Undergoing CT scans during childhood is associated with a small but significant risk of developing cancer, a large data linkage study from Australia has found. The absolute excess incidence rate for all cancers combined was 9.38 per 100,000 person-years at risk after nearly ten years’ follow-up.

A group of medical physicists, statisticians and epidemiologists compared data on 680,211 Australian patients aged up to 19 years who had had a CT scan, with around 10 million controls who had not been exposed to CT scanning, matched for age, year of birth and gender.

Median follow-up was 9.5 and 17.3 years for exposed and non-exposed individuals, respectively. For the first year after their scan, cancers in exposed patients were not included in the comparison. This lag period excluded those therefore who were having a CT scan as part of the work up for an already suspected cancer.

A variety of cancers were found to occur more commonly in exposed patients. These included solid tumours (e.g. of the digestive organs, urinary tract, ovary and melanoma) and lymphoid and haematopoietic cancers. Overall, 57,254 cancers were identified among the non-exposed subjects and 3,150 in the exposed group. This equated to 24% more cancers in the exposed group; an excess of 608 cancers (147 brain, 356 other solid tumours and 105 lymphoid and haematopoietic cancers).

The overall incidence rate ratio (IRR) in the exposed patients was 1.24 (95% CI:1.20-1.29, P < 0.001). An extra 0.16 increase in IRR was seen with each extra CT scan exposure (18% of patients had more than one CT scan).

Reanalysis of the data using longer lag periods still showed increased IRRs for the exposed group at 1.21 (95% CI:1.16-1.26) and 1.18 (95% CI:1.11-1.24) with lags of five and ten years, respectively (P < 0.001 for both).

Observational studies are prone to bias and in this study the original decision to perform a CT scan, which was not random, posed a concern. The lag period requirement would have reduced this risk of course. However, looking at brain cancer several observations suggest that the increased risk observed is real.

The brain was imaged in 60% of all scans. Brain cancer risk was greatest between one and four years after the scan. Although IRRs dropped thereafter they remained significantly raised 15 years later.

Brain cancer was higher after brain CT scanning than after abdominal or chest scans. The IRR of brain cancer after a brain scan was 2.44 (95% CI:2.12-2.81). Testing reverse causation the IRR of brain cancer after a non-brain CT scan was 1.51 (95% CI:1.19-1.91). Patients less than five years old at the time of their scan had a higher IRR than older children (P < 0.001 for trend for all cancers in under five year olds). However, the bottom line estimates were more reassuring; one extra cancer for every 1,800 scans, and one extra brain cancer per 4,000 brain CT scans performed.

Although the dose of ionising radiation used in CT scans has dropped in the past decade, concerns still remain that it increases the risk of cancer, especially in children.

The authors conclude that their results are ‘generally consistent with the linear no threshold theory (that is, there is no threshold dose below which there is a zero risk)’.

Every CT scan, especially in children, needs strong clinical justification. Proposed scanning for relatively minor indications should, say the authors, follow only after the use of objective decision analysis tools.

As ever, the lowest effective dose of radiation should be deployed.

Dr Chris Barclay