What is a “good” treatment decision?: Decisional control, knowledge, treatment decision-making, and quality of life in men with clinically localized prostate cancer

Abstract

Objective

We explored whether active patient involvement in decision making and greater patient knowledge are associated with better treatment decision making experiences and better quality of life (QOL) among men with clinically localized prostate cancer. Localized prostate cancer treatment decision-making is an advantageous model for studying patient treatment decision-making dynamics as there are multiple treatment options and a lack of empirical evidence to recommend one over the other; consequently, it is recommended that patients be fully involved in making the decision.

Methods

Men with newly diagnosed clinically localized prostate cancer (N=1529) completed measures of decisional control, prostate cancer knowledge, and their decision-making experience (decisional conflict, and decision-making satisfaction and difficulty) shortly after they made their treatment decision. Prostate cancer-specific QOL was assessed 6-months after treatment.

Results

More active involvement in decision making and greater knowledge were associated with lower decisional conflict and higher decision-making satisfaction, but greater decision-making difficulty. An interaction between decisional control and knowledge revealed that greater knowledge was only associated with greater difficulty for men actively involved in making the decision (67% of sample). Greater knowledge, but not decisional control predicted better QOL 6-months post-treatment.

Conclusion

Although men who are actively involved in decision making and more knowledgeable may make more informed decisions, they could benefit from decisional support (e.g., decision-making aids, emotional support from providers, strategies for reducing emotional distress) to make the process easier. Men who were more knowledgeable about prostate cancer and treatment side effects at the time they made their treatment decision may have appraised their QOL as higher because they had realistic expectations about side effects.

For most men diagnosed with clinically localized prostate cancer, there are multiple clinically appropriate intervention or management strategies, but they involve tradeoffs between side effects.^{1, 2} Consequently, there is consensus in the medical, academic, and policy communities that “good” treatment decisions for clinically localized prostate cancer are informed and consistent with patients’ preferences and values.^3–6 Good treatment decisions can also be set apart by their outcomes– satisfaction with the decision and little or no regret about the choice,⁶ and, potentially, better psychological well-being in survivorship.⁷ To increase the likelihood that treatment decisions are consistent with patient priorities and values, the current standard of care requires physicians, at the very least, to involve patients in collaborative decision-making.⁸ According to models of shared decision-making,^9–11 patients and physicians should collaboratively identify the problem to be solved. Patients should be fully informed about treatment options, benefits and drawbacks, clinical indicators and recommendations and have opportunities to clarify and communicate preferences to their physicians. There should also be periodic feedback to check understanding on the part of patients and follow up with respect to implementation of the decision.

Beyond the ethical argument for patient involvement in treatment decisions, some, although not all¹² empirical evidence indicates that patient involvement fosters better treatment decisions.¹³ Greater patient participation in treatment decisions has been associated with higher satisfaction with the decision, at least among younger prostate cancer patients,¹⁴ and higher quality of life among breast cancer survivors.¹³ Lack of control over treatment decisions has been associated with more decisional conflict in HIV/AIDS patients making medication decisions.¹⁵

The shared decision-making paradigm assumes that patients are knowledgeable about treatment options and potential risks and benefits.^{9, 16} However, to date, the idea that knowledge improves treatment decision-making, especially when patients are actively involved in decision making, has largely been assumed. Kaplan and colleagues studied 70 men, most of whom had low socioeconomic status and found that lower knowledge was associated with higher decisional conflict;¹⁷ however, it is unknown whether this would be true of a larger sample that varies with respect to socioeconomic status. In a second sample that included, but was not limited to prostate cancer patients, prostate cancer patients’ knowledge about treatment was not associated with a measure of concordance between values/preferences and treatment choice (decisional dissonance).³ Shared treatment decision-making is becoming institutionalized through policy and practice and resources are continuing to be allocated for developing decision-aids,^{18, 19} a primary component of which is increasing men’s knowledge.²⁰ A valuable addition to the science and practice of shared decision making would be improving our understanding of the extent to which knowledge contributes to decision-making experiences and survivorship outcomes, especially among those actively involved in making a treatment decision.

The goal of the present study was to test whether men’s control over the treatment decision and level of prostate cancer knowledge were associated with better treatment decision-making experiences and better well-being in survivorship.

