|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2001 American Society for Clinical Oncology Variation in Delivery of Palliative Radiotherapy to Persons Dying of Cancer in Nova Scotia, 1994 to 1998ByFrom the Nova Scotia Cancer Registry and Nova Scotia Cancer Centre of Queen Elizabeth II Health Sciences Centre, and School of Health Services Administration, Dalhousie University, Halifax, Nova Scotia, Canada. Address reprint requests to Grace M. Johnston, MHSA, PhD, School of Health Services Administration, 5599 Fenwick St, Dalhousie University, Halifax, Nova Scotia, Canada B3H 1R2; email: grace.johnston{at}dal.ca
PURPOSE: To examine sociodemographic and clinical variables associated with provision of palliative radiotherapy (RT) to persons dying of cancer. METHODS: The Nova Scotia Cancer Registry was used to identify 9,978 adults who were dying of cancer between 1994 and 1998 in the Canadian province of Nova Scotia. RT records from between April 1992 and December 1998 were obtained from the provincial treatment database. Multivariate analysis identified factors associated with two sequential decisions determining provision of palliative RT in the last 9 months of life: likelihood of receiving an RT consultation with a radiation oncologist and, given a consultation, likelihood of being treated with palliative RT. RESULTS: The likelihood of having a consultation decreased with age (20 to 59 years v 80+ years: odds ratio [OR], 4.43 [95% confidence interval, 3.80 to 5.15]), increased with community median household income (> $50,000 v < $20,000: OR, 1.31 [1.02 to 1.70]), was higher for residents closer to the cancer center (< 25 km v 200+ km: OR, 2.47 [2.16 to 2.83]), increased between 1994 and 1998 (OR, 1.34 [1.16 to 1.56]), varied by cause of death (relative to thoracic cancers, head and neck: OR, 1.75 [1.31 to 2.33]; gynecologic: OR, 0.35 [0.27 to 0.44]), and was greater for those who had prior RT (OR, 2.20 [1.89 to 2.56]). Similar associations were observed when outcome was the provision of palliative RT given a consult, with one notable exception: prior RT was associated with a lower likelihood of receiving palliative RT (OR, 0.48 [0.40 to 0.58]). CONCLUSION: Variations observed in delivery of palliative RT should prompt further investigation into equity of access to clinically appropriate, palliative radiation consultation and treatment.
PALLIATIVE MEDICINE has evolved into a distinct medical specialty in the past three decades, and palliative radiotherapy (RT) is an important branch of palliative medicine.1 Approximately half of all North American cancer patients receive RT during the course of their disease, and roughly half of these courses are devoted to palliation.2 Despite the significance of palliative RT, utilization and epidemiologic analyses have been limited to descriptions of the resources necessary for service delivery.3-6 No consensus has been reached on the best measurement of palliative RT outcomes and to whom and under what conditions palliative RT should be administered.3,7-9 It is accepted that the need for palliative RT will grow as the population ages and the count of newly diagnosed cancer cases increases.10 Accommodating this demand will require coordinated planning using quality data. Canadian health care is covered by a universal, single-payer insurance system administered by each of the 10 provinces.11 Despite the absence of explicit barriers to care, geographic, socioeconomic, and age-related variations in the use of curative or adjuvant RT have been shown to exist in Canada12,13 and in other countries with similar health care systems.4 Only one utilization study of palliative RT in Canada was identified in the literature. It focused on regional variations for a single disease, non–small-cell lung cancer.3 RT in the province of Nova Scotia is administered under a centralized system. The provincial cancer center, based in Halifax, provided more than 98% of the palliative RT to Nova Scotia residents during the study time period. Halifax is also the home of a comprehensive palliative care program (PCP). More than 90% of the patients registered in the PCP have cancer. In a previous study, major variations were found in referral rates to the PCP for those dying of cancer in Nova Scotia.14 Higher PCP registration rates were found among younger patients, patients who lived closer to the PCP, those who survived more than 6 months beyond diagnosis, and patients who received palliative RT. The purpose of this current study was to determine whether sociodemographic, diagnostic, and health system–related variations existed in the delivery of palliative RT to adults who died of cancer in Nova Scotia between 1994 and 1998. This analysis prompts discussion on equity in the delivery of palliative RT and supports additional studies that ascertain the optimal levels of efficacious, palliative consultation and treatment.
