• Open Access

Patterns of surgical care for prostate cancer in NSW, 1993-2002: rural/urban and socio-economic variation

Authors


Correspondence to:
Andrew Hayen, School of Public Health A27, University of Sydney 2006. Fax: (02) 9351 5109; e-mail: ahayen@health.usyd.edu.au

Abstract

Objectives: Prostate cancer is the most commonly registered cancer in Australian men, yet there is a lack of information about its management. We described the patterns and trends in the surgical treatment of men with prostate cancer in New South Wales (NSW).

Method: We used probabilistic record linkage to link cancer registry data with hospital admissions. All NSW men diagnosed with prostate cancer between 1993 and 2002 were eligible for the study. Rates of radical prostatectomy, bilateral orchidectomy and transurethral resection of the prostate were calculated. Factors affecting the probability of undergoing these procedures were examined using log-binomial regression.

Results: Between January 1993 and December 2002, 38,712 men were diagnosed with prostate cancer in NSW. Of these, 33,200 (85.8%) cancer registry records were linked to at least one hospital admission record. Men resident in rural areas at diagnosis (RR = 0.69, 95% Cl 0.65-0.73) and men resident in more socio-economically disadvantaged areas (RR =0.83, 95% Cl 0.78-0.89 for most disadvantaged) were significantly less likely to undergo a radical prostatectomy after adjusting for age and disease stage. While orchidectomy rates fell significantly during the period, rates were significantly higher in rural and lower socio-economic areas after adjusting for age and stage.

Conclusion and implications: Further investigation is needed to understand the reasons for the variation in the surgical patterns of care for prostate cancer so that interventions can be implemented to ensure appropriate access for all men.

After non-melanocytic skin cancer, prostate cancer is the most commonly diagnosed cancer in men in New South Wales (NSW, male population 3.2 million). It accounted for 29% of newly diagnosed cancers in males in 2004 and was the second most common cause of cancer death, accounting for 13% of cancer deaths in males.1

In Australia, there is a lack of information about trends in patterns of care for prostate cancer. Compared with more accessible areas of NSW, it has been documented that there is a poorer five-year relative survival after diagnosis of men resident in rural and remote areas.2 The extent to which they are related to differences in patterns of care is unknown.3,4

The aim of this study was to describe the patterns and trends in the surgical treatment of men with prostate cancer in NSW using record linkage between the NSW Central Cancer Registry and hospital admissions data. We explored the trends between 1993 and 2002 and examined whether prostate cancer patterns of care varied across regional or socio-economic groups.

Methods

Data

Men resident in NSW and diagnosed with prostate cancer between 1993 (the first full year for which linked data were available) and 2002 were included. Data on the surgical treatment of prostate cancer were obtained by linking the population-based NSW Central Cancer Registry (CCR) with the Admitted Patient Data Collection (APDC). The APDC is a census of inpatients treated in public and private NSW hospitals. Records from the APDC for all inpatient episodes from July 1992 to June 2003 were included. Radiotherapy, the other main form of curative treatment for prostate cancer, is usually administered on an outpatient basis and so is not routinely recorded in the APDC.

The two datasets were linked by the NSW Department of Health using the Automatch software for probabilistic record linkage.5 The Cancer Council NSW, NSW Department of Health and Cancer Institute NSW ethics committees approved the project.

ICD-9-CM procedure codes were used for hospital separations for the years 1992/3 to 1999/2000, after which MBS-Extended (MBS-E) procedure classifications in ICD-10-AM were used. We determined whether men had a radical prostatectomy (ICD-9-CM codes 60.5, MBS-E codes 37209-00, 37210-00 and 37211-00) and bilateral orchidectomy (ICD-9-CM code 62.41, MBS-E code 30641-01).

Stage information was defined as the spread of disease at diagnosis recorded by the Central Cancer Registry according to a simple classification of localised, regional, distant or unknown spread at diagnosis.6

Place of residence at the time of diagnosis was assigned to the corresponding statistical local area and to the area health service using boundaries defined by the NSW Health Department in 1996. Health areas in Sydney, the Central Coast, Illawarra and Hunter were designated as ‘urban’, and the remainder were classified as ‘rural’ health areas.