We hypothesized that patients who exert more control over the decision-making process and are more knowledgeable about prostate cancer and treatment side effects would experience less decisional conflict and will be more satisfied with the decision-making process. However, given evidence that many men and their families experience prostate cancer treatment decision making as challenging,^21–23 we predicted that greater engagement in the process, indicated by higher decisional control and knowledge, would be associated with greater decision-making difficulty. It is possible that knowledge has a larger impact on treatment decision making to the extent that patients are actively involved in the treatment decision-making process. We therefore hypothesized-that decisional control would moderate associations between knowledge and the decisional outcomes and that knowledge would be more strongly associated with decisional outcomes for men who made the decision actively or collaboratively than for men who had little or no input in the decision.

Quality of life ratings are appraisals that could be influenced by a variety of patient beliefs. Cancer patients who expect to experience a given side-effect are indeed more likely to experience the side effect.²⁴ We reasoned that the degree to which men feel that they are responsible for their treatment decisions and outcomes may also influence their post-treatment appraisals of side-effects. We know when they make hypothetical treatment decisions, people justify their choices by inflating the probability of treatment success.²⁵ After they have been treated, people may engage in similarly biased information processing and appraisals to reduce cognitive dissonance between their choice and their actual outcomes (i.e., side effects). We expected that being more knowledgeable and participating to a greater extent in treatment decision-making would reduce decisional conflict and increase decision-making satisfaction and this in turn would be associated with higher ratings of QOL. Our reasoning was that the same men who were most satisfied with their decision would be motived, even if they experienced side effects, to continue to believe that they had made the best possible decision under the circumstances. Six months after treatment, in order to justify their decision, these men might minimize any side effects that they were experiencing and report higher QOL.

Methods

Procedure

Study procedures were Institutional Review Board-approved. Data for the current study are from a multi-site longitudinal observational study of men who were recruited shortly after being diagnosed with clinically localized prostate cancer. Participants were recruited from five clinical facilities (2 academic cancer centers and 3 community practices) between July 2010 and September 2014. They were typically recruited at the follow-up visit after having a positive biopsy or when seeking a second opinion. Men would typically have received some type of decision counseling at these consultations, although many would also seek additional opinions after the consultation. For the present analyses we used data (demographic and clinical) from a baseline questionnaire that was completed at, or shortly after consent, and prior to the start of treatment. We also used data (decisional control, knowledge, and decision-making outcomes) from a treatment questionnaire that was completed after participants made their treatment decision but before they started treatment. For some men, making the decision took some time. An average of 26.5 days passed between the receipt of the baseline questionnaire and the treatment decision-making questionnaire. Prostate cancer-specific QOL was ascertained from a questionnaire administered and returned 6-months after treatment.

Participants

We approached 3337 participants and of these, 74.2% (n = 2476) were enrolled in the study. Participants were given the first survey after informed consent and completed it in the clinic (39.3%), at home and returned it by mail (59.9%) or with a research staff member over the phone (0.8%). The response rate for the first questionnaire was 81.1% (n = 2008). Participants (n = 1654) were only included if they also completed a second survey enquiring into their treatment decision-making experience. This survey was completed after they made their decision, but prior to treatment. Decisional experiences were analyzed for 1529 participants who had data on the variables included in the decisional conflict, decision-making satisfaction, and difficulty multivariable models. Of these participants, 1342 had 6-month follow up data at the time the data were analyzed.

Measures

Predictor Variables

Prostate cancer knowledge was assessed with a 17 (range = 0–17) item prostate cancer and treatment side effects knowledge scale. Participants responded true/false/don’t know to 13 items from Diebert et al. that assess general prostate cancer knowledge (e.g., ‘a man can have prostate cancer without having any pain or symptoms’)²⁶ and 4 author-created items added to assess knowledge of treatment side effects of radical prostatectomy and external beam radiation (e.g., ‘radiation treatment of prostate cancer can cause rectal pain or discomfort’). “Don’t know” responses were recoded as incorrect. The total number of correct responses were summed to generate scores. Internal reliability was not calculated as this scale does not assess a single underlying construct.

Decisional control was assessed with a question that asked participants to report how much control they had over their treatment decision using response options adapted from Degner and Sloan’s (1992) assessment of decision-making role preferences²⁷ (1 = ‘My doctor(s) made the decision with little input from me’/2 = ‘My doctor(s) made the decision but seriously considered my opinion’/3 = ‘My doctor(s) and I made the decision together’/4 = ‘I made the treatment decision after seriously considering the opinion of my doctor(s)’/5 = ‘I made the treatment decision with little input from my doctor(s)’. Responses for the first two and the last two options were collapsed yielding passive, collaborative, and active decision-making categories.

