICAL LABORATORIES: A SYSTEMATIC REVIEW

Abstract Background: Clinical laboratories affect most medical decisions, yet quality threats, diagnostic interpretation errors, and inefficient workflows delay care and increase risk. We aimed to synthesize data from original research on interventions and systems that improve laboratory quality, diagnostic accuracy, safety of laboratory interpretation, and workflow performance in clinical laboratories. Methods: A PRISMA-aligned systematic review was conducted using PubMed Central as the mandatory full-text source. We included original studies evaluating quality improvement, automation, or decision support affecting measurable laboratory outcomes. Two reviewers performed screening and extraction. Due to heterogeneity of designs and outcomes, results were synthesized narratively. Results: Ten original studies met eligibility. Lean-based redesign in emergency and core laboratory pathways reduced turnaround time (TAT) and improved flow. Digital monitoring integrated with Lean Six Sigma was associated with reduced intra-laboratory TAT. Automation interventions improved timeliness and efficiency, including tube sorting, registration, total laboratory automation (TLA) performance and predictability, TLA system fusion decreasing prolonged out-of-range TAT, and microbiology automation markedly shortening TAT for negative reports. Quality-indicator programs quantified preanalytical error burdens and targeted improvement opportunities. A prospective cohort study of an AI decision-support tool for laboratory interpretation reported clinically relevant accuracy and high safety sensitivity for urgent, emergency cases. Conclusions: Across varied settings, workflow redesign (Lean), automation (preanalytic modules and TLA), and structured quality-indicator monitoring consistently improved operational performance and highlighted actionable error sources. Emerging AI decision support may enhance diagnostic safety, but broader validation is needed.