A Population-based Study of the Fractionation of Palliative Radiotherapy for Bone Metastasis in Ontario

Int. J. Radiation Oncology Biol. Phys., Vol. 69, No. 4, pp. 1209–1217, 2007Copyright � 2007 Elsevier Inc.

Printed in the USA. All rights reserved0360-3016/07/$–see front matter

doi:10.1016/j.ijrobp.2007.04.048

CLINICAL INVESTIGATION Palliation

A POPULATION-BASED STUDY OF THE FRACTIONATION OF PALLIATIVERADIOTHERAPY FOR BONE METASTASIS IN ONTARIO

WEIDONG KONG, M.SC., JINA ZHANG-SALOMONS, M.SC., TIMOTHY P. HANNA, M.D.,AND WILLIAM J. MACKILLOP, M.B.CH.B., F.R.C.R., F.R.C.P.C.

Division of Cancer Care and Epidemiology, Queen’s University Cancer Research Institute, Kingston, Ontario, Canada

Purpose: To describe the use of palliative radiotherapy (PRT) for bone metastases in Ontario between 1984 and2001 and identify factors associated with the choice of fractionation.Methods and Materials: Electronic RT records from the nine provincial RT centers in Ontario were linked to theOntario Cancer Registry to identify all courses of PRT for bone metastases.Results: Between 1984 and 2001, 44,884 patients received 74,432 courses of PRT for bone metastases in Ontario.The mean number of courses per patient was 1.7, and 65% of patients received only a single course of PRT for bonemetastasis. The mean number of fractions per course was 3.9. The proportion of patients treated with a single frac-tion increased from 27.2% in 1984–1986 to 40.3% in 1987–1992 and decreased thereafter. Single fractions wereused more frequently in patients with a shorter life expectancy, in older patients, and in patients who lived furtherfrom an RT center. Single fractions were used more frequently when the prevailing waiting time for RT was longer.There were wide variations in the use of single fractions among the different RT centers (intercenter range, 11.8–62.3%). Intercenter variations persisted throughout the study period and were not explained by differences in casemix.Conclusions: Despite increasing evidence of the effectiveness of single-fraction PRT for bone metastases, most pa-tients continued to receive fractionated PRT throughout the two decades of this study. Single fractions were usedmore frequently when waiting times were longer. There was persistent, unexplained variation in the fractionationof PRT among different centers. � 2007 Elsevier Inc.

Palliative radiotherapy, Bone metastasis, Practice variations, Fractionation, Population-based study.

INTRODUCTION

Despite earlier diagnosis and more effective curative treat-

ment, more than 40% of patients diagnosed with cancer in

North America today will go on to die of their disease (1).

Quality of life is the highest priority for most patients with in-

curable cancer, but this is often compromised by severe pain

as patients approach the end of their lives (2). Bone metasta-

sis is the commonest cause of severe pain in advanced cancer,

and up to 75% of patients with breast, lung, and prostate can-

cer have been found to have bone metastases at postmortem

(3). Palliative radiotherapy (PRT) has an important role in the

management of bone metastases. Approximately 60% of pa-

tients experience meaningful improvement in their pain, and

approximately 30% obtain complete pain relief (4).

120

Several randomized clinical trials (5–10) and two subse-

quent systematic reviews (4, 11) have shown no difference

in overall pain response, or complete pain response, between

a single fraction of 8–10 Gy and a short, fractionated course

of treatment (usually 20 Gy in 5 fractions or 30 Gy in 10 frac-

tions). Moreover, no significant difference has been demon-

strated in time to response or duration of response in those

studies that have evaluated these outcomes (4). There is little

evidence of any difference in toxicity between a single frac-

tion and a short, fractionated course of RT (4). On the other

hand, a few studies have shown a small but significant in-

crease in the risk of a subsequent pathologic fracture among

those treated with a single fraction (4, 8–10), and several

studies have found a higher rate of retreatment among pa-

tients treated with a single fraction (4, 6, 8–10). The higher

Reprint requests to: William J. Mackillop, M.B.Ch.B., Divisionof Cancer Care and Epidemiology, Queen’s University CancerResearch Institute, 10 Stuart Street, Level 2, Kingston, ON K7L3N6, Canada. Tel: (613) 533-6000, ext. 78509; Fax: (613) 533-6794; E-mail: [email protected]

Supported by grants from the Canadian Institutes of HealthResearch, Cancer Care Ontario, and the Ontario Cancer ResearchNetwork (W.J.M.).

Acknowledgments—The authors thank the professional staff andmanagement of all the Ontario radiotherapy departments forproviding access to their treatment records; Dr. Eric Holowaty forpermitting use of the Ontario Cancer Registry for the purposesof this study; and Mrs. Beverley Shortt for her skill and patiencein preparation of the manuscript.

Conflict of interest: none.Received Jan 10, 2007, and in revised form April 24, 2007.

Accepted for publication April 30, 2007.

9

1210 I. J. Radiation Oncology d Biology d Physics Volume 69, Number 4, 2007

retreatment rate may reflect a greater need for retreatment, but

because none of the trials of the fractionation of PRT has

been double-blinded, the higher retreatment rate may merely

reflect radiation oncologists’ (ROs) greater readiness to re-

treat after a single fraction (12). Although surveys of patients’

views generally have shown a preference for shorter courses

of treatment (13–15), a recent study has shown that most pa-

tients choose fractionated treatment when they are told about

the higher risk of fracture and the increased likelihood of

retreatment (16). The optimal fractionation of PRT for bone

metastases, therefore, remains controversial (16, 17).