Outcome Variables

Decisional conflict was assessed with the 16-item Decisional Conflict Scale.²⁸ Subscales assess the degree to which participants felt informed, supported, and uncertain about the decision, experienced values clarity and thought the decision was effective. Participants responded to 16 questions such as, ‘are you clear about which benefits matter most to you?’ using a 5-point Likert-type response format. According to the author instructions, scale scores were computed by summing item values, dividing by 16 and multiplying by 25 to yield scores potentially ranging from 0 to 100,²⁹ with higher scores indicating greater decisional conflict (α = 0.89). Reliability for the informed, values clarity, support, uncertainty, and effective decision subscales were α = 0.82, 0.90, 0.66, 0.59, and 0.80, respectively.

Decision-making satisfaction was assessed using a modified version of the Satisfaction with Decision Scale³⁰ that included four (α = 0.87) of the original six items. Participants responded to four statements using a 5-point Likert-type response format (1=strongly disagree, 5=strongly agree). Scores were averaged (range=1–5) and higher scores indicate higher decision-making satisfaction.

Decision-making difficulty was assessed with 3 items (α = 0.72)²³ for which participants rated the extent to which they agreed (1=strongly disagree, 5= strongly agree) with the following statements: ‘making the decision about the type of treatment to have was stressful,’ ‘it was difficult to make the decision about what treatment to have,’ and ‘knowing the opinions of family members made it more difficult for me to decide what kind of treatment to have.’ Scores were summed (range=3–15), with higher scores indicating a more difficult decision-making process.

Quality of Life (QOL) was assessed with the Expanded Prostate Cancer Index Composite (EPIC), a 50-item prostate cancer health-related quality of life (HRQOL) scale (α≥0.82).³¹ The EPIC assesses both function (how frequently one has been affected by a treatment-related side effect during the previous 4 weeks) and bother by urinary, bowel, sexual, and hormone-related side effects (‘how big a problem’ were these side effects) during the same time-period. We used outcome scores that combined both function and bother. The hormonal scale was not analyzed for this study as relatively few of our participants received androgen deprivation therapy. Scores for each domain can range from 0 to 100.

Covariates

Research assistants recorded the site at which participants were recruited. Participants self-reported years of education completed (high school or less, some college, college, and beyond college), income (<$25,000, $25,000–49,999, $50,000–74,999, $75,000–99,999, and ≥$100,000, marital status (married/cohabitating versus single/never married/divorced/widowed), employment status (full-time/part-time/unemployed/retired), age at diagnosis, and race/ethnicity (non-Hispanic White/non-Hispanic Black/Hispanic/other). They completed the MacArthur perceived social status assessment by rating their standing in their community on a ladder graphic representing social standing from low to high.³² Each ladder rung corresponded to a response option ranging from 1 to 10. Treatment received (active surveillance vs. surgery vs. external beam radiation, proton, or brachytherapy) was ascertained via self-report and verified via chart abstraction in the majority of cases.

Statistical Analyses

We conducted multivariable linear regressions with robust standard errors to test for adjusted associations between decisional control and knowledge and the outcomes, including decisional conflict subscales. We estimated the association between decisional control and knowledge with a Pearson correlation, and evaluated whether the two interacted to predict the outcomes using multivariable linear regressions with robust standard errors. Recruitment site, years of education, race/ethnicity, age at diagnosis, marital status, employment status, and perceived social status were also included as covariates in all multivariable models. Given there were so few sites (2 comprehensive cancer centers and 3 community facilities) we did not attempt to make any comparisons on the basis of type of facility or geographical location. Income was not included in the model due to a high percentage of missing data (14.4%). Furthermore, there were moderate to strong correlations between income and social status (r = 0.36, p < .001), and income and education (r = 0.44, p < .001), which were included in the multivariable models. We tested whether level of prostate cancer knowledge at the time of the decision and decisional control predicted QOL six months after treatment or the initiation of active surveillance. These tests were performed on a subset of participants, as only 1342 participants had six-month data at the time the analyses were conducted and some participants had missing data on variables included in the models. As baseline QOL and type of treatment received have significant influences on QOL after treatment, we controlled for these, along with the covariates included in the other multivariable models.

