Reducing first appointment delays for electron radiotherapy patients by improving the treatment planning pathway: a quality improvement project

Problem

The Clatterbridge Cancer Centre (CCC) National Health Service (NHS) Foundation Trust is one of the UK’s regional tertiary (specialist) hospital trusts. We have three sites in the northwest of England: in Liverpool city centre and its suburb of Aintree, and across the River Mersey at Clatterbridge on the Wirral. CCC provides highly specialized cancer care to both inpatients and outpatients, including pioneering chemotherapy, immunotherapy, gene therapy, haemato-oncology and radiotherapy. Our catchment is 2.4 million people across Cheshire, Merseyside and the surrounding areas, including the Isle of Man.1

Radiotherapy uses ionising radiation to kill or shrink cancerous tumours. The most common technique is to direct a beam of photons (mega voltage (MV) X-rays or gamma rays) or subatomic particles (electrons or protons) from outside the body using a linear accelerator (linac) or (for protons) a cyclotron. (The alternative to external beam radiotherapy (EBRT) is to place a radiation source inside the body, eg, brachytherapy.2) Most EBRT is photon therapy, but a small number of patients receive electron beam therapy (or, for very few, proton beam therapy).3 We have 10 linacs across our three sites, 2 of which are commissioned to provide electron radiotherapy. In 2021, we administered 728 electron treatments (fractions).

At CCC, all patients receiving MV photon therapy have a CT scan to aid planning. However, the majority of our electron patients do not, since clinicians deem it unnecessary in most cases; this group is described locally as non-scanned electron patients—this is the target group for our quality improvement (QI) project, referred to simply as ‘patients’ from now on. These make up approximately 1% of our total treatment workload at CCC.

Prior to this improvement project, our electron treatment planning care pathway had a number of tasks which required manual calculations and manual data entry. The small patient numbers meant that these tasks were unfamiliar to staff, the tasks having been automated on other EBRT pathways to improve patient safety. As a result, these manual tasks caused delays, producing significant time pressures at the patient’s New Start—their appointment for their first radiotherapy treatment (first radiation fraction). Further stress and delays at the New Start arose from having to determine the required setup parameters while the patient was present and adjust the treatment plan accordingly. Amending the treatment plan, at this stage and under pressure, can also introduce inadvertent errors.

The primary aim of this project was to reduce delays at New Start, and so improve the experience for both patients and staff. We hoped that as a result of changes, 90% of New Start appointments would commence within 30 min of their allotted time by August 2021, a substantial improvement over the previous level of 33%. There would be consequent secondary impacts: reducing staff stress and increasing patient safety.

We used the Model for Improvement (MfI) and its Plan–Do–Study–Act (PDSA) cycles.4–6 This approach has been used successfully in other NHS clinical science areas: life sciences7–10 and physiological sciences.11

Background

The majority of radiotherapy patients receive photon EBRT for the treatment of deep-seated tumours; however, for a subset of patients with superficial (shallow) tumours, electron EBRT is a more suitable treatment.12 The majority of these patients have non-melanoma skin cancer, which often appears on sun-exposed areas of skin such as the head, neck and other extremities, although any superficial tumour can be treated with electrons.12

For such superficial tumours, the advantage of electrons over photons is that they provide a high dose at the skin, followed by rapid dose fall-off deeper into the body (beyond the tumour).12 Tissue-equivalent material may be placed on top of the tumour to increase the skin dose to the desired level—usually 90–95% of the prescribed dose.12 Electrons are often used to treat lesions between 1 and 5 cm deep with a single radiation field; treatment is relatively simple compared with photons.12 The lesions are, by their nature, superficial and can be palpated and marked up in clinic, so our clinicians often choose not to authorise a CT planning scan for these patients.

The majority (95%) of our electron patients receive treatment to either their head and neck or to their extremities, and it is policy at CCC to immobilise these patients in a thermoplastic cast during treatment to limit movement. (Other common body radiation treatment sites are the breast and sternum for which the patient does not have a cast.) The cast is made and checked in the Mould Room as part of the planning pathway, as shown in figure 1.

Request permission

Previously, there had been numerous incremental changes to this patient care pathway. For example, in 2017, a paperless process was introduced, which improved both efficiency and safety (through staff familiarity, since all our other EBRT treatments were already paperless).