A number of mail surveys done over the last two decades

have shown wide variations in ROs’ stated fractionation

practices (18–22). There is less information available about

the pattern of prescribing of PRT, based on the analysis of

treatment records. However, two prospective nationwide

studies of fractionation in Sweden (23) and the United King-

dom (24) have also shown considerable diversity of practice.

The primary objective of this study was to determine the ex-

tent to which single fractions for bone metastases had been

adopted in Ontario over the two decades in which most of

the relevant clinical trials were published, and to identify fac-

tors associated with the use of single fractions in the general

cancer population of Ontario. A previous study showed a de-

cline in the use of single fractions at one Ontario cancer cen-

ter between 1998 and 2002, but trends elsewhere in Canada

have not previously been reported (25).

In the 1980s and 1990s the increased incidence of cancer

caused by the aging of the population, combined with new

indications for RT, led to rapid increases in demand for

RT (26). In some publicly funded systems, the capacity of

RT programs did not increase rapidly enough to keep pace

with the increased demand. This resulted in long waiting lists

for RT in many parts of the world, including Ontario (27). As

a result, patients who required PRT have often had to wait

longer for treatment than their doctors believed to be appro-

priate (28). Delays in RT may be caused by shortages of

resources at any step in the RT process, but in Ontario

inadequate treatment machine time has usually been the

rate-limiting problem (26). Patients who need PRT have,

therefore, had to compete for limited treatment resources

with patients who need radical or adjuvant treatment. Radia-

tion oncologists’ choices of fractionation have the potential

to modify demand for treatment machine time, and there is

evidence that such decisions have a large effect on machine

workload in Ontario (29). A previous Australian study had

shown that shorter palliative courses were used when waiting

times were longer (30). We therefore hypothesized that ROs

might modify their prescribing practices in response to the

demand for machine time in their department. The second

objective of this study was to determine whether ROs adopt

more parsimonious fractionation schemes when prevailing

waiting times are longer.

We have previously shown that access to PRT in particular

(31) varies across socioeconomic strata in Ontario. Specifi-

cally, we have shown that younger patients and residents of

wealthier communities are more likely to receive PRT in

the last two years of life than older, poorer people (31).

The third objective of this study was to determine whether

ROs modify their fractionation in response to the socioeco-

nomic factors that have been shown to affect the utilization

of PRT.

METHODS AND MATERIALS

Sources of dataRT records. Between 1984 and 2001, all RT in Ontario was pro-

vided by eight Cancer Care Ontario regional cancer centers and the

Princess Margaret Hospital in Toronto. Since the early 1980s, each

of these provincial centers has kept a computerized record of all ra-

diation treatments in a standard format, which includes intent of

treatment, number of fractions administered, total dose, dates of first

and last radiation treatments, and body region irradiated. All courses

of PRT delivered for bone metastases between January 1, 1984 and

December 31, 2001 were extracted from clinical databases of the

RT centers and linked to individual patients, in the Ontario Cancer

Registry. Two centers did not have detailed treatment records for the

entire study period, but both had gathered sufficient information

about RT visits to allow us to describe all courses of RT in terms of

treatment intent, start date, volume irradiated, and number of frac-

tions. Before 1991, simultaneous palliative treatments directed to

two or more target volumes that were delivered in the same number

of fractions were sometimes recorded as a single computerized re-

cord. We therefore opted to include all simultaneous palliative treat-

ments as a single course to avoid artificial variations in workload due

to differences in the way that the data had been recorded. That is, if

two or more records of PRT had the same start date and the same

number of fractions, they were counted as a single course for the

purposes of this study. A random sample of 1,122 charts of patients

showed that the RT database was more than 95% complete and more

than 99% accurate with respect to the variables used in this analysis.

Clinical and demographic data. The Ontario Cancer Registry

(OCR) is a population-based registry that routinely collects informa-

tion about the demographic and clinical characteristics for all inci-

dent cases of cancer in Ontario. The design and operation of the

registry has been described in detail elsewhere (32). The OCR pro-

vided us with the following information about all cases treated with

RT: primary diagnosis, date of diagnosis, date of birth, vital status,

date of death, and place of residence at diagnosis. The completeness

of cancer registration at OCR, as measured by capture–recapture

methodology, is greater than 95% for all sites combined (33).

Socioeconomic status. Statistics Canada provided us with de-

scriptors of community socioeconomic status at the level of the cen-

sus enumeration area and census subdivision, which we linked to

each case in the OCR according to the person’s place of residence

at diagnosis, as described previously (34).

Study populationFigure 1 summarizes how we selected the 74,432 courses of PRT

for bone metastases from the total of 346,107 courses of external

beam RT administered between 1984 and 2001. It also illustrates

how we identified the 41,924 completed, first courses of PRT, which

were included in the analysis of factors associated with choice of

fractionation.