Results

Participant Characteristics

Demographic and clinical characteristics of the sample are reported in Table 1, along with mean knowledge scores as a function of participant characteristics. The majority of the sample was Non-Hispanic White (81.6%), married (84.0%), and more than half had a college degree or greater (57.4%). Mean age at diagnosis was 63.1 (SD = 7.9). Most of the men reported that they had made the decision on their own or with their physicians’ input (actively; 66.8%) or collaboratively with their physician(s) (26.4%). A minority reported that their physicians made their decisions with or without their input (passively; 6.5%). The mean knowledge score across the sample was 11.72 (SD = 3.26) out of 17. Mean decisional conflict was 8.05 (SD = 10.36) out of 100, mean decision making satisfaction was 4.55 (SD = 0.53) out of 5, and mean difficulty was and 8.71 (SD = 2.72) out of 15. Among those with 6-month follow up data and confirmed treatment type data, 22.3% (n=294) received active surveillance, 26.8% (n = 353) received radiation, and 51.0% (n = 672) received surgery. Mean urinary, sexual, and bowel QOL scores were 83.9 (SD=14.6), 40.0 (SD = 27.0), and 93.3 (SD = 9.2) out of 100.

Table 1.

Participant characteristics and mean prostate cancer knowledge as a function of participant characteristics (N=1,529)

Characteristic	N	% or mean (SD)	Mean knowledge (SD/95% CI)
Education
≤High school	444	29.04	10.45 (3.37)
Some college	207	13.54	11.20 (3.47)**
College degree	408	26.68	12.27 (2.99)***
Graduate degree	470	30.74	12.66 (2.83)***
Marital status
Not married/cohabitating	245	16.02	11.05 (3.63)
Married/cohabitating	1284	83.98	11.84 (3.16)***
Income
<25,000	86	6.57	9.14 (4.01)
25,000–49,999	147	11.23	10.55 (3.08)**
50,000–74,999	196	14.97	11.68 (3.08)***
75,000–99,999	185	14.13	11.94 (2.87)***
≥100,000	695	53.09	12.46 (2.93)***
Race
Non-Hispanic White	1248	81.62	11.95 (3.18)
Non-Hispanic Black	157	10.27	10.43 (3.28)***
Hispanic	102	6.67	10.71 (3.57)***
Other	22	1.44	12.36 (3.16)
Employment status
Full time	782	51.14	12.22 (3.03)
Part time	112	7.33	11.63 (2.99)
Unemployed	39	2.55	11.56 (2.94)
Retired	596	38.98	11.08 (3.49)***
Age at diagnosis	1529	63.14 (7.90)	b = −0.09*** (−0.11, −0.07)
Perceived social status	1529	6.78 (1.67)	b = 0.23*** (0.13, 0.33)
Treatment received	1319
Active surveillance	294	22.29	11.80 (3.60)**
Radiation	353	26.76	11.00 (3.29)
Surgery	672	50.95	12.31 (2.86)***
QOL
Urinary	1340	83.92 (14.56)	b = 0.02*** (0.01, 0.03)
Sexual	1318	39.98 (27.00)	b = 0.01 (−0.00, 0.02)
Bowel	1342	93.33 (9.17)	b = 0.06*** (0.04, 0.07)

Notes: Percentages for a given variable will not sum to 100% if cases were missing data for the variable. Referent groups for comparisons of knowledge as a function of participant characteristic were ≤ high school, not married, income <25,000, non-Hispanic White, being employed full time, receiving radiation;

^*p<.05,

^**p<.01,

^***p<.001.

Differences in knowledge as a function of participant characteristics can also be found in Table 1. There were also a number of differences in decisional control as a function of participant characteristics. Men who were married were more likely to have made the decision collaboratively (RR = 2.28, 95% CI 1.35, 3.84, p = .002) and actively (RR = 2.04, 95% CI = 1.27, 3.27, p = .003) than passively compared to those who were unmarried. Those who were employed part-time (RR = 0.44, 95% CI = 0.21, 0.93, p = .03) or retired (RR = 0.53, 95% CI = 0.34, 0.83, p = .006) were less likely to have made the decision actively than passively compared to those who were employed full-time. Older age was associated with lower likelihood of making the decision actively (RR = 0.96, 95% CI = 0.94, 0.99, p = .005) compared to passively. There were no differences in how men made the decision as a function of education, race/ethnicity, recruitment site, or perceived social status. Correlations between predictors and outcomes are shown in Table 2.

Table 2.

