Original Article

Clinical Prioritisation for Curative Radiotherapy: A LocalWaiting List Initiative

J. M. Martin, G. Ryan, G. Duchesne

Division of Radiation Oncology, Peter MacCallum Cancer Centre, St Andrews Place, East Melbourne, Australia

ABSTRACT:Aims: Waiting time for radiotherapy is a major problem in radiation oncology practice. The aim of this paper is to present theexperience of the Peter MacCallum Cancer Centre in trialling a number of strategies to reduce patient waiting times.Materials and methods: All patients starting megavoltage radiotherapy with curative intent in three separate 1-week blocks had theirwaiting times recorded. The cohorts were each 8 weeks apart and were before (September), during (November) and after (January)the introduction of a priority points system.Results: Median waiting time was 35 days in September, 42 days in November and 31 days in January. The number of extremelylong waits (>90 days) decreased to 1 by January. Significantly more patients were pre-booked for treatment in January (27/51)compared with September (17/65; P=0.003) and November (12/65; P<0.001). Pre-booked patients had shorter waiting timescompared with patients who was not pre-booked (P<0.0001). Difficulties at one particular treating location contributed to thelonger median waiting times in November. Although there had no significant difference in waiting time in non-breast unit patientsbetween the three cohorts, there was a decrease in waiting times in breast unit patients, especially between November and January(P=0.0008). There was no significant increase in delay to starting treatment in other treating units, resulting in more equitable accessacross all units.Conclusions: A combination of encouraging pre-booking and the introduction of a priority points system has led to a decrease inwaiting times, especially among breast unit patients. Martin J. et al. (2004). Clinical Oncology 16, 299–306

� 2004 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.

Key words: Curative radiotherapy, treatment delay, waiting lists

Received: 14 July 2003 Revised: 3 December 2003 Accepted: 31 December 2003

Introduction

Undue waiting to commence radiation therapy is one ofthe most pressing concerns facing radiation oncologytoday [1,2]. Outside factors and influences within ourpractice have been responsible for its growth. Recentyears have seen developing recognition of the impor-tance of radiation therapy as an essential modality inthe treatment of cancer, with evidence to suggest thatbetween 50–60% of patients will benefit from the modal-ity at some time during the course of their illness [3,4]. Inaddition, our ageing population has an increasing cancerincidence, in turn increasing the need for radiationtherapy. Within our practice, treatments have becomelonger and more complex as worldwide shortages ofprofessional staff have developed, notably among theradiation therapy workforce [5–7].

One of the effects of a waiting list is to increaseanxiety among patients and staff [8]. Patients are uncer-tain about their progress through the list, and have fearsof being dropped off the list and the progression of theirdisease. Staff spend increasing amounts of time fieldingcalls from concerned patients and relatives, and sufferthe increasing pressure associated with a back-log ofwork. Prompt treatment is perceived as necessary toprevent primary tumour progression, the evolution ofmetastatic disease and possible tumour re-growth afterprimary surgery. A delay between planned surgery andradiation therapy in some circumstances may lead topoorer outcomes [9] and poorer rates of local controland metastasis-free survival, although these findings arenot always consistent [10–14].

Faced with increasing delays in commencing patientson curative radiation therapy, we initiated a pilot studyto trial a new waiting list strategy based, as far aspossible, on evidence of the clinical efficacy of themodality. The aims were as follows: (1) to eliminate asfar as possible excessively long waiting times; (2) toreduce levels of stress among staff and patients; and to

Author for correspondence: Dr Jarad Martin, Division of RadiationOncology, Peter MacCallum Cancer Centre, St Andrews Place,East Melbourne, Victoria 3002, Australia. Tel: +61-3-9656 1111;Fax: +61-3-9656 1424; E-mail: Jarad.Martin@petermac.org

Clinical Oncology (2004) 16: 299–306doi:10.1016/j.clon.2003.12.008

0936-6555/04/040299+8 $30.00/0 � 2004 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.

(3) ensure equitable access based on clinical priorityacross different treatment facilities and disease sites.