Study Design The study population consisted of all adults who died of cancer in Nova Scotia between 1994 and 1998. The outcome of interest was the provision of palliative RT in the final 9 months of life. This was investigated through examination of a pair of sequential treatment decisions: first, whether a patient received an RT consultation with a radiation oncologist in the final 9 months of life; and second, given a consultation, whether a patient went on to receive palliative RT. Factors investigated for associations with these treatment decisions included age at death, sex, community median household income (MHI), distance to the provincial cancer center, year of death, microscopic confirmation of the cancer diagnosis, type of cancer causing death, time from cancer diagnosis to death, and previous provision of palliative or nonpalliative RT.
Data Sources Statistics Canada. The median household income and longitude/latitude of each federal enumeration area (EA) in the province were obtained from the 1996 electronic national census data files provided by Statistics Canada.16 Statistics Canada also provided postal code conversion files and an accompanying SAS program (SAS Institute, Cary, NC)17-19 that were used to convert postal codes of the study subjects to the corresponding EAs. A one-to-one assignment of postal codes to EAs was not possible where a single postal code covered multiple EAs. In such cases, the SAS program randomly assigned each study subject to one of the EAs within the postal code. The random assignment was weighted by the population size of each EA.
Study Subjects A death registration or autopsy record was the only source of information for 599 persons who died during the study period. Since palliative RT was not a possible option for these people, they were removed from the analysis. Thus, the data analysis reported herein is based on the 9,978 adult cancer deaths between 1994 and 1998 for those individuals who were diagnosed with cancer before death.
Definitions of Variables A course of RT was defined as palliative if there was a palliative intent code provided by the treating radiation oncologist or if 10 or fewer fractions were administered. Ten fractions was selected as a logical cut point based on the distribution of fractions for courses with a palliative intent code, as shown in Fig 1. Of the RT courses with a palliative intent code that were administered between 1994 and 1998, 91.7% had 10 or fewer fractions. Only a small number of study subjects had courses with both a nonpalliative intent code and 10 or fewer fractions (3.2% of the entire sample, or 13.4% of those who received palliative RT by our definition).
Figure 2 is a cumulative percentile plot of the time between treatment with palliative RT and death, with the time of death indicated as zero on the horizontal axis. For patients who received multiple courses of palliative RT, only the final course was considered, ie, the course closest to death. More than one fourth (28.6%) of terminally ill cancer patients were administered palliative RT in the last 2 years of life, with most of these patients receiving it within 9 months of death (24.0%). Thus, palliative RT within 9 months of death was selected as the outcome measure of access to palliative RT and is referred to as the terminal period.
RT records were available from the provincial cancer center database from April 1992 onward, allowing at least 21 months of RT information on all study subjects. Previous RT was defined as RT that occurred between 9 and 21 months before death, ie, the year before the terminal period, regardless of the intent code or number of fractions administered. A sensitivity analysis using a 4-year window for previous RT was performed to examine the robustness of conclusions using the 1-year window. For the 4-year cutoff, the study population had to be limited to deaths between 1997 and 1998. As shown in Fig 3, where the beginning of the terminal period is indicated as zero on the horizontal axis, for those previously administered RT, approximately two thirds (65.7%) received it within a year of the terminal period.