Correlations between predictor and outcome variables

	Decisional Control	Knowledge	Decisional conflict	Decision-making satisfaction	Decision-making difficulty	Sexual Quality of Life	Urinary Quality of Life	Bowel Quality of Life
Decisional control	1.00
Knowledge	0.22***	1.00
Decisional conflict	−0.16***	−0.16***	1.00
Decision-making satisfaction	0.10***	0.10***	−0.54***	1.00
Decision-making difficulty	0.10***	0.15***	0.28***	−0.23***	1.00
Sexual Quality of Life	0.05	0.17***	−0.07*	0.08**	−0.01	1.00
Urinary Quality of Life	−0.03	0.05	−0.09***	0.09**	−0.06*	0.43***	1.00
Bowel Quality of Life	0.07*	0.16***	−0.10***	0.07**	−0.06*	0.27***	0.35***	1.00

^*p<.05,

^**p<.01,

^***p<.001

Multivariable Analyses

Decisional control and decision-making experiences

Adjusted models of decision-making outcomes as a function of decisional control are found in Table 3. Participants who made the decision collaboratively (b = −4.88, 95% CI = −7.88, −1.88, p = .001) or actively (b = −6.62, 95% CI = −9.47, −3.78, p < .001) reported less decisional conflict than those who were passive (Model 1). For most decisional conflict subscales (informed, values clarity, support, effectiveness, and uncertainty), making the decision more actively compared to being passive, was significantly associated with better (lower) decisional conflict sub-scale scores. Comparisons between collaborative and passive, and active and passive decision makers all yielded significant results (p-values ≤ .05) except three effects for uncertainty and effectiveness that only reached a trend level (p ≤ .07). Collaborative (b = 0.24, 95% CI = 0.12, 0.36, p < .001) and active decision makers (b = 0.25, 95% CI = 0.14, 0.36, p < .001) were more satisfied with the decision-making process than those who were passive (Model 2). However, active (b = 0.72, 95% CI = 0.22, 1.22, p = .005), but not collaborative decision makers, reported more difficulty than those who were passive (Model 3).

Table 3.

Multivariable analyses of decisional control and decision-making outcomes

Predictors	Model 1 b (95% CI) n=1517	Model 2 b (95% CI) n=1524	Model 3 b (95% CI) n=1515
	Outcomes
	Decisional conflict	Decision-making satisfaction	Decision-making Difficulty
Decisional control
Collaborative	−4.88** (−7.88, −1.88)	0.24*** (0.12, 0.36)	0.25 (−0.28, 0.79)
Active	−6.62*** (−9.47, −3.78)	0.25*** (0.14, 0.36)	0.72** (0.22, 1.22)
Education
Some college	0.91 (−0.90, 2.72)	0.04 (−0.05, 0.12)	−0.12 (−0.56, 0.32)
College degree	0.84 (−0.59, 2.26)	0.04 (−0.04, 0.11)	−0.06 (−0.43, 0.31)
Graduate degree	0.26 (−1.15, 1.68)	0.06 (−0.01, 0.14)	−0.18 (−0.55, 0.19)
Married/cohabitating	−2.81** (−4.41, −1.21)	0.11* (0.03, 0.19)	−0.69*** (−1.07, −0.31)
Race
Non-Hispanic Black	1.20 (−0.77, 3.15)	−0.06 (−0.15, 0.03)	0.15 (−0.29, 0.60)
Hispanic	0.82 (−1.51, 3.14)	−0.06 (−0.17, 0.06)	0.48 (−0.19, 1.15)
Other	3.69 (−3.07, 10.45)	−0.05 (−0.30, 0.20)	0.94 (−0.21, 2.09)
Employment status
Part time	1.79 (−0.77, 4.35)	−0.04 (−0.16, 0.08)	0.32 (−0.26, 0.88)
Unemployed	−1.77 (−5.70, 2.17)	0.06 (−0.11, 0.23)	0.56 (−0.20, 1.32)
Retired	−1.05 (−2.42, 0.32)	0.03 (−0.04, 0.10)	0.25 (−0.10, 0.60)
Age	0.06 (−0.03, 0.16)	−0.00 (−0.01, 0.00)	−0.06*** (−0.08, −0.04)
Perceived social status	−0.26 (−0.60, 0.08)	0.01 (−0.00, 0.03)	−0.02 (−0.11, 0.07)

Notes: Referent groups were passive decisional control, ≤ high school, not married, non-Hispanic White, and a full time employment status;

^*p<.05,

^**p<.01,

^***p<.001. Facility at which participants were recruited was included in all of the models; however, output for this variable was not included as comparisons between sites are arbitrary.

