De-implementation of routine laboratory testing in a surgical intensive care unit: lessons from a two-phase quality improvement initiative

Excessive laboratory investigations in the intensive care units (ICUs) contribute to increased risk of line infections, iatrogenic anaemia, environmental burden and healthcare costs without clear mortality benefit. Despite Choosing Wisely recommendations advocating clinically indicated testing, overutilisation of laboratory tests in ICUs remains common. In our surgical ICU, ordering practices were driven by unclear consultant expectations, reassurance-seeking, junior doctor inexperience, ease conferred by order sets and limited awareness of value-based care.

We conducted a two-phase quality improvement initiative between December 2023 and December 2025 to reduce laboratory utilisation. Root causes were identified via clinician surveys, visualised using an Ishikawa diagram and prioritised using Pareto charting. The primary outcome was the number of laboratory investigations ordered per patient-day. Counterbalance metrics included ICU mortality and length of stay (LOS).

Interventions were mapped to a driver diagram. Phase 1 focused on education, visual reminders, encouraging shared decision-making, nursing empowerment to clarify orders and mid-phase audit and feedback. Phase 2 incorporated system-level strategies, including electronic prompts to reconcile investigation orders, frequent digital reminders and engagement of new junior doctors. Data were analysed using run charts according to standard healthcare improvement rules for special cause variation.

During Phase 1, median laboratory investigations increased from 12.9 to 13.3 tests per patient-day despite comparable illness severity. ICU mortality was similar in the pre-intervention period (12.7%) and during our first study phase (12.9%), without a consistent temporal association with interventions. In Phase 2, laboratory utilisation returned to baseline levels with reduced ICU LOS and mortality. No sustained reduction was observed.

In conclusion, educational and behavioural interventions alone were insufficient to de-implement entrenched ordering practices in a high-acuity ICU environment. Sustainable reduction likely requires stronger system redesign, including order-set modification and embedded decision support. Reporting null effects in de-implementation initiatives provides important insights into the structural and cultural barriers to reducing low-value care.