Family Practice Vol. 20, No. 3, 294-303
© Oxford University Press 2003
Screening |
GP and patient predictors of PSA screening in Australian general practice
a Division of Population Health, South Western Sydney Area Health Service,
b School of Public Health, University of Sydney, and
c Surgical Outcomes Research Centre (SOuRCe), Central Sydney Area Health Service, Sydney, Australia.
Correspondence to Professor Jeanette E Ward, Director, Division of Population Heath, South Western Sydney Area Health Service, Locked Bag 7008, Liverpool NSW 1871, Australia; Email: Jeanette.Ward{at}swsahs.nsw.gov.au
Gattellari M, Young JM and Ward JE. GP and patient predictors of PSA screening in Australian general practice. Family Practice 2003; 20: 294–303.
Received ; Revised 26 November 2002; Accepted 13 January 2003.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Objective. We determined GP and patient variables associated first with men’s prior uptake of prostate-specific antigen (PSA) screening and, subsequently, its initiation during an ‘index consultation’ in Australian general practice.
Methods. From the practices of 60 GPs, we recruited a sample of 423 male patients aged 40–70 years. In a waiting room questionnaire completed before their ‘index consultation’ (retrospective component), men reported their previous PSA screening status. We obtained demographic and clinical data, including the presence of lower urinary tract symptoms (LUTS). Men also were mailed a questionnaire 2 days after their ‘index consultation’ to ascertain whether the GP had discussed PSA screening (prospective component) for prostate cancer and other behaviours. GPs themselves completed questionnaires eliciting demographic and practice characteristics as well as their propensity to screen and understanding of the evidence about PSA testing. GP and patient study variables were modelled simultaneously in analyses.
Results. Of those 348 men consulting with their regular GP, 80 (23.0%) reported previously having had a PSA screening test. Men were significantly and independently more likely ever to have had PSA screening if their regular GP reported a propensity to initiate screening [adjusted odds ratio (AOR) = 2.27, 95% confidence interval (CI) 1.23–4.20; P = 0.009]. GP age also was independently associated with men’s PSA screening status [chi-squared (3) P < 0.0001] as was men’s age and severity of LUTS (AOR = 2.38, 95% CI 1.58–3.57, P < 0.0001 and AOR = 1.79, 95% CI 1.00–3.19, P = 0.004, respectively). Current smokers were less likely ever to have had a PSA screening test (AOR = 0.34, 95% CI 0.16–0.69; P = 0.003). Discussion of PSA screening in their ‘index consultation’ was recalled independently more often by older men (AOR = 1.46, 95% CI 1.00–2.13; P = 0.04), those with moderate/severe LUTS (AOR = 1.94, 1.07–3.49; P = 0.04), those whose GP had performed or discussed a cholesterol test (AOR = 2.26, 95% CI 1.03–4.92; P = 0.04) and those whose GP had postgraduate training in family medicine (AOR = 3.13, 95% CI 1.23–8.00; P = 0.02).
Conclusion. In the absence as yet of compelling evidence that PSA screening will prolong life or enhance its quality, our findings identify GP and patient factors that could be targeted to modify PSA screening.
Keywords. Evidence-based practice, general practice, PSA screening.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
During the last decade, the incidence of prostate cancer has increased substantially,1,2 attributed in part to the growing use of the prostate-specific antigen (PSA) blood test for detection of asymptomatic disease.1 Almost half of Australian men aged 60–69 years have had at least one PSA test.3 In Europe, participation in the screening trials evaluating the efficacy of PSA testing is estimated to be
45%.4 While randomized controlled trials are underway,5 there currently is no compelling evidence that early detection reduces disease-specific mortality and improves quality of life.6,7
While the apparent simplicity of PSA screening may explain its appeal, it belies its potential for harm. Adverse consequences are not trivial, including psychological distress associated with false-positive test results, risks related to diagnostic work-up, the potential for detecting clinically indolent cancer not destined to threaten life, and side effects associated with treatments.6,7 Equipose over treatment also persists.6,7 International studies such as the ProtecT in the UK aim to evaluate the value of treatment for early-stage prostate cancer using a rigourous randomized study design.8
Previous studies investigating factors associated with PSA screening are limited, since only patient predictors of screening were examined.9,10 Given that GPs initiate discussions for at least half of PSA screening tests performed,9 the predictive effects of patient variables in the presence of GP factors merit investigation. Hence, we undertook this study to determine the relative influence of both GP and patient predictors of prior PSA screening history in Australian general practice. We also included a prospective component to explore further GP and patient predictors of screening-related behaviours during a specific ‘index consultation’.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Setting
Sydney is the urban capital of New South Wales, the most populous state in Australia in which
7000 GPs provide health care, reimbursed through publicly-subsidized fee-for-service mechanisms.11
Recruitment of GPs
As reported elsewhere,12 we approached all GPs in our local region in Central Sydney to participate in a project on quality of care.