The time from diagnosis with cancer to death was analyzed as a dichotomous variable: less than 6 months or 6+ months. This factor was associated with the provision of palliative care in a previous study.14 To specify the community of residence for each study subject, the postal code on the death certificate was used to derive an EA of residence for 87.1% of the subjects. Of the remaining subjects, 12.1% were linked to an EA from the postal code on the cancer incident record, 0.3% were linked to a county of residence from the death certificate, and 0.4% were linked to a county from the incident record. Among study subjects with a postal code recorded both at diagnosis and death, 75.6% had the same postal code. There was no postal address available for eight subjects. The Statistics Canada 1996 census MHI of the EA or county of residence was used to characterize the community in which each subject lived. This MHI does not relate directly to the socioeconomic attributes of an individual but rather to the area in which the individual resides.20 MHI reported herein should be viewed as relating to the resources and characteristics of the region of residence. Such ecologic measures are good proxies for the socioeconomic status of the individual, although they may underestimate associations.21,22 The distance from the study subject’s place of residence to the provincial cancer center was calculated using the latitude/longitude coordinates of the centers of the EA of residence and the EA containing the cancer center.23 If the EA of residence for a study subject was unavailable, then the latitude/longitude coordinates of the county of residence were used. These distances were measured "as the crow flies" and are not the actual travelling distance to the cancer center. A death was considered to be due to cancer if cancer was an underlying or contributing cause of death on the death registration. These ICD-9 cause-of-death codes were grouped into one of 10 mutually exclusive categories.24
Validity Check of Cause of Death To check on the validity of the cause-of-death codes, ICD-9 diagnosis site codes were similarly categorized. If a study subject had been diagnosed with cancer at multiple primary sites, only the initial site was considered. Table 1 shows that agreement was in excess of 90% for cancers of the gastrointestinal tract (90.2%), thoracic cancer (94.3%), and cancer of the CNS (92.4%). With the exception of head and neck cancer (62.5%) and diagnosis with an unknown primary tumor (49.4%), all other diagnoses groups had greater than 75% agreement. Agreement for diagnoses that were microscopically confirmed was similar to agreement for those that were not microscopically confirmed.
When there were disagreements between the cancer site at diagnosis and the cancer cause of death, anatomically adjacent and metastatic sites were typically involved. For example, 18.8% of the head and neck cancers at diagnosis were coded as thoracic cancer at death. Among the cancers identified as gynecologic at diagnosis, 7.1% were coded with a gastrointestinal and 4.0% with a thoracic cause-of-death code. Among cancers identified as genitourinary at diagnosis, 7.7% were gastrointestinal and 7.9% were thoracic at death. Among the breast cancers identified at diagnosis, 5.1% were coded as gastrointestinal and 4.1% as thoracic at death. Among the unknown primary tumors at diagnosis, 12.5% were coded as gastrointestinal and 12.0% as thoracic at death. This implies validity in the cancer cause of death being related to the cancer site at diagnosis, although the cancer cause of death may not always be the actual primary cancer site where the disease originated.
Analysis
Univariate associations between each of the treatment decisions and the above-mentioned factors were explored through two-dimensional tables and the
Description of the Study Population Table 2 describes the study population by a number of factors. Not surprisingly, it was an elderly group, with a median age of 72 years; 53.5% were male. Most resided in communities where the MHI was between $20,000 and $35,000 (54.1%). Almost half of all subjects lived within 100 km of the provincial cancer center (49.3%), with more than one fourth living within 25 km (27.8%). (The maximum distance in the province to the cancer center was 353 km.) Annual cancer mortality stayed roughly constant at about 2,000 deaths per year, ranging from a low of 1,939 in 1995 to a high of 2,054 in 1996. The most common cancer causes of death were thoracic (mostly lung; 27.8%), gastrointestinal (26.7%), genitourinary (mostly prostate; 11.2%), and breast (8.5%). Approximately one third (36.4%) died within 6 months of their diagnosis. Approximately one tenth (10.6%) had received RT in the year before the terminal period.
Classifying RT During the Terminal Period Table 3 displays a classification of RT administered during the final 9 months of life. Of the 9,978 study subjects, 63.3% did not visit a radiation oncologist and were not treated with RT during the terminal period. The remainder (36.7%) had an RT consultation, and roughly one fourth (24.0%) of the study subjects went on to be treated with palliative RT.