The electronic treatment records that were available to us de-

scribed the treatment administered rather than the treatment pre-

scribed. Approximately 5% of these records represent courses of

treatment that had not been completed. For example, we assume

that records of 300 cGy in 1 fraction, 600 cGy in 2 fractions, and

Fractionation of palliative RT in Ontario d W. KONG et al. 1211

Total Number of Courses of

External Beam RT (1984-2001)

346,107

Courses of Adjuvant/Radical RT163,069

Courses with Missing Informationabout Intent or Primary Site

569

Total Number of Courses of PRT182,469

Courses of PRTto Other Body Regions

99,806

Number of Courses of PRT to Spine(N=40,724) or Other Bones(N=41,939); Total 82,663

Total Number of Courses of PRT

for Bone Metastases

74,432

Courses of PRTfor Malignant SpinalCord Compression

8,231

Number of the 1st Courses of PRTfor Bone Metastases

44,223

Number of Completed 1st Courses

of PRT for Bone Metastases

41,924

Subsequent Courses30,209

Incomplete Courses2,299

WORKLOAD

ANALYSIS

ANALYSIS OF

FRACTIONATION

Fig. 1. Palliative radiotherapy (RT) for bone metastases in Ontario: the study population. The flow chart illustrates howcourses of palliative RT were identified and selected for analysis.

900 cGy in 3 fractions, and so on, represent uncompleted courses of

3,000 cGy in 10 fractions and that records of 400 cGy in 1 fraction,

800 cGy in 2 fractions, and so on, represent uncompleted courses of

2,000 cGy in 5 fractions. All such records were included in the anal-

ysis of workload but excluded from the analysis of fractionation

because of uncertainty about the actual prescription.

Definition of median prevailing waiting timeWe created the concept of ‘‘prevailing waiting time’’ to serve as

an indicator of the overall level of demand for RT, in relation to sup-

ply, at a given center, at a specific point in time. Recognizing that,

under conditions of restraint, many different types of cases compete

for finite RT resources, we chose to measure waiting time from first

visit to the RT center to the start of RT for any purpose (radical, ad-

juvant, or palliative). The median prevailing waiting time at a given

center on a given date was calculated from the distribution of wait-

ing times from the date of first visit to that RT center to the date of

start of RT for all cases seen for the first time at that center in the

preceding 60 days. A matrix was created that described the median

prevailing waiting time at each center on each day of the study.

We attributed to each patient the prevailing waiting time at the cen-

ter where she/he was treated, on the first day of his/her PRT for

bone metastasis.

For much of the study period, one provincial RT center had a pol-

icy of not accepting new referrals unless machine time was available

to allow RT to start within 2 weeks. This policy systematically lim-

ited waiting times from consultation to treatment by shifting the

delay to the preconsultation period. Prevailing waiting times, as

defined above, therefore, did not reflect the workload pressure in

that department. Cases treated at this center were, therefore, ex-

cluded from the component of the analysis in which we explored

the effects of prevailing waiting times on choice of fractionation.

1212 I. J. Radiation Oncology d Biology d Physics Volume 69, Number 4, 2007

Statistical analysisWe performed a multivariate analysis to examine associations be-

tween patient-related, disease-related, and health system–related

factors, and the decision to use a single fraction as opposed to a frac-

tionated course of treatment. Modified Poisson regression was used

rather than logistic regression because this method provides relative

risks, which are more readily interpreted than odds ratios (35, 36).

We performed a second multivariate analysis to examine

associations between patient-related, disease-related, and health

system–related factors, and the decision to use a shorter course of

fractionated treatment (2–5 fractions) as opposed to a longer course

of fractionated RT ($6 fractions). Single fractions were excluded

from this analysis.

The primary analyses included courses given over the entire

study period (1984–2001), but we also performed secondary anal-

yses restricted to more contemporary cases treated between 1999

and 2001.

RESULTS

WorkloadBetween 1984 and 2001, 44,884 patients received 74,432

courses of PRT in Ontario. The mean number of fractions per

course was 3.9, creating a total treatment machine workload

of 293,934 fractions. Table 1 shows that approximately three

quarters of all courses were administered to patients in the

50 to 80-year age group. Breast, lung, and prostate cancer ac-

counted for two thirds of the courses of PRT administered for

bone metastases. Figure 2a shows that 65% of patients

received only a single course of PRT for bone metastases.

Figure 2b shows that the majority of courses were adminis-

tered as 1, 5, or 10 fractions; more protracted courses of

PRT for bone metastases were used only very rarely.

The number of courses administered each year increased

slightly over the study period but did not keep pace with

increasing incidence of cancer and mortality from cancer.

Between the 1984–1986 and 1999–2001 periods, the mean

number of courses of PRT for bone metastases per incident

case decreased significantly, from 0.12 to 0.08 (p < 0.0001),

and the mean number of courses per cancer death decreased

significantly, from 0.17 to 0.13 (p < 0.0001).

Table 1. Workload of palliative radiotherapy for bonemetastases in Ontario (1984–2001)

Total no. (%)of courses

Total no. (%)of fractions

GenderMale 35,340 (47.5) 140,181 (47.7)Female 39,092 (52.5) 153,753 (52.3)

Age (y)<50 10,997 (14.8) 47,316 (16.1)50–59 14,836 (19.9) 61,137 (20.8)60–69 22,703 (30.5) 90,002 (30.6)70–79 19,859 (26.7) 74,731 (25.4)80–89 5,996 (8.1) 20,587 (7.0)