Prostate cancer knowledge and decision-making experiences

Adjusted models of the associations between knowledge and decision-making outcomes are found in Table 4. Being more knowledgeable about prostate cancer was associated with lower decisional conflict (b = −0.49, 95% CI = −0.68, −0.29, p < .001) (Model 4). Having more knowledge was associated with lower scores on all of the decisional conflict subscales except uncertainty (p-values ≤ .02). Having more knowledge was associated with higher decision-making satisfaction (b = 0.01, 95% CI = 0.00, 0.02, p = .03) (Model 5). Being more knowledgeable was associated with greater decision-making difficulty (b = 0.12, 95% CI = 0.08, 0.17, p < .001) (Model 6).

Table 4.

Multivariable analyses of knowledge and decision-making outcomes

Predictors	Model 4 b (95% CI) n=1522	Model 5 b (95% CI) n=1529	Model 6 b (95% CI) n=1520
	Outcomes
	Decisional conflict	Decision-making satisfaction	Decision-making Difficulty
Knowledge	−0.49***(−0.68, −0.29)	0.01* (0.00, 0.02)	0.12*** (0.08, 0.17)
Education
Some college	1.19 (−0.60, 2.98)	0.03 (−0.05, 0.12)	−0.20 (−0.64, 0.25)
College degree	1.04 (−0.39, 2.47)	0.04 (−0.04, 0.12)	−0.17 (−0.55, 0.20)
Graduate degree	0.70 (−0.74, 2.14)	0.06 (0.01, 0.14)	−0.35 (−0.73, 0.03)
Married/cohabitating	−2.76**(−4.38, −1.14)	0.11** (0.03, 0.19)	−0.73*** (−1.10, −0.36)
Race
Non-Hispanic Black	0.69 (−1.30, 2.68)	−0.05 (−0.14, 0.04)	0.28 (−0.16, 0.73)
Hispanic	0.65 (−1.66, 2.96)	−0.06 (−0.17, 0.06)	0.54 (−0.13, 1.21)
Other	3.55 (−2.94, 10.03)	−0.05 (−0.30, 0.21)	0.97 (−0.19, 2.13)
Employment status
Part time	2.12 (−0.45, 4.69)	−0.05 (−0.16, 0.07)	0.27 (−0.30, 0.84)
Unemployed	−1.33 (−5.48, 2.82)	0.05 (−0.13, 0.23)	0.48 (−0.28, 1.24)
Retired	−0.94 (−2.32, 0.43)	0.02 (−0.05, 0.09)	0.24 (−0.10, 0.59)
Age	0.04 (−0.05, 0.13)	−0.00 (−0.01, 0.00)	−0.05*** (−0.08, −0.03)
Perceived social status	−0.26 (−0.60, 0.08)	0.01 (−0.00, 0.03)	−0.03 (−0.12, 0.06)

Notes: Referent groups were passive decisional control, ≤ high school, not married, non-Hispanic White, and a full time employment status;

^*p<.05,

^**p<.01,

^***p<.001

Covariates and decision-making outcomes

There were consistent patterns in relations between covariates and decision-making outcomes across Models 1–6. Married men fared better with respect to all decision-making outcomes and older age was associated with experiencing lower decision-making difficulty.

Decisional control moderates associations between knowledge and decision-making outcomes

There was a significant interaction between decisional control and knowledge predicting decision-making difficulty (b = 0.28, 95% CI = 0.13, 0.43, p < .001), but not predicting decisional conflict or decision-making satisfaction. The interaction predicting decision-making difficulty was due to there being an association between having greater knowledge and decision-making difficulty in men who made the decision actively (b = 0.17, 95% CI = 0.11, 0.23, p < .001), but an association between greater knowledge and less decision-making difficulty in men who were passive (b = −0.17, 95% CI = −0.35, 0.00, p = .05). Of note, there was no relationship between prostate cancer knowledge and decision-making difficulty among men who made the decision collaboratively (b = 0.05, 95% CI = −0.04, 0.14, p = .28).

Knowledge, decisional control, and quality of life

We modeled 6-months post-treatment QOL after controlling for treatment choice, baseline QOL, and demographic and clinical characteristics (results not shown in tables). Prostate cancer knowledge predicted sexual QOL six months post treatment (b = 0.49, 95% CI = 0.13, 0.86, p = .008)¹. Decisional control did not (ps > .22). Knowledge also predicted bowel QOL (b= 0.18, 95% CI = 0.04, 0.32, p = .01), whereas decisional control did not (ps > .21). Neither knowledge nor decisional control predicted urinary quality of life (ps > .43).