Methods

A points system to accord patients appropriate clinicalpriority when prescribed for curative treatment wasdeveloped on the basis of available evidence of efficacy,or consensus opinion of the likely patient benefit ifevidence was inconclusive. Each specialist unit wasasked to rank the priority for their patients to starttreatment from five (being most urgent) to one (theleast urgent), and then had to justify its scoring in anopen forum involving the entire radiation oncologydepartment (Table 1).

Patients who were enrolled on a clinical trial earnedone extra point (patients offered trial participation werenot given this information). In addition, for each weekthat a patient remained on the waiting list, they gainedan arbitrary extra half point to a maximum of fourpoints, allowing recognition of delay as well as clinicalstatus in determining their starting time. This was toredress the situation before the trial in which non-urgentpatients were delayed repeatedly at the expense of thosejudged urgent by their treating clinician. Each week, thecurative patient starting slots at each site were allocatedto the patients with the highest scores according to thepoints system, taking into account clinical priority andthe duration of waiting to date. Typically, the number ofpoints needed to start treatment was in the order of7–7.5, although this depended on the number of patientsin the system and the case-mix.

At the time of initial assessment, each patient wasgiven a letter from the Chief Executive Officer outliningthe new waiting list structure, waiting times at eachcentre and contact people who could be reached forfurther discussion. A dedicated staff member was intro-duced to keep patients up-to-date with their progressthrough the list. Complaints regarding the waiting listbefore and after the implementation of the points systemwere monitored through the patient advocate. Thisperson had the role of initial hospital contact foranybody who had a treatment-related concern.

Definitions

Prescription date was the date that the prescription wassubmitted. Ready for care (RFC) date was the date thatthe patient would ideally commence radiotherapy (e.g.after completion of neoadjuvant hormones for prostatecancer or after recovery from surgery). For manypatients, the RFC date was the same as the prescriptiondate (e.g. definitive radiotherapy for non-small cell lungcancer). Some patients were pre-booked, defined as aninstance in which the prescription specified an RFC datein the future; these patients were assigned their clinicalpriority scores, and accrued the extra delay points asabove during the interval. Start date was the date that

treatment was first delivered. Waiting time was thereforedefined as the period between the RFC date and thestart date.

This scoring system was implemented across fourtreatment sites, including urban, suburban and ruralunits, in September 2002. After an 8-week trial period,interim perceived problems with inconsistent scoringwere rectified. The pilot period ended on 18 November2002. A further follow-up period was examined toascertain the ongoing impact of this system.

Three time-points were used: initial (September),interim (November) and follow-up (January), eachseparated by an 8-week period. On each occasion, a5-day (Monday to Friday) period was used. All patientsstarting treatment in each week were identified throughthe centralised radiotherapy computer database. Treat-ment prescriptions and patient notes were individuallyreviewed to ensure appropriate capture of all patients tobe treated with megavoltage radiotherapy with curativeintent. This method of defining waiting time thereforereflects the actual delay from an ideal treatment startdate. The alternative method of looking at all patientscurrently on the waiting list has the problems of notdescribing an end point (i.e. start date), and possiblemultiple counting in different time periods. Datarecorded included prescription date, ready for care date,treatment start date, treating unit (defined as per Table1), location of treating unit and whether the patient waspre-booked. Key parameters assessed were the mean andmedian waiting time, maximum waiting time, unit-specific factors, effect of pre-bookings and the effect oftreatment location.

Notes were made of other potential confoundingfactors, such as workforce variations, extraordinarybookings (e.g. excessive total body irradiation commit-ments) or machine downtime during the study periods.

Statistics

Owing to the presence of long waiting time outliers, thedistributions are all skewed to the right and, because ofthis, fail normality tests. Therefore, non-parametric testswere used. The Kruskal–Wallis test was used with thenull hypothesis that the medians were equal between thethree groups. The Mann–Whitney test was used forpaired comparisons of medians. Chi-squared tests wereused to assess row and column independence in contin-gency tables. For binominal data, such as pre-booking, atest of two proportions was used. In all cases, thesignificance level of the tests was set at the P=0.05level. All statistics were calculated using the statisticalsoftware package Minitab v13.32 (Minitab Inc).

