The relation between cancer patient treatment decision-making roles and quality of life


  • Pamela J. Atherton MS,

    Corresponding author
    1. Department of Health Sciences Research, Division of Biomedical Statistics and Informatics, Section of Cancer Center Statistics, Mayo Clinic, Rochester, Minnesota
    • Corresponding author: Pamela J. Atherton, MS, Harwick 836, Mayo Clinic, 200 First Street SW, Rochester, MN 55905; Fax: (507) 266-2477;

    Search for more papers by this author
  • Tenbroeck Smith MA,

    1. Behavioral Research Center, Intramural Research Department, American Cancer Society, Atlanta, Georgia
    Search for more papers by this author
  • Jasvinder A. Singh MD,

    1. Department of Medicine, Division of Rheumatology, University of Alabama, Birmingham, Alabama
    2. Department of Epidemiology, School of Public Health, University of Alabama, Birmingham, Alabama
    Search for more papers by this author
  • Jef Huntington MPH,

    1. Intermountain Healthcare, Institute for Health Care Delivery Research, Salt Lake City, Utah
    Search for more papers by this author
  • Brent B. Diekmann BS,

    1. Department of Health Sciences Research, Division of Biomedical Statistics and Informatics, Section of Cancer Center Statistics, Mayo Clinic, Rochester, Minnesota
    Search for more papers by this author
  • Mashele Huschka BS, RN,

    1. Department of Health Sciences Research, Division of Biomedical Statistics and Informatics, Section of Cancer Center Statistics, Mayo Clinic, Rochester, Minnesota
    Search for more papers by this author
  • Jeff A. Sloan PhD

    1. Department of Health Sciences Research, Division of Biomedical Statistics and Informatics, Section of Cancer Center Statistics, Mayo Clinic, Rochester, Minnesota
    Search for more papers by this author

  • We acknowledge the cooperation and efforts of the Minnesota Cancer Surveillance System (Minnesota's cancer registry) and the cancer survivors, their physicians, and their loved ones who contributed to the collection of these data. The authors assume full responsibility for analyses and interpretation of these data.



The objective of this study was to explore relations between patient role preferences during the cancer treatment decision-making process and quality of life (QOL).


One-year cancer survivors completed a survey in 2000 as part of a larger survey conducted by the American Cancer Society. The current report was based on survey respondents from Minnesota (response rate, 37.4%). Standardized measures included the Profile of Mood States (scores were converted to have a range, from 0 to 100, with 100 indicating the best mood), the Medical Outcomes Survey 36-item short-form health survey (SF-36) (standardized scores), and the Control Preferences Scale. Patients' actual and preferred role preference distributions and concordance between roles were compared with QOL scores using 2-sample t test methodology.


The actual role of survivors (n = 594) in cancer care was 33% active, 50% collaborative, and 17% passive. Their preferred role was 35% active, 53% collaborative, and 13% passive. Overall, 88% of survivors had concordant preferred and actual roles. Survivors who had concordant roles had higher SF-36 Physical Component Scale (PCS) scores (P < .01), higher vitality (P = .01), less fatigue (P < .01), less confusion (P = .01), less anger (P = .046), and better overall mood (P = .01). These results were similar among both women and younger individuals (aged <60 years). Survivors who had active actual roles had higher PCS scores (P < .01), less tension (P = .04), and higher vitality (P = .04) than survivors who were either collaborative or passive. No differences existed in QOL scores according to preferred role.


Survivors who experienced discordance between their actual role and their preferred role reported substantial QOL deficits in both physical and emotional domains. These results indicate the need to support patient preferences. Cancer 2013;119:2342–2349. © 2013 American Cancer Society.


