Making decisions about treatment for localized prostate cancer
Objective To describe the decision-making processes used by men diagnosed with localized prostate cancer who were considering treatment.
Patients and methods Men newly diagnosed with localized prostate cancer from outpatient urology clinics and urologists' private practices were approached before treatment. Their decision-making processes and information-seeking behaviour was assessed; demographic information was also obtained.
Results Of 119 men approached, 108 (90%) were interviewed; 91% reported non-systematic decision processes, with deferral to the doctor, positive and negative recollections of others' cancer experiences, and the pre-existing belief that surgery is a better cancer treatment being most common. For systematic information processing the mean (sd, range) number of items considered was 4.19 (2.28, 0–11), with 57% of men considering four or fewer treatment/medical aspects of prostate cancer. Men most commonly considered cancer stage (59%), urinary incontinence (55%) and impotence (51%) after surgery, and low overall mortality (45%). Uncertainty about probabilities for cure was reported by 43% of men and fear of cancer spread by 37%. Men also described uncertainty about the probabilities of side-effects (27%), decisional uncertainty (25%) and anticipated decisional regret (18%). Overall, 73% of men sought information about prostate cancer from external sources, most commonly the Internet, followed by family and friends.
Conclusions In general, men did not use information about medical treatments comprehensively or systematically when making treatment decisions, and their processing of medical information was biased by their previous beliefs about cancer and health. These findings have implications for the provision of informational and decisional support to men considering prostate cancer treatment.
Localized prostate cancer is a clinical situation in which, because there is uncertainty about the ideal treatment, the patient's participation in decision making is considered to be critical [1,2]. However, many patients prefer to defer the responsibility for making the decisions about cancer treatment to their clinician . This is particularly so for patients who are men, older and less well educated, with 32–58% of men with prostate cancer preferring that their clinician decide the treatment on their behalf [4,5]. There is also evidence that men may use other non-systematic processes, e.g. previous lay beliefs about health, in deciding their treatment for prostate cancer. For example, Mazur and Hickman  investigated men's preferences for treatment for localized prostate cancer, using a hypothetical prostate cancer scenario that was biased against surgical treatment. The older patients who were asked to choose a treatment preferred surgical therapy, irrespective of the survival benefits. These authors suggested that this finding reflected a lay belief that in the case of cancer, radical surgery was better than conservative treatments. These studies raise important questions about men's use of expert opinions and lay beliefs in their decision making relative to the clinical information about prostate cancer and its treatment. In particular, while informing patients about treatment alternatives and risks is increasingly advocated, the extent to which patients actually use this information in their decision making is less clear.
To date, research on how men make decisions about prostate cancer treatments has been limited largely to hypothetical and retrospective reports [6–10]. Thus, an understanding of how men make treatment decisions at the time of actual diagnosis of localized prostate cancer is needed. Accordingly, the present study used a systematic method of eliciting men's decision-making processes, termed Verbal Protocol Analysis (VPA) . This method was developed specifically by cognitive psychologists to study the underlying strategies and utilities used by decision makers. It involves the use of a structured interview with ‘probe’ questions designed to elicit information from the respondent without biasing either the retrospective recall or the decision processes. A transcription is then coded for ‘prescribed meaning’ units by a researcher not involved in the initial interview. This method was applied to men who were newly diagnosed with localized prostate cancer and who were considering treatment. In addition, the men's information-seeking behaviour, as related to their cancer diagnosis and decision making, were examined with a later interview.
Patients and methods
Participants were men newly diagnosed with localized prostate cancer recruited from two hospital urology clinics and four urologists' private practices in Queensland, Australia. Inclusion criteria were that the men must: (i) have been newly diagnosed with localized prostate cancer and be considering treatment with curative intent; (ii) be able to read and speak English; (iii) have no history of head injury or dementia; (iv) have no concurrent psychiatric illness; and (v) no other concurrent cancer. Men were referred to the project by their urologists if they were judged at the time of diagnosis to have localized prostate cancer suitable for treatment with curative intent, and had no evidence of metastatic disease on scans and X-rays. In all, 131 men were referred consecutively to the project over a 20-month period. Of these men, 12 were ineligible for the study and thus 119 were invited to participate, of whom 111 provided consent. Of these men, two were geographically too remote to participate in the assessment protocol and one man's data was unusable because the audio-tape was inaudible; thus 108 men participated (90% response).
The men's decision-making processes were assessed using VPA , whereby participants were asked to say aloud their thoughts related to their decision making about prostate cancer treatment. This procedure allows researchers to identify which pieces of information are being used in a decision task. VPA also differentiates the use by participants of lay beliefs and non-systematic processes, compared with systematic information processing . The men's responses to the VPA task were audio-taped and later converted verbatim into a typewritten transcript for protocol analysis.