Results

Overall Waiting Times

Table 2 summarises the various waiting time parameters.The Kruskal–Wallis test, comparing the three datasets,

300 CLINICAL ONCOLOGY

Table 1 – Priority rating points system for curative radiotherapy

CL

INIC

AL

PR

IOR

ITISA

TIO

NF

OR

CU

RA

TIV

ET

HE

RA

PY

301

was highly significant for a difference in rank medians(P=0.007). Figure 1 shows that the November datasethad a significantly longer waiting time than the Januaryset (P=0.0032), and had a difference approaching signifi-cance compared with the September dataset (P=0.0544).The decrease in waiting times between January andSeptember was not significant (P=0.1165). The medianwaiting time in September of 35 days increased inNovember to 42 days because of changes in workinghours at one centre, but decreased again in January to31 days despite continued reduced working hours.

Prolonged Waiting

A few patients starting in September and Novemberexperienced waits of longer than 90 days (Table 3),none of whom were pre-booked. In September, allthree patients were at treatment centre A whereas, in

November, six out of eight were at centre D. There wereno prolonged waits at either centre B or C. Centre Ahad only commenced treatment in April, and was stillstarting a substantial back-log of patients prescribedtreatment before the centre opened. All six of thepatients who had to wait 109 days or longer were fromthe breast unit. The urology unit contributed five of thesix remaining patients, reflecting the perceived non-urgent clinical priority. The extremely long waits had allbut disappeared by January.

Effect of Pre-booking

The proportions of pre-bookings in each cohort were asfollows: 17 out of 65 in September, 12 out of 53 inNovember and 27 out of 51 in January. A Chi-squaredtest is significant for dependence between pre-bookingand month (P<0.0005). The corresponding tests of twobinomial proportions show that January had signifi-cantly more pre-bookings than either September orNovember (September vs January: P=0.003; Novembervs January: P<0.001; November vs September:P=0.290). Figure 2 shows that patients who were pre-booked had significantly shorter waiting times thanpatients who were not (P<0.0001). This was despite thefact that pre-booked individuals starting in Novemberexperienced significantly longer delays than in the othertwo time periods (September P=0.0266 and JanuaryP=0.0066). There were no significant differences in wait-ing times for the patients who were not pre-bookedbetween the three cohorts (Kruskal–Wallis P=0.705).

Effect of Treating Location

The longer waits in November were solely a result ofreduced working hours at location D, illustrated in Fig.3. This difference in waiting times between location Dand the other sites did not reach conventional statisticalsignificance (P=0.0895) because of the small numbers ofpatients involved.

Table 2 – Selected demographics on each cohort

Month

September November January

Number 65 65 51Median 35 42 31Mean 37.8 47.8 31.2Minimum 0 0 0Maximum 118 121 92

WT (days) 5 25 45 65 85 105 125

January 27% 37% 75% 90% 96% 100% 100%

November 3% 15% 55% 80% 87% 92% 100%

September 5% 29% 71% 89% 94% 98% 100%

0 10 20 30 40 50 60 70 80 90 100 110 120

0

10

20

30

40

50

60

70

80

90

100

Waiting time (days)

Cum

ulat

ive

per

cent

January

SeptemberNovember

Fig. 1 – Cumulative frequency histogram comparing the wait-ing times for the three time cohorts. Mann–Whitney statistics:September vs November: P=0.0544; November vs January:P=0.0032; September vs January: P=0.1165. The waiting timesdeteriorated in November because of reduced working hoursbut, despite this, continuing improvements in waiting timeswere being achieved by January.

Table 3 – Characteristics of patients waiting greater than 90 days tocommence radiotherapy

Patient Waiting time Location Unit Pre-booked Month

1 91 A Urology No September2 91 A Urology No September3 118 A Breast No September4 99 D Urology No November5 104 D Urology No November6 105 D Urology No November7 109 D Breast No November8 109 D Breast No November9 110 A Breast No November10 117 A Breast No November11 121 D Breast No November12 92 A Benign No January

302 CLINICAL ONCOLOGY

Unit-specific Data

The three units starting most patients were breast(n=66), gastrointestinal/sarcoma (n=33) and urology(n=28). The latter unit only started four patients in theJanuary data set week, which made it unsuitable forvalid comparison. Figure 4 shows the cumulative fre-quency graphs of waiting times for the breast cancerpatients, showing a significant improvement betweenNovember and January, and a non-significant benefitacross the whole time period. There were no significantdifferences in the curves for the gastrointestinal andnon-breast units (Kruskal–Wallis P values 0.32 and0.68, respectively). This was despite the continuedpressure of reduced working times at location D inJanuary.

