CT scans save lives every day. They catch strokes in time, spot hidden internal injuries, and guide urgent decisions in theatres and emergency rooms.

Yet a major new analysis suggests the radiation from today’s levels of computed tomography use could carry a bigger long term cancer impact than previously recognised.

The study, published in the peer reviewed journal JAMA Internal Medicine, estimates that CT examinations performed in 2023 may lead to roughly 103,000 future cancers over the lifetimes of the people scanned. That is a significant number. About one in twenty of all new cancer diagnoses if current practice continues unchanged.

The research team drew on one of the largest collections of real world CT data in the United States. They used the University of California San Francisco International CT Dose Registry, which pools detailed information from 143 hospitals and imaging centres across 20 states. Every exam record includes the technical specifics: the energy settings, scan length, whether the study used a single phase or several passes, and the patient’s size and age.

Using these inputs, scientists reconstructed organ doses with Monte Carlo simulations. Think of it as a virtual twin of each scan, estimating how much radiation reached the lung, colon, thyroid and other organs. These dose profiles were then put through the US National Cancer Institute’s risk tool, which is grounded in the widely accepted BEIR VII models for radiation induced cancer.

The team scaled the risk to national volumes using an industry survey of CT usage and cross checks from major radiology registries, then excluded scans performed in the last year of life. That last step matters, because those patients are unlikely to live long enough to develop a radiation related cancer.

The numbers are striking. About 93 million CT examinations were performed in 2023 on 61.5 million people in the US. Adults accounted for the vast majority of scans. Children received roughly 3.3 percent of examinations yet face a higher risk per scan because their tissues are more sensitive and they have more years ahead. Even so, adults contributed nine in ten of the projected cancers because they undergo far more imaging.

Which cancers make up the bulk of the projection? Lung tops the list at around 22,400 cases. Colon follows with about 8,700. Leukaemia appears next at about 7,900, then bladder and stomach. For women, breast cancer is prominent too, with an estimated 5,700 cases related to CT radiation over time.

These estimates reflect patterns in what gets scanned most often and how much radiation different organs receive. Abdomen and pelvis examinations drive the largest share overall. They account for an estimated 39,100 future cancers. Chest CTs contribute around 22,700. Spine about 12,900. Head around 12,500.

Among children, head CT features most heavily. In adults, abdomen and pelvis dominate, in part due to multiphase studies that repeat imaging during one appointment to capture contrast at different times. Multiphase protocols deliver greater diagnostic detail in selected situations. They also increase dose when used broadly.

Risk is not flat across ages. It is highest for infants and falls with increasing age at exposure. For example, girls younger than one year had an estimated 20 cancers per 1,000 examinations. By later adolescence, that risk per exam drops to about 2 per 1,000. Adults have lower risk per exam than children. Yet the huge number of adult scans outweighs the per exam difference. In absolute terms, adult imaging accounts for roughly 93,000 of the 103,000 projected cancers.

How certain are these estimates? The researchers report 90 percent uncertainty limits of 96,400 to 109,500 for the central projection. They also ran sensitivity analyses to test major assumptions. Lowering estimated doses by 20 percent dropped the projection to about 79,900 cancers. Raising doses by 20 percent increased it to about 126,600.

Adjusting national scan volumes by plus or minus 10 percent shifted totals accordingly. Using a higher national estimate of paediatric CT share increased the overall projection by 11 percent and raised the fraction attributable to childhood imaging to nearly a quarter. The scale of the effect remains large across scenarios.

This is not the first attempt to quantify the risk. A widely cited analysis from 2009 estimated about 29,000 future cancers from 2007 CT usage in the US alone.

So why are current projections three to four times higher? Several factors stand out. First, there are more scans now. Volumes have risen by more than 30 percent since 2007, surpassing pre pandemic levels.

Secondly, today’s modelling is more granular. The new study explicitly counts multiphase imaging, which occurs in almost three in ten examinations. Earlier work could not, due to limited data. Thirdly, dose estimates here are based on examination level technical details from more than 120,000 real scans matched to modern computational body models, not just protocol averages or survey figures. Lastly, scans were grouped by both body region and clinical purpose, which better tracks dose differences than broad categories alone.