Primary sitesBreast 23,448 (31.5) 88,702 (30.2)Lung 14,668 (19.7) 52,704 (17.9)Prostate 12,329 (16.6) 45,600 (15.5)Other GU* 3,819 (5.1) 17,831 (6.1)Myeloma 6,005 (8.1) 21,160 (7.2)L & Ly 3,367 (4.5) 18,807 (6.4)Colorectum 2,849 (3.8) 12,986 (4.4)Other GIz 1,460 (2.0) 5,631 (1.9)Gynecologic 1,323 (1.8) 6,652 (2.3)Head and neck 1,081 (1.5) 4,201 (1.4)Other sites 4,083 (5.5) 1,9660 (6.7)

Year of treatment1984–1986 11,590 (15.6) 54,174 (18.4)1987–1989 12,231 (16.4) 44,892 (15.3)1990–1992 12,052 (16.2) 43,504 (14.8)1993–1995 12,703 (17.1) 49,467 (16.8)1996–1998 13,394 (18.0) 54,249 (18.5)1999–2001 12,455 (16.7) 47,648 (16.2)

Total 74,432 293,934

* Other genitourinary cancers, ICD-9 codes 186-189.y Lymphoma and leukemia, ICD-9 codes 200-202 and 204-208.z Other gastrointestinal cancers, ICD-9 codes 150-152 and 155-159.

1 2 3 4 5 6No. of Courses Per Case

1030

5070

Perc

enta

ge o

f Tot

al T

reat

ed C

ases

(%)

5 10 151 20No. of Fractions Per Course

020

4060

Perc

enta

ge o

f Tot

al T

reat

ed C

ours

es (%

)

(a) (b)

Fig. 2. Palliative radiotherapy (RT) for bone metastasis in Ontario: frequency distributions of courses per case and frac-tions per course. (a) Frequency distribution of the number of courses of palliative RT for bone metastases administered percase, among patients who received at least one course of treatment in Ontario between 1984 and 2001. (b) Frequency dis-tribution of the number of fractions per course of palliative RT for bone metastases between 1984 and 2001.

Fractionation of palliative RT in Ontario d W. KONG et al. 1213

Factors associated with choice of fractionationWe examined associations between patient-related factors,

disease-related factors, and environmental factors, and the

choice of fractionation in the first course of PRT administered

for bone metastases in each of the 41,924 patients. Tables 2

and 3 show the results of both the univariate and multivariate

analyses of factors associated with the choice of fractionation

for bone metastasis. The first column of these tables shows the

proportion of patients in each subgroup of cases who received

a single fraction. The second column shows the mean number

of fractions per course prescribed for each subgroup of cases.

The third column shows the results of a modified Poisson re-

gression analysis performed to identify factors independently

associated with the choice of a single fraction as opposed to

a fractionated course of PRT. The fourth column shows the

results of a second Poisson regression performed to identify

factors independently associated with choice of a shorter

course of fractionated PRT (2–5 fractions), as opposed to

a longer course of $6 fractions. The results shown in Tables

2 and 3 were derived from the analyses of the entire study sam-

ple. A secondary analysis of patients treated between 1999

and 2001 gave very similar results (data not presented).

Table 2. Patient-related and disease-related factors associated with choice of fractionation

Univariate analyses Multivariate analyses

Factor (n)

Proportion (%)of courses

given as a singlefraction

Mean no. offractions

per course

Single fractionvs. fractionated

treatment,RR (95% CI)y

2–5 fractionsvs. $6 fractions,*

RR (95% CI)y

All combined (44,223) 34.3 4.4Gender

Male (22,611) 34.7 4.3 1 1Female (21,612) 33.9 4.4 1.06 (1.02–1.11) 1.01 (0.99–1.03)

Age (y)<50 (6,028) 28.8 5.0 0.91 (0.85–0.97) 0.94 (0.92–0.97)50–59 (8,238) 31.3 4.6 0.96 (0.91–1.01) 0.95 (0.93–0.97)60–69 (13,429) 33.1 4.4 1 170–79 (12,400) 36.8 4.1 1.09 (1.05–1.14) 1.01 (0.99–1.03)80– (4,128) 44.9 3.7 1.31 (1.25–1.38) 1.02 (0.99–1.05)

SES quintileLowest (8,530) 36.3 4.2 1 12nd (7,645) 35.1 4.3 0.99 (0.95–1.04) 0.99 (0.97–1.02)3rd (7,445) 34.3 4.4 1.00 (0.95–1.05) 1.00 (0.98–1.02)4th (6,869) 34.5 4.3 0.97 (0.93–1.03) 0.99 (0.97–1.02)Highest (7,228) 33.3 4.4 0.97 (0.91–1.02) 1.00 (0.97–1.02)

Time to death (mo)#1 (5,565) 48.3 3.1 2.12 (2.00–2.25) 1.19 (1.16–1.23)1–2 (5,457) 38.3 3.9 1.53 (1.44–1.63) 1.14 (1.10–1.17)2–3 (4,177) 38.2 4.1 1.52 (1.42–1.62) 1.12 (1.09–1.16)3–6 (7,622) 34.4 4.3 1.29 (1.21–1.37) 1.07 (1.04–1.10)6–12 (6,976) 30.6 4.7 1.09 (1.02–1.16) 1.03 (1.00–1.06)>12 (10,159) 27.9 5.0 1 1Alive or unknown (4,267) 28.2 5.0 0.52 (0.49–0.56) 0.98 (0.95–1.01)