Several steps were taken to ensure the integrity of the data. First, men completed the VPA task immediately after informed consent had been obtained and before any other data were collected, to minimize possible reactivity from other assessments. Second, instructional guidelines were adhered to when eliciting the verbal report, to ensure that the assessment task did not prompt the men to re-code or alter their decision processes . These guidelines included: the researcher not introducing content that the man had not spontaneously recalled; using only paraphrasing and reflection or restatement of the original question to elicit clarification or expand thoughts the man described; using the same words or phrases to discuss treatments that the man had used; using minimal encouraging comments that were not negatively or positively loaded; not using probing ‘why’ or ‘how’ questions. The assessments were undertaken by a psychologist (S.S.) on the research team and a Masters-qualified cancer-care counsellor working under the psychologist's supervision.
In the verbal task men were asked; (i) to describe to the researcher what they were thinking about in making their decision about their prostate cancer treatment; (ii) what was the most important factor they had considered; and (iii) what had been helpful or unhelpful in making their decision (Appendix 1).
After the verbal task a further interview was conducted where participants were asked several demographic questions, whether they had yet made a decision about their medical treatment, and what sources of information they had accessed other than the advice provided by their clinician. Participants later completed several self-reported written psychometric measures for an ongoing prospective study.
The coding scheme for data analysis of the VPA was developed by the research team primarily as a deductive procedure to reflect ‘meaning’ units of interest. First, lay beliefs and biases in decision-making found in previous research and in clinical practice were identified and defined. Second, categories were constructed for the clinical aspects of prostate cancer that would be relevant to a man making a treatment decision. Finally, the lead researchers selected a random subset of verbal protocols that were examined for specific ‘meaning’ units that had not been identified previously, but no such units were found.
In all, six categories of meaning were defined that included: non-systematic processes, systematic information processing, fear of cancer spread, anticipated decisional regret, contextual uncertainty and decisional uncertainty (Appendix 1). The category of non-systematic processes included common decision making ‘rules of thumb’, or heuristics. These included deferral of decision-making responsibility to the doctor (the ‘expert opinion’ heuristic) ; previous experiences and memories (availability heuristic) ; and lay beliefs about cancer treatments and cancer causes . The category of systematic information processing assessed the extent of consideration of medical information, e.g. treatment side-effects, and the clinical aspects of prostate cancer, e.g. cancer stage and grade, as evidence in making a treatment decision. The categories of ‘fear of cancer spread’ and ‘anticipated decisional regret’ were included as possible factors considered by men in their decision making [16,17]. A general category of contextual uncertainty included uncertainty about the probability of cure and the incidence of treatment side-effects. Finally, a category was included for decisional uncertainty. Units from each category formed the basis for segmenting the transcripts, with transcripts coded for the presence of each distinct unit of meaning.
Following previous studies, 30 verbal protocols were randomly selected from the total sample and coded by two independent coders for reliability analysis [18,19]. Inter-rater reliability was assessed using Spearman's rho for ordinal codes and Cohen's kappas for binary codes. For non-systematic processes Spearman's rho was 0.791–0.935 (P < 0.001). For all other category units, Cohen's kappas were calculated with a median (range) score of 0.93 (0.76–1.0) (P < 0.001). On this basis it was concluded that the coding scheme had acceptable reliability. Accordingly, the remaining transcripts were coded by a research assistant who was unaware of the study hypotheses.
The mean (sd, range) age of participants was 62 (8.58, 39–80) years and 89% were married. As their highest level of educational attainment, 15% of men had primary school education (year 7), 39% had completed junior high school (year 10), 14% had completed senior high school (year 12), 18% had attended vocational training and 14% had completed university studies. Of all participants 59% were currently retired, 14% were managers or professionals and the remaining 27% were employed as tradesmen, in clerical positions, in production or transport, or as labourers. At the time of assessment the men were a mean (sd, range) of 4.22 (4.59, 0.2–24) weeks after diagnosis.
In all, 91% of men reported using non-systematic processes in their decision making. Of the total sample, 48% of men reported that they would choose the treatment recommended by their doctor. Consistent with this preference, some men described feeling distressed when their doctor would not make the decision for them. For example, one man reported: “Its hard because you rely on the doctor to tell you what to do and they leave it to you. That's one of the hardest parts.”
Similarly, men described using strategies to elicit the doctor's view, e.g. asking the doctor what he would do if he had prostate cancer. Other men described being relieved when they felt their doctor agreed with or approved of the decision they had made.