The majority of these subjects had treatment fulfilling both criteria of our definition of palliative RT: a palliative intent code and/or 10 or fewer fractions (19.4%). Smaller proportions received 10 or fewer fractions with a nonpalliative intent code (3.2%) or more than 10 fractions with a palliative intent code (1.4%). During the terminal period, less than 3% of the study sample received RT that satisfied neither of the palliative RT criteria. The outcome of interest was the receipt of palliative RT in the terminal period before death from cancer. We considered intervals of 6 months, 9 months, and 1 year before death for the terminal period. Approximately 24.0% of those dying from cancer received palliative RT in the final 9 months of life, but only an additional 1.9% of study subjects were administered palliative RT in the interval between 9 and 12 months before death (Fig 2). Six months has been used in other studies28,29; using this time period, 21.2% of subjects received palliative RT. The number of fractions administered was associated with the length of time until death. Among the study subjects receiving palliative RT within 6 weeks of death, 47.6% had four or fewer fractions, 40.4% had five fractions, and only 12.1% had six to 10 fractions. In comparison, among those who had their last palliative RT more than 102 days before death, only 28.3% had four or fewer fractions, 45.0% had five fractions, and 26.7% had six to 10 fractions.
Sequential Decisions Associated With Provision of Palliative RT in the Terminal Period There were major site-related variations observed for both the RT consultation and treatment decisions. Consultations were given more frequently to those dying of cancer of the head and neck (66.7%), thorax (55.0%), CNS (53.2%), and breast (48.6%) than for other cancers, particularly hematologic (6.5%) and gastrointestinal (17.2%) cancers. Variations were also observed in the rates of those who went on to be administered palliative RT. After a consultation, study subjects were more likely to receive palliative RT if they were dying from genitourinary cancer (65.6%), unknown primary tumor (64.3%), or other cancers (68.2%). They were less likely to receive palliative RT for cancers of the CNS (29.4%) and gynecologic system (39.4%). Microscopic confirmation of the tumor at diagnosis was strongly associated with receiving an RT consultation and moderately associated with going on to be treated with palliative RT. There was a small but significant association between time from diagnosis to death and the receipt of an RT consultation, with those living at least 6 months beyond diagnosis receiving a consultation 30% more frequently than those who died within 6 months. There was no association between time from diagnosis to death and the proportion of subjects who went on to be administered palliative RT. RT in the year before the terminal period was strongly associated with receiving an RT consultation in the terminal period. However, those who had previously been treated with RT were less likely to be administered palliative RT after their consultation. The proportion of study subjects treated with palliative RT during the final 9 months of life increased annually from 20.2% in 1994 to 26.8% in 1998. This trend coincided with a general increase in the use of RT in Nova Scotia: the number of courses of RT administered by the provincial cancer center increased by approximately 9.7% per year. No substantial change was observed in the time between consultation and treatment, which indicates that the cancer center was able to cope with the increasing demand. In 1994, 36% of persons received palliative radiation within 14 days of their consultation; in 1998, 33% of persons received treatment in the same time period. Further evidence that the center was able to meet the demand was that out-of-province RT, primarily at the New Brunswick border, remained stable over time at approximately 30 curative and palliative cases per year. Multivariate analysis. The RT consultation and treatment decisions were further examined using logistic regressions, with all covariates entered into the models simultaneously. The odds ratios from these models are listed in Table 4.
The conclusions drawn from the multivariate analysis were generally the same as those drawn from the univariate analysis. There were a few exceptions: after controlling for all other factors, time from diagnosis to death had no association with the likelihood of receiving an RT consultation in the final 9 months of life. Also, the association with community MHI was altered. The odds ratios of MHIs above $20,000 were all similar but significantly different from MHIs below $20,000. Hence, although the income-associated gradient was explained by other variables, the likelihood of not receiving an RT consultation as well as treatment seemed to be associated with poverty. In a sensitivity analysis, three definitions of palliative RT were used: palliative intent code alone, both palliative intent code and 10 or fewer fractions, and both palliative intent code and five or fewer fractions. There were no significant differences in the observed associations. Prior RT information was not available for the 21.9% of study subjects diagnosed before April 1, 1992. A sensitivity analysis was carried out by steadily lengthening the time interval for previous access to the RT system. This required that the years of death defining the study sample be simultaneously decreased. Table 5 describes the proportion of subjects who received palliative RT in each of the periods, as well as odds ratios from separate logistic regressions modeling the likelihood of receiving an RT consultation and the likelihood of going on to be administered palliative RT. The associations between the provision of RT before the terminal period and each of the sequential treatment decisions during the terminal period remained virtually unchanged, regardless of the length of the terminal period.