GPs first completed a 16-page questionnaire, comprising questions about a variety of preventive care issues, thereby ensuring that those about PSA testing did not appear to be the primary focus. GPs were asked to respond to a clinical case scenario in which a 58-year-old male patient attended to have his ears syringed. GPs were asked to indicate whether they would be ‘highly likely’ or ‘somewhat likely’ to discuss PSA testing opportunistically with this patient or if they ‘would not discuss’. This question measured GPs’ ‘propensity to screen’.
Four questions next were included to assess GPs’ perception of the evidence. GPs were asked to indicate if there was ‘enough evidence to support’, ‘enough evidence against’ or ‘inconclusive’ evidence for PSA testing in asymptomatic men to reduce premature mortality from prostate cancer. We then asked GPs to indicate their perception of the evidence for its use for men with lower urinary tract symptoms (LUTS). GPs also were asked to indicate their perceptions of the evidence when PSA testing was combined with a digital rectal examination (DRE) in asymptomatic men and, next, for men with LUTS.
Our GP questionnaire ascertained demographic and practice characteristics. In addition, we asked GPs to indicate postgraduate training in general practice, their membership of the Royal Australian College of General Practitioners (RACGP), membership of the Australian Medical Association (AMA) and membership of their local GP division. GP divisions are geographically defined areas which are responsible for local educational activities.
Patient recruitment
A consecutive sample of male patients aged 18–70 years who consulted participating GPs during a specified 1-week period were asked to complete a questionnaire while waiting to see their GP. Patient recruitment for GPs practising part-time occurred on an equivalent number of consecutive working days. As with the GP questionnaire, PSA testing was not over-emphasized. Patients were considered ineligible if they were unable to read English or were too sick to participate.
Men first were asked to clarify whether they were seeing either their (i) regular GP; (ii) a GP previously consulted who was not their regular GP; or (iii) a GP never before consulted. Current health status was assessed using a validated 5-point scale (‘excellent’ to ‘poor’).13 Men provided the reason(s) for their visit, using a checklist of 25 items. Reasons were grouped into clinical symptoms (n = 16), preventive care (n = 6) or men’s health (n = 3). The latter grouping comprised attendance for either a ‘prostate problem’, a prostate cancer screening test or, for those with prostate cancer, prostate cancer surveillance.
Next, the presence of LUTS was assessed using the validated 7-item American Urological Association Symptom Index (AUA-SI).14–16 Responses to these seven items were summed14 to produce an overall score (0–35). Using this overall score, we applied the standardized classification for symptom severity of Barry et al.:14 ‘mild’ (0–7), ‘moderate’ (8–19) or ‘severe’ (20+) to describe men’s symptom severity. Men then indicated the degree to which they were satisfied with their current level of urinary functioning, using a validated 7-point ‘Quality-of-life’ assessment scale (‘delighted’ to ‘terrible’).14
Men’s previous participation in prostate cancer screening was then ascertained. First, they were asked if they ever had a rectal exam, blood test (prostate-specific antigen or PSA) or both a rectal exam and a blood test to screen for prostate cancer. Secondly, those who had ever been screened were asked to report when they last had a screening test (within the last year, between 1 and 5 years ago or more than 5 years ago).
Participants were also asked relevant demographic questions and then sealed their completed questionnaire in an envelope and placed it into a security box, before seeing their GP.