In all, 47% of men described thinking of people they knew who had had negative experiences with cancer or cancer treatments (‘negative availability’ heuristic), and using these examples to decide about their own treatment. However, these examples often included people who had other cancers, e.g. women with breast cancer, where the clinical relevance of this experience for the man's individual case is low. Men also revealed content inaccuracies in their understanding of the differences between non-surgical cancer treatments, e.g. chemotherapy and radiation therapy. Thus, some men described examples of people they knew who had unpleasant experiences with chemotherapy as a reason for having excluded radiation therapy.
By comparison, 31% of men considered positive recollections of people with cancer (‘positive availability’ heuristic) as a factor in their decision making. Similarly to negative recollections, these examples frequently were of people who had not had prostate cancer. Men also reported lay beliefs about cancer treatments, in particular the belief that surgery was the best way to cure cancer (34%). Radiation therapy as a treatment category was described as being an uncertain way to cure cancer by 12% of men and 11% thought the best way to cure cancer was with alternative remedies such as dietary changes and vitamins. Only 5% of men reported lay beliefs about cancer causes.
Systematic information processing
The treatment side-effects most commonly considered by men were incontinence and impotence from surgery, and bowel problems and skin burns from radiation therapy (Table 1). Impotence was considered infrequently by men as a side-effect of radiation therapy, and some men reported choosing radiation therapy based on the belief that by comparison with surgery this treatment would preserve their potency. The clinical aspects of prostate cancer considered by men included the stage of their cancer (59%) and the low overall mortality rate associated with prostate cancer (45%). Only 20% of men reported considering the grade of their cancer and their PSA level (21%) in their decision making. The extent of men's systematic processing of the treatment/clinical aspects of prostate cancer was calculated by summing the number of units mentioned in the category of systematic information processing. The mean (sd, range) score was 4.19 (2.28, 0–11), with 57% of men considering four or fewer treatment/clinical aspects of prostate cancer in their decision making (Table 1).
Table 1. Men's systematic information processing of treatment side-effects
|Incontinence (55)||Bowel problems (30)|
|Impotence (51)||Skin burns (21)|
|Surgical mortality (12)||Travelling for treatment (15)|
|Hospital stay (7)||Impotence (9)|
|Pain (7)||Incontinence (8)|
|Social disruption (4)||Pain (4)|
Other decision thoughts
In all, 37% of men described their fear of the cancer spreading and 43% of men described uncertainty about the probabilities of obtaining a cure. To a lesser extent, men described uncertainty about the probabilities of side-effects (27%) and a quarter of men described being uncertain about the best treatment choice. Finally, 18% of men described their fear of making a decision they would later regret.
Interview: information-seeking behaviours and choices
In all, 73% of men had sought information about prostate cancer from sources other than their doctor, most commonly the Internet (40% of all men), followed by family or friends (32%). Of the total group, 23% of men had sought second opinions, most often for an alternative urological assessment or radiation therapy consultation. Of those men who not yet sought a second opinion, 26% intended to do so, primarily for radiation therapy. Finally, 19% of men had joined a prostate cancer self-help group.
At the time of assessment half the men had made a treatment decision. In an unprompted recall of their urological consultation, most men reported that their clinician had discussed at least three treatment options with them, that included watchful waiting (71%), radiation therapy (87%) and radical prostatectomy (92%). Ultimately, 56% of men elected to have a radical prostatectomy, 19% chose radiation therapy and 25% chose watchful waiting.
In the present study men reported high rates of information-seeking about their prostate cancer from other than their clinician. However, despite this, systematic processing was limited on average to only four treatment/clinical aspects of treatment. This finding contrasts with a recent study by Feldman-Stewart et al., where men previously treated for localized prostate cancer who considered 59 items of information about prostate cancer treatment rated on average 32 items as necessary to make a decision. Thus, although men with prostate cancer may retrospectively report needing much information and may show high levels of information seeking, the extent to which patients actively process treatment and clinical information at diagnosis is more limited.
In addition, men's use of non-systematic processes in their decision-making about treatment for localized prostate cancer was pervasive, and revealed content inaccuracies in their understanding of different treatments. This was particularly evident in their understanding of radiation therapy and often arose from the disproportionate effect of previous experiences or beliefs about cancer. Half of the men in the present sample intended to defer decision-making responsibility to their clinician. Consequently, when men found that the responsibility for this decision lay primarily with them, decision making became a source of present and anticipated distress. In summary, the present findings suggest first, that patients do not use information about medical treatments comprehensively or systematically when making treatment decisions, and second, patients' processing of medical information is biased by their previous beliefs about cancer and health.