The evolution of modern interdisciplinary cancer care and the emergence of personalized medicine have increased the presence and importance of patient-centered communication.[1] Research has demonstrated that 2-way communication, shared understanding, and trust between patients and health care providers are paramount to the success of treatment.[2-4] Such research also has indicated that patients desire information regarding their medical condition and available treatment options that so medical decision making can occur.[5-9] In response, cancer care in the United States has made a concerted effort to become more patient-focused and collaborative.[3, 10]

Much has been written about how to provide information and engage patients in making decisions. The optimal approach depends on the extent of participation (active, collaborative, or passive) the patient wants or needs.[7, 8, 11-15] Various studies have explored the degree of involvement patients want. In 1996, Beaver et al reported that the majority (52%) of newly diagnosed breast cancer patients preferred a passive role.[16] This finding was challenged in 2006, when Hack et al reported that only 22% of breast cancer patients preferred a passive role.[17] Late-stage disease and tumors of the reproductive system, especially among men, tend to be associated with a more passive approach by patients.[15, 18]

A large, multisample study involving cancer patients with a variety of tumors indicated that roughly 25% of patients prefer to play an active role in treatment decision making, 50% prefer a collaborative role, and 25% prefer to have physicians make decisions for them.[19] Thus, it is known that patients' preferred role in medical decision making varies among individuals[15, 20] but is relatively trait-like (ie, it does not change) over time.[15, 18, 19] Coulter summarized that the desire to participate varies according to age, educational attainment, disease severity, and cultural background.[5]

Past studies of cancer survivor participants have reported associations between quality of life (QOL) and satisfaction with control of or involvement in health care.[21, 22] Griggs et al reported that treatment satisfaction was associated with an increase in mental health.[23] In the current report, we used population-based data to describe decision-making preferences, demographics, and QOL of cancer survivors. Our objective was to determine whether QOL is impacted by patient concordance or discordance in preferred and actual decision-making roles.


Eligibility and Recruitment

The American Cancer Society's (ACS) Study of Cancer Survivors (SCS) is a longitudinal, population-based study of adult cancer survivors that was designed to examine physical and psychosocial adjustment to cancer and changes in QOL. The cancer survivors for that study were drawn from 11 state cancer registries. The study had the approval of the Institutional Review Board (IRB) of Emory University, and additional IRB approvals were obtained in each state. Details of the complete study design and analysis have been previously reported.[24] The current article reports on results from Minnesota survivors using the SCS survey as administered at 1 year postdiagnosis and conducted by the ACS, the Minnesota Cancer Surveillance System (Minnesota's cancer registry), and the Mayo Clinic (IRB 0-2462-01).

Patients with newly diagnosed cancer were selected from the state cancer registry and screened for eligibility. To be eligible for participation, patients had to be aged ≥18 years; had to be diagnosed with 1 of the 10 most common cancers (prostate cancer, female breast cancer, lung cancer, colorectal cancer, bladder cancer, non-Hodgkin lymphoma, skin melanoma, kidney cancer, ovarian cancer, and uterine cancer); and had to have stage I through IV disease at the time of diagnosis. Patients were ineligible if they were identified as mentally incompetent by their physicians or if they were institutionalized or incarcerated at the time of recruitment. Physicians identified in the state cancer registry were notified that their patient had been sampled for this study and were given an opportunity to update patient eligibility. Subsequently, survivors were consented and surveyed by mail and telephone. The overall response rate in Minnesota, including physician notification and survivor recruitment, was 37.4%. The methodological implications of this low response rate were discussed in detail in a report by Smith et al.[24] Informed consent was obtained from all patients who participated in this investigation.


Patient decision-making preferences were measured by the Control Preferences Scale (CPS) (Fig. 1).[25] This 2-item tool allows the patient to record their actual (item 1) and preferred (item 2) roles in decision making. The goal of developing the CPS was to provide clinicians with a tool for assessing patient role preferences and experiences to facilitate communication.

Figure 1.

The Control Preferences Scale is illustrated.

Patient health-related QOL was measured using the 36-item Medical Outcomes Study short-form health survey (SF-36), which is a validated self-reported tool composed of questions regarding health status, feelings, and ability to do usual activities as recollected during the last 4 weeks.[26, 27] The SF-36 measure is comprised of 2 summary scales. The Physical Component Scale (PCS) is composed of physical functioning, physical role functioning, bodily pain, and general health indexes. The Mental Component Scale (MCS) is composed of vitality, social functioning, emotional role functioning, and mental health indexes. The summary scales and indexes are standardized and age-adjusted.