Prospective component
Men were mailed a follow-up questionnaire within two working days of the ‘index consultation’. We asked them to recall if, during this recent consultation, the GP had:
- Discussed screening for prostate cancer
- Given them any information to read about prostate cancer
- Advised them to have a PSA test (prostate-specific antigen)’ and/or
- Performed a PSA test.
As serum cholesterol testing provides an opportunity for GPs to request serum PSA, we also asked them to report whether the GP discussed or performed a cholesterol (blood lipid) check. For all questions assessing recall of GP behaviours, men could indicate ‘yes’, ‘no’ or ‘can’t remember’.
All data were collected in 1999. Only data obtained from men aged between 40 and 70 years are reported here.
Statistical analysis
Analysis of GP responses..
Frequencies for GP responses to the case scenario measuring their propensity to screen and the four items assessing their perceptions of the evidence for PSA screening first were determined.
Multivariate logistic regression analysis then was applied to identify predictors of GPs’ responses. We dichotomized GPs’ propensity to screen as follows: ‘highly’ and ‘somewhat likely’ or ‘would not discuss’. GPs who incorrectly indicated ‘sufficient evidence to support use’ were compared with those who gave other responses. For these analyses, univariate logistic regressions first were performed for each potential predictor, and those associated with the outcome at the P < 0.25 level17 were entered into the multivariate analysis. Significance tests were based on the log-likelihood ratio. The goodness and strength of fit for the final models were determined by the c-statistic and the Hosmer and Lemeshow goodness of fit test.17 These analyses were conducted using SPSS (version 9.0).18
Men’s previous participation in PSA screening.. A man was considered to have ever had a PSA screening test if he identified having had a PSA screening test, either alone or in combination with DRE.
Five variables were considered potential predictors of previous participation in PSA screening for our multivariate analysis: men’s age (continuous), main language spoken at home (English versus other), highest educational level attained (up to school certificate versus HSC/ tertiary educated), employment status (full-time versus other) and marital status (living as married versus other). Although our data ascertained current LUTS rather than previous LUTS, we also modelled AUA-SI scores and quality-of-life assessment (delighted/pleased/mostly satisfied versus other) on the assumption that symptoms most commonly remain constant or worsen over time. Data collected from community and patient samples suggest that LUTS severity remains constant over time or worsens in the majority of cases, respectively.19,20
The categorization of men as having ‘mild’, ‘moderate’ or ‘severe’ urinary symptoms is standard for researchers, and is purely descriptive. Continuous scores were used in the analysis, as is standard.14 Hence, a score of 0 essentially denotes an asymptomatic state.
GPs’ demographic and practice characteristics were also included in the analysis. In addition, a GP’s propensity to screen using PSA testing and their perceptions of the evidence also were considered as potential predictors.
To provide a valid assessment of the relationship between GP characteristics and patient self-reported participation in PSA screening, we conducted this analysis only for those men who were consulting their regular GP for the ‘index consultation’.
Recall of GP behaviour during the ‘index consultation’.. Predictors of men’s recall of four GP behaviours during the index consultation were also determined, using the same variables outlined above. In addition, whether or not men reported their GP as having discussed and/ or performed cholesterol (blood lipid) testing during the index consultation was also considered a potential predictor. However, we considered only consultations in which cholesterol testing featured for men who did not attend for a ‘men’s health issue’. In this manner, we aimed to minimize the possibility that cholesterol testing was discussed and/or performed as a result of discussions about prostate cancer screening.
Data from all men were included in this analysis irrespective of whether or not they were consulting their regular GP.
Modelling principles and strategy.. Potential predictors associated with all outcomes of interest at the P < 0.25 level were first entered into the models. Continuous variables were modelled as categorical if a non-linear association with the outcome was evident, with categories formed using quartile cut-off ranges.17 To aid the interpretation of findings where patient age emerged as a significant predictor of the outcomes, we present the odds ratio based on every 10 year increase in age. For LUTS, we report the odds ratio comparing men with ‘moderate’ or ‘severe’ LUTS with those who had ‘mild’ symptoms.
Once the baseline model was thus determined, a non-automated backward selection procedure was adopted whereby the least significant term was removed and the analysis repeated until only significant variables remained in the model (P < 0.05).