Complaints

In the 9 months before the instigation of the pointssystem, five informal complaints and three formal com-plaints were made about waiting times for radiotherapy.In the 4 months during which the points system wasbeing established, four informal complaints weremade. In the 5 months since then, there have been nocomplaints.

The staff member assigned to speak to patients tonotify them of their progression through the waiting listfound that such pre-emptive contact was not alwayswell received by patients. It was found that patientsbecame anxious about having received such contact. Herimpression was that patients felt fully informed of thesituation during their initial consultation, and felt sup-

ported by receiving the letter from the CEO to quantifymaximal waiting times and avenues to seek redress ifproblems were perceived.

Radiotherapy Staff Reaction

The heads of radiotherapy at two of the treatmentlocations were interviewed, as were the staff membersresponsible for managing the waiting list. Throughoutthe study period, the main treating facility was staffed at90% capacity. The main perceived advantage of thepoints system was the removal of arbitrariness frompatient waiting times. Although previously these staffmembers had been pressured to move patients forwardin the queue for a variety of reasons, the firm structureallowed them more security in the organisation of thewaiting list. This led to fewer enquiries from anxiouspatients. It was felt that this produced a less stressfulworking environment for these staff members.

Other Variables

Radiation therapy staffing numbers increased slightlyacross the entire study period, but not enough toincrease patient throughput or to open a machine closeddue to staff shortages. There were no significantfluctuations in extraordinary workload, and machineup-times were steady. In November, the patientthroughput at location D had dropped by 20% becauseof reduced radiation therapist working hours introducedin October. The lower number of cases starting in the

WT (days) 6 7-20 21-34 35-48 49-62 63-76 77-90 91-104 105

January 27% 8% 25% 14% 12% 6% 6% 2% 0%

November 3% 9% 27% 19% 17% 11% 2% 3% 8%

September 8% 14% 27% 27% 11% 2% 6% 3% 2%

0 14 28 42 56 70 84 98 112

0

10

20

30

Waiting time (days)

Per

cen

t

January

SeptemberNovember

Fig. 2 – Waiting times divided into 2-week blocks showing thepercentage of each cohort waiting for each time block. ByJanuary, the new system increased the number of patientsstarting in a timely fashion and had decreased the number oflong waits. A smaller proportion of patients were waiting anintermediate length of time.

WT (days) ≤5 ≤25 ≤45 ≤65 ≤85 ≤105 ≤125

Prebooked 30% 55% 91% 98% 100% 100% 100%Not prebooked 2% 14% 55% 81% 89% 95% 100%

0 10 20 30 40 50 60 70 80 90 100 110 120

0102030405060708090100

Waiting time (days)

tnecrepevitalu

muC Pre-bookedNot pre-booked

Fig. 3 – Cumulative frequency histogram comparing patientswho were pre-booked with patients who were not. Patientswho were pre-booked were able to commence treatment moreclosely to their planned date compared with patients who werenot. Mann–Whitney statistics: pre-booked vs not-pre-booked;P<0.0001.

CLINICAL PRIORITISATION FOR CURATIVE THERAPY 303

selected week in January was a result of a number ofcurative cases starting in the preceding 2 weeks after theChristmas/New Year break.

Discussion

Queuing theory dictates that once the prescription rateexceeds the service rate, even for a brief period, a waitinglist will form [15]. The pilot stage of the implementationof a priority points system for a radiotherapy waiting listseems to have been well received and has deliveredseveral intriguing results. This paper describes ourexperience using three ‘snap-shots’ of waiting times asthe system was introduced. The factors influencing wait-ing times during this period were complex and dynamic,and some had the potential to increase rather thandecrease waits. However, the overall effect seems to havebeen to reduce long waiting times, making access moreequitable.