Disease groupBreast (10,889) 32.5 4.3 1.00 (0.95–1.06) 1.05 (1.03–1.08)Lung (10,873) 39.5 3.8 1 1Prostate (6,757) 34.7 4.1 1.18 (1.12–1.24) 1.04 (1.02–1.07)Other GUz (2,497) 27.9 5.1 0.81 (0.75–0.87) 0.89 (0.86–0.93)Myeloma (3,009) 38.9 4.0 1.36 (1.27–1.46) 0.99 (0.96–1.03)L & Lx (2,465) 27.4 5.9 0.78 (0.71–0.87) 0.66 (0.61–0.70)Colon & rectum (2,204) 31.6 4.8 0.87 (0.80–0.94) 0.90 (0.86–0.93)Other GIk (1,136) 38.6 4.0 1.01 (0.92–1.11) 0.97 (0.92–1.02)Gynecologic (971) 30.6 5.4 0.81 (0.70–0.92) 0.80 (0.74–0.87)Head and neck (762) 36.4 4.2 1.06 (0.94–1.20) 0.96 (0.90–1.03)Other sites (2,660) 28.1 5.3 0.86 (0.80–0.93) 0.87 (0.83–0.91)

Body region treatedSpine (22,151) 25.4 4.7 1 1Other bones (22,072) 43.3 4.0 1.86 (1.79–1.92) 1.04 (1.02–1.07)

Abbreviations: RR = relative risk; CI = confidence interval; SES = socioeconomic status.* Choice of fractionation was studied in 20,335 courses of fractionated treatment. Singles were excluded.y Relative risks estimated based on modified Poisson regression models.z Other genitourinary cancers, ICD-9 codes 186-189.x Lymphoma and leukemia, ICD-9 codes 200-202 and 204-208.k Other gastrointestinal cancers, ICD-9 codes 150-152 and 155-159.

1214 I. J. Radiation Oncology d Biology d Physics Volume 69, Number 4, 2007

Table 3. Health system–related factors associated with choice of fractionation

Univariate analyses Multivariate analyses

Factor (n)

Proportion (%)of courses

given as a singlefraction

Mean no.of fractionsper course

Single fractionvs. fractionated

treatment,RR (95% CI)y

2–5 fractionsvs. $6 fractions,*

RR (95% CI)y

All combined (44,223) 34.3 4.4Cancer center

A (5,059) 32.7 4.6 1.00 (0.94–1.06) 0.96 (0.93–0.98)B (6,727) 31.7 4.3 0.94 (0.89–0.99) 1.02 (0.99–1.04)C (6,936) 34.5 4.1 1 1D (3,159) 62.3 3.2 1.82 (1.73–1.92) 0.97 (0.93–1.01)E (5,732) 19.4 5.1 0.61 (0.57–0.65) 0.95 (0.93–0.98)F (12,153) 41.3 4.0 – –G (1,021) 11.8 6.8 0.44 (0.37–0.52) 0.67 (0.62–0.72)H (1,750) 16.5 6.0 0.62 (0.56–0.70) 0.88 (0.84–0.92)I (1,686) 29.2 4.2 0.81 (0.74–0.88) 1.02 (0.98–1.05)

Distance to cancer center (km)#5 (5,862) 31.0 4.5 0.99 (0.94–1.04) 0.99 (0.96–1.01)5–25 (12,063) 32.9 4.4 1 125–75 (8,326) 37.1 4.1 1.10 (1.05–1.15) 0.98 (0.96–1.00)75– (7,440) 38.9 4.1 1.18 (1.13–1.24) 0.99 (0.97–1.02)

Median prevailingwaiting time (wk)z

<3 (3,026) 12.5 6.2 1 13–4 (3,316) 26.0 5.3 1.66 (1.42–1.93) 1.02 (0.96–1.09)4–5 (4,268) 35.5 4.4 1.99 (1.71–2.31) 1.12 (1.06–1.19)5–6 (6,996) 38.2 4.0 1.92 (1.65–2.24) 1.13 (1.06–1.20)6–7 (5,629) 34.9 4.1 1.88 (1.62–2.19) 1.12 (1.06–1.19)7–9 (6,043) 30.2 4.4 1.87 (1.60–2.19) 1.08 (1.02–1.15)$9 (2,792) 33.8 4.3 2.14 (1.81–2.52) 1.03 (0.97–1.11)

Year of treatment1984–1986 (6,750) 27.2 5.2 0.92 (0.81–1.04) 0.80 (0.75–0.85)1987–1992 (13,920) 40.3 4.1 1.26 (1.21–1.31) 0.98 (0.96–1.00)1993–1998 (15,694) 31.6 4.4 1 –1999–2001 (7,859) 35.4 4.1 1.18 (1.12–1.24) 1.07 (1.04–1.10)

Abbreviations as in Table 2.* Choice of fractionation was studied in 20,335 courses of fractionated treatment. Singles were excluded.y Relative risks were estimated based on modified Poisson regression models.z Patients from cancer center F were not included.

Patient-related factors. Table 2 shows that the patient’s

gender was not associated with any difference in fraction-

ation in the univariate analysis, but once other factors were

included in the model it seems that women were slightly

more likely to receive a single fraction than men. The propor-

tion of patients who received a single fraction increased

steadily with increasing age, from 28.8% in patients aged

<50 years to 44.9% in patients aged >80. The mean number

of fractions per course decreased steadily from 5.0 in those

younger than 50 to 3.7 in those older than 80. These trends

were found to be highly significant in the multivariate analy-

ses that controlled for the other factors listed in Table 2. Older

patients were significantly more likely to receive a single

fraction. Among those who received fractionated treatment,

older patients were more likely to receive a shorter course

of treatment. The socioeconomic status of the community

in which the patient resided was not associated with any sig-

nificant difference in the use of single fractions or in choice of

fractionation, among those who received fractionated PRT,

once the other factors listed in Table 2 were included in the

model.