For the outcomes of men’s previous participation in PSA screening and recall of GP behaviour, we used generalized estimation equations (GEEs) to account for dependencies arising from clustered patient sampling by GP. We specified an exchangeable correlation structure underlying intra-class correlations. GEE analysis allows one simultaneously to assess the significance of covariates at the individual (i.e. male patients) and cluster (i.e. GP) levels. We used PROC GENMOD in SAS (Version 8.0).21
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Table 1
displays the demographic and practice characteristics of the 60 participating GPs. Age, gender and practice type of our sample were similar to GPs practising elsewhere in Australia and in the state of NSW11 (Table 1).
|
GP self-report of likelihood to discuss PSA testing opportunistically
In response to the hypothetical case scenario of a 58-year-old male patient presenting to the surgery to have his ears syringed, 19 of the 60 GPs (32%) reported they would be ‘highly likely’ to introduce a discussion about PSA. Just under half (n = 26, 43%) said they would be ‘somewhat likely’ to discuss PSA testing, while 14 (23%) said they would not discuss it. One GP did not respond to this question. None of the GP practice or demographic characteristics listed in Table 1
were significantly and independently predictive of GPs’ propensity to discuss PSA screening. GPs in our study were significantly more likely to discuss PSA screening than a national sample surveyed in 1996, however22 [chi-squared (2) = 9.96; P = 0.007].
Perceptions of the evidence supporting PSA screening for prostate cancer
Fifteen per cent (n = 9) incorrectly indicated that there is ‘enough evidence’ to support PSA blood testing in asymptomatic men (Table 2). Significantly more GPs incorrectly indicated that there was ‘enough evidence’ to support its use in men with LUTS (50%) (McNemar’s test, P < 0.001). Similarly, significantly more GPs indicated that there was ‘enough evidence’ for PSA screening alone and in combination with DREs, for men either with or without LUTS (83 and 37%, respectively) (both McNemar’s test P < 0.001) (Table 3).
|
|
GPs who were not members of their local general practice division and who had not undergone postgraduate training in general practice were significantly and independently more likely to indicate that there was enough evidence to support the use of PSA and DRE tests for asymptomatic men [adjusted odds ratio (AOR) = 4.76, 95% confidence interval (CI) 0.89–25.50, P = 0.04 and AOR = 4.27, 95% CI 1.03–17.64, P = 0.04) [c = 0.69; Hosmer Lemeshow chi-squared (2) = 0.29; P = 0.87]. GPs who were not members of the AMA were significantly more likely to indicate that the evidence supported the use of PSA in men with LUTS (OR = 3.29, 95% CI 1.05–10.31; P = 0.04).
Patient recruitment
Of 1160 male patients attending during the study period, 441 were not aged between 40 and 70 years old. Of the remaining 719 males aged 40–70 years, 186 were ineligible because they could not read or write English (n = 126), had forgotten their reading glasses (n = 40), were feeling too ill to participate (n = 19), or because the identity of the GP they were due to consult was unknown (n = 2). Of the 531 eligible male participants aged 40–70 years attending during the study period, 423 consented to participate (79.7% consent rate). The ages of consenting and non-consenting patients were not significantly different (mean 54.1 versus 54.1 years) [t (529) = 0.01, P = 0.99]. Of the 60 participating GPs, six did not see any men aged 40–70 years of age. These GPs did not differ significantly from those 54 who did, by practice size (P = 0.17), age (P = 0.95) and sex (P = 0.19). The number of male patients aged 40–70 years seeing one of these 54 GPs ranged from one to 25 per GP [median 6.5, interquartile range (IQR) 2–11.25, mode = 2].
Patient demographics and LUTS
Men in our study were an average of 54.1 years of age (SD 9.16). Only 18 patients (4.3%) had never previously consulted with the GP they were about to see (Table 3).
The age distribution of residents within Central Sydney was comparable with that of the study sample23 (Table 3).
AUA-SI scores were low in our sample (median 4.0; IQR = 1–7.25) (Table 3). Almost three-quarters of men (n = 308, 72.8%) had ‘mild’ LUTS (Table 3). Increasing age was significantly associated with increasing AUA-SI scores (
= 0.31, P < 0.001) and lower levels of satisfaction with urinary functioning ( = 0.24, P < 0.001).
