Choosing the right scan: measuring appropriateness in modern imaging

Modern radiology operates at a scale that makes traditional oversight increasingly difficult. Drawing on international audit data and recent evidence, this article examines how measuring imaging appropriateness can help close the gap between regulatory intent and everyday clinical practice.

The contemporary practice of clinical radiology sits at a precarious intersection between technological brilliance and systemic inefficiency. As the cornerstone of modern medicine, diagnostic imaging informs the vast majority of critical clinical pathways. However, the scale of this enterprise has reached a magnitude that challenges our traditional frameworks of clinical governance. With over five billion imaging studies performed globally each year, the demand for radiologic expertise has outpaced the human capacity for manual oversight. This has led to a pervasive gap between the perceived value of imaging and the reality of its clinical application. In this environment, the philosophy of the management theorist Peter Drucker that one cannot manage what one cannot measure finds its most urgent medical application.  Importantly, these challenges should not be interpreted as a failure of individual clinicians or radiology departments. They reflect the scale and complexity of modern imaging practice, where referral volumes, time pressure, and fragmented digital systems have outpaced the tools available to support consistent, evidence-based justification at scale.

Concerns regarding appropriateness in medical imaging, particularly within computed tomography (CT), are not merely logistical. It is a profound issue of patient safety and systemic sustainability. In the UK, radiology teams operate under the rigorous Ionising Radiation (Medical Exposure) Regulations (IR(ME)R) 2017. Despite this, the operational reality of justification remains a significant challenge. Every unjustified exposure to ionising radiation carries a cumulative stochastic risk. This reality is often overlooked in the pursuit of rapid diagnostic answers. Beyond the individual radiation burden, poor justification drives a significant waste of finite resources. It needlessly inflates waiting lists and delays care for patients with genuine clinical need.

This inefficiency creates a bottleneck that compromises the entire diagnostic pathway. The volume of inappropriate scans directly competes with the timely delivery of life-saving interventions. To bridge the gap between regulatory intent and clinical practice, the radiological community must move beyond subjective justification. We must transition toward a framework where evidence based guidelines actively drive referral and vetting decisions.  This shift must be underpinned by a measurement based approach to the continuous audit of radiology testing. Such a framework requires a return to first principles.

The foundation of justification

Justification is the primary principle of radiation protection. It is a mandatory safety check designed to ensure that the individual health benefit of a medical exposure clearly outweighs the potential detriment. Under international frameworks and UK law, this intellectual process is divided into two core tenets: the Net Benefit Principle and the Alternatives Principle. The former requires a case-by-case assessment of clinical risk versus benefit. The latter mandates the prioritisation of non-ionising modalities, such as MRI or Ultrasound, where they provide equivalent diagnostic utility.

Despite these clear mandates, evidence suggests that CT is responsible for more than half of the total medical radiation exposure to citizens in many advanced economies. The responsibility for justification is shared between the referrer and the practitioner. However, we have historically failed to integrate evidence-based guidelines, such as the RCR’s iRefer or the ESR’s iGuide, into the referral workflow. This has led to high rates of inappropriate utilisation. In day-to-day practice, radiologists frequently mitigate these risks through protocol modification, vetting, and informal correction of referrals. However, reliance on individual expertise rather than measurable system design limits scalability, auditability, and assurance at population level.

Quantifying stochastic risk: the 2025 JAMA projections

The most significant recent scientific contribution to our understanding of population-level radiation risk is the modelling study published in JAMA Internal Medicine in April 2025 by Smith Bindman et al. This research provided a sobering quantification of the potential long term impacts of current CT utilisation patterns. Utilising data from a multicentre sample, the study projected the lifetime cancer incidence resulting from 93 million CT scans performed in a single year.

The projections indicate that approximately 103,000 individuals from that cohort will develop radiation induced cancer during their lifetime. Perhaps most critical for healthcare policy is the conclusion that if current practices continue, CT associated cancers could eventually account for 5% of all new cancer diagnoses annually in the United States. While these projections reflect population level modelling rather than individual patient risk, they highlight the importance of ensuring that every exposure delivers a clear net clinical benefit when applied across millions of examinations. While UK utilisation patterns differ, the biological principles remain identical. Adult abdominal and pelvic CT scans were the highest contributors to this risk, accounting for 38% of projected cancers.

Table 1: Projected lifetime radiation induced cancer incidence (JAMA 2025 model)

The EU JUST CT study: a multinational benchmark

The European co-ordinated action on improving justification of computed tomography (EU JUST CT) represents the first multinational effort to audit justification practices using a common methodology. Conducted between 2021 and 2024, the study collected over 6,700 referrals from seven Member States to assess appropriateness rates (AR). The study revealed deep disparities in guideline aligned imaging and optimal test selection.

Denmark led the cohort with an AR of 85.9%, while Greece recorded the lowest at 57.9%. These results suggest that justification is not a static legal requirement. Instead, it is a clinical skill that correlates strongly with national efforts in education and institutional quality assurance. A recurring theme was the appropriateness gap between specialists and general practitioners. In Belgium, for example, specialists achieved an AR of 80%, while general practitioners scored 53%. This disparity likely reflects the specialist’s deeper familiarity with a concentrated range of clinical pathways. This contrasts with the immense breadth of presentations encountered by general practitioners in primary care. Modern radiological technology is dauntingly complex, and the criteria for test selection evolve rapidly. It is clear that primary care clinicians require more robust, real-time support to navigate the diagnostic landscape effectively.