Disease-related factors. Table 2 shows that patients with

a shorter life expectancy at the start of PRT were more likely

to receive a single fraction than those with a longer life expec-

tancy. Those with a shorter life expectancy were also more

likely to receive a shorter rather than a longer course of frac-

tionated PRT. These associations proved to be strong and

significant when other factors in Table 2 were included in

the model. The primary site of the cancer was also associated

with the choice of fractionation, although the magnitude of

these associations generally decreased once life expectancy

had been included in the model. For example, according to

the univariate analysis, patients with lung cancer were

more likely to receive a single fraction than those with breast

cancer, but this difference disappeared entirely once life ex-

pectancy had been included in the regression model. The

strength of the association with primary site was strongest

for the lymphomas, which were least likely to receive a single

Fractionation of palliative RT in Ontario d W. KONG et al. 1215

fraction and most likely to receive a longer course of treat-

ment if fractionated PRT was used. Single treatments were

used much less frequently when the spine was included in

the treatment volume.

Health system–related factors. We also investigated asso-

ciations with other factors unrelated to the individual

patients’ needs. Table 3 shows that the choice of fractionation

varied from one Ontario cancer center to another. The propor-

tion of patients who were treated with a single fraction ranged

from 11.8% to 62.3%, and the mean number of fractions per

course ranged from 3.2 to 6.8. The multivariate analyses con-

firmed that there were significant intercenter differences in

the proportion of cases who received a single fraction and

that there were also significant differences in the proportion

of patients who received shorter or longer courses of fraction-

ated PRT. Table 3 also shows that patients whose place of

residence was furthest from a RT center were most likely to

receive a single fraction. Among those who received fraction-

ated PRT, the distance from home to RT center was not asso-

ciated with the choice of a shorter or longer course of treatment.

We also tested the specific hypothesis that ROs would

adopt more parsimonious fractionation schemes at times

when there were long waiting times for RT in their depart-

ment. Table 3 shows that the prevailing waiting time for

RT, as defined in Methods and Materials, was indeed associ-

ated with choice of fractionation. The proportion of patients

treated with a single fraction was highest when waiting times

were longer. The mean number of fractions was highest when

waiting times were shortest. The regression models showed

that this was almost entirely due to an increase in the use of sin-

gle fractions when waiting times were longer. If fractionated

PRT was chosen, the prevailing waiting time had very little im-

pact on the choice of a longer or shorter fractionation scheme.

The prevailing waiting time, as defined in Methods and Mate-

rials, should not be equated with actual waiting time for PRT.

Temporal trends and geographic variations in fraction-ation. Table 3 shows that in the early part of the study period,

the overall proportion of Ontario patients treated with a single

fraction increased significantly, from 27.2% to 40.3%. This

trend subsequently reversed, and in the most recent period

(1999–2001), only 35.4% of patients received a single frac-

tion. Among patients treated with fractionated PRT, there

was a steady increase in the use of shorter courses of treat-

ment over the study period. However, these province-wide

statistics oversimplify the real picture. Figure 3 shows that

temporal changes in fractionation varied among the different

cancer centers, although the downward trend was ubiquitous.

Figure 3 also shows that the degree of intercenter variation

in the mean number of fractions per course diminished over

time. Nonetheless, as shown in Fig. 4, large variations in frac-

tionation persisted in the most recent period for which

full information was available (1999–2001).

DISCUSSION

The main finding of this study was that, despite increasing

evidence to support the use of single fractions, the majority

of patients in Ontario continue to receive fractionated RT for

bone metastases. Over the whole study period only approxi-

mately one third of all courses were administered as a single

fraction. The use of single fractions peaked at 40% around

1990 and subsequently declined. The rate of use of single

fractions in the most recent period in our study (1999–

1001) was 35%, which is very similar to the rate of 36%

Treatment Year

Mea

n N

umbe

r of F

ract

ions

Per

Cou

rse

Individual Cancer Center

All Cancer Centers In Ontario

1984-86 1987-89 1990-92 1993-95 1996-98 1999-2001

03

69

1215

18

Fig. 3. Temporal trends in the fractionation of palliative radiotherapy for bone metastases in Ontario (1984–2001). Thegraph shows changes in the mean number of fractions per course at each of the radiotherapy centers in Ontario between1984 and 2001.

1216 I. J. Radiation Oncology d Biology d Physics Volume 69, Number 4, 2007

2550

75

(a) (b) (c)

2550

75

(d) (e) (f)

1 5 10 15

025

5075

(g)

1 5 10 15 20

(h)

No. of Fractions Per Course1 5 10 15 20

(i)

Perc

enta

ge (%

)

20

Fig. 4. Intercenter variations in fractionation of palliative radiotherapy (RT) for bone metastases in Ontario (1999–2001).The nine panels show frequency distributions of the number of fractions per course of palliative RT for bone metastasesat each of the RT centers in Ontario during the 1999–2001 period.

reported in the United Kingdom in 2003 (24) and the rate of

37% reported in Sweden in 2001 (23). This report adds to

a growing literature that shows that the choice of fractionation

of PRT for bone metastasis is influenced not only by patient-

and disease-related factors (25) but also by health system–

related factors (22).