In 2019, further changes were made, including the introduction of plan approval by a medical physics expert (MPE)13 again to improve safety and bring this pathway into line with other EBRT treatments, all of which have either plan approval by the responsible clinician or an MPE under a clearly defined protocol.14 While this meant patient plans were approved within the ARIA radiation therapy management record-and-verify software system,15 it also meant that these approved patent plans had to be ‘unapproved’ to allow the final treatment parameters to be captured and added.

This pathway is different from all other types of EBRT, where patients have a CT scan. The CT scan allows appropriate treatment parameters (gantry, collimator and couch positions) to be determined at the planning stage. Non-scanned electron patients do not have a CT scan, so a treatment simulation has to be performed to determine setup positions while the patient is on the radiotherapy couch at their New Start appointment. To enable the newly determined treatment parameters to be captured, the previously approved plan needs to be unapproved. This unapproval with the patient on the bed causes stress and concern to staff. Time pressures caused by delays earlier in the planning pathway add to this stress. The consequences were delays at New Start plus 13 events (10 of which were plan not ready) over 21 patients requiring recording in our patient safety incident reporting system (Datix).16 17 These delays are consistent with the literature which demonstrates that electron patients experience longer delays than other EBRT patients.18

Similar QI projects specifically related to radiotherapy planning are rare. In cervical cancer treatment, a QI project reduced planning times for brachytherapy19 and another increased the proportion of cases treated within a target total time for prolonged chemoradiation.20 Price et al suggest how routinely collected data on healthcare outcomes could be used in a PDSA-type way to iterate towards better trade-offs between lung tumour reduction and sparing collateral damage to the heart.21

Measurement

For this project, we established a QI team comprising medical physicists and both pretreatment and treatment radiographers. Initially, a process map was developed from care paths recorded within ARIA. Figure 1A shows all steps in this part of the pathway. As shown, there are three teams responsible for tasks: physicists, radiographers and clinical oncologists. During previous lean-style waste removal6 work on this pathway, many non-value adding steps22 had been addressed. However, there were some still necessary non-value adding steps,23 shaded orange, that needed to be further investigated. These were: the decision point for planning/no planning (‘Electron No Scan?’) and the treatment expert practitioner (TEP) sign-off (‘TEP Sig’).

To assess whether changes constituted an improvement, we established a set of metrics and analysed them using statistical process control (SPC) charts24 25 with the NHS Excel XmR (individuals) template,26 modified to display case sequence on the X axis rather than dates (see figure 2). We also monitored safety incidents (through the Datix reporting system) to check for any inadvertent safety issues.

Request permission

Our chosen outcome metric (OM) was the delay experienced by the patient from their given appointment time to the time their treatment started. Baseline data (the 21 successive cases from September 2019 to September 2020) had an average delay of 52.4 min with considerable process variation (see figure 2, top). The upper process limit (UPL), which shows the expected maximum extent of random variation (mean+3 SDs), was around 3 hours. Our (internally agreed) target was to keep delays below 30 min. In 10 of these 21 cases, the treatment plan was not ready 30 min before the scheduled New Start time.

The physicists then conducted a root causes analysis6 to investigate causes of treatment delay at New Start caused by planning issues. shows an Ishikawa (cause-and-effect or fishbone) diagram6 27 of potential causes of these delays. This highlighted manual data entry and a lack of trained staff. To investigate these issues further, we analysed Datix incidents recorded over the same time period. shows a Pareto chart5 6 of these 13 incidents. The main causes were plan not ready and manual data entry (together causing over 90% of the incidents). These initial analyses motivated the QI project described in this paper.

To monitor the impact of changes on important parts of the planning pathway, we established a set of three process metrics (PMs), as shown on figure 1, with baseline values as shown on the SPC charts in figure 2:

PM1: the time taken to complete the initial planning by the radiographer, including any prior wait. The baseline was 46.0 hours, UPL nearly 160 hours.

PM2: the time taken to complete the Mould Room Checks, including any prior wait. The baseline was 64.9 hours, UPL over 250 hours.

PM3: the time taken to complete the Final Checks by the radiographer, as shown on the relevant parts of figure 1A,B, including any prior wait. The baseline time was 23.6 hours with UPL around 135 hours. One patient was an outlier, initially delayed due to waiting for a skin graft to heal and then further delayed due to our electron-capable treatment machines being moved from our Wirral site to our Liverpool site.

We set internal targets of 12 hours for the initial plan to be completed and 24 hours for each of Mould Room Checks and Final Checks to be completed.

