Discussion

This retrospective study compared the set-up accuracy of carbon fibre and foam headrests in patients undergoing IMRT for HNC. With 1,000 image datasets analysed across five degrees of freedom, both headrest types demonstrated equivalent performance in maintaining patient position within clinically acceptable tolerances. This was reflected in the results (Appendix A), with only one significant finding, and in the boxplot analyses (Figures 2–6), which demonstrated tight clustering in both groups, confirming low variability. The only statistically significant finding—a 0.1° reduction in yaw random error in the foam headrest group—was well below the 1° rotational threshold typically considered clinically meaningful in HNC radiotherapy. ^{Reference van Herk4,13} It would have been valuable to calculate the difference in planning target volumes (PTV) margins between the headrests using Van Herk’s formula ^{Reference van Herk4} ; however, due to limitations—absence of delineation, intrafraction and registration data, and no measurement of residual errors prior to treatment—this was not possible. Future studies could undertake a representative margin recipe to assess the potential impact of headrest choice on treatment outcomes.

These findings build on earlier studies examining headrest design. Howlin et al. ^{Reference Howlin, O’Shea and Dunne9} and Houweling et al. ^{Reference Houweling, van der Meer and van der Wal10} reported that while customised or vacuum-moulded supports may marginally reduce set-up errors, such benefits often fail to translate into meaningful clinical gains, particularly in the era of daily image guidance. Our study adds value by directly comparing standard carbon fibre and foam headrests, addressing a gap where most studies focused on bespoke or modified supports. ^{Reference Howlin, O’Shea and Dunne9–Reference Rodrigues, Veen and van Egmond11,Reference Androjna, Marcius and Peterlin14}

Carbon fibre has long been favoured in radiotherapy for its physical properties—low density, high strength and excellent radiotranslucency—leading to less beam attenuation and greater dosimetric consistency. ^{Reference Njeh, Parker and Spurgin6,Reference Olson, Phillips, Eng, Lenards, Hunzeker and Lewis7,Reference Soriani, Strigari and Petrongari15} This is supported by our dosimetric findings (Appendix B), which showed the carbon fibre headrest demonstrated greater dose conformity and less attenuation than the foam headrest. These qualities are particularly important in high-precision techniques such as SABR and proton therapy, where dosimetric uncertainty can significantly affect clinical outcomes. ^{Reference Billiet, Vingerhoed and Van Laere16,Reference Olch, Gerig and Li17}

Despite these advantages, headrest comfort plays a crucial role in preventing intrafraction motion. Prolonged treatment times, often required in HNC radiotherapy due to complex dose modulation, can cause patient discomfort and increase motion during delivery. ^{Reference Dhillon, Erler and Poon18,Reference Kang, Lee and Chan19} While this study did not use formal patient-reported outcome measures (PROMs), radiographer feedback indicated that foam headrests may be perceived as more comfortable—an observation supported in previous literature comparing immobilisation devices. ^{Reference Houweling, van der Meer and van der Wal10,Reference Rodrigues, Veen and van Egmond11,Reference Lamprecht20} The literature indicates that patient comfort is sometimes overlooked in favour of accuracy and this is mirrored in other similar studies. ^{Reference Cheng and Wang21–Reference Goldsworthy, Latour and Palmer23} Mask tightness may also affect comfort; carbon fibre headrests do not compress under the patient’s weight, potentially increasing mask tightness and discomfort. Future studies should incorporate comfort assessment when evaluating HNC immobilisation.

Operational and financial considerations are also relevant. Foam headrests require replacement approximately every three years due to deterioration from daily disinfectant cleaning, whereas carbon fibre products offer greater durability and resistance to deformation. ^{Reference Leech, Coffey and Mast5,24} Suppliers quoted an average of £344 for a set of three foam headrests and £910 for a carbon fibre equivalent; brand names are not included, as the quotation is commercially sensitive. Radiographer feedback (Table 2) suggested carbon fibre headrests are favoured for operational consistency. Although foam may appear more cost-effective initially, carbon fibre could prove more cost and operationally effective long term, though the initial investment may challenge smaller centres.

The study’s findings—namely, the lack of clinically significant differences in set-up accuracy—raise an important practical consideration: is it justifiable to adopt carbon fibre headrests based solely on marginal dosimetric improvements? For standard fractionation with daily IGRT, the answer may be no. However, for stereotactic treatments, re-irradiation cases, or facilities exploring couch attenuation modelling, carbon fibre remains an appealing option.

There are limitations. The retrospective design introduces bias. Another potential bias is the mould room staff’s headrest selection; carbon fibre headrests were newly introduced in the department prior to data collection. The primary researcher (>3 years’ experience) examined the imaging datasets to reduce inter-observer variation as shown in numerous RT studies. ^{Reference Huger, Graff and Harter25–Reference Devereux, Frantzis and Sisson27} However, intra-observer bias may exist, as the researcher was not blinded to headrest type. While intra-observer variation in image analysis has been evaluated in numerous RT studies, ^{Reference Rodgers, Hales and Whiteside28–Reference Belshaw, Agnew and Irvine31} quantifying it was outside this study’s scope.

While the sample size was considerable, the absence of a formal power analysis limits definitive conclusions regarding headrest equivalence. Intrafraction motion was not evaluated—something that could influence outcomes with tighter PTV margins. Only the first 20 fractions were analysed, as weight loss and foam stability may affect displacements over time.

At the time, the host radiotherapy department used a combination of 3D CBCT and 2D kV imaging. Although dual image verification is acceptable in On Target 2 guidelines, ^{Reference McNair, Franks and van Herk1} CBCT verification is superior for detecting anatomical and physiological changes for corrective action before radiotherapy is delivered. ^{Reference Li, Zhu and Zhang32,Reference Kim, Pawlicki and Le33} All online displacements were corrected; offline checks corrected only systematic errors (>0.8 cm). No systematic errors were corrected via isocentre shifts. Future research should quantify differences between modalities, as this may skew set-up results. Future studies analysing foam and carbon fibre headrests or other immobilisation should use 3D CBCT verification to assess six-degrees-of-freedom, which may provide nuanced information on patient positioning.

Future research should include randomised controlled trials over the full course of treatment with PROMs, assessment of intrafraction stability through post-treatment imaging and long-term durability testing to quantify economic implications.