Palliative medicine has evolved into a distinct and important specialty in recent years, attracting increased attention and health care resources,1 although medical education on end-of-life care remains inadequate.30 While other parts of Canada faced capacity constraints31 and the Nova Scotia cancer center weathered a shortage of physical resources, the number of radiation oncologists remained stable at an average of 7.4 per year with a range of from six to eight full-time equivalents. Because an increasing demand was being met, the variations identified in this study raise questions concerning the equality of access to palliative RT referral. The age-associated gradients in both the RT consultation and treatment decisions were striking. Elderly study subjects were far less likely to receive a consultation with a radiation oncologist in the final 9 months of life than younger study subjects, which indicates a lower referral rate to the cancer center. The elderly were also less likely to be administered palliative RT after their consultation. Age has been strongly associated with the receipt of RT in other studies,12,13,32 even after functional status is controlled for.32 Age has also been shown to be associated with referral to a PCP.14,33 It has been suggested that age in and of itself should not be a factor in the decision to treat a tumor with RT and that, in the absence of comorbidity, the elderly receive as much benefit from such treatment as a younger population.34,35 Several theories have been offered in explanation of this association, including impaired functional status, patient preferences, and a lower rate of referral to an oncologist by the primary physician.12,13,32 Greater education for both the patient and primary physician of the positive indications and potential benefits of RT might reduce the steepness of the age-associated treatment gradient. After other covariates were controlled for, including physical proximity to the cancer center and the cancer cause of death, there was an inverse association between residence in an impoverished area and both the receipt of an RT consultation and the later administration of palliative RT. A similar association has been documented for adjuvant RT in the treatment of breast cancer in Ontario.13 Low income has been shown to be associated with a myriad of health problems,36-38 so the presence of comorbidity in those from poor neighborhoods may explain part of the association. MHI was strongly correlated with census education measures. For example, the Pearson correlation between MHI and adult high school education rate at the level of the EA was 0.605, which implies that residents of impoverished areas tend to be less educated than the rest of the population. This might result in less understanding of the potential benefits of palliative RT and a reluctance to be treated in this way. Greater dissemination of information on these benefits focused on the province’s poorest communities might reduce the magnitude of this variation. Meals and overnight accommodations are provided for those being treated with RT at the provincial cancer center, and financial assistance is available to the indigent to help with other travelling expenses. However, additional expenses must be incurred by most patients travelling to the cancer center, and this may be a deterrent for some. The likelihood of a consultation with a radiation oncologist during the terminal period was inversely associated with the distance of the community of residence from the cancer center. It is probable that those in the painful, final stages of their illness might not wish to make a journey lasting up to 6 hours to receive treatment. The association was not present for the administration of palliative RT after a consultation. It seems that distance from the cancer center fails to become a consideration once a patient has entered the RT system. A second cancer clinic was opened in Sydney, Cape Breton, in late 1998. Only five study subjects received palliative RT from this new center during the study period. Future studies will be able to ascertain the extent to which this new center overcomes the apparent inequity in access to palliative RT. Both the RT consultation and treatment decisions were strongly associated with the type of cancer causing death. Consultations with a radiation oncologist were most commonly given to those dying from cancer of the head and neck, thorax, or genitourinary system, while palliative RT was most commonly administered to those dying from cancer of the breast or thorax. In fact, almost half of those who were treated with palliative RT during the terminal period died of cancer of the thorax (1,116 out of 2,395 subjects; 46.6%). Conversely, subjects dying from hematologic, gastrointestinal, or gynecologic cancers were less likely to receive RT consultations, whereas subjects dying from hematologic, CNS, or gynecologic cancers were less likely to be administered palliative