Reasons for GP visit
Most men (n = 353, 83.5%) were presenting with clinical symptoms. Fifteen men (3.5%) were attending for a urinary tract problem. None of these had had prostate cancer. One half of the sample (n = 212, 50.1%) presented for preventive care, most commonly a blood pressure check (n = 140, 33.1%). Just over 10% (n = 47, 11.3%) presented for preventive care only.
Of those three men (0.7%) in our sample known to have prostate cancer, all three presented for follow-up (surveillance) of their prostate cancer.
Of the 420 men who did not have a previous diagnosis of prostate cancer, 28 (6.7%) were presenting for a ‘men’s health issue’. Fifteen of these 28 men indicated they were attending for a ‘prostate problem’. Thirteen men out of the total sample of 420 (3.1%) were attending for a prostate cancer screening test. All 13 men presenting for prostate cancer screening nominated at least one other reason unrelated to men’s health for their visit (median 2, IQR 1–4.5). Eight (62%) of these 13 men were also attending for other preventive health checks. A similar proportion (50.9%) of men not attending for prostate cancer screening were attending for other preventive health checks [chi-squared (1) = 0.57, P = 0.45].
Previous participation in prostate cancer screening
Almost one-quarter (n = 100, 23.6%) of men reported in our waiting room survey that they had ever had a PSA test, either alone (n = 31) or in combination with a ‘rectal examination’ (n = 69). Of these, 42% reported having a PSA test within the last year.
Predictors of previous participation in PSA screening.
Of the 423 men in our sample, 348 (82.3%) indicated that they were visiting their regular GP. Of these 348, 10 did not provide data on their screening status. Thus, data for 338 men were available to examine GP and patient predictors of ever having had a PSA test.
Multivariate analysis modelling of both patient and GP variables simultaneously demonstrated that older patients and those with currently higher AUA-SI scores were significantly and independently more likely to have ever had a PSA screening test (P = 0.0001 and P = 0.004, respectively) (Table 4). An OR of 2.38 for ever having had a PSA test was associated with every 10 year increase in age (95% CI 1.58–3.57), whereas, compared with men with ‘mild’ LUTS, men with ‘moderate’ or severe’ LUTS had increased odds of 1.79 (95% CI 1.00–3.19) of having ever been tested. While the lower limit of the confidence interval suggests a minimal effect of LUTS, the odds could be 3.5 times greater amongst men with ‘moderate’ or ‘severe’ LUTS.
|
In addition, smokers were significantly and independently less likely (AOR = 0.34, 95% CI 0.17–0.70, P = 0.003) to have ever had a PSA test than non-smokers (Table 5
). Furthermore, patients whose regular GP self-reported a propensity to discuss PSA testing opportunistically with patients were also significantly more likely to have reported ever having had a PSA test (AOR = 2.27, 95% CI 1.23–4.20). GP age also was independently associated with patient self-reported screening status (P < 0.0001). However, the effect of GP age was not clearly linear, with higher rates of PSA screening evident amongst men who consulted with GPs who were relatively younger or older (Table 4).
|
As AUA-SI scores were correlated with men’s satisfaction with urinary functioning (
= –0.26, P < 0.001), we re-ran the analysis to determine if satisfaction with urinary functioning would emerge as a significant predictor in the absence of our testing for the effect of LUTS. The results were replicated. Satisfaction with urinary functioning did not emerge as an independent and significant predictor.
Recall of PSA screening-related behaviour during the ‘index consultation’
Of the 407 (96.2%) men consenting to receive a follow-up questionnaire, 294 responded (72.2%). Responders were significantly older compared with non-responders (mean 54.7 versus 52.7 years) [t (421) = 2.10, P = 0.04].
Recall of GP behaviour about prostate cancer screening was uncommon. Of those 291 not already known to have prostate cancer, 31 (10.7%) recalled their GP discussing screening for prostate cancer during the index consultation. Eight (2.7%) recalled receiving written information about prostate cancer. Twenty-nine patients (10.0%) recalled their GP advising them to have a screening test for prostate cancer. Twelve men (4.1%) recalled the doctor performing a PSA test. However, only one of these 12 men received written information.