The findings are not an indictment of CT. Far from it. Experts stress that CT is often the right test. In emergencies like suspected stroke, major trauma or pulmonary embolism, speed and accuracy are critical. CT delivers both. For cancer diagnosis and staging, it is central to care. For many patients, skipping the scan would be a far greater risk. The message is to use CT when it will change management, and to tailor dose so images are good enough for diagnosis at the lowest reasonable exposure.

There are practical steps for patients and clinicians. Ask whether the scan is necessary today. Will it change treatment? Could ultrasound or MRI, which do not use ionising radiation, provide the answer?

Share prior imaging to avoid repeats. If a child needs a CT, check that paediatric protocols will be used. For radiology teams, scrutinise multiphase protocols and default settings. Many abdomen and pelvis studies can be done single phase without losing crucial information. Modern scanners have dose modulation and advanced reconstruction that preserve image quality at lower dose. These tools should be fully used, especially for smaller patients.

From a public health perspective, the projected 5 percent share of cancers linked to CT, if current practice persists, places medical imaging among other modifiable risks. Alcohol is associated with about 5 percent of cancers in the US. Excess body weight about 8 percent.

As with those exposures, the aim is not elimination but harm reduction. Less unnecessary scanning. Optimised protocols. Better tracking and feedback. National dose registries and quality programmes can support this shift by benchmarking departments, highlighting outliers and promoting best practice. Commissioners and payers can reinforce change by discouraging low value imaging and rewarding adherence to evidence based guidelines.

The study also lifts the lid on a quiet but important part of risk science. The models used, set out by BEIR VII and implemented in the National Cancer Institute’s Radiation Risk Assessment Tool, draw from decades of research, including follow up of Japanese atomic bomb survivors and groups exposed through medical or occupational sources.

These models estimate excess lifetime cancer risk given an organ dose at a particular age, adjusted for competing risks like death from other causes. There remains debate about the exact size of risk at very low doses and low dose rates, and whether risk differs by radiation type. The authors acknowledge these uncertainties and include them explicitly. They also note that if low energy X rays from CT cause more biological damage per unit dose than assumed, actual risks could be nearer the upper end of their ranges.

Limitations deserve a clear airing. The projections rely on transferring risks from historical cohorts to contemporary populations. The life expectancy of patients undergoing imaging may be shorter, though the team tried to account for this by excluding scans in the last year of life and testing a two year exclusion.

Categorising scans can misclassify a minority of examinations, although validation suggests high accuracy and any errors are unlikely to swing national totals. CT guided procedures such as biopsies were not included and can involve additional exposure. National practice patterns may differ from the registry’s sample, although the sensitivity checks help bracket that uncertainty.

So what does this mean for you? If a clinician recommends a CT, ask about purpose and alternatives, but do not be fearful. The immediate benefit to diagnosis and treatment is often decisive. In urgent care, CT is routinely the best choice and delays can be dangerous.

For elective imaging, a conversation helps ensure the scan is justified. Bring records of recent imaging to appointments. Keep a simple personal log. Small actions that can avoid duplication.

For healthcare providers, the study is a prompt. Review high dose protocols. Reconsider routine multiphase series without clear indication. Embed paediatric imaging expertise. Track doses and provide feedback to referrers. Strengthen shared decision making with patients. The path forward is not about rationing care. It is about smarter, safer use.

The bottom line is clear. The study gives the most comprehensive contemporary estimate of how much cancer risk could accrue from current CT use in the US. About 103,000 projected cancers from 93 million scans in one year. Abdomen and pelvis and chest exams in adults dominate the total, with head CTs most consequential in children. The risk per scan is small for any one person. The collective effect is not. Thoughtful imaging, optimised dose, and fewer unnecessary scans can preserve the life saving power of CT while reducing avoidable harm.

In a world where CT is both indispensable and ubiquitous, that balance matters. We do not need to fear the machine. We need to respect it, use it wisely, and make every scan count.

FAQ