RT after a consultation. Certain types of cancer are more amenable to palliative RT treatment than others.2 The most common sites of metastatic disease that are likely to benefit from palliative RT are bone and brain. Bone metastases most frequently arise from cancers of the breast, lung, prostate, thyroid, and kidney, whereas brain metastases are most likely to develop in patients with breast and lung cancer.2 This corresponds well with study results. Prior RT was associated with a greater likelihood of receiving an RT consultation during the final 9 months of life, which suggests that previous access to the provincial RT system increases the probability of a consultation with a radiation oncologist being incorporated into end-of-life care. However, those treated with prior RT who had a consultation were less likely to go on to be administered palliative RT. It is possible that for a number of study subjects, treatment with radical RT decreased the advisability of palliative RT at a later time. This study provides baseline data on consultation and treatment decisions associated with the administration of palliative RT to adults dying of cancer in the province of Nova Scotia. Several sources of variation were identified, but because of a paucity of literature on the epidemiology of palliative RT, it is difficult to make comparisons with other studies or clinical benchmarks for efficacious care. Future studies on the reasons for and modifications of these sources of variation might address the impact of the Cape Breton cancer center on regional variations in RT rates and the socioeconomic and age barriers in the delivery of cancer care. In Nova Scotia, this study is one component of a much larger project on the totality of palliative services for those dying of cancer and further review of RT in general. Family and specialist physician payment claims, as well as hospital separation data, have now been incorporated into our database for analysis. Plans are underway for the addition of narcotic analgesic and chemotherapy data. With these data, referral patterns and substitution effects, eg, narcotics for palliative RT, will be studied in the years to come. As other geographic areas undertake similar analyses, a more comprehensive understanding of the factors related to optimal palliative treatment delivery will develop.
Supported in part by funding for C.J.B. from Cancer Care Nova Scotia, Nova Scotia, Canada.
1. MacDonald N: Palliative care: An essential component of cancer control. CMAJ 158: 1709-1716, 1998[Abstract] 2. Kirkbride P: The role of radiation therapy in palliative care. J Palliat Care 11: 19-26, 1995 3. Dixon P: Palliative radiotherapy in Canada. Can J Oncol 6: 74-83, 1996 (suppl) 4. Stevens G, Firth I: Audit in radiation therapy: Long-term survival and cost of treatment. Australas Radiol 41: 29-34, 1997[Medline] 5. Stevens G, Firth I: Clinical audit in radiation oncology: Results from one centre. Australas Radiol 40: 47-54, 1996[Medline] 6. Anonymous: A prospective survey of radiotherapy in Sweden. Acta Oncol 35:47-56, 1996 (suppl) 7. Priestman TJ: Palliative radiotherapy in the UK. Can J Oncol 6: 69-73, 1996 (suppl) 8. Duncan G, Duncan W, Maher EJ: Patterns of palliative radiotherapy in Canada. Clin Oncol 5: 92-97, 1993 9. Coia LR, Owen JB, Maher EJ, et al: Factors affecting treatment patterns of radiation oncologists in the United States in the palliative treatment of cancer. Clin Oncol 4: 6-10, 1992 10. National Cancer Institute of Canada: Canadian Cancer Statistics 1999. Toronto, Canada, 1999 11. The Canada Health Act. Statutes of Canada, Chapter 6, sect 12, 1984 12. Tyldesley S, Zhang-Salomons J, Groome PA, et al: Association between age and the utilization of radiotherapy in Ontario. Int J Radiat Oncol Biol Phys 47: 469-480, 2000[Medline] 13. Paszat LF, Mackillop WJ, Groome PA, et al: Radiotherapy for breast cancer in Ontario: Rate variation associated with region, age and income. Clin Invest Med 21: 125-134, 1998[Medline] 14. Johnston GM, Gibbons L, Burge FI, et al: Identifying potential need for cancer palliation in Nova Scotia. CMAJ 158: 1691-1698, 1998[Abstract] 15. Puckett CD (ed): The Educational Annotation of ICD-9-CM. Reno,NV, Channel, 1986 16. Statistics Canada: 1996 Census: Profile of Enumeration Areas, Profile Series—Nova Scotia. Ottawa, Canada, Geography Division, Statistics Canada, 1997, Catalogue No. 95F0185XDB96001a 17. Statistics Canada: Postal Code Conversion File: May 1998 Postal Codes—Reference Guide. Ottawa, Canada, Geography Division, Statistics Canada, 1998, Catalogue No. 92F0027XDB. 18. Statistics Canada: Postal Code Population Weight File: May 1996 Postal Codes—Reference Guide. Ottawa, Canada, Geography Division, Statistics Canada, 1998, Catalogue No. 93F0040XDB 19. Wilkins R: Geocodes/PCCF Version 2 User’s Guide: Automated Geographic Coding Based on the Statistics Canada Postal Code Conversion File. Ottawa, Canada, Health Statistics Division, Statistics Canada, 1996
20.