In total, 45 men (15.5 %) recalled any one of these four behaviours relating to PSA screening. Attending for a ‘men’s health issue’ was highly correlated with patient recall of any such GP behaviour (70.0 versus 13.1%) [chi-squared (1) = 41.37, P < 0.001].
Men were significantly more likely to recall any one of these behaviours if the GP consulted had completed postgraduate training in family medicine (AOR = 3.13, 95% CI 1.23–8.00; P = 0.01). In addition, GPs were significantly and independently more likely to perform these behaviours if patients self-reported the GP discussing and/or performing cholesterol (blood lipid) testing (AOR 2.26, 95% CI 1.03–4.92; P = 0.04). While the lower confidence limit suggests a small increase, the upper limit suggests a strong association. Independent patient predictors of these behaviours were LUTS and age status. Men who had relatively higher AUA-SI scores and who were older were significantly and independently more likely to report their GP as having discussed and/or performed a PSA test (AOR = 1.94, 95% CI 1.07–3.49, P = 0.04 and AOR = 1.46, 95% CI 1.00–2.13, P = 0.04, respectively) (Table 5). Satisfaction with urinary functioning was not a significant predictor, even when tested in the absence of AUA-SI scores (Table 5).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
To our knowledge, ours is the first study to quantify the influence of patient and GP variables simultaneously on men’s PSA screening status in a general practice setting. We first obtained retrospective data on screening status from patients themselves to determine GP and patient variables associated with previous participation. GPs’ propensity to discuss PSA testing was confirmed as an independent predictor of men’s previous participation in PSA screening.
In contrast, GPs’ beliefs about the evidence underpinning PSA screening were not predictive. This suggests that merely knowing the evidence will be insufficient to align GP behaviour with evidence-based practice. GPs may perform PSA tests as a defensive practice to minimize medico-legal threats perceived inherent in missing opportunities to detect cancers at an early and potentially curable stage.24,25 GPs may screen men in order to meet patient expectations, which have developed in a context of pervasive community faith in prostate cancer screening despite a lack of evidence of its efficacy.26 Furthermore, while a minority of GPs are aware of the paucity of evidence supporting PSA screening, they may be disinclined to abandon screening should it eventually be proven to be effective. Finally, even if GPs divulge the evidence to patients, it may be insufficient to quell men’s enthusiasm for screening.26
Of GP demographic characteristics, only GP age was predictive of previous PSA screening. However, the effect of age was not clearly linear. Different reasons may underlie variations in screening rates across the age range of GPs. Older GPs may view cancer detection as personally salient and, therefore, may be more willing to screen similarly aged patients. Yet it remains unclear why younger GPs are more likely to screen than their ‘middle-aged’ counterparts. One conclusion is that younger GPs are more likely to engage in preventive health care, perhaps irrespective of the evidence for its value.27
With respect to patient variables, age and LUTS significantly and independently predicted PSA screening history, even when adjusting for GP characteristics. These findings are consistent with previous research.9,10 More compellingly, these findings were also reproduced in our prospective component. While the association was modest, men’s age and LUTS may provide ‘cues to action’ for PSA screening. What remains unclear, however, is whether GPs or men themselves are prompted by these cues. While men recalling GPs discussing and/or performing PSA testing in their index consultation were more likely to have attended for a men’s health issue, the majority recalling these behaviours had attended for other reasons. For example, discussions and/or performance of cholesterol testing may be another cue for PSA screening. GPs may perform a cluster of preventive health activities in the one consultation. Indeed, 55% of PSA tests ordered in Australia are done in combination with other pathology tests.28 While previous research suggests that GPs are just as likely to initiate discussions about PSA testing,9 we do not know whether this is the case for this sample. While it is possible that patients may have been cued into initiating discussions as a result of completing questions about PSA testing in the waiting room questionnaire, we deliberately embedded these questions among a larger series to avoid such ‘cuing’.
While advancing age is a recognized risk factor for prostate cancer,1,2 systematic reviews of available evidence have not established an incontrovertible link between LUTS and prostate cancer.19,29 LUTS is prevalent in both men and women over the age of 40.19 An evidence-based approach for the management of LUTS in Australia does not recommend PSA testing to detect prostate cancer.19 Our findings underscore the need for educational materials reassuring both men and their GPs that LUTS does not usually pose a serious threat to men.