Controlling for treatment type, baseline quality of life and other covariates, the indirect effect of knowledge on bowel QOL through decision making difficulty was small but reliable (−0.2, 95% CI = −0.05, −0.00, p=.04) and the indirect effect of knowledge on bowel QOL through decisional conflict was not significant (0.02, 95% CI = −0.00, 0.04, p=.07). The indirect of knowledge on sexual QOL through decisional conflict was not significant (−0.00, 95% CI = −0.06, 0.06, p=.94).² We did not test a mediation model for decision-making difficulty and sexual quality of life because the two were not associated. Decision-making satisfaction did not predict sexual or bowel QOL (ps > .11), therefore we did not test if it mediated relationships between knowledge and QOL.

Of all covariates in the multivariable models, the type of treatment men had received had the largest impact on QOL. Adjusting for knowledge and covariates, compared to men on active surveillance, men who had received surgery had lower sexual (29 points lower), urinary (10 points lower), and bowel (1 point lower) QOL. Men who received radiation also had significantly lower sexual (16 points lower), bowel (4 points lower), and urinary (4 points lower) QOL than men followed with active surveillance.

Discussion

The hypothesized main effects for decisional control/knowledge and the decision-making outcomes were supported. For men deciding how to treat their prostate cancer, the more decisional control they had, the less decisional conflict they experienced and the more satisfied they were with the decision-making process. However; more actively involved men rated the decision as having been more difficult. We found a similar pattern of results for knowledge; men who were more informed about prostate cancer reported less decisional conflict and greater decision-making satisfaction, but greater difficulty with the decision-making process.

Knowledge and decisional control interacted to predict decision-making difficulty; being more knowledgeable was only associated with experiencing more treatment decision-making difficulty for men who were most actively involved in making their treatment decision. However, this was the largest subgroup of men. They may have been more active in the decision-making process and consequently sought prostate cancer information more widely and extensively and spent more effort and time comparing treatment options. While these individuals may ultimately have low decisional conflict and may be satisfied with the decision-making process, they may also have found decision-making difficult and stressful, given how effortful it was. Another interpretation of the results is that men’s judgments of their decision-making satisfaction and conflict are motivated by a need to reduce cognitive dissonance or discrepancies between their attitudes and behavior. When people put more effort into a task they often evaluate it more favorably;³³ in this case, if they put considerable effort into making the treatment decision, they may perceive their decision as better justified and are more satisfied with the decision-making process. Although we hope that the decision-making conflict and satisfaction measures are capturing truly well-justified decisions, future research could attempt to rule out the cognitive dissonance hypothesis by measuring both patient knowledge and decision-making effort and determining their independent effects on decisional conflict and decision-making satisfaction.

Across all multivariable models, being married was associated with greater decision-making satisfaction and lower decisional conflict and decision-making difficulty. Marital status was unrelated to QOL. Married men diagnosed with PCa are more likely to choose more aggressive treatment, in particular surgery, than unmarried men.^{34, 35} Their family roles and wives’ social influence may lead them to prioritize choosing a treatment that they believe will maximize their chance of cure and longevity. In contrast, some unmarried men may place relatively greater value on sexual function, making potential erectile dysfunction more threatening to their identities and lifestyles. Consequently, married men may not be as conflicted about the treatment choice as single men who may weigh the costs of treatment more heavily and therefore have a more negative decision-making experience. Married men likely also have a more positive decision-making experience than unmarried men because of the social support afforded by marriage.

Breast cancer survivors who play a more active role in decision-making have been found to report higher physical and social QOL compared to those who were less active¹³ as well as better psychological well-being after treatment.³⁶ Whether there is a relationship between decisional control and QOL has hitherto not been tested in men with prostate cancer. Decisional control was not associated with QOL in our sample. A reason for the divergent findings may be that the women in Hack et al.’s study were reassessed 3 years after surgery; whereas our sample was only 6-months post-treatment when we assessed QOL. Perhaps over time, as side effects emerge or fail to subside, prostate cancer patients perceptions of their QOL might come to be more influenced by misgivings about treatment decision-making outcomes and processes. Being more knowledgeable about the disease and treatment side effects was associated with higher sexual and bowel QOL six months after treatment; however, little or no part of these relationships were explained by the decision-making experience. Perhaps more knowledgeable men make more positive appraisals of QOL because they have more realistic expectations about the likelihood and time-course of side effects. Consistent with this idea, educational interventions that improve prostate cancer knowledge have been shown to reduce bother by sexual problems³⁷ and side effects of radiation.³⁸