These findings present challenges for the current trend towards making increasing informational resources available to patients, and encouraging patient autonomy in decision making. While decision aids appear promising in assisting clinicians and patients to make shared decisions, these aids are often expensive to produce, may be time-consuming to use, and findings about their benefits are as yet inconclusive [21,22]. Recently Goel et al. undertook a randomized trial of the benefits of a decision aid for women with breast cancer, and found it produced no significant improvements in women's anxiety, knowledge about breast cancer or decisional satisfaction. These authors suggested that this may be partly because women already had a preferred treatment in mind at baseline. Thus, decision aids may not counter the influence of previous lay beliefs about health on the decision-making process. Further, current decision aids may not take into account the limitations in patients' ability or preparedness to systematically process medical information about treatments at diagnosis [24,25].
The present results reinforce the influential role of the clinician in patient decision-making. They also suggest that patient education that is tailored to individual patients' concerns may be more economical and responsive to each person's capacity for information processing in the context of a threatening medical scenario such as cancer. Finally, the finding that patients' reliance on lay beliefs may bias their decision-making suggests that challenging lay beliefs about cancer where the relevance to the patient's current diagnosis is low may be needed.
A limitation of the present study is that all patients had been seen by urological surgeons and so may have received a more detailed and positive impression of surgery than for radiation therapy. However, the nature of the lay beliefs reported in this study are consistent with previous research. In addition, men in the present study appeared to have been informed about the negative consequences of surgery and ultimately 44% of men chose not to undergo surgery. Future research to assess men's decision processes after consultation for radiation therapy may ascertain the durability of negative lay beliefs about this treatment.
The assessment of patient decision-making will continue to be an important area of research as health consumerism advances and therapeutic options for patients with serious illness increase. However, to date there is little understanding of how patients use health information in their decision making . The present study is the first to document how men actually decide on the treatment for localized prostate cancer. Significant advances in decision support for this and other patient groups may not emerge until an understanding of patients' actual decision behaviours and preferences is integrated into practice.
We gratefully acknowledge the assistance of the Queensland Cancer Fund and the Royal Brisbane Hospital and Redcliffe Hospital in the undertaking of this study. This study was undertaken as part of the requirements for Suzanne Steginga's doctoral candidature.
S. Occhipinti, School of Applied Psychology (Health), Griffith University, Mt. Gravatt Campus, Nathan QLD 4111, Australia.
Appendix 1: Examples of category units
Expert opinion heuristic
Definition: judgement influenced by the rule ‘experts can be trusted’
Example: ‘I’ve always been a great believer in doctors. Whatever the doctor tells me to do that's the way I'll go. They know what they are doing, I don't.'
Definition: when a judgement or decision is mediated through an instance that comes easily to mind. May refer to an event with a positive or negative outcome.
Example of an availability heuristic with a positive outcome: ‘My mother had breast cancer and she lived for about 45 years afterward (surgery) … In that case I’d prefer to have it out because my mother, she had both breasts removed.
Example of an availability heuristic with a negative outcome: ‘Well with the radiation therapy, I wasn’t fussed on that … my auntie died from cancer and after her radium; it wasn't a very pleasant experience.
Lay beliefs about cancer treatments
Definition: Belief about cancer treatments that is not based on scientific data.
Example: ‘It was a little bit foreign I thought (radiation therapy). A little bit foreign to the human body … I think that the fact that surgery is much more prominent, its been part of our lives, all of our life really. It’s not foreign.
Lay beliefs about cancer causes
Definition: Belief about cancer causes that is not based on scientific data.
Example: ‘I have been fairly run down physically over the last few years. I do feel that this contributes to getting cancer.
Systematic information processing
Definition: Consideration of treatment side-effects and clinical aspects of cancer.
Example: ‘The surgeon said, well you’re going to be impotent – that's a worry! At my stage of life it's of less importance than if I was 20 years younger, or 10 years younger. But it's still a concern.'
Fear of cancer spread
Definition: Fear of the prostate cancer metastasizing.
Example: ‘It’s early stages at the moment but in five years time who knows? It might have blown up and gone into the bone and all sorts of things, and I thought I can't wait – so I'll have it done now.
Anticipated decisional regret
Definition: The fear of making a decision that will later be regretted.
Example: ‘My doctor gave me three options and what appeals to me is to do nothing. But then I could come to regret that in years to come.’
Definition: Uncertainty about the probabilities of cure or side-effects.
Example: ‘Well, one doctor told me the percentages were 60/40 cure with radium therapy, and then another doctor said chop that back to 50/50. So then I went to see the doctor at the radium institute and he said 80%. So it’s bit confusing isn't it?'
Definition: Feeling uncertain about what choice to make.
Example: ‘One minute I think, I’ll go surgery. Then I read something else and I think no, that's not the way the go. I just don't where I am at with this to tell you the truth.'