That smokers were less likely to have a history of having undergone a PSA screening test may reflect systematic bias against smokers in the provision of preventive care.
Given that the patient and GP characteristics of this sample were similar to the wider population and earlier findings that Australian patients recruited from general practices are representative of the area in which they live,30 our findings are likely to be representative of local general practice. However, the validity of our findings may be limited by our reliance on patient self-report for determining PSA screening. One American study of 221 patients in general practice found concordance between two data sources, namely medical records and patient self-report, in 71% of cases, with no significant differences in the rates of PSA testing between data sourced from patient self-report or medical records.31 In addition, other research has demonstrated that errors in self-report are due in part to problems patients have in recalling tests that occurred within a specific time period, as would be the case when asked to recall, for example, if tested within the last 2 years.32 The authors advise that reports of life-time screening, as ascertained in our study, may be the most accurate self-reported measure of past screening. However, other research indicates that medical records should not be relied upon as a ‘gold standard’ of actual practice.33 Stange et al.33 determined that the sensitivity of PSA testing of medical records was 78% when assessed against direct observation. Hence, medical records are also subject to inaccuracies.
However, our finding of a significant association between patient self-reported participation and GPs’ propensity to screen provides evidence of convergent validity of our measure. Further, it suggests that assessing GP propensity to screen as a proxy for behaviour may be a valid outcome for GP education programmes.
|
Acknowledgments |
|---|
We thank the participating GPs, patients and receptionist staff. We also thank Tracey Bruce for diligent research assistance. MG is supported by an Australian Post-graduate Award. The Central Sydney Area Health Service Ethics Committee (RPAH zone) approved the study protocol.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
1 Smith D. Prostate Cancer in New South Wales in 1972–1994. Sydney: Cancer Council of New South Wales, 1998.
2 Ries L, Eisner M, Kosary C et al. SEER Cancer Statistics Review, 1973–1997. (NH publication number 00-2789). Bethseda (MD): National Cancer Institute, 2000.
3 Smith DP, Armstrong BK, Saunders R. Patterns of Prostate Specific Antigen (PSA) Testing in Australia in 1992 to 1996; An Examination of Medicare Data. Sydney: MSW Cancer Council, 1998.
4 Beemsterboer P, Kranse R, de Koning HJ, Habbema, JD, Schröder FH. Changing role of three screening modalities in the European randomized study of screening for prostate cancer. Int J Cancer 1999; 84: 437–441.[CrossRef][Web of Science][Medline]
5 Schroder FH, Kranse R, Rietbergen J, Hoedemaeke R, Kirkels W. The European Randomized Study of Screening for Prostate Cancer (ERSPC): an update. Members of the ERSPC, Section Rotterdam. Eur Urol 1999; 35: 539–543.[CrossRef][Web of Science][Medline]
6 Ferrini R, Woolf SH. American College of Preventive Practice Policy: screening for prostate cancer in American men. Am J Prev Med 1998; 15: 81–84.[CrossRef][Web of Science][Medline]
7 British Association of Urological Surgeons Working Party. Diagnosis and management of early prostate cancer. Report of a British Association of Urological Surgeons Working Party. BJU Int 1999; 83: 18–33.[CrossRef][Web of Science][Medline]
8 Donovan J, Mills N, Smith M et al. Quality improvement report: improving design and conduct of randomised trials by embedding them in qualitative research: ProtecT (prostate testing for cancer and treatment) study. Br Med J 2002; 325: 766–770.
9 Slevin TJ, Donnelly N, Clarkson JP, English DR, Ward JE. Prostate cancer testing: behaviour, motivation and attitudes among Western Australian men. Med J Aust 1999; 171: 185–188.[Web of Science][Medline]
10 Bunting PS, Goel V, Williams JI, Iscoe NA. Prostate-specific antigen testing in Ontario: reasons for testing patients without diagnosed prostate cancer. Can Med Assoc J 1999; 160: 70–75.[Abstract]
11 Commonwealth Department of Health and Aged Care. General Practice in Australia, 2000. Canberra: Australian Government Printing Service, 2000.
12 Young JM, D’Este C, Ward JE. Improving family physicians’ use of evidence-based smoking cessation strategies: a cluster randomization trial. Prev Med 2002; in press.