Patient mood was measured using the Profile of Mood States (POMS),[28] which consists of 37 items, each rated on a scale from 0 to 5, with 0 indicating “no” and 5 indicating “always.” Each item asks the patient whether he or she has experienced a particular feeling in the past week. Sample items are “unhappy,” “lively,” “worn-out,” “tense,” “angry,” and “unable to concentrate.” The POMS produces an overall score and a score for each of 6 subscales: anger-hostility, confusion-bewilderment, depression-dejection, fatigue-inertia, tension-anxiety, and vigor-activity.

Statistical Considerations

All assessments were scored using the appropriate algorithms. To improve interpretability and comparability, the POMS total score and subscale scores were converted to a point scale from 0 to 100, in which 100 represented the best possible mood or QOL; so, a higher number indicates better mood or QOL. Both the PCS and the MCS are norm-based with a mean score of 50 and a standard deviation of 10. Higher scores are indicative of better health status.[27] Responses to the CPS were used to categorize patients' actual and preferred roles as active, collaborative, or passive.[15] Furthermore, actual and preferred CPS responses were compared. If the actual role and the preferred role were the same, then the patient's role agreement was categorized as concordant. If the roles were not the same, then the patient's role agreement was categorized as discordant.

Role preference distributions were compared with demographic categories and QOL scores using the Fisher exact test, the 2-sample t test, and Kruskall-Wallis methodology, as appropriate. General linear modeling techniques were used to determine any relations between QOL scores with PCS, MCS, and POMS mood as dependent variables and with baseline characteristics and CPS scores as independent variables.

In 594 observations, with approximately 7.4 times as many patients in the concordant group than in the discordant group, a 2-sample t test (with a 0.05 2-tailed test of significance) had 80% power to detect a mean difference equal to 0.36 times the pooled standard deviation of the 2 groups, suggesting a standardized effect size of 0.36.[29]


Patient Population

There were 594 eligible patients. They completed the CPS and at least 1 of the POMS or the SF-36 questionnaires approximately 1 year after diagnosis. Baseline characteristics indicated that survivors were predominantly aged >60 years (54%), were women (54%), had breast cancer (23%) or prostate cancer (21%), and had received chemotherapy (32%), radiation therapy (30%), or surgery-resection (66%) (Table 1). The majority of patients (59%) had at least 1 comorbidity. Asthma (9%), depression (8%), diabetes (7%), heart condition (8%), high blood pressure (31%), or other (12%) were the most frequently reported.

Table 1. Patient Baseline Demographics, N = 594
CharacteristicNo. of Patients (%)
Age group, y 
<60273 (46)
≥60321 (54)
Women330 (55.6)
Men264 (44.4)
White578 (97.5)
Black3 (0.5)
Other12 (0.8)
Cancer type 
Bladder19 (3.2)
Breast134 (22.6)
Colorectal90 (15.2)
Kidney27 (4.5)
Lung78 (13.1)
Non-Hodgkin lymphoma31 (5.2)
Ovarian30 (5.1)
Prostate126 (21.2)
Skin/melanoma23 (3.9)
Uterine36 (6.1)
Cancer treatment 
Surgery, resection392 (66)
Chemotherapy188 (31.6)
Radiation179 (30.1)
Bone marrow transplantation4 (0.7)
Hormone therapy73 (12.3)
Immunotherapy14 (2.4)

Control Preferences Scale Responses

Te CPS results indicated that 33% of survivors actually played an active role in treatment decision making, 50% played a collaborative role, and 17% played a passive role (Table 2). The role preferred had approximately the same distribution. The patient population had a distribution of role agreement of 523 concordant (88%) and 71 discordant (12%). The κ statistic in this sample indicated that there was 80% more agreement than expected by chance alone. Of the patients who reported discordance, 52 (73%) played a role that was more active than preferred, and 19 (27%) played a role that was less active than preferred. There were no demographic differences between concordant patients and discordant patients.