Table 2: Comparison of CT justification audit results (EU JUST CT)

Analysis by body region revealed that certain clinical indications are prone to systemic use of CT outside optimal first line pathways. Spine imaging had the lowest appropriateness rates, often falling below 40% and as low as 20% in Hungary. This is primarily driven by the use of CT for degenerative conditions, where MRI is the clinically superior and safer alternative. 

Referral quality at scale

A fundamental prerequisite for justification is the provision of adequate clinical information. Deficiencies in referral quality typically reflect time pressure, fragmented workflows, and system design rather than a lack of clinical insight or intent. If the clinical indications are missing or vague, the practitioner cannot perform the necessary benefit-risk assessment. The EU JUST CT study revealed a crisis of referral quality, with unscorable rates reaching 26% in some regions.

To standardise the measurement of referral quality, the Reason for Exam Imaging Reporting and Data System (RI RADS) has been developed. RI RADS grades imaging requests on a scale from A to D. Research using this system has shown that Adequate Requests (Grade A) are exceptionally rare. They account for only 1% to 4% of total referrals in some hospital settings. In contrast, Grade D deficient requests typically account for 53% to 68% of all imaging orders.

Table 3: The RI RADS classification system

Outcome impact of guidelines

The clinical impact of adhering to guidelines was recently quantified in a study by Tay et al. (2025) regarding cervical spine imaging. The results demonstrated that guidelines act as highly specific filters for significant pathology. The positivity rate was highest (4.3% to 7.2%) for referrals classified as appropriate. Critically, the positivity rate for studies classified as not appropriate was 0%. Imaging that fell outside guideline aligned pathways did not meaningfully alter subsequent patient management.

From manual audits to AI-powered system insights

The recurring theme across these studies is the limited ability of current healthcare systems to consistently measure the performance of their diagnostic pathways at scale. Traditional clinical audit is manual, labour intensive, subjective, and difficult to sustain in high volume environments. By the time manual results are analysed, clinical practice may have changed. This renders the data historically interesting but operationally less useful. The purpose of automated audit is therefore visibility and learning, not retrospective judgement of individual clinical decisions.

To achieve net benefit at scale, the radiology profession must transition to automated data audits. Artificial intelligence (AI) offers a solution to the audit bottleneck. Automated retrospective audit tools like xWave Insight can offer a broad system insight into institutional performance.

Retrospective large-scale audits can provide unique insights into the diagnostic enterprise. They enable educational targeting by identifying specific patterns of poor referral quality or clinical inappropriateness. This allows departments to deploy surgical upskilling programmes. These tools also facilitate waitlist optimisation. Referrals currently on waiting lists can be analysed to surface missed urgent cases or identify duplicates and poorly justified requests. Finally, continuous measurement ensures governance and compliance with IR(ME)R 2017 without labour intensive manual surveys. Mapping the diagnostic pathway reveals precisely where suboptimal justification is clogging the system. This allows for a more rational allocation of limited resources.

Strategic recommendations

The data synthesised in this exploration reveals a stark reality. Poor appropriateness rates and deficient referral quality drive a quantifiable public health risk. They also contribute significantly to waiting lists and health system inefficiency. To address this system level challenge, we must revisit the Drucker principle that measurement is the prerequisite for management.

Primary among these recommendations is the implementation of automated large scale auditing. Before deploying active solutions to improve appropriateness, we must establish an objective baseline of current performance. This retrospective insight allows us to identify specific gaps in clinical practice and referral completeness that manual surveys cannot capture. By first measuring these deficits, we can ensure that subsequent interventions effectively improve both appropriateness and referral quality.

Furthermore, the integration of digital clinical decision support at the point of care is essential. Evidence based guidelines such as iRefer must transition from passive reference materials to active digital gateways. By embedding these guidelines into the referral and vetting workflow, we ensure that the Alternatives Principle is considered before a patient is scanned.

Conclusion

The future of radiology lies in our ability to combine the precision of scan choice with the precision of our scan acquisitions. By embracing measurement as the foundation for improvement, we can ensure that every study provides the maximum diagnostic benefit with the minimum possible harm. Radiology is uniquely positioned to lead this transition, combining clinical expertise with data driven governance to improve patient safety and system efficiency. The path forward is scientific, quantitative, and data driven. We must close the gap between perception and reality for the ultimate benefit of the patient.

References:

Smith-Bindman R, Chu PW, Azman Firdaus H, et al. Projected Lifetime Cancer Risks From Current Computed Tomography Imaging. JAMA Intern Med. 2025;185(6):710–719. doi:10.1001/jamainternmed.2025.0505

Singer C, Saban M, Luxenburg O, et al. Computed tomography referral guidelines adherence in Europe: insights from a seven-country audit. Eur Radiol. 2025 Mar;35(3):1166-1177. doi: 10.1007/s00330-024-11083-x

Abedi A, Tofighi S, Salehi S, et al. Reason for exam Imaging Reporting and Data System (RI-RADS): A grading system to standardize radiology requisitions. Eur J Radiol. 2019 Nov;120:108661. doi: 10.1016/j.ejrad.2019.108661

Tay, Y.X., Foley, S.J., Killeen, R. et al. Positivity rates and subsequent patient dispositions after utilisation of cervical spine imaging referral guidelines in Singapore. Insights Imaging 16, 170 (2025). https://doi.org/10.1186/s13244-025-02048-9

Insight Platform. https://www.xwave.ie/xwave-insight-for-providers.  Accessed Feb 9 2026.

The content on this page is provided by the individuals concerned and does not represent the views or opinions of RAD Magazine.