13 Ware J. MOS SF36 Health Survey: Manual and Interpretation Guide. Boston: The Health Institute, 1993.
14 Barry MJ, Fowler FJ Jr, O’Leary MP et al. The American Urological Association symptom index for benign prostatic hyperplasia. The Measurement Committee of the American Urological Association. J Urol 1992; 148: 1549–1557.[Web of Science][Medline]
15 Lawrence K. Measurement properties of the AUA symptom score: a methodological clarification. Br J Urol 1996; 77: 175–180.[Web of Science][Medline]
16 Gray M. Psychometric evaluation of the international prostate symptom score. Urol Nurs 1998; 18: 175–183.[Medline]
17 Hosmer DW, Lemeshow S. Applied Logistic Regression. Brisbane: John Wiley and Sons, 2001.
18 Norusis JM/SPSS Inc. SPSS for Windows-based System User’s Guide Release 9.0. Chicago: SPSS Inc., 1999.
19 NHMRC. Clinical Practice Guidelines: The Management of Uncomplicated Lower Urinary Tract Symptoms in Men. Canberra: Australian Government Printing Service, 2000.
20 Jocobsen SJ, Girman CJ, Guess HA, Rhodes T, Osterling JE, Leiber MM. Natural history of prostatism: longitudinal changes in voiding symptoms in community dwelling men. J Urol 1996; 15: 595–600.
21 SAS Institute. SAS Procedures Guide, Version 8.0. North Carolina: SAS Institute, 1999.
22 Ward J, Young J, Sladden M. Australian general practitioners’ views and use of tests to detect early prostate cancer. Aust NZ J of Public Health 1998; 22: 374–380.
23 Public Health Division. The Health of the People of New South Wales—Report of the Chief Health Officer, 2002. Sydney: NSW Department of Health, 2002.
24 Girgis S, Ward JE, Thomson CJH. General practitioners’ perceptions of medicolegal risk. Using case scenarios to assess the potential impact of prostate cancer screening guidelines. Med J Aust 1999; 171: 362–366.[Web of Science][Medline]
25 Austin OJ, Valente S, Hasse LA, Kues JR. Determinants of prostate-specific antigen test use in prostate cancer screening by primary care physicians. Arch Fam Med 1997; 6: 453–458.
26 Wolf AM, Nasser JF, Schorling JB. The impact of informed consent on patient interest in prostate-specific antigen screening. Arch Intern Med 1996; 156: 1333–1336.
27 Rolfe IE, Pearson SA. Screening recommendations in general practice: a survey of graduates from different medical schools. Med J Aust 1996; 165: 14–17.[Web of Science][Medline]
28 Pathology Services Tables Committee. Report on the use of the prostate specific antigen (PSA) test April 2000 http://www.health.gov.au/haf/branch/dtb/psa.htm Last accessed June 12, 2002.
29 Young JM, Muscatello DJ, Ward JE. Are men with lower urinary tract symptoms at increased risk of prostate cancer? A systematic review and critique of the available evidence. BJU Int 2000; 85: 1037–1048.[CrossRef][Web of Science][Medline]
30 Jordan TR, Price JH, King KA, Masyk T, Bedell AW. The validity of male patients’ self-reports regarding prostate cancer screening. Prev Med 1999; 28: 297–303.[CrossRef][Web of Science][Medline]
31 O’Dell KJ, Volk RJ, Cass AR, Spann SJ. Screening for prostate cancer with the prostate-specific antigen test: are patients making informed decisions? J of Fam Pract 1999; 48: 682–688.[Web of Science][Medline]
32 Driver B, Britt H, O’Toole B, Harris M, Bridges-Webb C, Neary S. How representative are patients in general practice morbidity surveys? Fam Pract 1991; 8: 261–268.
33 Stange KC, Zyzanski SJ, Smith TF et al. How valid are medical records and patient questionnaires for physician profiling and health services research? A comparison with direct observation of patients visits. Med Care 1998; 36: 851–867.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. K. Gwede and R. J. McDermott Prostate Cancer Screening Decision Making Under Controversy: Implications for Health Promotion Practice Health Promot Pract, January 1, 2006; 7(1): 134 - 146. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||