Some of the observed variation in the use of single frac-

tions in Ontario seems to reflect variations in patients’ needs,

or at least in the doctor’s perception of the patient’s needs.

Patients with a very short life expectancy were most likely

to receive a single fraction, and single fractions were used

more frequently when the field did not include the spine.

The strong association between age and fractionation that

persisted after controlling for life expectancy is rather less

easy to explain. Although advanced age per se is not a contra-

indication to PRT (37–39), it has previously been shown that

the use of PRT declines more rapidly with age than would be

expected on the basis of the observed decline of functional

status with increasing age (40). Those previous findings sug-

gested that there might be age bias in selection of cases for

treatment. The present findings suggest that there may also

be age bias in the choice of fractionation.

We did not find any association between socioeconomic

status and the number of fractions administered. In a previous

study, we had found that patients from poorer communities

were less likely to receive PRT (31). This seemed surprising

in a publicly funded health system, but patients of higher

socioeconomic status may be more persistent and successful

in pursuing specialist referrals. The results of the present

study suggest that once patients have gained access to the

RT system their socioeconomic status does not affect the

RO’s choice of fractionation.

Some of the variation in the use of single fractions may be

understood as a relatively appropriate response to variations

in the accessibility of RT. Single fractions were used more

frequently in patients who lived further from the nearest RT

facility. Single fractions were also used more frequently when

prevailing times were longer, confirming the results of a previ-

ous Australian study (30). The fact that single fractions are used

more frequently when access is constrained suggests that some

ROs may consider this approach acceptable but suboptimal.

We found large variations in the use of single fractions

among different RT centers in Ontario. These variations per-

sisted after controlling for characteristics of the individual

case and after controlling for the workload pressure on the

individual department and are difficult to justify in the con-

text of a centrally managed, publicly funded RT system.

They presumably reflect differences in individual RO’s be-

liefs about the appropriateness of using single fractions

and/or the emergence of different ‘‘schools of thought’’ about

this issue at different centers. These divergent beliefs may be

caused by variations in the RO’s interpretation of the litera-

ture concerning the efficacy of single fractions, or they may

reflect differences in their willingness to modify their frac-

tionation in the interests of optimizing access to RT when re-

sources are limited.

In 2003, the Program in Evidence-Based Care of Cancer

Care Ontario published a guideline recommending a single

Fractionation of palliative RT in Ontario d W. KONG et al. 1217

8-Gy treatment as ‘‘the standard dose-fractionation schedule

for symptomatic and uncomplicated bone metastases’’ (41).

It remains to be seen whether the publication of this guideline

will have any impact on practice. We plan to repeat this anal-

ysis to address that question whenever we are able to obtain

access to contemporary data.

REFERENCES

1. Jemal A, Siegel R, Ward E, et al. Cancer statistics 2006. CACancer J Clin 2006;56:106–130.

2. Portenoy RK, Thaler HT, Kornblith AB, et al. Symptom prev-alence, characteristics and distress in a cancer population.Qual Life Res 1994;3:183–189.

3. Manoso MW, Healey JH. Treatment of metastatic cancer. In:DeVita VT, Hellman S, Rosenberg SA, editors. Cancer: Princi-ples & practice of oncology. 7th ed. Phildelphia: LippincottWilliams & Wilkins; 2004. p. 2368–2381.

4. Sze WM, Shelley MD, Held I, et al. Palliation of metastatic bonepain: Single fraction versus multifraction radiotherapy—A system-atic review of randomised trials. Clin Oncol 2003;15:345–352.

5. Gaze MN, Kelly CG, Kerr GR, et al. Pain relief and quality of lifefollowing radiotherapy for bone metastases: A randomised trial oftwo fractionation schedules. Radiother Oncol 1997;45:109–116.

6. Hartsell WF, Scott CB, Bruner DW, et al. Randomized trial ofshort- versus long-course radiotherapy for palliation of painfulbone metastases. J Natl Cancer Inst 2005;97:795–804.

7. van den Hout WB, van der Linden YM, Steenland E, et al.Single versus multiple-fraction radiotherapy in patients withpainful bone metastases: Cost-utility analysis based on a ran-domized trial. J Natl Cancer Inst 2003;95:222–229.

8. Price P, Hoskin PJ, Easton D, et al. Prospective randomisedtrial of single and multifraction radiotherapy schedules in thetreatment of painful bony metastases. Radiother Oncol 1986;6:247–255.

9. Cole DJ. A randomized trial of a single treatment versus con-ventional fractionation in the palliative radiotherapy of painfulbone metastases. Clin Oncol 1989;1:59–62.

10. Neilsen OS, Bentzen SM, Sandbert E, et al. Randomized trial ofsingle dose versus fractionated palliative radiotherapy of bonemetastases. Radiother Oncol 1998;47:233–240.

11. McQuay HJ, Collins SL, Carroll D, et al. Radiotherapy for thepalliation of painful bone metastases. Cochrane Database SystRev 2000;2:CD001793.

12. Shakespeare TP, Thiagaraian A, Gebski V. Evaluation of thequality of radiotherapy randomized trials for painful bonemetastases. Cancer 2005;103:1976–1981.