Krieger N: Overcoming the absence of socioeconomic data in medical records: Validation and application of a census-based methodology. Am J Public Health 92: 703-710, 1992 21. Mustard CA, Derksen S, Berthelot JM, et al: Assessing ecologic proxies for household income: A comparison of household and neighbourhood level income measures in the study of population health status. Health Place 5: 157-171, 1999[Medline] 22. Anderson RT, Sorlie P, Backlund E, et al: Mortality effects of community socioeconomic status. Epidemiology 8: 42-47, 1997[Medline] 23. Fraleigh JB: Calculus with Analytic Geometry ( ed 3 ). Reading, MA, Addison-Wesley, 1990 24. Brundage MD, Dixon PF, Mackillop WJ, et al: A real-time audit of radiation therapy in a regional cancer centre. Int J Radiat Oncol Biol Phys 43: 115-124, 1999[Medline] 25. Nova Scotia Health Act. R.S., code 195, sect 1, part VII, 101.1-101.2, 1989 26. Gaudette LA, LaBillois T, Bao R-N, et al: Quality assurance for the Canadian Cancer Registry, in Symposium ’96 Nonsampling Errors: Proceedings. Ottawa, Canada, Minister of Industry, 1997, pp 229-238, Statistics Canada, Catalogue No. 11-522-XPE 27. Agresti A: An introduction to categorical data analysis. New York, NY, John Wiley & Sons, 1996 28. Riley GF, Potosky AL, Lubitz JD, et al: Medicare payments from diagnosis to death for elderly cancer patients by stage at diagnosis. Med Care 33: 828-841, 1995[Medline]
29.
Christakis NA, Escarce JJ: Survival of Medicare patients after enrollment in hospice programs. N Engl J Med 335: 172-178, 1996
30.
Rabow MW, Hardie GE, Fair JM, et al: End-of-life care content in 50 textbooks from multiple specialties. JAMA 283: 771-779, 2000 31. Mackillop WJ, Fu H, Quirt CF, et al: Waiting for radiotherapy in Ontario. Int J Radiat Oncol Biol Phys 30: 221-228, 1994[Medline] 32. Brown JS, Eraut D, Trask C, et al: Age and the treatment of lung cancer. Thorax 51: 564-568, 1996[Abstract] 33. Hunt R, McCaul K: Coverage of cancer patients by hospice services: South Australia, 1990 to 1993. Aust N Z J Public Health 22: 45-48, 1998[Medline] 34. Crocker I, Prosnitz L: Radiation therapy of the elderly. Clin Geriatr Med 3: 473-481, 1987[Medline] 35. Zacharia B, Balducci L: Radiation therapy of the older patient. Hematol Oncol Clin North Am 14: 131-167, 2000[Medline]
36.
Smith GD, Neaton JD, Wentworth D, et al: Socioeconomic differentials in mortality risk among men screened for the Multiple Risk Factor Intervention Trial: I. White men. Am J Public Health 86: 486-496, 1996
37.
Smith GD, Wentworth D, Neaton JD, et al: Socioeconomic differentials in mortality risk among men screened for the Multiple Risk Factor Intervention Trial: II. Black men. Am J Public Health 86: 497-504, 1996
38.
Kunst AE, Groenhof G, Mackenbach JP, et al: Occupational class and cause specific mortality in middle aged men in 11 European countries: Comparison of population based studies. BMJ 316: 1636-1642, 1998 Submitted June 5, 2000; accepted April 16, 2001. This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2001 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 statistic.