The index of relative socio-economic disadvantage (IRSD) was used to construct the socio-economic (SES) groups.7

Age-standardised percentages of procedure types by year were calculated using men diagnosed with prostate cancer in 1993 as the standard population. We used binomial regression with a log link function to examine the associations between the probability of undergoing a procedure and age-group at diagnosis (less than 60 years, 60-69 years and 70 years or older), year of diagnosis, spread of disease at diagnosis, socio-economic status group and urban/rural health area of residence at diagnosis.8 For the analysis of orchidectomies, we used Poisson regression with a log-link function and an empirical co-variance structure as the log-binomial model failed to converge.8

Results

Between January 1993 and December 2002, 38,712 men diagnosed with prostate cancer in New South Wales were identified by the CCR as eligible for this study. Of these, 85.8% were linked to at least one inpatient (APDC) record.

Radical prostatectomy

There were 6,124 radical prostatectomies recorded during the study period. The crude and age-standardised percentages of men undergoing radical prostatectomy increased between 1993 and 2002 (Figure 1a). Of men diagnosed in 2002, 875 or 22% were recorded as having had a radical prostatectomy.

Figure 1.

Crude and age-standardised percentage of men undergoing (a) radical prostatectomy and (b) bilateral orchidectomy by year of diagnosis, NSW, 1993-2002.

After adjusting for all other variables, men resident in rural areas at diagnosis had significantly lower rates of radical prostatectomy than those resident in urban areas (Table 1). When compared to men resident in the least disadvantaged socio-economic areas (quintile 1), men from more disadvantaged areas had lower rates of radical prostatectomy after controlling for stage and age. Younger men were more likely to undergo a radical prostatectomy than were older men. Men with distant disease were least likely to have a radical prostatectomy.

Table 1.  Crude and adjusted relative risks and 95% confidence intervals (CI) for undergoing a radical prostatectomy or bilateral orchidectomy by patient and disease characteristics.
 LevelnPer centCrude relative risk95% CIAdjusted relative risk+95% CIp-value
  1. Note:

  2. + Adjusted for all variables shown in the table

Radical Prostatecomy
Area of residence at diagnosisUrban5,07918.111<0.001
 Rural1,0459.80.540.51-0.580.690.65-0.73 
Socioeconomic quintile1 (Least disadvantaged)2,10623.511<0.001
 21,05614.80.630.59-0.680.890.85-0.94 
 399814.90.630.59-0.680.930.89-0.98 
 41,17812.60.540.50-0.570.870.82-0.92 
 5 (Most disadvantaged)78611.90.510.47-0.550.830.78-0.89 
Age at diagnosis< 60 years2,18149.811<0.001
 60-69 years3,45527.80.560.54-0.580.700.67-0.72 
 70+ years4882.20.040.04-0.050.070.06-0.08 
Spread at diagnosisLocalised4,17527.011<0.001
 Regional80845.71.691.60-1.791.181.14-1.23 
 Distant291.40.050.04-0.080.090.07-0.14 
 Unknown1,1125.70.210.20-0.230.320.30-0.34 
Year of Diagnosis1 year increase  1.071.06-1.081.041.03-1.04<0.001
Bilateral orchidectomy
Area of residence at diagnosisUrban1,8906.711<0.001
 Rural1,0179.61.421.32-1.531.361.26-1.47 
Socioeconomic quintile1 (Least disadvantaged)5015.611<0.001
 25958.41.501.33-1.681.301.17-1.46 
 35638.41.501.34-1.691.291.15-1.45 
 46256.71.201.07-1.351.060.94-1.19 
 5 (Most disadvantaged)6239.41.691.51-1.891.301.15-1.46 
Age at diagnosis< 60 years751.711<0.001
 60–69 years5224.22.451.93–3.121.971.55–2.49 
 70+ years2,31010.56.165.90–7.734.713.75–5.91 
Spread at diagnosisLocalised9706.311<0.001
 Regional1206.81.080.90-1.301.381.16-1.64 
 Distant37218.42.932.63-3.272.161.94-2.41 
 Unknown1,4457.41.191.10-1.281.000.92-1.08 
Year of Diagnosis1 year increase  0.760.74- 0.770.770.76-0.78<0.001

Orchidectomy

During the study period there were 2,907 bilateral orchidectomies used in the treatment of prostate cancer. The percentage of men with prostate cancer treated by orchidectomy steadily declined from almost 20% of men in 1993 to under 2% in 2002 (Figure 1b).

After adjusting for other variables, patient characteristics associated with an increased likelihood of orchidectomy included being resident in a rural area, being aged 60 years or older, and having advanced stage of disease at diagnosis (Table 1). Men resident in the more disadvantaged socio-economic areas had higher rates of orchidectomy than men resident in the least disadvantaged areas (RR=1.30, 95% CI 1.15-1.46 for the most disadvantaged area).

