Recommendations for additional imaging (RAIs) play a critical role in the diagnostic process but have historically varied widely among radiologists and are often ambiguously phrased, leading to potential overuse of imaging and missed follow-ups. Recognising this challenge, a large academic hospital initiated a series of structured, technology-driven interventions aimed at improving the clarity, appropriateness and follow-through of RAIs. By comparing outcomes with a control site in the same healthcare system, which continued with standard practices, the study provided robust evidence on how such interventions can reshape radiology workflows and enhance patient safety.
Improving Clarity and Reducing Variability in RAIs
Initial concerns focused on the excessive variability and vagueness in RAI language among radiologists, which reduced the actionability of such recommendations. The first intervention, introduced in 2017, sought to address this by educating radiologists about inter-radiologist variability and the importance of using unambiguous, actionable language. This educational effort was supported by faculty meetings and leadership involvement, encouraging the adoption of clearer terminology in radiology reports.
Subsequent interventions built on this foundation. In 2018, a pilot programme introduced a closed-loop communication tool in the thoracic division, enabling referring providers to confirm, modify or reject RAIs. A dedicated safety net team helped ensure the completion of necessary follow-ups. In 2018 and 2019, feedback reports were issued to individual radiologists, comparing their RAI rates with peers in the same subspecialty. These reports heightened awareness and accountability, motivating further behavioural changes.
The fourth intervention, launched in 2019, expanded the communication tool across all divisions, accompanied by monthly reporting and a diagnostic certainty scale. This comprehensive programme, named Addressing Radiologist Recommendations Collaboratively (ARRC), marked a pivotal shift toward standardising RAIs across the department.
Measurable Impact on RAI Rates, Actionability and Resolution
The interventions yielded notable improvements across multiple dimensions. Over the eight-year study period, the RAI rate at the intervention site dropped by 44%, while remaining unchanged at the control site. Each successive intervention phase corresponded with a further reduction in RAI frequency, with the largest decrease occurring during the ARRC implementation.
Recommended Read: Rising Workloads in Diagnostic Radiology
RAIs at the intervention site also became markedly more actionable. Reports that included all key elements—modality, time frame and rationale—rose from just 5.6% to 42.3%. In contrast, the control site saw no significant change in actionability. The use of ambiguous or conditional language, as well as recommendations offering multiple or alternative options, also declined substantially at the study site, further reinforcing the clarity and utility of recommendations.
Resolution of RAIs improved in parallel. At the end of the study period, over 84% of actionable RAIs were either completed or scheduled at the study site, compared with just under 60% at the control site. These improvements suggest that clearer and more structured RAIs are more likely to be acted upon by referring providers, thereby enhancing patient follow-up and safety.
Factors Contributing to Success and Future Considerations
The success of the programme can be attributed to its multifaceted nature and strong institutional support. Leadership engagement, continuous performance feedback, mandatory fields in the ARRC tool and active communication between radiologists and referring clinicians created a system that not only encouraged better practices but also ensured their execution. The presence of a dedicated safety net team played a crucial role in bridging gaps in communication and care coordination, helping to close the loop on clinically necessary follow-ups.
The study also highlighted some influencing factors on RAI rates. For example, the involvement of radiology trainees was associated with higher RAI rates, possibly due to their heightened vigilance and tendency to err on the side of caution. On the other hand, greater radiologist experience was modestly linked to lower RAI rates. These findings suggest opportunities for targeted support or training based on experience level and workflow structures.
While the outcomes were compelling, the interventions were implemented in a single institution with a dedicated radiology quality and safety team, which may limit generalisability. Furthermore, as the interventions were introduced sequentially, their individual effects are best interpreted as part of a cumulative strategy rather than isolated efforts. Future research is needed to determine the feasibility and impact of deploying select elements in different institutional contexts.
The study demonstrated that targeted, technology-enabled interventions can significantly improve the quality and impact of radiologist recommendations for additional imaging. By reducing unnecessary imaging, enhancing recommendation clarity and increasing follow-through on clinically necessary RAIs, the initiative holds promise for broader adoption in radiology departments aiming to improve diagnostic precision and patient outcomes. Such structured approaches provide a practical model for aligning radiology practices with value-based care principles.
Source: Radiology
Image Credit: iStock