The November snap-shot showed longer waitingtimes than the other two samples, and by January themean waiting time had decreased again, being margin-ally shorter than in September. The bulge in Novemberresulted from a decrease in working hours at onelocation, increasing particularly the number of peopleexperiencing extremely long waits. Despite this reduc-tion in hours continuing in January, the long waits werealmost completely abolished, and the median had re-turned to less than the September levels, suggestinga real benefit of the prioritisation system. November alsosaw fewer patients starting who had been pre-booked.Additionally, the maximum waiting times has been

reduced from 18–12 weeks through this process. Thepresence of the long waiting times acted as a stimulusto educate referring practitioners to send patientsthrough early, particularly if undergoing planned pre-radiotherapy treatment. The proportion of pre-bookedpatients has increased significantly as a result, allowingtheir start dates to be better planned by giving someweight to their length of time in the system as well astheir clinical status.

One of the key observations over the time period isthe more rapid treatment of breast cancer patients inparticular. It seems that both the frequency of pre-booking in this group, and the impact of the pointssystem are responsible for this. The net improvement inwaiting times for breast unit patients has occurred in thesetting of static waiting times for other units as a whole,and illustrated particularly by the gastrointestinalunit. What is reassuring is that, with occasionalexceptions, the average waiting times for the otherpatients have not increased significantly, resulting inmore equitable access for all patients. Figure 5 illustratesthe change in the distribution of patient starts: risein the percentage of patients starting within 2 weeks,fewer patients starting at around the month mark,a comparable group starting around 6 weeks and areduction in the long tail.

The decrease in both formal and informal complaintsreflects the benefits of a better informed patient popu-lation. Pre-emptive contact, however, was not wellreceived, and was ceased at the completion of the trialperiod. Patients seemed to prefer to initiate contact

WT (days) 5 25 45 65 85 105 125

Other Location 2% 16% 58% 88% 96% 96% 100%

Location D 7% 13% 47% 53% 60% 80% 100%

0 10 20 30 40 50 60 70 80 90 100 110 120

0

10

20

30

40

50

60

70

80

90

100

Waiting time (days)

Cum

ulat

ive

per

cent

Other location

Location D

Fig. 4 – Cumulative frequency histogram comparing locationD with the other three treating locations in November. Therewere significantly greater delays at this site than at the otherthree sites, because of reductions in staffing hours, that causedthe inflation in the November figures overall. Mann–WhitneyStatistics; P=0.0895.

WT (days) 5 25 45 65 85 105 125

January 27% 45% 82% 82% 95% 100% 100%

November 0% 4% 33% 67% 79% 79% 100%

September 5% 30% 65% 85% 90% 95% 100%

0 10 20 30 40 50 60 70 80 90 100 110 120

0

10

20

30

40

50

60

70

80

90

100

Waiting time (days)

Cum

ulat

ive

per

cent

January

SeptemberNovember

Fig. 5 – Cumulative frequency histograms of waiting times forthe breast unit patients. The November inflation is againillustrated but, despite the continued pressure, the maximumwaiting times were shortened by January compared withSeptember, with 100% commencing treatment by 90 daysrather than 120 days.

304 CLINICAL ONCOLOGY

themselves via the avenues set out for them in the letterthey received from the CEO.

Another confounding factor in interpreting the effectsof the policy was the seasonal variation in patient flow.The pre-Christmas period was busy, but a number ofcurative courses were planned to start at the beginningof January, reducing the number of starters in theJanuary snap-shot week. We were unable to control forthis effect in our analysis, but such variations occurfrequently in normal practice, and hence the selection ofevenly spaced weeks was considered valid despite theChristmas effect. Ideally, another time of year wouldhave been selected to avoid this problem, but theurgency of the situation denied us this opportunity.Neither did we control for the effect of palliativecourses – a set proportion of starting slots was reservedfor palliative cases, with single fractions being fitted inad hoc. There seemed to be no significant effects onpalliative waiting times, although these frequently felloutside best practice guidelines.