Table 2. Patient Control Preferences Scale Distribution, N = 594
Control Preferences Scale SummaryNo. of Patients (%)
Actual role played 
Active196 (33)
Collaborative299 (50)
Passive99 (17)
Role preferred 
Active208 (35)
Collaborative312 (52.5)
Passive74 (12.5)
Role agreement 
Discordance71 (12)
Concordance523 (88)
Discordant category 
Role played was more active than preferred52 (73)
Role played was less active than preferred19 (27)

Control Preferences Scale Results and Quality-of-Life Scores

Kruskall-Wallis analysis indicated that patients who reported an actual active role had higher SF-36 PCS scores (mean, 46.5 vs 43.7 for a collaborative role and 42.5 for a passive role; P < .01), a higher SF-36 physical function index (mean. of 51.3 vs 49.5 for a collaborative role and 47.9 for a passive role; P = .04), higher POMS tension/anxiety scores (mean, 82.8 vs 81.4 for a collaborative role and 76.3 for a passive role; P = .04), and higher POMS vitality scores (mean, 50.8 vs 48.6 for a collaborative role and 44.4 for a passive role; P = .04) (Figs. 2, 3). Survivors' actual role was not related to SF-36 MCS scores, the POMS total score, or other POMS subscale scores. Survivors' preferred role was not related to any mean QOL scores.

Figure 2.

Profile of Mood States (POMS) subscale scores are illustrated according to the role played. Statistically significant differences occurred for scores on the tension/anxiety (T/A) and vigor/activity (V/A) subscales. P values are based on the Kruskal-Wallis nonparametric test. A/H indicates anger/hostility subscale; C/B, confusion/bewilderment subscale; D/D, depression/dejection subscale; F/I, fatigue/inertia subscale.

Figure 3.

Scores on the Medical Outcomes Survey 36-item short-form health survey (SF-36) are illustrated according to the role played. The Physical Component Scale (PCS) score was statistically significantly different between roles played. P values are based on the Kruskal-Wallis nonparametric test. MCS indicates Mental Component Scale.

The primary objective of the current study was to determine whether patient concordance or discordance of preferred and actual roles had an impact on QOL. Survivors who had concordant roles reported better QOL scores than those who had discordant roles. Specifically, they had higher mean SF-36 PCS scores (45.0 vs 40.5; P < .001) and better mean POMS mood scores (77.4 vs 73.1; P = .01) (Table 3). SF-36 index analyses indicated better scores in concordant patients for physical functioning (50.2 vs 47.5; P = .02), bodily pain (51.0 vs 47.9; P = .01), general health (50.5 vs 45.6; P < .01), and vitality (49.1 vs 45.4; P = .01). POMS subscales indicated that concordant patients had less anger (88.0 vs 84.9; P = .046), higher vitality (49.5 vs 42.3; P = .01), less fatigue (70.2 vs 60.0; P < .01), less confusion (84.1 vs 79.5; P = .01), and better overall mood (POMS total score). In a subset analysis, younger concordant patients (aged <60 years) and concordant women had similar results compared with their discordant counterparts. Only older discordant patients had significantly lower PCS scores (38.9 vs 43.0; P = .03) and vitality scores (41.6 vs 51.3; P = .02) compared with older concordant patients; and only discordant men had worse fatigue compared with their concordant counterparts (62.1 vs 72.4; P = .02). There were no differences in any QOL scores within the discordant group when we compared patients who played a role that was more active than preferred (n = 52) versus those who played a role that was less active than preferred (n = 19).

Table 3. Concordance/Discordance in Preferred and Actual Control Preferences Scale Roles With Quality of Life Scores
 Quality-of-Life Score: Mean ± SD 
  1. Abbreviations: MCS, Mental Component Scale; PCS, Physical Component Scale; POMS, Profile of Mood States; SD, standard deviation; SF-36, Medical Outcomes Survey 36-item short-form health survey.

MeasureConcordance, N = 523Discordance, N = 71Total, N = 594P
PCS45.0 ± 11.040.5 ± 10.744.5 ± 11.1.001
Physical functioning index50.2 ± 9.647.5 ± 10.149.8 ± 9.7.02
Physical role functioning47.9 ± 10.845.2 ± 11.247.5 ± 10.9.05
Bodily pain51.0 ± 9.347.9 ± 9.150.6 ± 9.3.01
General health50.5 ± 10.445.6 ± 10.849.9 ± 10.6<.01
MCS51.4 ± 10.051.1 ± 9.951.4 ± 10.0.65
Vitality49.1 ± 10.045.4 ± 10.848.6 ± 10.1.01
Social functioning50.6 ± 9.248.1 ± 10.350.3 ± 9.4.07
Emotional role functioning49.0 ± 10.749.2 ± 10.049.0 ± 10.6.80
Mental health51.1 ± 8.950.5 ± 9.451.0 ± 9.0.74
Total POMS77.4 ± 14.973.1 ± 15.076.9 ± 15.0.01
Anger/hostility subscale88.0 ± 14.884.9 ± 15.987.6 ± 15.0.046
Confusion/bewilderment subscale84.1 ± 14.779.5 ± 16.683.6 ± 15.0.01
Depression/dejection subscale86.5 ± 17.185.3 ± 17.086.4 ± 17.1.28
Fatigue/inertia subscale70.2 ± 23.160.0 ± 26.669.0 ± 23.8.002
Tension/anxiety subscale81.3 ± 17.879.3 ± 19.181.0 ± 18.0.47
Vigor/activity subscale49.5 ± 22.242.3 ± 23.848.7 ± 22.5.01