13. Slotman BJ, van der Wal G, Kregar S, et al. Patients’ apprecia-tion of single fraction radiotherapy for painful bone metastases.Palliat Med 2004;18:72–73.

14. Szumacher E, Llewellyn-Thomas H, Franssen E, et al. Treat-ment of bone metastases with palliative radiotherapy: Patients’treatment preferences. Int J Radiat Oncol Biol Phys 2005;61:1473–1481.

15. Chow E, Andersson L, Wong R, et al. Patients with advancedcancer: A survey of the understanding of their illness and expec-tations from palliative radiotherapy for symptomatic metasta-ses. Clin Oncol (R Coll Radiol) 2001;13:204–208.

16. Shakespeare TP, Lu JJ, Back M, et al. Patients’ preference forradiotherapy fractionation schedule in the palliation of painfulbone metastases. J Clin Oncol 2003;21:2156–2162.

17. Kachnic L, Berk L. Palliative single-fraction radiation therapy:How much more evidence is needed? J Natl Cancer Inst 2005;97:786–788.

18. Priestman T, Bullimore JA, Godden TP, et al. The RoyalCollege of Radiologists’ Fractionation Survey. Clin Oncol1989;1:39–46.

19. Coia LR, Hanks GE, Martz K, et al. Practice patterns of pallia-tive care for the United States 1984-1985. Int J Radiat OncolBiol Phys 1988;14:1261–1269.

20. Duncan G, Duncan W, Maher E. Patterns of palliative radiother-apy in Canada. Clin Oncol 1993;5:92–97.

21. Ben-Josef E, Shamsa F, Williams AO, et al. Radiotherapeuticmanagement of osseous metastases: A survey of current pat-terns of care. Int J Radiat Oncol Biol Phys 1998;40:915–921.

22. Lievens Y, Kesteloot K, Rijnders, et al. Differences in palliativeradiotherapy for bone metastases within Western Europeancountries. Radiother Oncol 2000;556:297–303.

23. Moller TR, Bronsson B, Ceberg J, et al. A prospective survey ofradiotherapy practice 2001 in Sweden. Acta Oncologica 2003;43:387–410.

24. Williams MV, James ND, Summers ET, et al. National surveyof radiotherapy fractionation practice in 2003. Clin Oncol 2006;18:3–14.

25. Haddad P, Wong RKS, Pond GR, et al. Factors influencing theuse of single vs multiple fractions of palliative radiotherapy forbone metastases: A 5-year review. Clin Oncol 2005;17:430–434.

26. Mackillop WJ, Zhou S, Groome PA, et al. Changes in the useof radiotherapy in Ontario 1984-1995. Int J Radiat Oncol BiolPhys 1999;44:355–362.

27. Mackillop WJ, Fu H, Quirt CF, et al. Waiting for radiotherapyin Ontario. Int J Radiat Oncol Biol Phys 1994;30:221–228.

28. Mackillop WJ, Zhou Y, Quirt CF. A comparison of delays in thetreatment of cancer with radiation in Canada and the UnitedStates. Int J Radiat Oncol Biol Phys 1995;32:531–539.

29. Dixon PF, Mackillop WJ. Could changes in clinical practicereduce waiting lists for radiotherapy? J Health Svcs Res Policy2001;6:70–77.

30. Franklin C IV, Poulsen M. How do waiting times affect radiationdose fractionation schedules? Radiat Oncol 2000;44:428–432.

31. Huang J, Zhou S, Groome P, et al. Factors affecting the useof palliative radiotherapy in Ontario. J Clin Oncol 2001;19:137–144.

32. Clarke EA, Marrett LD, Kreiger N. Cancer registration inOntario: A computer approach. IARC Sci Publ 1991;95:246–257.

33. Robles SC, Marrett LD, Clarke EA, et al. An application ofcapture-recapture methods to the estimation of completenessof cancer registration. J Clin Epidemiol 1988;41:495–501.

34. Mackillop WJ, Zhang-Salomons J, Groome PA, et al. Socioeco-nomic status and cancer survival in Ontario. J Clin Oncol 1997;15:1680–1689.

35. McNutt LA, Wu C, Xue X, et al. Estimating the relative riskin cohort studies and clinical trials of common outcomes. AmJ Epidemiol 2003;157:940–943.

36. Zou G. A modified Poisson regression approach to prospectivestudies with binary data. Am J Epidemiol 2004;159:702–706.

37. Souchon R. Palliative radiation oncologic therapy: Is patient’sage a determining factor of feasibility? A 1-year analysis(1997) at a radiotherapy clinic of an academic teaching hospital.Strahlenther Onkol 1999;175:218–224.

38. Mitsuhashi N, Hayakawa K, Yamakawa M, et al. Cancer inpatients aged 90 years or older: Radiation therapy. Radiology1999;211:829–833.

39. Pignon T, Scallier P. Radiotherapy in the elderly. Eur J SurgOncol 1998;24:407–411.

40. Tyldesley S, Zhang-Salomons J, Groome PA, et al. Associationbetween age and the utilization of radiotherapy in Ontario. Int JRadiat Oncol Biol Phys 2000;47:469–480.

41. Wu JS, Wong RK, Lloyd NS, et al. Radiotherapy fractionationfor the palliation of uncomplicated painful bone metastases—An evidence-based practice guideline. BMC Cancer 2004;4:71.