There are a number of other methods for waiting listreduction, some of which were used concurrently in oursetting [16]: (1) use of a less refined points scale [17].Previously this institution used a six-point scale rangingfrom ‘emergency’, through to ‘standard >15days’. It wasfound that the manner in which patients were assignedto a particular category was arbitrary. As a result, oncea waiting list formed, ‘points drift’ began to be observed,where clinicians would upgrade the category of eachpatient. This had the ultimate effect of compounding thewaiting list problem with large numbers of ‘semi-urgent’cases. The application of strict criteria in the currentstudy has prevented this; (2) limitation of radiotherapyfor palliative cases [18]. This was achieved in our settingin two manners: insisting on optimal medical manage-ment before the use of radiotherapy (e.g. maximalanalgesia for bony pain) and hypofractionation, a con-certed campaign through e-mail, posters and meetingpresentations for the use of single 8 Gy fractions forbony pain. Fractionation had to be argued for on acase by case basis, for example soft tissue disease,re-treatment and long life expectancy (e.g. breast andprostate carcinoma); (3) referral to other centres withless of a problem with waiting lists, thus geographicallymedianising the problem. As Peter MacCallum CancerCentre has a number of treating locations, this is anoption that can be pursued without necessarily sendingpatients elsewhere, but collaboration with all centreswithin Melbourne was encouraged; (4) introduction of awaiting list further upstream, such as in the time to theinitial outpatient appointment. A potential advantage ofthis approach is that there are lower patient expectationsfor immediate treatment as they are not yet ‘in thesystem’. However, this may inhibit appropriate triage,and does also not facilitate the optimal pre-booking ofpatients. Encouraging early referral for pre-bookingis also advantageous, as the radiation oncologist isinvolved earlier in the decision-making process forthe patient, enhancing a multi-disciplinary approach.

There was an initial fear that the relationship withreferring surgeons and medical oncologists may be cor-rupted by this points system [8]. What seems to havehappened is an education process where referrersnow send patients for an opinion at a much earlier stageof their treatment. Often, this occurs preoperatively,which allows optimal gross tumour volume recognitionas well as permitting planning of the waiting liststructure.

The overall reduction in median waiting times islikely to be due to a combination of these factors. Someimprovement may be due to the points system allowingthe combination of multiple queues into one moreefficient queue. The exact root cause is difficult topinpoint owing to the crisis nature of the situation,and hence the need for multiple, simultaneous andoccasionally poorly quantifiable interventions.

The points system has been well received by bothpatients and staff. For patients, there is the reassurancethat they are not ‘lost in the system’. There is also somequantification of patients with conditions with greaterneed receiving treatment in a timely manner. Uniformityof practice across all treatment sites has ensued. Stafffound that they now spend less time fielding calls fromdistressed patients and relatives, allowing stress reduc-tion and more time for clinical responsibilities. Themajor achievement of the system has been to reduce thetail of very long waiting times for the ‘routine’ breastand prostate cancer patients.

One limitation of the system is the degree of arbitrari-ness in deciding the points categories. Although there isusually some internal consistency within a unit, this israrely seen between units. Thus, a three-point urologypatient (prostatic carcinoma, intact prostate) is unlikelyto be truly equivalent on a priority basis to a three-pointhead and neck patient (melanoma). Some attempt torectify this by introducing a category six for head andneck patients (postoperative squamous cell carcinoma)has resulted in an influx of patients also enrolled onclinical trials for immediate treatment from this unit.Often, this exceeds the planning capacity of the unit,ironically leading to delays. This led to the abolition ofthe six-point category at the end of the pilot period. Analternative system might score patients across all unitsaccording to the chance of cure. However, this wouldalmost totally preclude the curative treatment of con-ditions with a relatively poor prognosis, such as stagethree non-small cell lung cancers. The availability ofnecessary evidence to make this process rigorous is alsolimited.