Generalized linear modeling techniques were used to determine whether baseline characteristics (age, sex, race, cancer treatment) and CPS-defined concordance were predictive of PCS, MCS, and mood scores. The results indicated that age was statistically significant in each model, and concordance was significant in the PCS and POMS models. The amount of variance accounted for by age and concordance was small (<5%) with R2 values <0.08, indicating that QOL scores were highly individualistic relative to the demographics that were recorded. Hence, little variation in the set of predictors explained the variation in the outcomes.


The current study of cancer survivors who had completed assessments within 1 year after diagnosis confirmed previously posited relations between the decision-making role and QOL domains. Discordance between actual and preferred roles resulted in poorer physical health and poorer mood (anger, confusion, fatigue, and vigor) but did not result in worse mental well being. Similar results were observed within the subsets of younger patient and women. Survivors whose actual role was active had less tension and more vigor. Poorer QOL for discordant survivors did not depend on the direction of the discordance. Previous research has demonstrated that treatment satisfaction and QOL are related.[4, 30, 31] Our findings go beyond this to support the hypothesis that role satisfaction, and not just treatment satisfaction, impacts patient QOL.

Dow et al[7] and Hodgkinson et al[13, 14] have indicated that making decisions regarding health care is an important need of patients. Patients wished to have some control,[7, 13] and most frequent unmet goals were in the domain of existential survivorship, as indicated on the Cancer Survivors Unmet Needs measure, which included items like “cope with changes to my beliefs” and “make decisions about my life in context of uncertainty.”[13] Breast cancer survivors report that being informed regarding medical decisions is 1 of the most important pieces of advice for newly diagnosed patients, thus confirming patient preference to play an active role in decision making.[11] A recent systematic review pointed to an increasing trend in patient preferences for shared decision making; in 50% of studies published before 2000, the majority of patients preferred to share decisions; this increased to 71% of studies published in 2000 or later.[32] Those studies, however, did not include any comparisons between decision making and QOL.

Satisfaction with treatment and care patterns has been associated with QOL. In particular, Griggs et al reported a correlation between satisfaction of information and vitality, mental health, and distress.[23] Satisfaction likely has several components, including patient-physician interaction, access to care, quality of care, and outcome. One important component of satisfaction in patient-physician interaction during cancer treatment is the degree to which a patient plays his or her preferred role in cancer treatment decision making. In our study, we observed better physical and emotional QOL in patients who played an active role and in those with concordance between the preferred role and the actual role. Similar to Griggs et al, our study demonstrated that concordance of roles was associated with higher vitality (POMS vitality/activity subscale) and better overall mood (POMS total score). Unlike the results reported by Griggs et al, in our study, mental health was not associated with an actual or preferred decisional role.

Regardless of the role preferred, survivors in our study who experienced their preferred level of input into the decision-making process reported better QOL and associated outcomes. Survivors with concordant actual and preferred roles may be more adherent to treatment protocols, resulting in higher QOL scores in some areas. I also may be possible that concordant roles led to improved knowledge, self-efficacy, and/or self-management in these patients, thus improving QOL.

A recent meta-analysis demonstrated that information needs were greatest among those who preferred an active role in treatment decisions.[33] Information satisfaction has been related to global QOL, physical well being, social well being, emotional well being, and functional well being in descriptive studies.[34] Arora et al observed that access to information was associated with better well being and higher perceptions of health competence.[35] Patients playing their preferred role may have more appropriate access to information, which may lead to better QOL. Other theories include the possibility that those who play a preferred role may find the experience less distressing. Individuals who prefer active or collaborative roles and do not attain them may become frustrated, whereas those who prefer a more passive role may become overwhelmed when physicians ask them to engage in decision making.