Discussion

We used routinely collected population health data to describe patterns and trends in the surgical treatment of prostate cancer in NSW. We identified regional and socio-economic differences in surgical management in NSW between 1993 and 2002. Men resident in highly accessible areas and in more socio-economically advantaged areas were more likely to have had a radical prostatectomy and less likely to have had orchidectomy.

These results confirmed and quantified some commonly accepted management practices for prostate cancer, namely that younger men and men with earlier stage disease were more often treated with radical prostatectomy than older men and men with advanced disease.

Linkage of routinely collected datasets is a cost effective and efficient way of obtaining population-based health information.9 The linked dataset used here was previously shown to have high accuracy and completeness for examining patterns of surgical care for breast cancer.10

Limitations of the present study include the lack of information on other forms of treatment for prostate cancer, particularly radiotherapy and androgen deprivation therapy. Approximately 75% of men had forms of treatment other than radical prostatectomy or bilateral orchidectomy or had no active treatment. Fifteen per cent of all prostate cancer cases were unable to be linked to an inpatient episode. Coding of stage was undertaken by medical coders at the Central Cancer Registry according to the method described by Jensen.6 Our finding that men with regional stage disease were more likely to have undergone radical prostatectomy than men with localised disease is counter-intuitive to recommended guidelines for the treatment of this disease and likely reflects an ‘upstaging’ effect11 based on pathological findings for men who underwent radical prostatectomy.

With these data we were unable to examine other recognised factors likely to determine men's suitability or preference for surgery, such as the clinical stage, grade, PSA level, any co-morbid conditions or patients' or doctors' preferences for the type of treatment.

Men from the more disadvantaged socio-economic groups were less likely to undergo a radical prostatectomy than men from the least disadvantaged socio-economic groups, a result consistent with studies from Western Australia,4 the US,12,13 and the UK.14 Men from urban areas were more likely than men from rural areas to have localised disease (42% vs 35%), and less likely to have unknown stage disease. Even after adjusting for age and stage differences, rates of prostatectomy were lower and rates of orchidectomy were higher in rural compared to urban areas, reflecting patterns similar to those found in other Australian studies.3,4 Reasons for this variation are unclear. Many areas of rural NSW had poorer accessibility to urologists, particularly during the early period of study, which may account for some of the disparity. However, if men in less accessible parts of the state were not receiving radical prostatectomy it is unlikely that they were being treated with radiotherapy, given that the only radiation oncology centres were located in urban areas and a standard course of external beam radiotherapy involves a seven-week series of daily treatments.

Rates of radical prostatectomy increased and orchidectomy fell between 1993 and 2002, reflecting similar trends observed in the US15,16 and Canada.17 A trend of medical androgen deprivation replacing orchidectomy, particularly in the treatment of later stage disease is likely, as data from the Pharmaceutical Benefits Scheme, a national database of government reimbursements for prescription medicines in Australia, show an increase in use of a number of the common anti-androgens and LHRH agonists during the period 1991 to 2002.18

Will the differences observed in patterns of care result in inequitable outcomes of care? Differences in prostate cancer mortality are likely to be due in part to differences in patterns of treatment. One randomised controlled trial19 has shown that radical prostatectomy reduces mortality from prostate cancer. Previous studies have demonstrated poorer survival in men with prostate cancer resident in less accessible areas of Australia.2,3,4 Although we found differences by socio-economic status in rates of radical prostatectomy, the extent to which this may be attributed to physical or financial barriers to access services or treatment is not possible to determine from this study.

This study found differences in the surgical treatment of prostate cancer according to socio-economic status and urban/rural area of residence. Further investigation is needed to better understand the reasons for these differences and whether mortality outcomes are affected by differences in patterns of care. Irrespective of these outcomes, access to appropriate treatment for all men with prostate cancer regardless of place of residence or socio-economic status is necessary.

Acknowledgments

The authors would like to thank the staff of the Centre for Epidemiology and Research at the NSW Department of Health for assistance with the data linkage and provision of the data. The NSW Central Cancer Registry provided the data on newly diagnosed cancer cases. The Cancer Registry is funded by the NSW Department of Health and managed by the Cancer Institute NSW under a performance and funding agreement.

Andrew Hayen began work on this paper while employed by the NSW Department of Health.

Ancillary