The influence of pre-booking would seem to favourunits in which adjuvant treatment can be deliveredbefore the radiotherapy [19]. Examples include hormonedeprivation for prostate cancer and chemotherapy forbreast cancer. Thus a patient can have their initialassessment and prescription, and then receive activetreatment while serving out their artificial waiting time.This puts units in which adjuvant treatment is unavail-able at a disadvantage and hence the possibility of

CLINICAL PRIORITISATION FOR CURATIVE THERAPY 305

pre-booking is slim. Although we did not do so in thepilot, it might be fairer to share the burden of waiting byholding back the accumulation of points until the readyfor care date.

Acknowledgements. The authors would like to thank Loris Jonesand Greg Leslie for their feedback on changes to waiting list manage-ment; Joanne Moss, the Patient Advocate for her information regard-ing patient complaints; Carol Hart for her input about pre-emptivelycontacting patients; Claire Fitzpatrick and Judy Andrews for theircomments on the implementation of the points system in theirtreatment centres.

References1 Mackillop WJ, Fu H, Quirt CF, et al. Re: ‘Waiting for radio-

therapy in Ontario’. Int J Radiat Oncol Biol Phys 1995;32:895–896.2 James ND, Robertson G, Squire CJ, et al. A National Audit of

Radiotherapy in Head and Neck Cancer. Clin Oncol 2003;15:41–46.

3 Kenny L. Special report: radiotherapy summit 2000. Cancer Forum2000;26:173.

4 Penman A, Barton M. Access to radiotherapy: the gap betweenpolicy and practice. Cancer Forum 2002;26:173–176.

5 Turner S, Suchowerska N, Kenny L. The Australian radiationoncology workforce: forethought or afterthought? Current status,problems and solutions. Cancer Forum 2002;26:183–185.

6 Thomas SJ, Williams MV, Burnet NG, et al. How much surpluscapacity is required to maintain low waiting times? Clin Oncol2000;13:24–28.

7 Munro AJ. The costs of hyperfractionation: an economic analysisincorporating consideration of waiting time for treatment. ClinOncol 1995;7:162–167.

8 Mackillop WJ, Bates JHT, O’Sullivan B, et al. The effect of delayin treatment on local control by radiotherapy. Int J Radiat OncolBiol Phys 1996;34:243–250.

9 Ang KK, Trotti A, Brown BW, et al. Randomized trial addressingrisk features and time factors of surgery plus radiotherapy inadvanced head-and-neck cancer. Int J Radiat Oncol Biol Phys 2001;51:571–578.

10 Froud PJ, Mates D, Jackson JSH, et al. Effect of time intervalbetween breast-conserving surgery and radiation therapy on ipsi-lateral breast recurrence. Int J Radiat Oncol Biol Phys 2000;46:363–672.

11 Huang J, Barbera L, Brouwers M, et al. Does delay in startingtreatment affect the outcomes of radiotherapy? A systematicreview. J Clin Oncol 2003;21:555–556.

12 Thomas PRM, Tefft M, Compaan PJ, et al. Results of tworadiotherapy randomizations in the Third National Wilms’ TumorStudy (NWTS-3). Cancer 1991;68:1703–1707.

13 Slotman BJ, Meyer OW, Njo KH, et al. Importance of timing ofradiotherapy in breast conserving treatment for early stage breastcancer. Radiother Oncol 1994;30:206–212.

14 Murray N, Coy P, Pater JL, et al. Importance of timing forthoracic irradiation in the combined modality treatment of limited-stage small-cell lung cancer. The National Cancer Institute ofCanada Clinical Trials Group. J Clin Oncol 1993;11:336–344.

15 Cormack DV, Fisher PM, Till JE. A study of waiting times andwaiting lists for radiation therapy patients. Can J Oncol 1996;6:427–434.

16 Denham JW, Hamilton CS, Joseph DJ. How should a waiting listfor treatment be managed? Aust Radiol 1992;36:274–275.

17 Lamb D, Spry N. Re: Waiting lists for radiation treatment. AustRadiol 1992;36:347.

18 Franklin CIV, Poulsen M. How do waiting times affect radiationdose fractionation schedules? Aust Radiol 2000;44:428–432.

19 Stevens G. Waiting lists for radiation therapy. Aust Radiol 1993;37:228–229.

306 CLINICAL ONCOLOGY