Design

Reviewing the data, the QI team considered the third question of the MfI:

What changes can we make that will result in the improvements that we seek?

Strategy

Applying the MfI, we conducted PDSA cycles to test and refine the change ideas.

Results

QI interventions began in October 2020. From then until the end of our study period, 75 non-scanned electron patients were treated. We have reduced the mean delay at New Start (figure 2, OM) from 52.4±45.5 min to 27.2±26.8 min. This is within our internal target of 30 min, but there remains considerable variation with a relatively large UPL, so our performance is not as reliable (consistent) as we would like.

Prior to the QI project, 33% of patients started their first fraction within 30 min of their appointment time; for the first 12 months of the project, this increased to 63.2%; and in the final cycle (A5), this further improved to 69.2%. However, there are confounding factors on these timings, such as other delays and machine breakdown. Therefore, we also analysed the impact of planning by considering the number of patient plans actually ready by 30 min prior to the New Start appointment time. Prior to the QI project, 48% of plans complied with this; in the final cycle, we improved this to 92.3%. We consider this a major achievement and benefit to patients.

Additionally, prior to this project, there were 10 potential safety (Datix) incidents recorded (from a sequence of 21 patients) due to plans not being ready by the New Start time. This has been reduced to zero as a result of the changes made during this project.

We have reduced the time to complete initial planning (PM1) from 46.0 hours to our internal target (12 hours), on average. We reduced delays at other stages through staff engagement sessions, for example, prior to the project, Mould Room Checks (PM2) took an average of 64.9±57.4 hours. Through our PDSA cycles, we have reduced this to 25.0±23.0 hours, close to our target of 24 hours. This gives us more safety margin later in the process.

It is disappointing to note that Final Checks (PM3) now takes longer to complete than it did prior to the improvement project, initially 23.6±47.8 hours vs 47.1±42.5 hours in cycle A5, and we are far from our 24-hour target. This may be an unintended consequence of the extra time noted above: going beyond safety margin to overly relaxed. We intend to share these data with staff, as was effective for Mould Room Checks, in an effort to reinforce awareness and improve Final Checks completion times.

Moving away from manual data entry in ARIA (cycle B1) did not feed through directly to reduced delays at New Start (our OM), contrary to our predictions. We thought that by making plans more similar to other EBRT sites, and so increasing the pool of trained staff available to work on the pathway, we would reduce delays. While the number of trained staff has increased, the overall treatment delays (OM) have remained fairly constant since the introduction of the day 0 appointment in cycle A1. Looking at figure 2 suggests that, as noted above, shortened initial planning duration (early in the planning process) was countered by lengthened duration for Final Checks (at the end). However, these changes to the pathway have improved the safety of the planning process.

Lessons and limitations

Some changes were very successful—in particular the day 0 proxy appointment—but took several cycles of PDSA learning and refinement to embed effectively. Some of these refinements might have been built into the change idea design initially, but it is hard to foresee the impacts of such fundamental changes to a pathway like this.

As clear in table 1 and figure 2, in some PDSA cycles, we tested several different change ideas or refinements at the same time. This is not good QI practice as it makes the effects of different changes hard to disentangle.29 It was a compromise here due to the need to make progress in attempting improve performance and maintain QI momentum, even though data (cases) build up relatively slowly in this system.

Our aim was to reduce delays at New Start, as a proxy for improving patient experience. While by the end of the project, our treatment plan was now ready for over 92% of patients (vs 48% previously), we were at 69% of New Starts within 30 min of appointment time (vs 33% previously)—some way short of our aspiration of 90%.

Conclusion

This project aimed to improve patient experience through the proxy measure of reducing the delays they experience at their first radiotherapy treatment (New Start), inevitably anyway a stressful occasion. Consequent secondary outcomes were to reduce staff stress at New Start and to improve the safety of the hitherto rather manual and less familiar electron treatment planning pathway.

Our initial data collection showed that the majority of patients were experiencing material delays at New Start and that, in the majority of cases, plans were not ready in time for treatment. There had been previous attempts to improve workflow, but these focused on individual tasks rather than the entire pathway.

We judge the project a success: our change ideas have proved worthwhile. Delays at New Start have been reduced very considerably, and in particular the contribution of treatment planning to this. In addition, the non-scanned electron radiotherapy planning pathway is now better aligned with our other EBRT pathways at CCC, with unfamiliar and risk-prone manual tasks removed. This has enabled an increased pool of trained staff available to work on the pathway and has increased patient safety.