For this study, control preferences were recorded 1 year after treatment. A reasonable question is to ask whether the findings would have been different had the CPS had been recorded at various time points. We have observed very little change in CPS scores change over the short term (within 6 months). However, there are very limited data on the longitudinal nature of control preferences. In their study, Hack et al reported a 48% concordance rate in baseline role assumed versus role preferred 3 years later.[17] The CPS seems to be more trait-like than state-like.

When considering the clinical implications of our results, we examined minimally important differences. Although there are many methods for establishing and many factors influencing minimally important differences, previous work often has indicated that differences of a half standard deviation in health-related QOL measures are clinically important,[36] and variability in thresholds is underscored by another study suggesting that differences as small as 0.1 standard deviation may be important in certain situations.[37] In our study, statistically significant differences between concordant and discordant patients ranged in size from roughly a one-half standard deviation for the SF-36 PCS score to a little less than a one-third standard deviation for the POMS total mood score. This amounts to a 5-point difference in the SF-36 PCS (which reports a minimum clinically important difference in the range typically of 3-5 points) relative to a population norm of 50 and a 4-point difference in the POMS relative to a population norm of approximately 71. Results like these suggest that clinicians should encourage patients who are discordant to pursue a preferred role to improve QOL.

Current endeavors in augmenting patient-centered care, which involves systematic, evidence-based structures aimed at integrating the patient's perspective more formally into the decision-making process, involve the use of patient-reported outcomes (PROs) in a paradigm of treatment and referral (Fig. 4). Our results support the use of this paradigm. The model in Figure 4 indicates an integrated flow of information and decision-making processes between the patient and the health care team that will promote improved care, QOL, and patient outcomes. PROs are assessed regularly at each patient clinical appointment, and the results are communicated to the clinical team. Within our context, patient-centric care considers that patients who play a preferred role may receive a wide array of benefits. The entire health care system is moving toward patient-centric care, which means that our results are particularly timely. This is relevant to the Patient-Centered Outcomes Research Institute (PCORI) and its recent mandate to generate standardized methods and applications to improve patient-centered care.[38] Our data suggest that the benefits hypothesized by PCORI's focus on patient-centered care are empirically demonstrable.

Figure 4.

This chart illustrates the use of patient-reported outcomes (POMs) data to enhance patient-centered care.

There were methodological limitations of the ACS survey documented by Smith et al that mandate caution in broad interpretation of our findings. In particular, the relatively low response rate calls into question the generalizability of our results to the general adult cancer population. Furthermore, the results were recorded 1 year after diagnosis, which prevented the construction of a longitudinal profile that could be used to examine differences in our findings over time. Finally, our data were derived strictly from a data set comprised of patients from the upper Midwest in the United States.

In conclusion, patients who had discordance between their preferred and actual roles in cancer care reported substantial QOL deficits in both physical and emotional domains. The current results indicate that improved patient satisfaction with care and improved QOL may be achieved by meeting patient expectations with respect to the amount of input they have in making treatment decisions.


This work was supported through grants from the National Institutes of Health (CA 25224 and CA 37404). Dr. Singh is supported by research grants from the National Institute of Arthritis, Musculoskeletal, and Skin Diseases; the National Institute of Aging; the National Cancer Institute; and the Agency for Health Quality and Research Center for Education and Research on Therapeutics. He is also supported by the resources and the use of facilities at the Veterans Affairs Medical Center at Birmingham, Alabama. The American Cancer Society (ACS) Studies of Cancer Survivors were funded as an intramural program of research conducted by the ACS Behavioral Research Center.


Dr. Singh has received research grants from Takeda and Savient and consultant fees from Savient, Takeda, Ardea, Regeneron, Allergan, URL Pharmaceuticals, and Novartis. He also is a member of the executive of OMERACT, an organization that develops outcome measures in rheumatology and receives arms-length funding from 36 companies; a member of the American College of Rheumatology's Guidelines Subcommittee of the Quality of Care Committee; and a member of the Veterans Affairs Rheumatology Field Advisory Committee.