The field has embraced shared decision-making as the ideal for preference sensitive treatment decision making,^39–41 and shared decision-making is becoming institutionalized through health policy changes.¹⁸ Patients too, appear to have embraced shared or autonomous decision making; only 6.5% of our sample reported having been passive in their decision-making, a smaller proportion than previously reported.^{14, 42, 43} Consequently, it is important to continue to develop and invest in strategies that both increase patient knowledge and reduce the psychological burden of treatment decision making.⁴⁴ One important implication of our findings is that patient participation and knowledge are important for good prostate cancer treatment decisions, but they are not sufficient. The treatment decision may remain difficult for many men and their families. Treatment decision-making can be difficult for a third or more of men diagnosed with the disease.^{22, 23} Also, when shared decision-making involves family members, couple- or family-centered support may be beneficial⁴⁵ as family and other support people often have unmet informational needs.^{46, 47} Support at this point in the cancer care continuum could benefit many. In essence, interventions are needed to help patients and their families manage what may be the paradoxical nature of ‘good’ prostate cancer treatment decisions. In order to make an informed choice between two or more treatment options, patients presumably need a high level of knowledge about treatment procedures and potential side effects. They, therefore, are likely to need support gathering, processing, and integrating the information they need in order to perceive that they have an adequate basis for choosing between treatment options. Psychosocial support from nurses, social workers, and psychologists may be valuable.⁴⁸ As might be increased use of decision aids.^{49, 50}

In light of a recent meta-analysis of 14 decision-aid trials having revealed mixed effects for the effects of decision aids on outcomes such as decisional conflict and decision-making satisfaction,⁵⁰ supporting prostate cancer treatment decision-making in patients and their families remains an area in need of innovation and rigorous evaluation. Decision aids may be helpful for reducing information-seeking strain and some decision aids may help men integrate their values and preferences; however other strategies might also be considered for reducing decision-making difficulty. Health care providers can provide valued emotional support that may reduce decision-making stress.⁵¹ Decision-making self-efficacy is associated with lower decision-making difficulty²³ and can be successfully modified to improve outcomes for a range of challenging behaviors.⁵² Brief cognitive behavioral therapy strategies are effective at reducing negative mood states⁵³ that might make information seeking and decision-making more challenging. Finally, as an initial step, providers may consider enquiring into how their prostate cancer patients are coping with the treatment decision as a routine part of clinical care.

Limitations, and Future Directions

As with all cross-sectional designs, a limitation of our study is that one cannot infer that men’s decisional control and knowledge causally influenced their decision making. Men’s decision making experiences could have influenced the extent to which they were involved in making their decision. For example, the outcome measures may tap into men’s experiences of dissatisfaction with their physicians. Men who are unhappy with their interactions with their physicians may consequently become more involved in the decision. However, this is inconsistent with the association between decisional control and satisfaction with support from others during decision making which is assessed by the support subscale of the decisional conflict measure. Another possible issue is conceptual overlap between prostate cancer knowledge and the items from the decisional conflict scale that ask people to self-report their level of knowledge (e.g., ‘I feel I have made an informed choice’ or ‘I know the risks and side effects of each option.’) which are part of the informed and effective decision subscales of the decisional conflict measure. This overlap may be inflating associations between the two constructs; however, knowledge was associated with decisional conflict subscales that did not include self assessments of how informed men were (i.e., values clarity and support subscales). There was also overlap between the decision-making difficulty scale and one item from the Decisional Conflict Scale, “Is this decision easy for you to make?”. Whereas all the other items on the Decisional Conflict Scale were positively related with knowledge and decisional control. This item was negatively related to these constructs, similarly to the decision-making difficulty scale.

Our sample was quite well educated; one would expect prostate cancer knowledge to be considerably lower in less educated prostate cancer patients, as evident in the very low knowledge reported in studies with low-income, minority patients.^{26, 54} In samples including larger proportions of less educated men, greater variance in knowledge should result in greater variation in decision-making experiences.

Future work might consider the relationships between decision-making experiences, decisional regret and QOL. Decisional regret, which has been found to be associated with whether one was able to play one’s desired role in the decision-making process,^{55, 56} as well as decision-making satisfaction and decisional conflict⁵⁷, may be an important mediator of the influence of decision-making experiences on QOL.